- Email: [email protected]
Integration of virtual and on-line machining process control and monitoring
G Model
CIRP-1605; No. of Pages 4 CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
CIRP Annals - Manufacturing Technology jou rnal homep age : ht t p: // ees .e lse vi er . com /ci r p/ def a ult . asp
Integration of virtual and on-line machining process control and monitoring Y. Altintas (1)*, D. Aslan Manufacturing Automation Laboratory, Department of Mechanical Engineering, The University of British Columbia, BC V6T 1Z4, Canada
A R T I C L E I N F O
A B S T R A C T
Keywords: Cutting Monitoring Control
This paper presents a virtually assisted on-line milling process control and monitoring system. A part machining process is simulated to predict the cutting forces, torque, power, chip load and other process states. The simulated machining states are accessed by a real time monitoring system which detects the tool failure and adaptively adjusts the feed by predicting the forces from the feed and spindle drive motor current supplied by CNC. The integration of virtual simulation with real time measurements avoids false tool failure detection and transient overloads of the tools during adaptive control. The system has been implemented on a CNC machining centre for use in production. ß 2017 Published by Elsevier Ltd on behalf of CIRP.
1. Introduction The recent trend in manufacturing is to develop intelligent, selfadjusting and unattended machining systems to improve the productivity. As reported in past CIRP keynote articles, the reliability of such systems highly depends on the availability of industry friendly sensors [1] and robustness of methods to avoid false alarms and incorrect actions [2]. There have been major efforts in the field of sensor assisted machining. The primary applications focussed on tool wear and breakage monitoring, chatter detection and avoidance, adaptive control of the process forces and dimensional errors, thermal compensation of machines, collision avoidance, and spindle health monitoring. A variety of sensors have been used such as force, vision, acoustic emission, vibration, power, strain, thermocouples and laser devices depending on the application [2]. The reliability of all tool condition and machining process control systems have been suffering mainly because of having difficulties in installing practical and reliable sensors on the machine [3], and not being able to distinguish the actual machining process state from the effects of geometric changes along the tool path [4]. This article presents a new approach: the use of virtual part machining simulation to feed data which is needed to improve the robustness of on-line tool condition monitoring and cutting process control system. The approach is in accordance with Industry 4.0 principles which recommend the integration of digital simulation and sensory data to achieve intelligent manufacturing systems. The proposed approach has been demonstrated in simultaneous adaptive control of cutting forces and detection of tool failure as shown in Fig. 1. A real time communication link between the CNC and external computer has been developed. The actual
* Corresponding author. E-mail address: [email protected] (Y. Altintas).
motor current, velocity, and acceleration of each drive; tool centreposition of the machine; spindle speed and load; tangential feed and currently executed NC block number are obtained from the CNC in real time, and mapped to the virtually simulated data stored in the external computer. The cutting forces are indirectly measured from the feed drive motor current by compensating the disturbance of structural dynamics of ball screw and table through Kalman filters. The cutting forces in Cartesian axes are estimated by transferring the individual motor torque to tool tip.
Fig. 1. Parallel execution of virtual and real time system with information exchange.
The tool breakage is detected from the average spindle torque, and the load on the cutting tool is maintained at the desired level by adaptively controlling the feed rate. The chatter is detected and avoided similar to the method given in [5], hence it is not presented here. However, the locations of chatter events are mapped to tool path contained in Virtual model. The robustness of algorithms is ensured by sending the part geometry changes and average force patterns from the stored, virtual part machining system. The system has been experimentally proven on a CNC machining centre.
http://dx.doi.org/10.1016/j.cirp.2017.04.047 0007-8506/ß 2017 Published by Elsevier Ltd on behalf of CIRP.
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047
G Model
CIRP-1605; No. of Pages 4 Y. Altintas, D. Aslan / CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
2
2. Prediction of cutting forces from drive currents The installation of cutting force sensors to the machine would increase the cost and complexity; hence it is preferred to use sensory data available from the CNC directly [6]. The cutting forces are transmitted to feed drive motors as disturbance torque through the drive structure and servo amplifier as shown in Fig. 2. The friction and inertial loads are separated from the measured motor torque (Tm) measurements to predict the cutting torque (Tc) as [7]: dV T f T c ¼ T m J e dt
where l is the screw lead (pitch) and h is the efficiency. However, the cutting force acting on the tool tip is distorted by the structural dynamics of the feed drive system before reaching at the motor’s current amplifier as a disturbance torque. The Frequency Response Function (FRF) of the structural disturbance has been measured by applying impulse excitation at the table while measuring the current and angular position of the motor shaft from the CNC (Fig. 2b).
