A review on performance of artificial intelligence and conventional method in mitigating PV grid-tied related power quality events

Renewable and Sustainable Energy Reviews 56 (2016) 334–346

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews journal homepage: www.elsevier.com/locate/rser

A review on performance of artificial intelligence and conventional method in mitigating PV grid-tied related power quality events Ken Weng Kow a,n, Yee Wan Wong a, Rajparthiban Kumar Rajkumar b, Rajprasad Kumar Rajkumar a a b

Department of Electrical and Electronic Engineering, Faculty of Engineering, University of Nottingham Malaysia Campus, Malaysia Foundation in Engineering, University of Nottingham, Malaysia Campus, Malaysia

art ic l e i nf o

a b s t r a c t

Article history: Received 9 June 2015 Received in revised form 17 November 2015 Accepted 22 November 2015

Integration of renewable energy resources into power networks is the trend in power distribution system. It is to reduce burden of centralized power plant and global emissions, increase usage of renewable energy, and diverse energy supply market. However, solar photovoltaic which is a type of renewable energy resource, is found to generate peak capacity for a short duration only. Next, its output is intermittent and randomness. In addition, it changes behavior of power distribution system from unidirectional to bidirectional. As a result, it causes different types of power quality events to the power networks. Therefore, these power quality events are urged to be mitigated to further explore the potential of solar photovoltaic system. This paper aims to investigate negative impacts of photovoltaic (PV) grid-tied system to the power networks, and study on performance of artificial intelligence (AI) and conventional methods in mitigating power quality event. According to the surveys, power system monitoring, inverter, dynamic voltage regulator, static synchronous compensator, unified power quality conditioner and energy storage system are able to compensate power quality events which are caused by PV grid-tied system. From the studies, AI methods usually outperform conventional methods in terms of response time and controllability. They also show talent in multi-mode operation, which is to switch to different operation modes according to the environment. However, they require memory to achieve abovementioned tasks. It is believed that unsupervised learning AI is the future trend as it can adapt to the environment without the need of collecting large amount of data before the AI is implemented. & 2015 Elsevier Ltd. All rights reserved.

Keywords: PV Grid Tied System Power quality Mitigation Artificial intelligence High penetration

Contents 1. 2. 3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Power quality. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mitigation of power quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Power system monitoring . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Inverter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Dynamic voltage regulator (DVR) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4. Static synchronous compensator (STATCOM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5. Unified power quality conditioner (UPQC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6. Energy storage system (ESS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1. Future works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

n

Corresponding author. Tel.: þ 603 8924 8000/6017 2081856. E-mail address: [email protected] (Y.W. Wong).

http://dx.doi.org/10.1016/j.rser.2015.11.064 1364-0321/& 2015 Elsevier Ltd. All rights reserved.

334 335 336 336 337 338 340 341 343 344 344 344 344

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

1. Introduction Renewable energy resources are believed to be able to meet the energy challenges that are unable to be solved by traditional centralized power plants. These sources increase the variety of energy supply market, decrease the global emissions and increase long-term sustainable energy supply. However, renewable energy resources are only able to generate its maximum power capacity for a short duration. Consequently, mass research and development in the renewable energy resources sector are happening worldwide in order to increase productivity and effectiveness in using these energies. Among the renewable energy resources available today, the solar photovoltaic (PV) is the one in favor by most utility companies. Its inherent characteristics, free from pollution and availability in all sizes are further attracting interest from users. These advantages encourage the development of micro-grid PV systems. Micro-grid is an idea of supplying electrical power from a local renewable energy resource like a PV system. When the generated power exceeds consumption, extra power could be transmitted to other areas, and vice versa. Despite the benefits of the PV system, the downside of the PV system is that the output power highly relies on the solar irradiance and ambient temperature. As a result, the PV power production is stochastic and intermittent. This phenomenon becomes more serious when penetration of the PV system is high (up to gigawatt). Moreover, electrical devices are needed to be incorporated with PV arrays to be connected to the grid for voltage conversion. Thus, this process may create various types of power quality issues to the grid, which would affect the quality of the electrical power undesirably. Therefore, an urgent need is required to investigate potential power quality disturbances caused by the PV system. This is important to enhance the reliability of PV grid-tied system. Examples of poor power quality events are voltage swell, impulsive transient and harmonic interruption. Different levels of damage to equipment can be caused by these events [1]. Since a PV system is a combination of devices, it contributes different forms of power quality disturbances to the grid and affects quality of delivered power to consumer. Appropriate mitigations for this issue are urged to be found, as the impact of these disturbances is not only affecting an individual but a region. Power distribution systems are available for more than a century. These systems have undergone a continuous evolution till today. Generation, transmission, distribution and protection have been integrated into power networks for a safe and reliable system. However, integration of PV system to power networks has changed the flow of power from unidirectional to bidirectional. This behavior affects current power networks adversely. Publications show that existing protection environment is insufficient to deal with this change [1] [38]. Hence, compensator equipment and new grid codes are required to integrate as advanced protection and mitigation measurements to the PV grid-tied system. Artificial Intelligence (AI) is a type of human created intelligence based on machines or software. It is usually used to reduce human working load. Machine learning, which is one of the many areas in AI, is widely used in data analysis. It is able to assign commands from the results of data analysis by sets of algorithm. For example, vast amount of data is collected in power system monitoring for analytical purpose. However, it poses challenges to data analyser to identify potential power quality issues. This scenario can be solved by using AI techniques in a relatively shorter time. AI has been deployed in many countries to carry operations such as planning, controlling and management in power system operation [2]. It is believed that AI is able to deal with real life uncertainties in a short time. Hence, it greatly reduces the burden

335

of human. In addition, AI is well known in pattern analyzing and it could be used to control key protection components in a power network by learning from historical data. AI has also been implemented in controlling inverter to operate PV system autonomously [59,60]. This paper provides a comprehensive review on the topic of artificial intelligence’s (AI) and conventional methods’ performance in mitigating power quality events in PV grid-tied systems. Over 80 publications [1–87] are reviewed. The first sets of references [4–41] are on the causes and impact of power quality events due to PV grid-tied system. The second set [42–92] discusses methods that are being used in mitigating power quality events that are caused by PV grid-tied systems. From research findings, power system monitoring, inverter, dynamic voltage regulator (DVR), static synchronous compensator (STATCOM), unified power quality conditioner (UPQC), and energy storage system are approaches that are used in alleviating power quality issues. Performance of both conventional and artificial intelligence methods are reviewed and compared. This paper is divided into four sections. An introduction is briefed in Section 1. Section 2 describes causes and effects of power quality events in a PV grid-tied system. The following section discusses performances of artificial intelligence and conventional methods in mitigating PV grid-tied system related power quality events. Lastly, the concluding remarks are given in Section 4.

2. Power quality Power quality is a measure of the standard of delivered power. By delivering low quality electric power to the consumer, it could affect the accuracy of utility metering; cause malfunction to protective relays; cause destructive damage to equipment and others. Since a PV array is generating DC voltage and available in small sizes, unplanned installation of PV system to the grid could lead to power quality events. By increasing the number of PV systems in the grid, it contributes an observable amount of power to the grid. Therefore, any disturbances from PV systems could affect the region adversely. The examples of power quality events that occur due to PV system are power fluctuation, over voltage, etc. Power fluctuation is a phenomenon where the generated power is unstable. It is generally deemed that power fluctuation is one of the main issues with the PV system. This is due to its inherent characteristic of the photovoltaic cell, where the output strongly relies on the surrounding environments, irradiance and temperature. As a result, the power production is not constant and unstable. In order to validate this event, simulations and experiments were carried out [3–5]. From experiments, these events are usually being triggered during noon time (from 1000 to 1300) [5,13], when the clouds are moving rapidly. Since the cloud behaves as an obstacle to block the irradiance from the sun, active power generated from the PV system will be reduced greatly. In addition, sizing and topology of PV system are believed to be another culprit to power fluctuation [4,5]. It is because a PV system with lower capacity is having a relatively smaller area causing the percentage of the whole PV panel being shaded is large. From research finding [93], power fluctuation of a PV system’s standard is only available in technical requirement of Puerto Rico Electric Power Authority (PREPA) for interconnecting wind and solar generation. From PREPA, it only allows a fluctuation rate of 10% of rated capacity in 1-min. Other than power fluctuation, high current harmonic content (THDI) is also found in PV grid-tied systems. Kow [94] proved that third order LCL filter is unable to filter current harmonic effectively. In general, a PV system outputs DC voltage intrinsically. Hence, an inverter has to be incorporated to connect the PV

