Sensors for the food industry

Sensors for the food industry

Sensors for the food industry Eiji Ohashi and Isao Karube describe the collaborative Japanese development programme This article presents an overview ...

827KB Sizes 37 Downloads 288 Views

Sensors for the food industry Eiji Ohashi and Isao Karube describe the collaborative Japanese development programme This article presents an overview of the results of a five-year research project on sensors for the food industry in Japan. This project was a government-sponsered research programme involving the food industry and the instrumentation industry, coordinated by the Japanese Ministry of Agriculture, Forestry and Fisheries. The programme focussed on I7 food processing operations and quality control tasks, where the need for an on-line (or rapid at-line) measurement had been established. The instrument development work employed biosensors pH and gas sensors, optical-imaging devices, as well as near-infrared, microwave, ultrasound and mechanical resonance techniques, together with robotics and advanced data-processing techniques.

Introduction Food processing and its control technology has been developing along with the diversification of consumer demands. For the research and development of novel food processing systems which perform both smooth and flexible manufacturing operations, the Japanese Research and Development Association for Sensing Systems in the Food Industry (On-Line Sensor Association) was established in 1986 with the assistance of the Japanese Ministry of Agriculture, Forestry and Fisheries (MAFF). The On-Line Sensor Association carried out a five-year research and development project in order to develop sensors and systems in cooperation with 35 companies from the food industry, the instrumentation industry, the system engineering industry, etc. The project had 17 research subjects, and two or three companies were each specifically assigned a research subject as shown in Table I. Research efforts were roughly divided into the following phases: ?? collection

of fundamental data and survey of recent papers and patents;

Ohashi is from the Central Research Laboratory, Nippon Suisan Kaisha, Ltd. 559-6, Kitano. Hachioji, Tokyo 192, Japan

E.

and I. Karube from the Research Center for Advanced Science and Technology, University of Tokyo, 4-6-1, Komaba, Meguro-ku, Tokyo 153. Japan

0956-7135/93/040163-06

0

evaluation of conventional sensors and sensing systems; clarification of target function; selection of materials or systems for sensors; construction of prototype sensors and sensing systems; establishment of a total system for sensor application and controlling the food production line. All research was evaluated each year by an advisory group consisting of university professors, senior researchers from national institutions and consultants. Meat freshness sensor In the process of meat aging, particularly beef, simultaneous inspection of aging and decomposition is necessary. Development of a meat freshness sensor capable of simultaneous measurement of these two properties has long been awaited and Woller, 1982; (Wortberg Edwards et al., 1983). In response to such a demand, a proper control system using biosensors was investigated. At first, attempts were made to find novel chemical indicators using data mainly obtained from high chromatoliquid performance graphy (HPLC). A two-line flow injection system was then constructed for the rapid measurement of aging and decomposition. From the results of the experiments, it was clear that two diamine compounds, putrescine and cadaverine, were effective as indicators

1993 Butterworth-Heinemann Ltd

for the initial decomposition among amines generated from decarbonation of amino acids by bacteria. For aging, hypoxanthine and xanthine, generated by dissolution of ATP, were effective chemical indicators. A meat freshness sensor system was developed comprising two channels and two sensors. Small portions of beef were sampled over time and extracts of the sample were analysed by the system. The results showed good correlation with a conventional HPLC method; thus the sensor system proved effective for monitoring the aging of meat. Bacteria sensors meat products

for meat

and

In the manufacturing of meat products, bacterial inspection of the products and processes are carried out as part of the quality assurance programme. There have however, been drawbacks in the conventional inspection method, in that it requires considerable labour and time. A method for quick and simple detection of coliform colonies was therefore examined. A culture medium for detecting coliform colonies was developed with the aim of shortening the period of the growth initiator phase and cell generation, plus improvement of the formation of P-Dgalatosidase. A cylindrical culture container and a simple colony counter were developed to simplify the coliform counting operation. The latter was characterized by formation of a larger colony and the former was characterized in that the culture container was rotated to detect a colony by scanning with a linear array CCD device array. When a pigment binding substrate [5-bromo-3-indolyl -p - D-galactoside (X-gal)] was incorporated into the culture medium, cleaver colonies were formed and it was therefore possible to count a coliform colony after only 15 hours of culturing. Another method was developed substrate using a fluorescent [4-methylumbelliferylp -o-galactoside (bMUG)] incorporated into the culture medium. This method,

Food Control 7993 Volume 4 Number 4

183

Comment

Table 1 Summary of the research projects Name of member company

Research subject

Food manufacturer

Sensor manufacturer

I. Development

Ito Ham Foods Inc.

DKK Corporation

2. Development of bacteria sensors for meat and meat products

Nippon Meat Packers Inc.

Toyo Jozo Co. Ltd Nissui Pharmaceutical Ltd

3. Development of sensors for fish processing I

Nippon Suisan Kaisha Ltd

Chino Works Ltd

4. Development of sensor for fish processing II

Kibun Co. Ltd

NOK Corporation

5. Development of composition sensor for milk and milk products

Meiji Milk Products Co. Ltd

Fujihira Industries Co. Ltd

6. Development of sensors for lactic acid fermentation process

Morinaga Milk Products Co. Ltd

Toa Electronics Co. Ltd

7. Development of sensors for margarine manufacturing process

Snow Brand Milk Products Co. Ltd

Shimazu Corporation

8. Development of sensors for confectionary baking process

Meiji Seika Kaisha Ltd

Toshiba Corporation

9. Development of sensors for nut processing

Morinaga & Co. Ltd

Anzai Universal Laboratory Ltd

10. Development of sensors for noodle manufacturing process

Nisshin Flour Milling Co. Ltd

Fuji Electric Co. Ltd Fuji Electric Corporate Research and Development

11. Development of sensors for soy-sauce manufacturing processing

Kikkoman Corporation

Omron Tateishi Electronics co. Tateishi Institute of Life Science Inc.