Fd ðvÞ ¼ (1)
Fig. 2. (a) Disturbance transfer function (fd), (b) feed drive mechanism, (c) Kalman filter (fKL) compensation (Inom to F a ) (Tcf = coulomb friction). ˆ
where Je and V (rad/s) are the equivalent feed drive inertia felt at the motor and motor velocity, respectively. The motor torque is proportional to the current Tm = Kti where Kt [N m/A] is the torque constant and i [A] is the measured current. The friction torque is modelled as Tf = BeV + Tcf where Be and Tcf are the viscous and coulomb components, respectively. Lu Gre friction model has been used to capture the Coulomb friction, Stribeck effect, hysteresis and pre-sliding displacement [8] of the drive as shown in Fig. 3a. The friction and equivalent inertia of all drives have been identified from the velocity and current values supplied by CNC while moving the drives at different velocities. The cutting force is predicted from the cutting component of the torque as: 2p l
(3)
A sample measurement of disturbance FRF is shown in Fig. 3b for x axis of the machine. The modes at 40, 64 and 175 Hz distort the magnitude and phase, and deviate the magnitude from the desired unity, leading to incorrect measurement of the force when the tooth passing frequency (i.e. spindle speed times number of teeth) and harmonics are beyond 10 Hz. Although the current amplifier of the CNC has 1000 Hz bandwidth, unless compensated, the structural dynamics of the drive reduces the bandwidth of the force prediction from motor current to 10 Hz or 600 rev/min spindle speed when one tooth is used on the cutter. The disturbance FRF is fitted to a transfer function as:
Fd ðsÞ ¼
Fc ¼ Tch
T c ðvÞ T m ðvÞ
(2)
X T c ðsÞ ak ¼ 2 þ 2& v s þ v 2 T m ðsÞ s nk k k k
(4)
where ak, k, vnk are the residue, damping and natural frequency of mode k. The disturbance caused by the structural modes has been compensated by passing the measured motor torque from a Kalman filter which has a transfer function of [9]; T ðsÞ FKl ðsÞ ¼ ˆa ¼ T c ðsÞ
X
aKl
k
s2 þ 2& Kl vKl s þ vKl 2
(5)
where T a is the estimated actual torque transferred from tool tip to ˆ the drive and actual tool tip force (F a ) is found from Eq. [2]. ˆ The FRF of the cutting force prediction from the drive current is shown in Fig. 3b, where the compensated system magnitude approached to unity and the bandwidth has been increased to 200 Hz. As a result, the cutting forces can be predicted accurately from the current monitored from the CNC at tooth passing frequencies up to 200 Hz. A sample cutting force prediction is compared against the dynamometer measurements in Fig. 3c–d. 3. Virtual model assisted process monitoring and control The predicted cutting forces from feed drive motors are used for tool breakage detection and adaptive force control. The machine was a Quaser UX600 CNC machining centre. The real-time algorithms are run on an external PC (Intel1 CoreTM i7-3.40 GHz CPU, 8 GB RAM) which communicates with the Heidenhein CNC via TNC Ethernet connection. A multi-threaded real time code is developed in C++ using the LSV-2 communication protocol which collects commanded, noise free digital motor currents, drive speeds, tool centre point position, spindle speed, and tangential velocity between 330 Hz and 10 kHz sampling frequency. External PC can vary the spindle and feed speeds at 10 Hz interval which is sufficient for the targeted tool condition and process monitoring/ control tasks which are outlined as follows. 3.1. Virtual model – real time application integration
Fig. 3. (a) LuGre friction curve (X axis), (b) identified and Kalman filter compensated FRF, (c, d) comparison of estimated cutting forces from feed drive current and reference forces measured from dynamometer, axial depth = 4 mm, radial depth = 20 mm, feed = 0.2 mm/tooth; (c) 1000 rpm (ft = 33.3 Hz), (d) 2000 rpm (ft = 66.6 Hz).
Prior to the cutting operation, part machining process is simulated using MACHpro1 Virtual Machining System [10] to calculate cutter–workpiece engagement (CWE), cutting forces, torque, power and the machining process states along the toolpath [11]. These machining states are stored in a file and accessed by the real time machining process monitoring and control system as a virtual feedback to avoid false tool failure detection and to prevent transient force peaks during adaptive control. In addition, the
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047
G Model
CIRP-1605; No. of Pages 4 Y. Altintas, D. Aslan / CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
3
uncertainties in the virtual model, such as cutting force coefficients, are continuously calibrated from the real time measurements [12]. The proposed virtual model assisted process monitoring and control system has been demonstrated experimentally on an Aluminium 7050 part shown in Fig. 4. Facing, profiling and two slot milling operations have been performed with a two fluted 20 mm end mill with a regular pitch and 30 degrees helix angle. Spindle speed was selected as 1000 rev/min throughout the operations. The sharp changes in CWE conditions are shown for each operation in Fig. 4. The torque and the cutting forces are simulated along the tool path and stored in a file to assist on-line tool failure monitoring, adaptive control and chatter detection that are run simultaneously. Fig. 5. Robust detection of tool breakage at tooth period 500 with force prediction from virtual machining system.
3.3. Adaptive control
Fig. 4. A test part with various milling operations and corresponding area of tool– workpiece contact.