336

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

system to the grid. Inverter converts DC voltage by high switching frequency. As a consequence, high order harmonic content is present in the AC voltage. Notably, inverter is the culprit for the total harmonic distortion in the PV system [4,7–9]. In addition to inverters, researchers have found that the irradiation is magnifying the total harmonic distortion indirectly. There are papers show that the irradiation is inversely proportional to the THDI [4,9–10]. This phenomenon happens because the fundamental element of current experiences a steep decrease as compared to other higher order harmonic content. Nevertheless, Golovanov and his team [9] state that with the usage of simple single phase inverter, it could introduce more current harmonic to the power system. Moreover, Anwari et al. [8] showed that in reversed power flow operation, it could further pollute the grid. Power factor is a measurement of the efficiency of the electricity usage, where it ranged from 0 to 1. Lee [11] believes that phase displacement and harmonic distortion are reasons for low power factor in a PV grid-tied system. Inverters in a PV system are mostly operating in unity power factor. This is because variable power factor function of inverter increases possibility of islanding to the grid during high penetration level [11]. Therefore, most of the current PV system outputs active power to the grid only. With higher penetration of PV systems to the grid, it injects high amount of real power to the power system. It seems to aid the grid effectively. However, this is not the case. With purely real power being injected to the grid from PV system, it reduces active power from central power plants effectively. On the other hand, the need for reactive power remains. As a result, low power factor electrical energy will be supplied from the utility. This phenomenon can be observed from results in [6,7,19]. Farhoonea et al. [4] designed a PV system to inject certain amount of reactive power to the grid. During daytime the power factor is able to be maintained but falls back in the night. The next power quality issue is the frequency fluctuation. It happens when the supply and demand of a power system are unbalanced. As the output power of the PV system relies strongly on the irradiance which always fluctuates, the frequency fluctuation could happen from time to time. The situation is worsened when high penetration of PV system is used. This is because a high penetration PV system creates a significant drop of power during a power fluctuation event. Another potential power quality issue is voltage flicker. Variation in output power is the cause to a voltage fluctuation. Research paper [32] and [33] shows that PV systems are mostly invulnerable to voltage flicker events. However, this is not applicable to countries like Malaysia which has the highest frequency and intensity of moving cloud [12]. Moving cloud is behaving as an obstacle to shed PV system in high frequency and short duration. It causes output power of PV systems to vary largely. In consequence, the voltage level also varies according to the output power of the PV system. Other than cloud transient, voltage flicker is found while PV grid-tied system is operating under burst mode. Burst mode operation is implemented to inverter in order to obtain high efficiency measure, where this measurement is defined by California Energy Commission (CEC). Hence, inverters will halt operation while the required load power is less than generated power by the PV system. As a result, capacitors will charge and discharge from the PV system to confine voltage into a permissible level. Experimental and simulation of a PV system operates under burst mode had been carried out in [31]. Results show that the burst mode operation causes voltage flicker to a local distribution system. Voltage sag is a well-known power quality issue among consumers as it contributes more than 80% of the reported complaints worldwide [15,16]. A PV grid-tied system is believed to improve voltage profile of distribution systems. It is applicable to low penetration of PV system only as observed from [29]. Yamashita

et al. [17] showed that by increasing penetration of a PV system in a high voltage network, PV systems exacerbate the voltage sag issue and post-disturbance steady state voltage. This phenomenon shows that PV system deteriorates the grid instead of benefiting it. Hence, IEEE 1547 standard states that a PV system should disconnect from the grid autonomously when a fault is found [36]. Apart from the voltage sag event, voltage swell also occurs in PV grid-tied systems. Dugan et al. [19] finds that PV grid-tied systems prolongs the voltage swell scenario. Vegunta et al. [20] indicated that voltage swell happens frequently during summer and under low load condition in low voltage (LV) distribution network. In an unbalance distribution network, voltage swell would happen in a lightly loaded phase with high PV penetration [21]. In the meantime, reverse power flow is observed in the lightly loaded phase. Actual assessment had been carried out in an Australia residential area where it shows that the voltage swell happened in a 4% of PV penetration distribution network [41]. Unbalanced voltage is defined as the ratio of the negative sequence voltage component to positive sequence voltage component. Voltage in a three phase four wire system is unbalanced to each other. Unbalance factor confined by both NEMA, Brazil relation PRODIST Module 8 and IEEE 112-1991 is 2% [23,24]. Various researchers find that high penetration of single phase rooftop PV systems creates severe unbalance voltage to a distribution network [23,24]. Research finds that a single phase PV penetration level up to 85% could create unbalanced voltage phenomenon to a three phase distribution system [24]. However, this issue can be solved by considering balanced distribution network in a three phase system before PV installation [24]. Similar experiment is carried out in [22] and [30], where they show that by installing PV system without considering balanced distributing network could lead to voltage swell, current unbalance and fluctuate power factor in that particular phase. Alam et al. [25] found that unbalanced voltage increases neutral-to-ground voltage (NGV). In addition, conventional methods are unable to suppress NGV to permissible levels. Adverse impact of unbalance voltage increases in output power ripple of PV inverter due to negative sequence voltage. As a result, current is polluted by high order harmonic content [26]. Tuladhar [27] and Shou et al. [28] found that a minor unbalance in voltage could lead a three phase electrical motor to overheat. Researchers [13,14] found that inrush current occurs in a PV grid-tied system as well. In an electric power system, power transformer has to be energized during start up. This is considered as a critical event because a high value of current in transient, which is the inrush current, is experienced by the transformer. Inrush current is produced when the transformer behaves as a short circuit device while being energized [13]. As a matter of fact, the value of the inrush current could be ten times of the rated value [14]. For grid-connected PV systems, a transformer is usually used as a medium to connect PV system to the grid, to isolate PV systems from the grid. Therefore, switching on the PV system will inject inrush current to the grid.

3. Mitigation of power quality 3.1. Power system monitoring In general, utility companies monitor high voltage (HV) distribution network in order to identify fault in a short time [38]. An example is a Malaysian energy utility company installed power quality monitor system in five regions [86]. This is used to detect and record power quality events in order to take corrective action in the shortest possible time. Since power monitoring is an important parameter in mitigating power quality events, Kilter

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

et al. [39] initiates a guidelines for power quality monitoring. However, extensive works are required to complete it. Music et al. [40] integrated Supervisory Control and Data Acquisition (SCADA) and Automated Meter Reading (AMR), Power Quality Monitoring System (PQMS), and Electric Vehicle Management System (EVMS) as the Integrated Power Quality Monitoring System (IPQMS) to monitor a smart distribution system. It shows power networks are easier to be controlled and monitored through IPQMS. AI techniques have also been implemented in power system monitoring. Zang and Zhao [57] develop an algorithm to identity power quality disturbance using support vector machine. The algorithm is developed from wavelet transform and multi-layer support vector machines. Results show that the proposed algorithm is able to locate six types of disturbances accurately and efficiently [57]. Power quality classification has been carried out by Ding et al. as well. They utilized least square support vector machine theory in applied algorithm. An accuracy of 99% was shown in the outcome of the classification [58]. In addition, Chan et al. [67] proved the validity of AI in micro-grid fault detection classification. Radial basis function, decision tree, K- nearest neighbor, and Naïve Bayes are used to classify seven groups of power quality events. Results show that the Naïve Bayes is having the best performance, which has an average of 20% of error [67]. Next, Bentley et al. are able to show validity of an unsupervised neural network, self-organizing map to identify power quality disturbance source in an IEEE 6-bus system [87]. Mallesham, Mishra, and Jha [90] introduce automatic generation control of microgrid, which is equipped with diesel generator, fuel cell generator, aqua electrolyzer, battery, wind and solar energy. The gains of each source which are confined by generation rate constraint (GRC) are calculated by AI techniques. The GRC is a ramp rate limit of increasing or decreasing a power source to ensure power balance between load and sources. From results, it shows the frequency of microgrid without GRC fluctuates in 2 Hz, whereas microgrid with GRC maintains. In addition, it shows conventional PID controller, genetic algorithm, bacterial foraging optimization (BFO), and particle swarm optimization (PSO) are capable in maintaining frequency due to increase of wind power generated, increasing and decreasing of load. Among the methods, BFO is having the lowest integral time squared error overall [90]. Llanos et al. [91] implemented an on-line neural network for load prediction in an off-grid microgrid system. From the results, the load prediction from the SOM network is able to track the actual trend closely. Hence, it concludes that AI can be used for load prediction purpose. Other than Llanos et al, Loewenstern et al. [92] also took part in load forecasting. They compared five methods for very short term load forecasting (five minutes in advanced). It is because increasing penetration of renewable energy resources requires shorter term of load forecasting to ensure balance in between generation and load due to highly stochastic natural environments From these results, difference averaging with moving average method achieves the lowest mean absolute percentage error overall [92]. These research findings shows AI techniques' capabilities in classification and prediction. In addition, implementing AI can reduce work load and burden of system analyser in a centralised monitoring room. Although types of event and accuracy of prediction are low, AI shows its talent in controlling DG in an off-grid microgrid system and is able to predict the load accurately. Therefore, it is believed that AI is unable to be practically implemented in power system monitoring at this stage, however it could be used as a centralised controller for an off-grid microgrid system.

337

3.2. Inverter Power system configuration and additional compensator equipment are methods to mitigate power quality events [34]. In a PV grid-tied system, the inverter is the core component to vary output of a PV system. Therefore, it is believed that the configuration of the inverter is able to mitigate major power quality issues. Shen and Wang [35] design a cost effective three phase three wire half bridge single stage inverter to improve power factor and eliminate current harmonics in a PV grid-tied system. The proposed method is able to maintain power factor within 1.0 in both sufficient and insufficient power environments [35]. Huang, Xu and Yang [37] found that an inverter controlled by instantaneous active reactive control (IARC) and average active reactive control (AARC) has issues as it is unable to output sinusoidal current and outputting power with a 120-Hz ripples respectively. Therefore, they propose a flexible active and reactive power control to command operation of the inverter [37]. The new algorithm verifies the validity in reducing power fluctuation and current harmonics. However, it is found that the proposed algorithm is unable to be implemented in the PV system. It is because it can only reduce the power fluctuation parameter or current harmonic parameter at a time. In addition, the optimized reduced current harmonic parameter is unable to reduce harmonic content of current to meet the IEEE 519-1992 criteria. Another control scheme based on instantaneous power theory and hysteresis current is found in [18]. Results show that it is effective in regulating voltage at the point of common coupling. Moreover, it is able to recover voltage profile due to a fault with a shorter time. Hao and Xu [26] proposed a new control algorithm based on current reference for a PV system working under unbalanced voltage. Outputs of simulation prove that the validity in eliminating power fluctuation with increasing control parameter, k. However it has to sacrifice the balance of current waveform. Shou et al. [28] suggested current adjustment coefficients control strategy to alleviate output power fluctuation and total harmonic distortion of current under unbalanced voltage environment. Results show that power fluctuation and harmonic distortion can be alleviated by regulating current adjustment coefficient with second control mode. AI also shows its good performance in commanding equipments like the inverter. Dasgupta et al. [59] applied spatial iterative learning in an inverter controller. They found that common methods such as resonant controller, repetitive controller and proportional controller are ineffective in maintaining the voltage waveform. Experimental and simulation have been conducted to compare spatial iterative learning method with the common methods. Results show that the controller is able to converge load voltage error to zero within 0.65 seconds, perform auto synchronization while maintaining voltage profile automatically when frequency varies, and generate harmonic to inverter’s voltage to counter grid harmonic voltage to ensure unity of load voltage waveform. Inverter is also being used as frequency smoothing purpose [60]. Fuzzy reasoning with inputs of average frequency deviation, average insolation and variance of insolation are used to configure inverter of a PV system in a three phase islanding system. Results show that the proposed method is able to minimize frequency deviation to about 0.2-Hz. However, this is achievable by introducing 40% of power loss to the PV system and increasing usage of diesel generator by about 20% [60]. Al-Saedi et al. proposed particle swarm optimization algorithm in controlling the inverter of distributed generator for switching distributed generator from grid-connected to islanding mode. Simulation results show that the control algorithm is able to allow a micro-grid system to switch from grid-connected mode to islanding mode autonomously [66]. Next, it is able to maintain its voltage profile