Gc ðzÞ ¼
3.2. Tool failure detection Tool failure in milling is detected with the algorithm given in [7] but with a simulated feedback from the virtual machining system. The average cutting torque per tooth period (Tsa(m)) is evaluated from the spindle motor; T sa ðkÞ ¼
P X T sa ðiÞ i¼1
P
(6)
where P is the number of torque samples collected at tooth period (m). The average spindle torque per tooth period must remain constant if there is no change in the CWE geometry, and the cutter is free of tooth breakage and run-out. The slow varying CWE effects are removed by a first order auto-regressive (AR) time series filter applied on the differenced average torque as:
eðmÞ ¼ ð1u˜z1 Þ½T sa ðmÞT sa ðm1Þ
Adaptive control has been implemented to keep the peak forces (Fp(k)) at the desired level (Fr(k)) by manipulating the feed-rate in real time at spindle periods (k), see Fig. 6. The objective is to prevent shank breakage and constrain the bending deflections of the tool. The peak force at each spindle period is estimated from the feed drive motor current. The feed and current are read from the CNC. The discrete time transfer function (Gc(z)) between the peak cutting force (Fp(k)) and feed-rate (fk(k)) is expressed as: F p ðkÞ z1 Bðz1 Þ z1 ðb0 þ b1 z1 þ b2 z1 Þ ¼ ¼ f c ðkÞ Aðz1 Þ 1 þ a1 z1 þ a2 z2
(8)
where the parameters (b0, b1, b2, a1, a2) of the combined feed drive and milling process system (Gc(z)) are estimated with the recursive least-square (RLS) algorithm at each spindle period. An adaptive generalized predictive control (GPC) method is used to keep the maximum resultant cutting force at a desired level by manipulating the feed at each control interval (k) as: f c ðkÞ ¼ f c ðk1Þ þ Gac ðzÞ F r ðkÞF p ðkÞ (9) ˜ Control law (Gac ðzÞ) is adaptively adjusted through; Gac ðzÞ ¼ ˜ ˜ fGI gT =½fGI gT fGI g þ l where fGI g ¼ ðfF p gf f gÞ=Df c . fF p g contains ˆ ˆ
(7)
where u is the AR filter parameter and estimated adaptively from ˜ the measurements using the Recursive Least Squares (RLS) method at each tooth period. A tool breakage alarm is triggered whenever the residuals violate the set threshold (e(m) > LIMIT) [7]. However, this previously proposed algorithm has are liability problem: The filter (Eq. [7]) cannot distinguish the tool failure from the step changes in CWE geometry and gives false alarm at tooth periods 262–270 (Fig. 5). The constant breakage threshold cannot distinguish the failure from the step changes in the geometry along a straight path as shown in Fig. 4 (Block Geometry). However, the virtual machining system automatically calibrates the cutting force coefficient from the CWE information and measured force from the current in the beginning of cut, and predicts the average torque accurately as the CWE geometry changes. The residual threshold is adaptively adjusted (i.e. 40% of the predicted force), and alerts the tool condition monitoring system whenever both the measured and residual torque violate the limits. The virtual machining assisted algorithm does not give false alarm at step changes (tooth periods 262–270) but detects the tool failure imposed at tooth period 500 (Fig. 5). The algorithm is now reliable and can be used in production. The thresholds are now geometry independent can be set % of the predicted values by considering only the disturbances caused by the tool run out.
Fig. 6. Virtual model assisted GPC, adaptive control (AC) with force identified from drives; no AC (NoAC), conventional AC (AC) and AC with virtual machining assistance (AC with VF).
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047
G Model
CIRP-1605; No. of Pages 4 4
Y. Altintas, D. Aslan / CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
j step ahead prediction of peak cutting forces and {f} contains present and past measured peak forces, feed commands and recursively calculated polynomials whose details can be found in [13]. The weighting factor is usually selected as l 0.2. However, such an adaptive control system has not been adopted by industry due to large transient force peaks, which either breaks the tool or damages the part due to excessive deflections when the CWE geometry suddenly increases as shown in the test part (Fig. 4). When the tool enters into a cavity or the depth of cut is low, the corresponding peak force become slow; hence the adaptive controller generates maximum feed to increase the force towards the reference value (Eq. [9]). When the tool exits the cavity or the depth suddenly increases, the servo cannot reduce the feed quickly. The resulting high feed creates a very large cutting load spike that breaks the tool or damages the part as shown in Fig. 6. Hence, the intensive adaptive control research in the past has not been successfully used in practice. This problem is solved by bringing the CWE geometry conditions and simulated forces from the virtual machining system to the adaptive control algorithm ahead of sudden geometry changes (i.e. one second long look-ahead of tool path distance) as shown in Fig. 6. A flag ( ) is set to 1 whenever the CWE area is, i.e. 15%, more than the current one, and set to zero otherwise. When the flag value is 1, the incoming force predicted by the virtual machining system is fed into the controller as a disturbance force prior to the actual CWE change. As a result, the adaptive controller slows down the feed, and prevents force overloads and collision spikes; hence the adaptive controller becomes robust and practical to be used in production, especially for roughing operations. In addition, the flag is set to 2 when the tool is in the cavities, the controller is halted and the feed rate is kept at a safe level to prevent excessive feeds which may lead to collision.