338

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

within 10% of fluctuation, maintain supply of active power, and maintain frequency profile within 1.2% of fluctuation during switching stage. From research findings, advanced conventional methods are built on complex algorithm to control the inverter. It is usually built on instantaneous power theory for current harmonic elimination and it maintains voltage profile. From findings [37,26] and [28], it is to deal with unbalanced 3 phase power system. However, it is found that 3 phase PV system would not yield unbalance voltage to the power distribution system. In addition, single phase PV system could prevent the unbalance voltage scenario by proper planning before installation [23,24]. For artificial intelligence controlled based inverter, it is usually meant for multi-tasking which reduces harmonic content, maintains voltage and frequency profile and switches between grid-connected mode and islanding mode [66]. It is found that advanced conventional methods could achieve similar performance as in AI based methods. However, they require complex algorithm to solve limited issues. In contrast, structure of AI enables it to be used in different fields which diversify its usage. Furthermore, AI methods are able to carry out more duties on the inverter compared to advanced conventional methods. Table 1 shows the summary of inverter configurations in alleviating power quality issues.

3.3. Dynamic voltage regulator (DVR) Compensator equipment is an additional electronic device that is used to regulate power profile within permissible levels. It is usually being installed at the point of common coupling of PV system to the grid. Dynamic voltage restorer (DVR) device is a well-known circuit used in mitigating voltage sag caused by low active power injection. It is composed of a control circuit and power circuit, which includes an energy storage system, an inverter, a filter and a transformer [42]. Al-Mathnani et al. [43] develop a control algorithm in configuring dynamic voltage regulator to mitigate voltage sag in a power distribution system. The DVR is supported by a PV system with a battery. Results show that the DVR which incorporates a new algorithm and a phase locked loop is able to recover voltage sag event with 22% of reduced voltage level [43]. Al-Mathnani et al. [15] further show that the DVR is capable of suppressing voltage sag event without polluting voltage waveform by using 12 active switches in an 11 kV power distribution system. From the results, it is able to mitigate 22% of voltage sag with 100 W active power injection, within 0.1 milliseconds, and without polluting the voltage waveform. However, the current waveform is not shown in the result. Hence, current harmonic distortion level is an unknown from the paper [15].

Table 1 Summary of configuration of inverter. Source Control algorithm

Performance

Target

[35]

Direct source current shaping

3 Phase 3 wire system

[37]

Flexible active and reactive power control

[18]

Instantaneous power theory and hysteresis current control

[26]

Flexible positive and negative sequence control

[28]

Advanced instantaneous power theory

[59]

Spatial iterative learning

[60]

Fuzzy reasoning

[66]

Particle swarm optimization

1. Improve Power Factor to 1.0 2. Eliminates Current Harmonic in both sufficient and insufficient power environments 3. Reduced Cost (Eliminate MPPT) (Contains 4 active switches only) 1. Reduce output power fluctuation level by increasing total harmonic distortion of current. 2. Reduced output current harmonic by increasing output power fluctuation 3. Optimized performance in reducing output current harmonic (15%) is unable to meet IEEE 519-1992 criteria which is below 5% 1. Solve overvoltage during PV start up where it is able to maintain in the range of 0.975 pu to 1.025 pu 2. Maintain Voltage Profile within 2.5% with a load reduction of 30% 3. Prevent under voltage scenario after disconnection of PV system 4. Reduced recover time of voltage profile due to a fault by 0.055s 1. Reduce active and reactive output power fluctuation by altering Control parameter, K. 2. It has to sacrifice balance of current waveform to reduce the output power fluctuation rate 1. Current Adjustment Coefficient α and β 2. Maintain the power fluctuation rate within 10% 3. Reduce third harmonic content to 0.4% 4. Results are able to meet power fluctuation rate and THD rate criteria 1. Converge load voltage error to zero within 0.65s 2. Maintaining load voltage waveform in an environment of 36% Voltage sag, 50% 3rd harmonic content, 30% 5th harmonic content 3. Auto synchronization – able to maintain load voltage profile due to Variation of grid frequency 4. Inverter generates harmonic voltage to counter grid voltage Harmonic to ensure unity of load voltage waveform 5. Common methods such as resonant, repetitive, and proportional Controller are unable to perform above tasks 1. Minimized frequency deviation to about 0.2-Hz 2. Achieved Off-Grid Power Distribution System 3. Drawback: a)Introduce about 40% of power loss to the PV system b) Increase about 20% usage of diesel generator 1. Maintain voltage profile within 10 % of fluctuation 2. Maintain Frequency profile within 1.2% of fluctuation 3. Maintain supply of active power 4. Assign generate power order to second DG unit in order to sustain microgrid during islanding

Unbalanced 3 Phase 3 wire system

IEEE 13-bus test feeder

Unbalanced 3 phase system

Unbalanced 3 Phase System

Single Phase System

Three Phase Islanding System

Three phase system for grid-connected and islanding mode

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

Other than the inverter, AI is also being utilized in controlling compensator equipment. Farsadi et al. [61] implemented a fuzzy control algorithm to command DVR in a PV and wind farm hybrid system. The main aim is to use DVR as reactive power compensation device in this paper. It absorbs reactive power when the voltage level of renewable energy resources is higher than the grid and vice versa for reactive power injection. It is important to highlight that, the paper is using PV system without battery to ensure the quality of voltage waveform. Simulation results show that this approach is able to suppress voltage sags and swells event up to 50% constantly without disrupting the power distribution system [61]. In addition, the current waveform is showing that it has no issue on harmonic distortion [61]. Ramasamy and Thangavel [62] used in-phase voltage injection method controlled by a fuzzy logic controller to operate a DVR. Results show that the 700 W PV based DVR controlled by a fuzzy logic controller is not only capable in mitigating voltage sag up to 10% and swell up to 190%, but also provide uninterruptable power supply (UPS) in a low voltage (LV) distribution system. Next, the supply voltage and current from DVR shows it have no serious harmonic content. In addition, it has a faster detection rate than conventional method such as RMS method, peak voltage detection and RMS averaging method. Results further show that, conventional PI controller is unable to deal with transient changes but the proposed controller is able to deal with it. It injects 3.01% of harmonic content to the load but the fuzzy logic controller is injecting 2.42% only. Nevertheless, it has a higher feasibility as it is able to work under four different modes according to situation which are standby, active, bypass and power saver mode. Particle Swarm Optimization (PSO) is used by Kumar et al. [68] in dealing with the voltage sags with phase-jumps through DVR. Results show that the proposed system is able to mitigate voltage sag of 45% with phase jumps of 10–14°. Moreover, it takes only 45

339

iterations converge to the solution and injecting power without harmonic distortion. In addition, it shows conventional control method requires more power than PSO to compensate voltage sag event, which is 42.80% and 30.49% respectively for voltage sag of 20% [68]. Controlling DVR by PSO based phase advancement compensation scheme is proposed in [69]. Its main aim is to reduce ESS of the DVR. Single phase sag in a three phase load was used as the simulation scenario. It showed that phase advance compensation is able to reduce the power need by DVR to compensate voltage sag event up to 60% compared with In-phase injection method. In addition, PSO phase advance compensation scheme is further capable in alleviating voltage sag of up to 60% in any one phase lasting for 30 ms, with reduced energy of 94 W [69]. From the research findings, both conventional and AI based methods do have significant results in compensating power quality issues such as voltage sag event. In particular, AI based methods are able to compensate voltage sag and voltage swell events without distorting current harmonic. Papers show that AI method is better than conventional method due to its ability of mitigation with reduced energy, larger margin in compensating voltage sag and swell events, higher feasibility, and has no harmonic distortion in current [62] [68,69]. In terms of dynamic response, Ramasamy and Thangavel [62] have proven their proposed method is capable to mitigate with a shorter time compared with conventional methods. In addition, some scenarios can be solved by AI method only as shown in [62] (switch into four different modes according to the environments). Besides, it is also able to mitigate voltage swell event as in [62]. Conventional controller is usually having fixed control parameter. As a result, they are unable to have a good performance in non-linear system as their operating parameter had to be redesigned [62]. Although AI seems to outperform advanced conventional methods in terms of response time, feasibility, and durability, it requires memory to

Table 2 Summary of configuration of dynamic voltage regulator. Source Control algorithm

Performance

Target

[15]

Two Fast Vector Control

3 Phase 3 wire system (12 – Switches in parallel)

[43] [61]

Vector control Fuzzy logic

[62]

Fuzzy logic

[68]

Particle swarm optimization

[69]

PSO based phase advancement compensation

1. Uses 12 active switches to mitigate voltage profile which is sagging for 22% with a 100w active power injection within 0.1 ms. 2. It did not pollute the voltage waveform 3. Current waveform is not available from the paper; so, the distortion level of current waveform is an unknown yet 1. It is able to mitigate a voltage sag event with a 22% reduced of voltage 1. Reactive power absorption when voltage level of renewable energy resource is higher than the grid and vice versa for reactive power injection 2. It uses a PV system without battery to as power source to DVR 3. Current and voltage waveform does not have harmonic distortion issue 4. It is able to mitigate voltage sags up to 50% and voltage swell up to 150% event without disrupting the power distribution system 1. Mitigate voltage sag up to 10% ad swell up to 90% 2. It is able to use as uninterruptable power supply (UPS) 3. Faster detection rate than conventional method 4. It is able to deal with transient changes, but conventional controller is unable to deal with it 5. It injects 2.42% of harmonic content, but conventional controller injects 3.01% of harmonic content 6. High feasibility as it is able to change to four operation modes according to the environment 1. Proposed method is able to mitigate voltage sag up to 45% with Phase jumps of 10 to 14 degrees 2. It takes 45 iteration converge to solution 3. It injects power without harmonic distortion 4. It reduces the need of power by 12.31% to mitigate voltage sag of 20% which compared with a conventional pre-sag method 1. Phase advancement compensation (PAC)method is able to mitigate voltage sag of up to 60% in 1700kW power distribution system with reduce power of 240W compared with InPhase Injection method 2. PSO based PAC is able to compensate above event with reduced power of 94W which last for 30 ms.