the peak cutting forces remain constant without any overshoot, the false tool breakage detection does not occur at transient changes in the cutter–workpiece engagement. 4. Conclusion The tool condition monitoring and process control in machining has not been used in industry due to false alarms, poor robustness of algorithms and lack of practical cutting force sensors. Utilizing the recent advances in digital machining and CNC technologies, this paper presents a new approach by integrating virtual machining system with the on-line monitoring and control system. The forces are extracted from the feed and spindle drive motor current commands by compensating the distortion caused by the structural dynamic chain between the cutting and motor locations. The use of CNC inherent force sensing eliminated the need to mount costly and impractical sensors on the machine. Current online monitoring systems are blind to changes in the cutter– workpiece engagement and process, which prevents them to be robust against collision forces and false tool failure alarms. The proposed system brings such critical information from the Virtual Machining system, and enables the on-line monitoring and control algorithms to detect the tool failure, control the process more robustly. The application of the new approach has been proven in tool failure detection and process force control in the paper. However, the proposed method has wide applications which include the detection of chatter at specific tool path locations, collecting machining load and idle data and matching them to NC program for improved process planning, and auto-calibrating the process simulation system from on-line force and torque measurements. The proposed system is aimed to let the machine tool to self-adjust itself by making robust decisions with the virtual machining feedback.
3.4. Integrated virtual and on-line machining system Acknowledgement The adaptive control and tool breakage system has been simultaneously tested on profile and slot milling paths shown in Fig. 4. The measured peak cutting forces predicted from motor current, commanded feed speed, spindle torque received from CNC and tool breakage detection results are in Fig. 7. It can be seen that
This research is supported by NSERC CANRIMT2 Strategic Research Network partners. Industrial Technology Research Institute (ITRI) of Taiwan donated the CNC machining centre.
References
Fig. 7. Profiling and slotting operations with virtual feedback; adaptive control and tool breakage results.
[1] Tlusty J, Andrews GC (1983) A Critical Review of Sensors for Unmanned Machining. CIRP Annals-Manufacturing Technology 32(2):563–572. [2] Teti R, Jemielniak K, O’Donnell G, Dornfeld D (2010) Advanced Monitoring of Machining Operations. CIRP Annals-Manufacturing Technology 59(2):717–739. [3] Mo¨hring HC, Litwinski KM, Gu¨mmer O (2010) Process Monitoring with Sensory Machine Tool Components. CIRP Annals-Manufacturing Technology 59(1):383–386. [4] Klocke F, Kratz S, Veselovac D (2008) Position-oriented Process Monitoring in Freeform Milling. CIRP Journal of Manufacturing Science and Technology 1(2):103–107. [5] Smith S, Tlusty J (1993) Stabilizing Chatter by Automatic Spindle Speed Regulation. CIRP Annals-Manufacturing Technology 41(1):433–436. [6] Koike R, Ohnishi K, Aoyama T (2016) A Sensorless Approach for Tool Fracture Detection in Milling by Integrating Multi-axial Servo Information. CIRP AnnalsManufacturing Technology 65(1):385–388. [7] Altintas Y (1992) Prediction of Cutting Forces and Tool Breakage in Milling from Feed Drive Current Measurements. Journal of Industrial Engineering 114:386–392. [8] Astrom KJ, Canudas de Wit C (2008) Revisiting the LuGre Friction Model. IEEE Control Systems 28(6):101–114. [9] Park S, Altintas Y (2004) Dynamics Compensation of Spindle Integrated Force Sensors with Kalman Filter. Journal of Dynamic Systems Measurement and Control 126:443–452. [10] MACHPROTM (2011) Advanced Virtual Machining System, Manufacturing Automation Laboratory, The University of British Columbia, Canada. [11] Altintas Y, Kersting P, Biermann D, Budak E, Denkena B, Lazoglu I (2014) Virtual Process Systems for Part Machining Operations. CIRP AnnalsManufacturing Technology 63(2):585–605. [12] Nouri M, Fussell KB, Ziniti LB, Linder E (2015) Real-time Wear Monitoring in Milling Using a Cutting Condition Independent Method. International Journal of Machine Tools and Manufacture 89:1–13. [13] Altintas Y, Munasinghe WK (1996) Modular CNC Design for Intelligent Machining, Part 2: Modular Integration of Sensor Based Milling Process Monitoring and Control Tasks. Transactions of the ASME Journal of Manufacturing Science and Engineering 118(4):514–521.
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047
CIRP-1605; No. of Pages 4 CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
CIRP Annals - Manufacturing Technology jou rnal homep age : ht t p: // ees .e lse vi er . com /ci r p/ def a ult . asp
Integration of virtual and on-line machining process control and monitoring Y. Altintas (1)*, D. Aslan Manufacturing Automation Laboratory, Department of Mechanical Engineering, The University of British Columbia, BC V6T 1Z4, Canada
A R T I C L E I N F O
A B S T R A C T
Keywords: Cutting Monitoring Control
This paper presents a virtually assisted on-line milling process control and monitoring system. A part machining process is simulated to predict the cutting forces, torque, power, chip load and other process states. The simulated machining states are accessed by a real time monitoring system which detects the tool failure and adaptively adjusts the feed by predicting the forces from the feed and spindle drive motor current supplied by CNC. The integration of virtual simulation with real time measurements avoids false tool failure detection and transient overloads of the tools during adaptive control. The system has been implemented on a CNC machining centre for use in production. ß 2017 Published by Elsevier Ltd on behalf of CIRP.
1. Introduction The recent trend in manufacturing is to develop intelligent, selfadjusting and unattended machining systems to improve the productivity. As reported in past CIRP keynote articles, the reliability of such systems highly depends on the availability of industry friendly sensors [1] and robustness of methods to avoid false alarms and incorrect actions [2]. There have been major efforts in the field of sensor assisted machining. The primary applications focussed on tool wear and breakage monitoring, chatter detection and avoidance, adaptive control of the process forces and dimensional errors, thermal compensation of machines, collision avoidance, and spindle health monitoring. A variety of sensors have been used such as force, vision, acoustic emission, vibration, power, strain, thermocouples and laser devices depending on the application [2]. The reliability of all tool condition and machining process control systems have been suffering mainly because of having difficulties in installing practical and reliable sensors on the machine [3], and not being able to distinguish the actual machining process state from the effects of geometric changes along the tool path [4]. This article presents a new approach: the use of virtual part machining simulation to feed data which is needed to improve the robustness of on-line tool condition monitoring and cutting process control system. The approach is in accordance with Industry 4.0 principles which recommend the integration of digital simulation and sensory data to achieve intelligent manufacturing systems. The proposed approach has been demonstrated in simultaneous adaptive control of cutting forces and detection of tool failure as shown in Fig. 1. A real time communication link between the CNC and external computer has been developed. The actual
* Corresponding author. E-mail address: [email protected] (Y. Altintas).
motor current, velocity, and acceleration of each drive; tool centreposition of the machine; spindle speed and load; tangential feed and currently executed NC block number are obtained from the CNC in real time, and mapped to the virtually simulated data stored in the external computer. The cutting forces are indirectly measured from the feed drive motor current by compensating the disturbance of structural dynamics of ball screw and table through Kalman filters. The cutting forces in Cartesian axes are estimated by transferring the individual motor torque to tool tip.
Fig. 1. Parallel execution of virtual and real time system with information exchange.
The tool breakage is detected from the average spindle torque, and the load on the cutting tool is maintained at the desired level by adaptively controlling the feed rate. The chatter is detected and avoided similar to the method given in [5], hence it is not presented here. However, the locations of chatter events are mapped to tool path contained in Virtual model. The robustness of algorithms is ensured by sending the part geometry changes and average force patterns from the stored, virtual part machining system. The system has been experimentally proven on a CNC machining centre.
http://dx.doi.org/10.1016/j.cirp.2017.04.047 0007-8506/ß 2017 Published by Elsevier Ltd on behalf of CIRP.
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047
G Model
CIRP-1605; No. of Pages 4 Y. Altintas, D. Aslan / CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
2
2. Prediction of cutting forces from drive currents The installation of cutting force sensors to the machine would increase the cost and complexity; hence it is preferred to use sensory data available from the CNC directly [6]. The cutting forces are transmitted to feed drive motors as disturbance torque through the drive structure and servo amplifier as shown in Fig. 2. The friction and inertial loads are separated from the measured motor torque (Tm) measurements to predict the cutting torque (Tc) as [7]: dV T f T c ¼ T m J e dt
where l is the screw lead (pitch) and h is the efficiency. However, the cutting force acting on the tool tip is distorted by the structural dynamics of the feed drive system before reaching at the motor’s current amplifier as a disturbance torque. The Frequency Response Function (FRF) of the structural disturbance has been measured by applying impulse excitation at the table while measuring the current and angular position of the motor shaft from the CNC (Fig. 2b).