PV grid-tied system 3 phase micro-grid with wind and PV system

Single Phase 700W PV gridtied system

Three phase system

Balanced three phase system

340

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

achieve abovementioned performances. Hence, it is more expensive than conventional methods. Table 2 summarises the usage of DVR in mitigating power quality issues. 3.4. Static synchronous compensator (STATCOM) Static synchronous compensator or STATCOM is used as a source and sink of reactive power. It is usually used to rectify poor power factor and power system which has poor voltage regulation. Sumathi and Bansilal [72] found that voltage instability is usually caused by the power distribution system being unable to meet the reactive power demand. Therefore, STATCOM is able to compensate the voltage instability issue by operating as a source and sink of reactive power. It absorbs reactive power from grid when the grid voltage is greater than its nominal voltage, and vice versa for reactive power injection. Meng et al. [89] claimed that conventional PI and PID controllers are facing a challenging task in fine tuning to suit the highly nonlinear environment. However, it does not improve system stability. Chindurala et al. [44] tested a photovoltaic-static synchronous compensator (PV-STATCOM) in an IEEE-13 bus system for voltage regulation, harmonic and reactive power compensation. The proposed control system requires 7 different control blocks to achieve the tasks. Results show that the proposed system is able to reduce voltage unbalance factor to an acceptable range according to AS/ NZS 61000.4.30 standard, which is below 4%. However, conventional PV inverter is unable to meet the above criteria, where the voltage unbalance factor is 5.1%. In terms of voltage variation during the night and day, difference of RMS voltage level is within 6% which is able to meet the AS/NZS 61000.3.100 standards. Results further show that the proposed method is able to isolate the PV source. Not only that, but the PV-STATCOM unit is able to maintain the voltage profile within acceptable range (7 6%) for fault events, where the same event is unable to be solved by conventional methods. Lastly, it showed harmonic contents of the current and voltage waveform is about 0.7% and 0.5% respectively, whereas the conventional inverter is 7.5% [44]. It concluded the 900 kWp PV-STATCOM ameliorates power quality issue such as voltage unbalance, voltage sag and swell, and current harmonics [44]. Varma et al. [45] carried out a real life simulation and results show that PV-STATCOM is capable in regulating voltage profile and power factor. The simulated 10 kWp PV-STATCOM will be installed for real-life usage as well. The results show that, PV-STATCOM is able to regulate the power factor to above 0.9. The results further prove that it is able to regulate voltage within 5% for voltage swell and sag events. It took 4.5 s to improve the voltage dip from 0.75 p. u. to 1 p.u. It also shows that, reverse power flow is possible during daytime which could generate some tariff from the generated power. Although the proposed controller is able to make full use of the PV system throughout a day, it opposes IEEE 1547 standard, where a PV system is allowed to output unity power factor only. It is because other power factor operation modes use complex islanding operation. Similar simulation is carried out by Varma et al. [46] in real-time. Outcomes indicate that PV-STATCOM is able to regulate voltage within 1.5 cycles, total harmonic distortion of current and voltage are below 2%, and improve power factor from 0.76 to 1.00 within 0.5 cycle. It is important to highlight that, PVSTATCOM is outputting unity power factor at day time and varying power factor at night. Therefore it opposes IEEE 1547 standard as well. In AI, step least mean square algorithm is used in [65] to control dynamic STATCOM (DSTATCOM) for power factor correction in a three wire power distribution system. Simulation and experiment have been conducted to prove the validation. Results show that it is able to maintain power factor at about 0.8, reduce harmonic content of current from 23.71% to 2.24%. Besides, ANN is

proven to be able to control DSTATCOM to carry out similar tasks [71]. Multi-layer feed-forward supervised network is used by Yang et al. [71] to control the DSTATCOM. Results show that, the proposed control is able to improve power factor to 1.00 within 0.01 s. It claims that harmonic content is in the voltage and current waveform and it can be compensated by filter [71]. Sharma et al. [70] introduce usage of STATCOM as additional equipment in a micro-grid system. The STATCOM is used to enhance voltage regulation, efficiency and power factor of micro-grid by supplying solely reactive power. ANN is used as a controller to tune the STATCOM. Results show that it is able to eliminate oscillations caused by disturbances within 0.01 seconds. In addition, it aids the grid by providing additional reactive power during normal condition effectively. Real life simulation has been carried out to verify validity of supervised ANN to control STATCOM for voltage control mode and reactive power control mode in a 24 bus extra high voltage (EHV) Indian Power System [72]. STATCOM is used to compensate voltage sags and swell event in bus-24, which is the most sensitive bus in EHV distribution system. Although most AI applications are able to outperform conventional methods, the proposed algorithm by Sumathi and Bansilal [72] is only able to have the same performance as the conventional method. Meng et al. [89] introduce reinforcement learning adaptive (RFA) controller in tuning the PID for voltage sag event, and improving system stability. From results, it shows that the RFA controller is able to reduce fluctuation rate but voltage is 0.4 p.u.; however, conventional PID is fluctuating about 0.5 p.u. It also indicates that voltage sag event is able to be compensated perfectly. Particle swarm optimization (PSO) is employed to regulate voltage profile by optimal sizing and allocation of multiple STATCOM units [73]. Results show that optimal placement and sizing which is decided by PSO is able to reduce power losses by 20%, and regulate voltage to 75% for every bus voltage profile under loading factor of 1.6. It also show that without optimal sizing and placement of STATCOM into the same bus system and environments, bus 30 experiences voltage sag event. Besides that, results show that availability of STATCOM increases the stability limit by about 0.8 loading factor. STATCOM in [75] is applied to control magnitude of reactive power according to load demand. PSO is selected to tune a PI controller in order to achieve good dynamic response. By taking few parameters from the circuit, PSO is able to calculate the particle positions within one sampling period, which is 1/ 15,360 s. In addition, its dynamic response is 0.1 s and it is lesser than conventional method, which is about 0.16 s. It reduced the maximum voltage from 55.3 V to 54.6 V. It indicates that the proposed method is having a shorter response with smaller overshoot performance for transient event. Results further show that it is capable in dealing with light, medium and heavy load. It compares with conventional PI controller. It shows that PSO based controller is having a better performance than conventional methods in terms of response time and overshooting value [75]. From above research findings, it shows that AI is not only capable in tuning the compensator equipment, but also planning of sizing and allocation. Advanced conventional method outperforms AI method in terms of regulating voltage, and feasibility of controller, where it [44] shows its talents in regulating unbalance voltage factor to below 4%, maintain voltage profile within 76% during fault, low harmonic distortion of current and voltage waveform, and isolating PV source during fault events. The STATCOM could maintain power factor of a PV grid-tied system to above 0.90 in [44] and [45]. However, it opposes IEEE 1547 standard where PV grid-tied system is only allowed to output unity power factor. The advantage in AI method is its transient response. It achieves shorter time as compared to the advanced conventional method. The dynamic response in AI method [75] and advanced

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

341

Table 3 Summary of STATCOM. Source Control algorithm

Performance

Target

[44]

Single phase synchronous reference frame

IEEE 13 Bus system micro-grid with PV system

[45]

Current control (Hysteresis band modulation technique)

[46]

Current control (Hysteresis band modulation technique)

[65]

Least mean square

[71]

Artificial neural network

[70]

Artificial neural network

[72]

Supervised artificial neural network

[73]

Particle swarm optimization

[75]

Particle swarm optimization

1. Reduce voltage unbalance factor to below 4% 2. Maintain variation of RMS voltage between night time and daytime Within 6% 3. Isolate PV source during fault events and maintain voltage profile of the grid within 7 6% 4. Regulate Current and voltage THD to 0.7% and 0.5% respectively 5. Conventional PV Inverter is unable to achieve abovementioned tasks 1. Maintain power factor above 0.90 2. Inject available power to grid 3. Maintain voltage profile within 5% during high inductive load is being Start up, voltage sag and swell events 4. Good transient response which improve voltage dip from 0.75 p.u. to 1 p.u within 4.5s 5. Oppose IEEE 1547 standard, where the proposed method is able to output different power factor voltage to the grid 1. Regulate voltage profile within 1.5 cycles 2. Total harmonic distortion of current and voltage are less than 2% 3. Improve Power factor from 0.76 to 1.00 within 0.5 cycle 4. Oppose IEEE 1547 standard, where the proposed method is able to output different power factor voltage to the grid 1. Maintain power factor 2. Regulate voltage and current by injecting reactive power 3. High THD 1. Improve power factor to 1.00 within 0.01s 2. Harmonic content occurs in voltage and current waveform 1. Supply required reactive power to load to regulate voltage profile 2. Eliminate oscillation caused by disturbances within 0.01 s 1. Maintain bus voltage in the range of 0.9 p.u to 1.05 p.u. for different Load factor in voltage control mode and reactive power control mode 2. It is having a similar performance as conventional method 1. It is used for optimal sizing and allocation of multiple STATCOM units 2. Regulate voltage profile within 7 5% for all bus voltage whereas voltage sag event happens for the situation does not have a STATCOM unit 3. Reduced Power Losses by 20% compared with situation does not have a STATCOM unit 4. Increase stability limit by about 0.8 loading factor compared with situation does not have a STATCOM unit 1. Particle Position is able to be calculated within 1 sampling period, 1/ 15,360s 2. Its dynamic response is 0.1s, but conventional PI controller is 0.16s 3. It reduced overshoot voltage by 0.7V compared with conventional PI controller 4. It is capable in light, medium and heavy loading

conventional method [45] is 0.1 s and 4.5 s, respectively. Table 3 shows the summary of STATCOM. 3.5. Unified power quality conditioner (UPQC) In addition to DVR and STATCOM, shunt active power filter (SAPF) or alternatively unified power quality conditioner (UPQC) is believed to be able to alleviate power quality issues caused by PV grid-tied systems. SAPF is proposed by Hideaki Fujita and Horofumi Akagi in 1985 for harmonic elimination by integrating series active filter and shunt active filter [88]. It compensates reactive power, eliminate harmonic content and maintain DC link voltage. Similar to STATCOM, SAPF is able to integrate itself as part of a PV grid-tied system. Kinhal, Agarwal and Gupta [76] found that conventional controllers require precise mathematical model and are sensitive to parameter variation. Therefore, advanced methods based on AI should be developed to learn, remember and make decisions according to the environment and causes it to be insensitive to parameter variation [76]. Kumar and Sinha used state vector modulation to control UPQC for voltage stability enhancement [48]. Outcomes of the simulation show that it is able to mitigate voltage sag event of up to 7% perfectly. Synchronous Detection Method (SDM) and instantaneous power theory (P-Q theory) are compared for effectiveness in controlling SAPF [49]. Results indicate that UPQC reduces current harmonic content to 0% and maintains unity power factor [49]. It also shows that P-Q digital control theory is having a faster