Fd ðvÞ ¼ (1)
Fig. 2. (a) Disturbance transfer function (fd), (b) feed drive mechanism, (c) Kalman filter (fKL) compensation (Inom to F a ) (Tcf = coulomb friction). ˆ
where Je and V (rad/s) are the equivalent feed drive inertia felt at the motor and motor velocity, respectively. The motor torque is proportional to the current Tm = Kti where Kt [N m/A] is the torque constant and i [A] is the measured current. The friction torque is modelled as Tf = BeV + Tcf where Be and Tcf are the viscous and coulomb components, respectively. Lu Gre friction model has been used to capture the Coulomb friction, Stribeck effect, hysteresis and pre-sliding displacement [8] of the drive as shown in Fig. 3a. The friction and equivalent inertia of all drives have been identified from the velocity and current values supplied by CNC while moving the drives at different velocities. The cutting force is predicted from the cutting component of the torque as: 2p l
(3)
A sample measurement of disturbance FRF is shown in Fig. 3b for x axis of the machine. The modes at 40, 64 and 175 Hz distort the magnitude and phase, and deviate the magnitude from the desired unity, leading to incorrect measurement of the force when the tooth passing frequency (i.e. spindle speed times number of teeth) and harmonics are beyond 10 Hz. Although the current amplifier of the CNC has 1000 Hz bandwidth, unless compensated, the structural dynamics of the drive reduces the bandwidth of the force prediction from motor current to 10 Hz or 600 rev/min spindle speed when one tooth is used on the cutter. The disturbance FRF is fitted to a transfer function as:
Fd ðsÞ ¼
Fc ¼ Tch
T c ðvÞ T m ðvÞ
(2)
X T c ðsÞ ak ¼ 2 þ 2& v s þ v 2 T m ðsÞ s nk k k k
(4)
where ak, k, vnk are the residue, damping and natural frequency of mode k. The disturbance caused by the structural modes has been compensated by passing the measured motor torque from a Kalman filter which has a transfer function of [9]; T ðsÞ FKl ðsÞ ¼ ˆa ¼ T c ðsÞ
X
aKl
k
s2 þ 2& Kl vKl s þ vKl 2
(5)
where T a is the estimated actual torque transferred from tool tip to ˆ the drive and actual tool tip force (F a ) is found from Eq. [2]. ˆ The FRF of the cutting force prediction from the drive current is shown in Fig. 3b, where the compensated system magnitude approached to unity and the bandwidth has been increased to 200 Hz. As a result, the cutting forces can be predicted accurately from the current monitored from the CNC at tooth passing frequencies up to 200 Hz. A sample cutting force prediction is compared against the dynamometer measurements in Fig. 3c–d. 3. Virtual model assisted process monitoring and control The predicted cutting forces from feed drive motors are used for tool breakage detection and adaptive force control. The machine was a Quaser UX600 CNC machining centre. The real-time algorithms are run on an external PC (Intel1 CoreTM i7-3.40 GHz CPU, 8 GB RAM) which communicates with the Heidenhein CNC via TNC Ethernet connection. A multi-threaded real time code is developed in C++ using the LSV-2 communication protocol which collects commanded, noise free digital motor currents, drive speeds, tool centre point position, spindle speed, and tangential velocity between 330 Hz and 10 kHz sampling frequency. External PC can vary the spindle and feed speeds at 10 Hz interval which is sufficient for the targeted tool condition and process monitoring/ control tasks which are outlined as follows. 3.1. Virtual model – real time application integration
Fig. 3. (a) LuGre friction curve (X axis), (b) identified and Kalman filter compensated FRF, (c, d) comparison of estimated cutting forces from feed drive current and reference forces measured from dynamometer, axial depth = 4 mm, radial depth = 20 mm, feed = 0.2 mm/tooth; (c) 1000 rpm (ft = 33.3 Hz), (d) 2000 rpm (ft = 66.6 Hz).
Prior to the cutting operation, part machining process is simulated using MACHpro1 Virtual Machining System [10] to calculate cutter–workpiece engagement (CWE), cutting forces, torque, power and the machining process states along the toolpath [11]. These machining states are stored in a file and accessed by the real time machining process monitoring and control system as a virtual feedback to avoid false tool failure detection and to prevent transient force peaks during adaptive control. In addition, the
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047
G Model
CIRP-1605; No. of Pages 4 Y. Altintas, D. Aslan / CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
3
uncertainties in the virtual model, such as cutting force coefficients, are continuously calibrated from the real time measurements [12]. The proposed virtual model assisted process monitoring and control system has been demonstrated experimentally on an Aluminium 7050 part shown in Fig. 4. Facing, profiling and two slot milling operations have been performed with a two fluted 20 mm end mill with a regular pitch and 30 degrees helix angle. Spindle speed was selected as 1000 rev/min throughout the operations. The sharp changes in CWE conditions are shown for each operation in Fig. 4. The torque and the cutting forces are simulated along the tool path and stored in a file to assist on-line tool failure monitoring, adaptive control and chatter detection that are run simultaneously. Fig. 5. Robust detection of tool breakage at tooth period 500 with force prediction from virtual machining system.
3.3. Adaptive control
Fig. 4. A test part with various milling operations and corresponding area of tool– workpiece contact.
Gc ðzÞ ¼
3.2. Tool failure detection Tool failure in milling is detected with the algorithm given in [7] but with a simulated feedback from the virtual machining system. The average cutting torque per tooth period (Tsa(m)) is evaluated from the spindle motor; T sa ðkÞ ¼
P X T sa ðiÞ i¼1
P
(6)
where P is the number of torque samples collected at tooth period (m). The average spindle torque per tooth period must remain constant if there is no change in the CWE geometry, and the cutter is free of tooth breakage and run-out. The slow varying CWE effects are removed by a first order auto-regressive (AR) time series filter applied on the differenced average torque as:
eðmÞ ¼ ð1u˜z1 Þ½T sa ðmÞT sa ðm1Þ
Adaptive control has been implemented to keep the peak forces (Fp(k)) at the desired level (Fr(k)) by manipulating the feed-rate in real time at spindle periods (k), see Fig. 6. The objective is to prevent shank breakage and constrain the bending deflections of the tool. The peak force at each spindle period is estimated from the feed drive motor current. The feed and current are read from the CNC. The discrete time transfer function (Gc(z)) between the peak cutting force (Fp(k)) and feed-rate (fk(k)) is expressed as: F p ðkÞ z1 Bðz1 Þ z1 ðb0 þ b1 z1 þ b2 z1 Þ ¼ ¼ f c ðkÞ Aðz1 Þ 1 þ a1 z1 þ a2 z2
(8)
where the parameters (b0, b1, b2, a1, a2) of the combined feed drive and milling process system (Gc(z)) are estimated with the recursive least-square (RLS) algorithm at each spindle period. An adaptive generalized predictive control (GPC) method is used to keep the maximum resultant cutting force at a desired level by manipulating the feed at each control interval (k) as: f c ðkÞ ¼ f c ðk1Þ þ Gac ðzÞ F r ðkÞF p ðkÞ (9) ˜ Control law (Gac ðzÞ) is adaptively adjusted through; Gac ðzÞ ¼ ˜ ˜ fGI gT =½fGI gT fGI g þ l where fGI g ¼ ðfF p gf f gÞ=Df c . fF p g contains ˆ ˆ
(7)
where u is the AR filter parameter and estimated adaptively from ˜ the measurements using the Recursive Least Squares (RLS) method at each tooth period. A tool breakage alarm is triggered whenever the residuals violate the set threshold (e(m) > LIMIT) [7]. However, this previously proposed algorithm has are liability problem: The filter (Eq. [7]) cannot distinguish the tool failure from the step changes in CWE geometry and gives false alarm at tooth periods 262–270 (Fig. 5). The constant breakage threshold cannot distinguish the failure from the step changes in the geometry along a straight path as shown in Fig. 4 (Block Geometry). However, the virtual machining system automatically calibrates the cutting force coefficient from the CWE information and measured force from the current in the beginning of cut, and predicts the average torque accurately as the CWE geometry changes. The residual threshold is adaptively adjusted (i.e. 40% of the predicted force), and alerts the tool condition monitoring system whenever both the measured and residual torque violate the limits. The virtual machining assisted algorithm does not give false alarm at step changes (tooth periods 262–270) but detects the tool failure imposed at tooth period 500 (Fig. 5). The algorithm is now reliable and can be used in production. The thresholds are now geometry independent can be set % of the predicted values by considering only the disturbances caused by the tool run out.
Fig. 6. Virtual model assisted GPC, adaptive control (AC) with force identified from drives; no AC (NoAC), conventional AC (AC) and AC with virtual machining assistance (AC with VF).
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047
G Model
CIRP-1605; No. of Pages 4 4
Y. Altintas, D. Aslan / CIRP Annals - Manufacturing Technology xxx (2016) xxx–xxx
j step ahead prediction of peak cutting forces and {f} contains present and past measured peak forces, feed commands and recursively calculated polynomials whose details can be found in [13]. The weighting factor is usually selected as l 0.2. However, such an adaptive control system has not been adopted by industry due to large transient force peaks, which either breaks the tool or damages the part due to excessive deflections when the CWE geometry suddenly increases as shown in the test part (Fig. 4). When the tool enters into a cavity or the depth of cut is low, the corresponding peak force become slow; hence the adaptive controller generates maximum feed to increase the force towards the reference value (Eq. [9]). When the tool exits the cavity or the depth suddenly increases, the servo cannot reduce the feed quickly. The resulting high feed creates a very large cutting load spike that breaks the tool or damages the part as shown in Fig. 6. Hence, the intensive adaptive control research in the past has not been successfully used in practice. This problem is solved by bringing the CWE geometry conditions and simulated forces from the virtual machining system to the adaptive control algorithm ahead of sudden geometry changes (i.e. one second long look-ahead of tool path distance) as shown in Fig. 6. A flag ( ) is set to 1 whenever the CWE area is, i.e. 15%, more than the current one, and set to zero otherwise. When the flag value is 1, the incoming force predicted by the virtual machining system is fed into the controller as a disturbance force prior to the actual CWE change. As a result, the adaptive controller slows down the feed, and prevents force overloads and collision spikes; hence the adaptive controller becomes robust and practical to be used in production, especially for roughing operations. In addition, the flag is set to 2 when the tool is in the cavities, the controller is halted and the feed rate is kept at a safe level to prevent excessive feeds which may lead to collision.