3 phase blue water power building with 10 kwp PV system

London hydro building with PV system in real-time

Three phase system

Three phase system Wind-diesel power plant 24 Bus EHV Indian power system

IEEE 30 Bus power distribution system

Low voltage experimental

response time than SDM method as PQ compensate current harmonic within 1st cycle whereas SDM require 14 cycles. Bhargavi [50] introduces another type of UPQC controlled by voltage angle. Proposed UPQC is able to mitigate several power quality events such as swell up to 26%, sag up to 81%, and no severe harmonic contents are found in output of compensator equipment. It is found that the proposed device works efficiently when DC link capacitor provides adequate power. Although simulation model does not utilize photovoltaic system, it is believed that the device is applicable to PV systems as well [50]. Lee et al. present a Distribution-Unified Power Flow Controller (D-UPFC) [51,52]. It consists of an AC chopper, a transformer and it is installed after the pole transformer. The main function of the DUPFC is to regulate voltage sag and voltage swell events that are caused by reverse power flow from the PV system. Although results show that it is possible to alleviate voltage sag and swell due to PV grid-tied system, more work is required in controlling the device in order to implement in real life [51,52]. Sabo et al. [63] propose an artificial neural network (ANN) control scheme for SAPF. The proposed algorithm is found to be able to detect and reduce current harmonics from 33.59% to 3.92% [63]. Kumar and Sastry [74] introduce usage of fuzzy logic controller (FLC) and ANN to tune UPQC. Both approaches have a significant improvement compared to conventional methods, in terms of accuracy and response time [74]. ANN is applied to UPQC by Kinhal et al. for decreasing response time in compensating voltage and current harmonics purpose [76]. Experimental and simulation results

342

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

showed that the proposed method is able to stabilize DC-link current within a half cycle, whereas two and a half cycle for conventional PI controller. In addition, it reduces harmonic content from 27.82% to 0.59%, which is 0.11% lower than conventional PI controller [76]. However, similar results are unable to be obtained during experimental testing. It is believed that the issue is due to insufficient small sampling period [76]. The stability of ANN is found to be stronger than PI controller, where its DC-link current is kept steady at 2 A, but for a PI controller it fluctuates between 0.75 A to 0.95 A. In [64], Takagi-Sugeno (TS) fuzzy control scheme is used to command SAPF. It found that the algorithm is able to eliminate total harmonic distortion (THD) of current effectively, which reduces THD of current from 30.39% to 3.11% in the simulation. Moreover, it has a faster dynamic response and smaller overshoot for voltage sag events than conventional PI controller as it is able to converge within 0.02 s and drop of 30 V, whereas 0.08 s and drop of 45 V for conventional PI controller [64]. Abovementioned performances are proved in the experimental stage [64]. Shen et al. [77,78] apply PSO-Fuzzy to tune the UPQC to eliminate current harmonic content from 13.78% to 1.56% and compensate voltage sag event within 0.1 s. They found that this approach is

having higher efficiency than conventional instantaneous reactive power theory [77,78]. PSO-based adaptive neuro-fuzzy inference system (ANFIS) is proposed by Kumar et al. [79,80] to control UPQC to deal with unbalanced voltage sag with phase jump. The proposed method is used to identify angle of voltage injection to ensure minimum injection of real power by UPQC. Experiments and simulation have been carried out to prove the validity. They found that this approach is more economical because it is able to reduce the energy storage size [79,80]. In addition, it is able to compensate voltage sag event due to double line fault which contains severe harmonic content and phase jump, with 28.15% less power than UPQC with minimum real power injection method. It is able to converge to a constant value at the 78th iteration. Results have also proved it is able to mitigate voltage sag with phase jump due to double line to ground fault and line to line fault. Experimental results have proved the validity of aforementioned performance. Comparison of conventional Linear Quadratic Regulator (LQR)based and PSO-based UPQC controller is done in [85]. It shows that PSO outperforms LQR in reducing harmonic content of current and voltage, where PSO is able to reduce to 2.75% whereas 4.42% for LQR method. However, both methods are able to meet the IEEE 519-1992 standard.

Table 4 Summary of UPQC. Source Control Algorithm

Performance

Target

[48]

State vector modulation

Single phase system three phase system

[49]

Synchronous detection method and P–Q theory

[50]

Voltage angle control

[51]

D-UPFC Control

1. Maintain voltage profile after significant changes in load as in single phase and three phase distribution system 2. Able to mitigate a voltage sag event of up to 7% 1. Regulate THD to 0% 2. Maintain Unity Power Factor 3. Maintain Unity Displacement Power Factor 4. Zero amplitude in Neutral wire 5. P-Q digital theory has a faster response time in compensating current harmonic than SDM method, which are 1 cycle and 14 cycles respectively 1. Mitigatevoltage Sag up to 81% without incur THD 2. Alleviate Voltage Swell event up to 26% without incur THD 3. Work efficiently while DC link Capacitor provide adequate power 1. Has Potential in mitigating voltage sag and swell issue

[52]

D-UPFC control

1.

[63] [74] [76]

ANN Fuzzy Logic and ANN ANN

1. 1. 1. 2. 3. 4.

[64]

TS Fuzzy

[77]

PSO Fuzzy

[78]

PSO Fuzzy

[79]

PSO based Adaptive neuro-fuzzy inference system

1. 2. 1. 2. 1. 2. 1. 2. 3. 4.

[80]

PSO based Adaptive neuro-fuzzy inference system

[85]

PSO based feedback controller

1. 2. 1. 2. 3.

Three phase four wire system

Three phase system

Three phase PV grid-tied system Has Potential in mitigating voltage sag and swell issue Three Phase PV grid-tied System It is able to reduce current harmonics from 33.59% to 3.92% Single phase system Elminate THD in distribution system in short response time with high accuracy Three Phase System Stabilize DC-link Current within a half cycle whereas conventional PI controller requires Three phase system 4 and a half cycles Simulation results show it is able to reduce harmonic content from 27.82% to 0.59%, where this result is 0.11% lower than conventional PI controller However, same performance is unable to be obtained during experimental stage due to sampling period is impractical to apply in real life Higher stability than conventional PI controller as DC-link current kept to be steady in 2A, but PI controller fluctuates in between 0.75A to 0.95A during the experiment Reduce THD of current from 30.39% to 3.11% Three phase system Faster dynamic response (0.02s) and smaller overshoot (drop of 30V) for voltage sag event Reduce current Harmonic content from 13.78% to 1.56% Three phase system Compensate voltage sag event within 0.1s Reduce current Harmonic content from 13.78% to 1.56% Three phase system Compensate voltage sag event within 0.1s Compensate Voltage Sag event with 28.15% reduced power compared with UPQC minimum Three phase four wire real power injection method system It is more economical since storage size is reduced th converge to a constant value at the 78 iteration It is able to mitigate voltage sag with phase jump due to double line To ground fault, line to line fault. Compensate Voltage Sag event with 22% reduced power compared with UPQC minimum Three phase four wire system real power injection method It is able to mitigate voltage sag with phase jump due to double line To ground fault Three phase system Reduce THD of Current from 18.2% to 2.96% Reduce THD of Voltage from 3.63% to 0.44% It has a better current THD regulating than LQR method, where PSO is able to reduce to 2.75% whereas 4.42% for LQR

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

343

Table 5 Summary of ESS. Source Control algorithm

Performance

Target

[53]

Discrete fourier transform

PV grid -tied system

[54]

PI Controller - injecting Reactive power to the grid

[55]

Micro energy management system

[56]

—

[81]

Coordinated control

[82]

ANN

[83]

Genetic algorithm

[84]

Matrix real-coded genetic algorithm

[47]

Least mean square

1. Time-window is directly proportional to rated power of ESS 2. Shortest time-window’s ESS is 42.75 times smaller than the largest timewindow 3. Shorter time-window is able to track actual output power closely 4. Longest time-window’s output power is 3.3 times more than the smallest time-window 1. Reduce impact of fast cloud transient in voltage profile 2. Maintain voltage profile during sunny day, and cloudy day within 5% 3. Increase Inverter rating by 10% could further enhance the voltage profile 1. Suppress Output Power Fluctuation without sacrifice efficiency of the System 2. There is a drop of 1% for system with ESS, and it is believed able to be reduced 1. Mitigate power fluctuation for diversely located PV system 2. It is more cost efficient as larger size of battery cost more than smaller size of battery 3. Diverge PV plants with small ESS have same performance as centralize PV plant 4. Control theory have to be implemented to configure ESS in order to further enhance reliability of ESS to microgrid 1. Mitigate voltage rise due to high (50%) PV penetration 2. Shorter response time 3. Peak load shaving is achievable 4. Longer life cycle by limiting depth of discharge state-of-charge control 1. Reduced electrical cost by forecast day ahead locational Marginal price and available wind power in order to determine store energy of ESS 2. Second ESS unit is required to adjust mismatches between forecasted and actual wind power 1. Mitigate power fluctuation event in substation 2. Cost efficient, which save operation cost by 7197 RMB in a summer 1. Determine optimal parameters for ESS in microgrid 2. Vanadium redox battery is able to generate $ 13,708 in a year for 30 kW microgrid system 1. Eliminate current harmonic to below 5% 2. Mitigate voltage sag and swell up to 25% 3. Inverter control algorithm is based on output power predicted by least mean square