the peak cutting forces remain constant without any overshoot, the false tool breakage detection does not occur at transient changes in the cutter–workpiece engagement. 4. Conclusion The tool condition monitoring and process control in machining has not been used in industry due to false alarms, poor robustness of algorithms and lack of practical cutting force sensors. Utilizing the recent advances in digital machining and CNC technologies, this paper presents a new approach by integrating virtual machining system with the on-line monitoring and control system. The forces are extracted from the feed and spindle drive motor current commands by compensating the distortion caused by the structural dynamic chain between the cutting and motor locations. The use of CNC inherent force sensing eliminated the need to mount costly and impractical sensors on the machine. Current online monitoring systems are blind to changes in the cutter– workpiece engagement and process, which prevents them to be robust against collision forces and false tool failure alarms. The proposed system brings such critical information from the Virtual Machining system, and enables the on-line monitoring and control algorithms to detect the tool failure, control the process more robustly. The application of the new approach has been proven in tool failure detection and process force control in the paper. However, the proposed method has wide applications which include the detection of chatter at specific tool path locations, collecting machining load and idle data and matching them to NC program for improved process planning, and auto-calibrating the process simulation system from on-line force and torque measurements. The proposed system is aimed to let the machine tool to self-adjust itself by making robust decisions with the virtual machining feedback.
3.4. Integrated virtual and on-line machining system Acknowledgement The adaptive control and tool breakage system has been simultaneously tested on profile and slot milling paths shown in Fig. 4. The measured peak cutting forces predicted from motor current, commanded feed speed, spindle torque received from CNC and tool breakage detection results are in Fig. 7. It can be seen that
This research is supported by NSERC CANRIMT2 Strategic Research Network partners. Industrial Technology Research Institute (ITRI) of Taiwan donated the CNC machining centre.
References
Fig. 7. Profiling and slotting operations with virtual feedback; adaptive control and tool breakage results.
[1] Tlusty J, Andrews GC (1983) A Critical Review of Sensors for Unmanned Machining. CIRP Annals-Manufacturing Technology 32(2):563–572. [2] Teti R, Jemielniak K, O’Donnell G, Dornfeld D (2010) Advanced Monitoring of Machining Operations. CIRP Annals-Manufacturing Technology 59(2):717–739. [3] Mo¨hring HC, Litwinski KM, Gu¨mmer O (2010) Process Monitoring with Sensory Machine Tool Components. CIRP Annals-Manufacturing Technology 59(1):383–386. [4] Klocke F, Kratz S, Veselovac D (2008) Position-oriented Process Monitoring in Freeform Milling. CIRP Journal of Manufacturing Science and Technology 1(2):103–107. [5] Smith S, Tlusty J (1993) Stabilizing Chatter by Automatic Spindle Speed Regulation. CIRP Annals-Manufacturing Technology 41(1):433–436. [6] Koike R, Ohnishi K, Aoyama T (2016) A Sensorless Approach for Tool Fracture Detection in Milling by Integrating Multi-axial Servo Information. CIRP AnnalsManufacturing Technology 65(1):385–388. [7] Altintas Y (1992) Prediction of Cutting Forces and Tool Breakage in Milling from Feed Drive Current Measurements. Journal of Industrial Engineering 114:386–392. [8] Astrom KJ, Canudas de Wit C (2008) Revisiting the LuGre Friction Model. IEEE Control Systems 28(6):101–114. [9] Park S, Altintas Y (2004) Dynamics Compensation of Spindle Integrated Force Sensors with Kalman Filter. Journal of Dynamic Systems Measurement and Control 126:443–452. [10] MACHPROTM (2011) Advanced Virtual Machining System, Manufacturing Automation Laboratory, The University of British Columbia, Canada. [11] Altintas Y, Kersting P, Biermann D, Budak E, Denkena B, Lazoglu I (2014) Virtual Process Systems for Part Machining Operations. CIRP AnnalsManufacturing Technology 63(2):585–605. [12] Nouri M, Fussell KB, Ziniti LB, Linder E (2015) Real-time Wear Monitoring in Milling Using a Cutting Condition Independent Method. International Journal of Machine Tools and Manufacture 89:1–13. [13] Altintas Y, Munasinghe WK (1996) Modular CNC Design for Intelligent Machining, Part 2: Modular Integration of Sensor Based Milling Process Monitoring and Control Tasks. Transactions of the ASME Journal of Manufacturing Science and Engineering 118(4):514–521.
Please cite this article in press as: Altintas Y, Aslan D. Integration of virtual and on-line machining process control and monitoring. CIRP Annals - Manufacturing Technology (2017), http://dx.doi.org/10.1016/j.cirp.2017.04.047