Different approaches to control UPQC have been studied. Although conventional and AI methods are able to alleviate harmonic contents to below 5% and transient voltage disturbances, AI methods are believed to be a better method because they could work with imprecise inputs, handle non-linearity, robust, and do not need accurate mathematical model [64]. In addition, AI methods can identify and solve the power quality event with shorter response time and smaller overshoot compared with conventional methods [64]. Next, PSO based controllers are usually able to mitigate voltage sag with phase jump issue which is unable to be deal with advanced conventional method. Table 4 shows the summary of UPQC. 3.6. Energy storage system (ESS) Battery, super capacitor and flywheel are examples of energy storage systems. Power output of renewable energy resources is well-known for intermittency problem due to their inherent characteristics. For PV systems in particular, generation could drop up to 15% per second. Thus, it creates significant adverse impacts to the grid. Researches on energy storage systems as a technique to solve power fluctuation have been carried out. In [53], Cao et al. showed that the length of operational period has a strong relationship with ESS, where the shortest time-window of ESS is about 42.75 times smaller than the largest time-window and able to track output power closely. However, longer operational period has better performance in smoothing output power. The longest timewindow is able to output more available power to the grid, where the largest time-window’s output power is about 3.3 times more than the smallest time-window. It also shows details on the

50% PV penetration in a 33 Node radial distribution micro -grid PV grid -tied system PV grid -tied system

PV grid-tied system

Wind grid-tied system

Wind grid-tied system Variable sources micro grid

PV grid-tied system

effect of different time-window to storage capacity and available power to the grid [53]. Next, Abdelkarim et al. [54] found that reactive power is able to enhance voltage profile with energy storage system in a 50% PV penetration. The proposed method is able to prevent voltage profile fall to the boundary of voltage sag, and increase of inverter rating by 10% could further regulate the voltage profile. To conclude, results indicate that this method increases reliability of voltage profile in sunny and cloudy weather [54]. Noro et al. [55] found that energy storage systems are able to reduce fluctuation and maintain high efficiency of PV system, where efficiency has a drop of 1% for system with ESS only. Hossain and Ali [56] discover spatially diverge PV plant equipped with small energy storage is able to have the same performance as in the centralized PV plant. Besides, it is more cost efficient. Liu et al. [81] introduce a coordinated control to mitigate voltage swell event due to high penetration (50%) of PV system. Both experimental and simulation results showed that it is able to increase cycle life of the ESS, and decrease switching operation times, stress and losses to the ESS. Most importantly, peak load shaving function is achievable through this technique [81]. In order to further enhance reliability of energy storage system to the microgrid, control theory need to be implemented to configure ESS. AI is found to be able to control the ESS as well. ANN is used to forecast next day generated power and locational marginal price in [82]. The outcome of this is used to determine a store or a release of energy from ESS. Results show that it is able to ensure microgrid operates in acceptable power fluctuation level and cost saving region. In addition, it has a backup ESS to adjust mismatches between forecasted and actual wind power. Minimizing distributed generation (DG) substation’s power fluctuation level by ESS is carried out in [83]. It introduces a matrix real-coded genetic

344

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

algorithm (MRCGA) to control ESS to deal with power fluctuation. From the results, this algorithm is able to provide feasible multi ESS system operation strategies. It also shows that it is capable in reducing operation cost by 7197 RMB in summer and reducing power fluctuation in substation. MRCGA is used in [84] for optimal allocation and economic operation of ESS for microgrid purpose. It shows that Vanadium Redox Battery is able to generate $ 13,708 in a year for 30 kW microgrid system. Chen et al. [47] found that an ESS in a PV system has a good performance in suppressing harmonic content and compensating reactive power in PV grid-tied systems to regulate within standard. Results also show that it is able to aid inverter in maintaining voltage profile within regulated standard as well [47]. However, it is important to note that, artificial neural network is deployed at the first stage to predict output power of PV system and following control theories are determined according to this prediction. From research findings, AI outperforms conventional method in terms of predicting upcoming usage pattern [82,47]. It is found that, control system that is equipped with predicting function is able to increase financial benefit of the system. In addition, AI methods are able to carry out similar tasks that are assigned to the conventional methods such as mitigate voltage sag, swell events and battery operation planning. Therefore, AI techniques should be used in controlling ESS. Table 5 shows the summary of ESS.

4. Conclusion Free and clean sources from the photovoltaic cell have raised the interest of people since it was invented. With increasing government incentive and decreasing PV system costs, it reduces the entry barrier and further attracts the attention of consumers for the past decade. However, the increase penetration of PV system had caused power quality issues. To further unlock the potential of the photovoltaic cell, mitigations techniques to solve the power quality issues are urged to be found. Implementation of artificial intelligence and conventional methods in alleviating PV grid-tied system related power quality disturbances are studied. Power system monitoring had been used to monitor power distribution system in the past 50 years. It enables faults to be identified and rectified in relatively short time. AI is believed to be able to reduce working load and stress of system analyzer by integrating it into power system monitoring applications. In addition, advanced configuration of PV system is believed to be able to prevent some disturbances autonomously. Moreover, compensator equipment shows capability of eliminating several power quality issues. AI methods also show that it is able to have an outstanding performance than conventional method in terms of response time. Next, energy storage system is found to be able to solve power fluctuation event. With AI control scheme, ESS could further upgrade from passive device to active device. Although AI seems to outperform conventional methods in terms of controllability and response time, it is time consuming to train an AI machine. Limited functions are another drawback of it. It is only able to carry out the task that pre-designed for it. In addition, vast amount of data is required to train a neural network. Otherwise, it is unable to fulfill a given task. Next, the AI algorithms that are being used in the PV grid-tied systems are not active learning algorithm. The drawback of these kinds of algorithms is it may not be able to perform in high accuracy if new unseen data is to be tested. To overcome this drawback, active learning neural network is proposed to be adopted in this application. Active learning neural network is expected to learn new data incrementally without having to retrain the data. As a result, the system could learn new unseen data with low complexity.

4.1. Future works The trend in increasing penetration of PV system into PV gridtied system is increasing exponentially. However, most power quality issues caused by the PV grid-tied system remained unsolved. Further research and development in mitigating these issues should be focusing on the use of artificial intelligence. Most of the power quality issues had been solved passively by conventional methods where they operate after the occurrence of an event. Hence, an advanced power grid system which is equipped with active protection system is believed to be able to reduce the amount of conventional power protection circuit, reduce response time in mitigating power quality issue and prevent the occurrence of power quality issue by predicting the events. In addition to that, the active protection power grid system is believed to be achievable by integrating machine learning technique into conventional power grid system. Supervised learning AI had been widely deployed in various fields. This type of AI is insufficient to evolve current power grid system to active protection power grid system as it requires large labeled training data. In contrast, unsupervised learning AI is able to learn new incoming unlabeled data. However, high accuracy is difficult to be achieved as it is being trained without any support value (such as class information). Other than unsupervised learning AI, active learning AI is a type of AI which undergoes training while executing. Therefore, this type of learning has no issue with lack of training data because it is able to learn incrementally. As a result, it is believed that active learning based AI methods would be the next trend in active protection power grid system. Next, most of the current protection circuits react after the occurrence of power quality events. Regardless of the response time, the power grid system is damaged. On the other hand, AI can also be used as a prediction engine to foresee and subsequently mitigate the power quality event prior to its occurrence. This could reduce chances of power protection components being destroyed.

Acknowledgments This research work is funded by the Ministry of Science, Technology and Innovation (MOSTI) Malaysia, under grant code 03-02-12-SF0211.

References [1] Kusko A, Thompson MT. Power quality in electrical system. USA: McGraw-Hill; 2007. [2] Bevrani H, Hiyama T. Intelligent automatic generation control. USA: CRC Press; 2011. [3] Noro Y, Naoi S, Toba K, Kimura M, Minegishi T, Shimizu M et al. Power fluctuation suppression system for large scale PV. IEEE Transmission and Distribution Conference and Exposition (TD); 2012. p. 1–6. [4] Farhoodnea M, Mohamed A, Shareef H, Zayandehroodi H. Power quality analysis of grid-connected photovoltaic systems in distribution networks. Prz Elektrotech 2013:208–13 2a:p. [5] Marcos J, Marroyo L, Lorenzo E, García M. Smoothing of PV power fluctuations by geographical dispersion. Prog. Photovolt: Res. Appl. 2012;20:226–37. http://dx.doi.org/10.1002/pip.1127. [6] Patsalides M, Evagorou D, Makrides G, Achillides Z, Georghiou GE, Stavrou, et al. The effect of solar irradiance on the power quality behaviour of grid connected photovoltaic systems. In: Proceedings of the international conference on renewable energies and power quality; March 2007. p. 28–30. [7] González P, Romero-Cadaval E, González E, Guerrero MA. Impact of grid connected photovoltaic system in the power quality of a distribution network. In: Proceedings of the Second IFIP WG 5.5/SOCONET Doctoral Conference on Computing, Electrical and Industrial Systems, (DoCEIS); 2011. p. 466–73. [8] Anwari M, Hamid MI, Rashid MIM, Taufik. Power quality analysis of grid connected photovoltaic system in adjustable speed drive. In: Proceedings of

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

[9]

[10]

[11]

[12]

[13]

[14] [15]

[16]

[17]

[18]

[19] [20]

[21]

[22]

[23]

[24]

[25]

[26]

[27]

[28]

[29]

[30]

[31]

[32]

the IEE PES/IAS conference on sustainable alternative energy (SAE); 28–30 September 2009. p. 1–5. Golovanov N, Lazaroiu GC, Roscia M, Zaninelli D. Power Quality Assessment in Small Scale Renewable Energy Sources Supplying Distribution Systems. Energies 2013;6(2):634–-645. Kontogiannis KP, Vokas GA, Nanou S, Papathanassiou S. Power Quality field measurement of PV inverter. Int J Adv Res Electr Electron Instrum Eng 2013;2(11):5301–14. Lee G. How PV Grid-Tie Inverters Can Zap Utility Power Factor Renewable Energy World.com 2014.15th October 2009. Available from: URL: http://www. renewableenergyworld.com/rea/news/article/2009/10/how-pv-grid-tie-inver ters-can-zap-utility-power-factor Retrieved on 14th September 2014. R. Albarracin H. Amaris Power Quality in distribution power networks with photovoltaic energy sources. Available from: URL: https://www.google.com. my/url? sa ¼t&rct¼ j&q¼ &esrc ¼ s&source ¼ web&cd¼ 1&cad ¼rja&uact ¼ 8&ved ¼ 0CC EQFjAA&url¼http%3A%2F%2Fwww.researchgate.net%2Fpublication% 2F229038319_Power_Quality_in_distribution_power_networks_with_photo voltaic_energy_sources%2Flinks% 2F0fcfd507d36c9796ab000000&ei¼ 8SeHVJeVMIe8uAT2y4HgBw&usg ¼AFQj CNGgQnD2feM7dTynw6AM-4LJM3GTnQ. Retrieved on 15 August 2014. Tseng HT, Chen JF. Voltage Compensation Type Inrush Current Limtiter for Reducing Power Transformer Inrush Current. IET Electric Power Appl 2012;6 (2):101–10. Shrivastava S, Khan A, Mahor A. Transformer Inrush Current and Related Challenges. Int J Emerg Technol Adv Eng 2014;4(12):450–2. Mathnani AO, Shareef H, Mohammed A, Ali MAM. Power quanlity improvement using DVR with two fast vector control. In: Proceedings of 4th international power engineering and optimization conference. Shah Alam, Selangor, Malaysia: IEEE; 23-24 Jun 2010. p. 376–81. Gao T, Cao J, Xu, Y, Zhang H, Yu P, Yao S. From power quality to power experience. In: Proceedings of fourth international conference on networking and distributed computing, Los Angelas, USA; 2013. p. 116–20. Yamashita K, Kitauchi, Y, Kobayashi, H. Influence of voltage sags on the power system with high penetration of photovoltaic power generation. In: IEEE Power and Energy Society General Meeting; 2012. p. 1–7. Pyo GC, Kang HW, Moon SI. A new operation method for grid-connected PV system considering voltage regulation in distribution system. In: IEEE power and energy society general meeting – conversion and delievery of electrical energy in the 21st Century, Pittsburgh; 2008. p. 1–7. Dugan RC, McGranaghan MF, Santoso S, Beaty HW. Electrical power system quality. 3rd ed.. United States of America: The McGraw Hill Companies; 2012. Vegunta SC, Twomey P, Randles D. Impact of PV and load penetration on LV network voltages and unbalance and potential solutions. In: Proceedings of 22nd international conference on electricity distribution, Stockholm; 2013. p. 1–4. Slsm MJE, Muttaqi KM, Sutanto D. A comprehensive assessment tool for solar PV impacts on low voltage three phase distribution networks. In: Proceedings of 2nd international conference on the developments in renewable energy technology (ICDRET), Dhaka; 2012. p. 1–5. Yan R, Saha TK. Investigation of voltage variations in unbalanced distribution system due to high photovoltaic penetrations. In: Proceedings of IEEE power and energy society general meeting, San Diego; 24–29 July 2011. p. 1–8. Block PAB, Salamanca HLL, Teixeira MD, Dahlke DB, Shino OM, Donadon AR. Power quality analyses of a large scale photovoltaic system. In: Proceedings of fifth international renewable energy congress (IREC), Hammamet; 2014. p. 1–6. Pansakul C, Hongesombut K. Analysis of voltage unbalance due to rooftop PV in low volage residential distribution system. In: Proceedings of the international electrical engineering congress, Chonbuti; 2014. p. 1–4. Alam MJE, Muttaqi KM, Sutanto, D. Effectiveness of traditional mitigation strategies for neutral current and voltage problems under high penetration of rooftop PV. IEEE Power and Energy Society General Meeting (PES), Vancouver; 2013. p. 1–5. Hao H, Xu Y. Control strategy of PV inverter under unbalanced grid voltage sag. In: Proceedings of the IEEE energy conversion congress and exposition (ECCE), Pittsburgh; 2014. p. 1029–34. Tuladhar A. Power management of an off-grid PV inverter system with generatir and battery banks. IEEE power and energy society general meeting, San Diego; 2011. p. 1–5. Shou T, Wang H, Zhu T, Zhu L, Wang Q, Lou X, et al. Harmonic current suppression for three phase PV generation system under unbalanced grid voltage. In: Procedings of the Asia-Pacific power and energy engineering conference (APPEEC), Kowloon; 2013. p. 1–6. Shen Q, Zhu Y. Impacts of small photovoltaic power station on voltage sag in low voltage distribution network. In: Proceedings of the international conference on electrical and control engineering (ICECE), Yichang; 2011. p. 1585– 88. Al-Baik D, Khadkikar V. Effect of variable PV power on the grid power factor under different load conditions. In: Proceedings of 2nd international conference on electric power and enegy conversion system (EPECS), Sharjah; 2011. p. 1–5. Zhao Z, Lai J,Wu K,Yu W. Utility grid impact with high penetration PV microinverters operatinig under burst mode using simplified simulation model. In: Proceedings of IEEE Energy Conversion Congress and Exposition (ECCE), Phoenix; 2011. p. 3928–32. Stewart EM, Aukai TP, MacPherson SDJ, Quach BP, Nakafuji D, Davis R. A realistic irradiance-based voltage Flicker analysis of PV applied to Hawaii

[33]

[34]

[35]

[36] [37] [38]

[39]

[40]

[41]

[42] [43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

[54]

[55]

[56]

[57]

345

distribution feeders. In: Proceedings of IEEE Power and Energy Society General Meeting, San Diego; 2012. p. 1–7. Pan A, Tian Y, Zhao H, Yang X, Jin J. Power quality analysis of PV system of summer and winter CIRED 2012 Workshop Integration of Renewables into the Distribution Grid, Lisbon; 2012. p. 1–4. Rendroyoko I, Rusli M. Development of power quality control procedures and standards to control the connection of non-linear loads in electrical systems. In: Proceedings of 2nd International conference on electricity distribution, Stockholm; 2013. p. 1–6. Shen CL, Wang JY. A 3 phase 3 wire half bridge single stage grid connected PV Power inverter with direct source current shaping algorithm to improve power factor. In: Proceedings of IEEE region 10 conference (TENCON), Hong Kong; 2006. p. 1–4. Peng W, Baghzouz Y, Hadad S. Local load power factor correction by grid interactive PV inverter. IEEE Grenoble PowerTech, Grenoble; 2013. p. 1–6. Huang H, Xu Y, Yang L. Control scheme of PV inverter under unbalanced grid voltage, National Harbor; 2014. p. 1–5. Rahman FA. System wide power quality monitoring in Malaysia. In: Proceedings of 18th international conference on electricity distribution, Turin; 2005. p. 1–5. Kilter J, Meyer J, Howe B, Zavoda F, Tenti L, Milanovic JV, et al. Current practice and future challenges for power quality monitoring-CIGRE WG C4.112 perspective. In: Proceedings of IEEE 15th international conference on harmonics and quality of power (ICHQP), Hong Kong; 2012. p. 390–97. Music M, Bosovic A, Hasansphahic N, Avdakovic S, Becirovic E. Integrated power quality monitoring systems in smart distribution grids. In: Proceedings of IEEE international energy conference and exhibition, Florence; 2012. p. 501–6. Wong PKC, Barr RA, Kalam A. Voltage rise impacts and generation modeling of residential roof-top photo-voltaic systems. In: Proceedings of IEEE 16th international conference on harmonics and quality of power, Bucharest; 2014. p. 229–33. Choi SS, Li BH, Vilathgamuwa DM. Dynamic voltage restorer with minumum energy injection. IEEE Trans Power Syst 2010;15(1):51–7. Al-Mathani AO, Hannan MA, Al-Dabbagh M, Ali MAM, Mohamed A. Development of new control strategy for voltage sag mitigation. In: Proceedings of 2nd IEEE international conference on power and energy, Johor Baharu; 2008. p. 318–23. Chidurala A, Saha TK, Mithulananthan N. Power quality enhancement in unbalanced distribution network using solar-DSTATCOM. In: Proceedings of the Australasian universities power engineering conference, Hobart; 2013. p. 1–6. Varma RK, Sas B, Axente L, Vanderheide T. Vanderheide optimal 24-hr utilization of a PV solar system as STATCOM (PV-STATCOM) in a distribution network. In: Proceedings of IEEE power and energy society general meeting, San Diego. 2011; p. 1–8. Varma RK, Siavashi EM, Das B, Sharma V. Real-time digital simulation of a PV solar system as STATCOM (PV-STATCOM) for voltage regulation and power factor correction. In: Proceedings of IEEE electrical power and energy conference, London; 2012. p. 157–63. Chen X, Fu Q, Yu S, Zhou L. Unified control of photovoltaic grid-connecion and power quality managements. Workshop on Power Electronics and Intelligent Transportation System, Guangzhou; 2008. p. 360–65. Kumar R, Sinha AK. Unified power quality conditioner controller for enhancement of power system voltage stability using state vector modulation. In: Proceedings of the annual IEEE India conference, Kochi; 2012. p. 891–96. Kabir MA, Mahbub U. Synchronous detection and digital control of shunt active power filter in power quality improvement. In: Proceedings of IEEE power and energy conference, Illinois, Champaign; 2011. p. 1–5. Bhargavi RN. Power quality improvement using interline unified power quality conditioner. In: Proceedings of 10th international conference on environment and electrical engineering, Rome; 2011. p. 1–5. Lee K, Koizumi H, Kurokawa K. Voltage sag/swell controller by means of DIPFC in the distribution system. In: Proceedings of the conference record of the 2006 IEEE 4th world conference on photovoltaic energy conversion, Waikoloa; 2006 May; . 2. p. 3538–2430. Lee K, Koizumi H, Kurokawa K. Voltage control of D-UPFC between a clustered PV system and distribution system. In: Proceedings of 37th IEEE power electronics specialists conference, Jeju; 2006. p. 1–5. Cao M, Xu Q, Zeng O, Xu X, Yuan X. An energy storage system configuration method to stabilize power fluctuation in different operation periods, National Harbor; 2014. p. 1–5. Abdelkarim E, Aly MM, Abdelakher M, Ziadi Z, Senjyu T. Supersession of large penetration photovoltaic power transients using storage batteries. In: Proceedings of IEEE 10th international conference on power electronics and drive system (PEDS), Kitakyushu; 22–25 April 2013. p. 78–83. Noro Y, Naoi S, Toba K, Kumura M, Minegishi T, Shimizu M, et al. Power fluctuation suppression system for large scale PV. In: Proceedings of IEEE PES transmission and distribution conference and exposition (TD), Orlando; 710 May 2012. p. 1–6. Hossain MK, Ali MH. Small scale energy storage for power fluctuation minimization with spatially diverged PV plants. In: Proceedings of IEEE Southeastcon, Jacksonville; 2013. p. 1–6. Zang H, Zhao Y. Intelligent identification system of power quality disturbance global congress on intelligent system, Xiamen; 2009. p. 258–61.

346

K.W. Kow et al. / Renewable and Sustainable Energy Reviews 56 (2016) 334–346

[58] Ding Z, Zhu Y, Chen C. Comprehensive Evaluation of power quality based on meaningful classification. In: Proceedings of 5th international conference on critical infrastructure, Beijing; 2010. p. 1 4. [59] Dasgupta S, Sahoo SK, Panda SK. Design of a spatial iterative learning controller for single phase series connected PV module inverter for grid voltage compensation. In: Proceedings of the international power electronics conference, Sapporo; 2010. p. 1980–987. [60] Senjyu T, Datta M, Yona A, Sekine H, Funabashi T. A new method for smoothing output power fluctuations of PV system connected to small power utility. In: Proceedings of 7th international conference on power electronics, Daegu; 2007. p. 829 834. [61] Farsadi M, Shahrak AG, Tabrizi SD. DVR with fuzzy logic controller and photovoltaic for improving the operation of wind farm. In: Proceedings of 8th international conference on electrical and electronics engineering, Bursa; 2013. p.191–95. [62] Ramasamy M, Thangavel S. Photovoltaic based dynamic voltage restorer with energy conservation capability using fuzzy logic controller. In: Proceedings of the international conference on emerging trends in science, engineering and technology, Tiruchirappalli; 2012. p. 485–92. [63] Sabo A, Wahab NIA, Radzi MAM, Mailah NF. A modified artificial neural network (ANN) algorithm to control shunt active power filter (SAPF) for current harmonic reduction. In: Proceedings of IEEE conference on clean energy and technology, Langkawi; 2013. p. 348–52. [64] Shang Y, Wu A. TS-fuzzy-controlled shunt active-power filter for power quality improvement. In: Proceedings of the international conference on electrical and control engineering, Yichang; 2011. p. 1869–72. [65] Arya SR, Singh B, Chandra A, Al-Haddad K. Control of DSTATCOM using adjustable step least mean square control algorithm. In: Proceedings of IEEE fifth power India conference, Murthal; 19-22 Dec. 2012. p. 1–6. [66] Al-Saedi W, Lachowicz SW, Habibi D, Bass O. Power quality enhancement in autonomous microgrid operation using particle swarm optimization. Electrical power and energy system 42 (1); Nov. 2012. p. 139–49. [67] Chan PKP, Zhu J, Qiu Z, Ng WWY, Yeung DS. Comparision of different classifiers in fault detection in microgrid. In: Proceedings of the 2011 international conference on machine learning and cybernetics, Guilin; 2013. p. 1210–13. [68] Kumar RA, Kumar GS, Kumar BK, Mishra MK. Compensation of voltage sags and harmonics with phase jumps through DVR with minimum VA rating using particle swarm optimization. In: Proceedings of the world congress on nature biologically inspired computing, Coimbatore; 2009. p. 1361–66. [69] Raj PA, Kumar SS, Raja J, Sudhakaran M, Palanivelu TG. Particle swarm optimization based energy optimized dynamic voltage restorer. In: Proceedings of the Joint international conference on power system technolofy and IEEE power india conference, New Delhi; 2008. p. 1–6. [70] Sharma P, Hoff B, Meena R. ANN based reactive power control of an autonomous wind-diesel hybrid power plant using PMIG and SG. Power and energy systems: towards sustainable energy, Bangalore; 2014. p. 1–6. [71] Yang X, Zhong Y, Wan Y. A novel control method for DSTATCOM using artificial neural network. In: Proceedings of 5th international power electronics and motion control conference, Shanghai; 2006. p. 1–4. [72] Sumathi S, Bansilal. Artificial neural network application for control of STATCOM in power systems for both voltage control mode and reactive power mode. In: Proceedings of 3rd international conference on power systems, Kharagpur; 2009. p. 1–4. [73] Kumarasamy K, Raghavan R. Particle swarm optimization algorithm for voltage stability improvement using multiple STATCOM. In: Proceedings of the International conference on emerging trends in electrical engineering and energy management, Chennai; 2012. p. 235–42. [74] Kumar KSR, and Sastry SVAR. Application of PI, fuzzy logic and ANN in Improvement of power quality using UPQC. In: Proceedings of the international conference on sustainable energy and intelligent system, Chennai; 2011. p. 316–19.

[75] Liu C, Hsu Y. Design of a self-tuning PI controller for a STATCOM using particle swarm optimization. IEEE Trans Ind Electron 2010;57(2):702–15. [76] Kinhal VG, Agarwal P, Guptal HO. erformance Investigation of Neural Network Based Unified Power Quality Conditioner. IEEE Trans Power Deliv 2011;26(1):431–7. [77] Shen H, Wan J, Yuan C, Liu Y, Li G. Harmonic signal detection algorithm in parallel of UPQC studies based on PSO-fuzzy. In: Proceedings of the AsiaPacific power and energy enegineering conference, Wuhan; 2009. p. 1–4. [78] Shen H, Wan J, Liu Y, Yuan C, Li G. UPQC Signal detection algorithm studies based on PSO-fuzzy. In: Proceedings of 4th IEEE conference on industrial electronics and applications, Xi’an; 2009. p. 2236–40. [79] G.S. Kumar B.K. Kumar M.K. Mishra Mitigation of Voltage Sags with Phase Jumps by UPQC with PSO-BASED ANFIS IEEE Transaction on Power Delivery 26 4 2011 2761 2773. [80] Kumar GS, Kumar BK, Mishra MK. Power quality improvement by UPQC with minimum real power injection. In: Proceedings of 14th European confence power electronics and applications, Birmingham; 2011. p. 1–10. [81] Liu X, Aichhorn A, Li H. Coordinated control of distributed energy storage system with tap changer transformers for voltage rise mitigation under high photovoltaic penetration. IEEE Trans Smart Grid 2012;3(2):897–906. [82] Liu M, Quilumba FL, Lee W. Dispatch scheduling for a wind farm with hybrid energy storage based on wind and LMP forecasting. In: Proceedings of IEEE industry applications society annual meeting, Vancouver; 2014. 1–8. [83] Zhang MD, Xie J, Gan DQ. Optimal ESS operation strategy in distribution system with wind power. In: Proceedings of the international conference on sustainable power generation and supply, Hangzhou; 2012. p. 1–6. [84] Shen C, Duan S, Cai T, Liu B, Hu G. Optimal allocation and economic analysis of energy storage system in microgrids. IEEE Trans Power Electron 2011;26(10): 2762–73. [85] Karanki SB, Mishra MK, Kumar BK. Particle swarm optimization-based feedback controller for unified power-quality conditioner. IEEE Trans Power Deliv 2010;25(4):2814–24. [86] Rahman FA. System-wide power quality monitoring in Malaysia. In: Proceedings of 18th international conference and exhibition on electricity distribution, Turin, Italy; 2005. p. 1–5. [87] Bentley EC, Putrus GA, McDonald S, Minns P. Power quality disturbance source identification using self-organising maps. IET generation, transmission distribution; Oct. 2010 . p.1188–96. [88] Fujita H, Akagi H. The unified power quality conditioner: the integration of series and shunt active filter. IEEE Trans Power Electron 1998;13(2):315–22. [89] Meng X, Wang H, Zhao L, Zhang, H. Controlling study of D-STATCOM based on reinforcement learning adaptive PID. In: Proceedings of IEEE international conference of automation and logistics, Shenyang, China; 2009. p. 1208–11. [90] Mallesham G, Mishra S, Jha AN. Automatic generation control of microgrid using artificial intelligence techniques. In: Proceedings of IEEE power and energy society general meeting, San Diego; 2012. p. 1–8. [91] Llanos J, Saez D, Palma-Behnke R, Nunez A, Jimenez-Estevez G. Load profile generator and load forecasting for a renewable based microgrid using self organizing maps and neural network. In: Proceedings of the international joint conference on neural networks, Brisbane; 2012. p. 1–8. [92] Loewenstern Y, Katzir L, Shmilovitz D. The effect of system characteristics on very-short-term load forecasting. International school on nonsinusoidal currents and compensation, Lagow; 2015. p. 1–6. [93] Marcos J, Narvte L, Berazaluce I, Gonzalez R, Samuel J. Attenuation of power fluctuations in large PV power plants: the use of prediction to optimize the storage system. In: Proceedings of 28th European photovoltaic solar energy conference (PVSEC), Paris; Oct. 2013. [94] Kow KW, Wong YW, Rajkumar RK, Rajkumar RK. Power quality analysis for PV grid connected system using PSCAD/EMTDC. Int J Renew Energy Res 2015;5(1):121–32.