12. Development of sensors for oil and fat refining process

Fuji Oil Co. Ltd

Murakami Color Research Laboratory Nireco Corporation

13. Development of sensors for soup concentrate manufacturing process

Nichirei Corporation

DKK Corporation

14. Development of sensors for inspection of packaged meat products

Prima Meat Packers Ltd

Nippon Keisoku Kogyo Co. Ltd NIKKA Densok Ltd

15. Development of sensors for inspection of external appearance of packagings for milk’products

Meiji Milk Products Co. Ltd

Yamatake Honeywell Co. Ltd

16. Development of sensors for inspection of external appearance of bottle

Suntory Ltd

Fuji Electric Co. Ltd Fuji Electric Corporate Research and Development

17. Development of sensors for detecting metalic foreign materials

Ajinomoto Co. Ltd

Nissin Electronics Co. Ltd

of meat freshness sensor

using fluorescence, made it possible to detect coliform colonies within 6 hours. Because of the shorter detection time using fluorescence, a highly sensitive fluorescent detection system is now under development. Sensors for fish processing Research 184

was

undertaken

I to

Co.

develop a moisture sensor measuring small differences in content at high moisture concentrations and apply it to the manufacturing process of surimi (minced fish meat). The aim of the research was to develop a sensor capable of rapidly measuring moisture content in surimi within a range of 7.545% with an accuracy of *OS%, there-

Food Control 1993 Volume 4 Number 4

by allowing for fine adjustment of the process conditions. absorption A near-infrared method was employed as the measuring principle because of its ease in operation and maintenance plus its safety and hygiene advantages. After several trials, it was found that a wavelength of 1.2pm was suitable for the moisture measurement and that the reflection method was more accurate than the diffusion method. A quantitative surimi feeder was next developed. It had a surfacetreating mechanism to optically prepare the surface of surimi, so that an average moisture content of the whole surimi could be measured continuously. This surface treatment was combined with the moisture sensor to fabricate a surimi moisture-measuring apparatus. Results with this apparatus showed that it was possible, during the manufacturing process, to measure the moisture content of surimi in the range of 7545% with an accuracy of better than +0.3%. It was also possible to obtain this accuracy over time with varying moisture content of the products. Sensors for fish processing

II

The most important material for Japanese marine paste products is frozen surimi. In order to control the product quality, variation in the quality of frozen surimi must be measured during the manufacturing processing quickly and with high accuracy. In addition, the mixing process plays an important part because it has a great effect on the quality and yield of products. Control of this process at present depends on observations of the product by skilled operators. Research was undertaken to find a mixingprocess control sensor to obtain a stable product quality and to reduce the manufacturing cost. TO evaluate the quality of surimi, a pH electrode and a pH-stat method were examined which were capable of performing an accurate evaluation within a short time compared with the conventional method using gel-breaking strength. For the mixing control sensor, a method using a combination of electricity consumption and conductivity of a cantilever-type mixing motor was examined.

Comment

From the comparison between the data from the pH electrode versus the pH-stat method, it was determined that the pH-stat method could perform a more accurate measurement in a shorter time. In addition, there was a good correlation with the conventional gel-breaking strength method. On the other hand, it was realized that the mixing control sensor could control all the processes, including temperature control, through consumption of electricity by the motor for the period from idling to salt feeding, the conductivity for the period following the feeding of salt and sub-materials, and the power consumption rate of the motor till the end of mixing.

Composition sensors for milk and milk products Recently, several instruments using infrared light have been developed for milk analysis (Sato et al., 1987, 1988; Biggs, 1978, 1979; Young, 1978; Van de Voort, 1980). These researchers developed software for utilizing a continuous measuring apparatus, i.e. a multicomponent system (MMS by Foss Electric) with a built-in infrared milk analyser Milko-Scan, as an on-line composition sensor for measuring not only milk but also milk products including flavoured milk, ice-cream mix, using a cream etc. Calibration regression equation made it possible to measure the components of milk products containing vegetable fat, cane sugar and sugared syrup, as well as dairy components. It was necessary to dilute thick samples such as ice-cream mix and cream measurement. Consebefore quently, a sample liquid diluting system (based on sample weight) and a cup with a built-in rotor and a cup washer were designed to produce a fully automatic sample liquid preparation system which uses a robot to handle the cup and a pipette. Installation of this system in a position adjacent to the MMS made it possible to carry out a series of operations including sampling, dilution and mixing, plus measurement of components and washing of the cup in a fully automatic mode. It was found that, with this method, the MMS is applicable as a composition sensor in the manufacturing process of milk products,

including thick liquids that require dilution.

Sensors for lactic acid fermentation processes The quality of yoghurt products varies depending on fermentation conditions. The quality is unstable because the fermentation process has not yet been automated. Thus, the objective of this research was to r develop an on-line sensor for the automation of the fermentation process to stabilize and improve the product quality. In order to attain this objective, the following sensors were developed: a fully automatic pH sensing system tolerant to cleaning in place (CIP) ; a sensing system, or yoghurt meter, was developed using an enzyme electrode and a whole lactic acid measuring device with a dilution conductivity method and an automatic sampling system; optical isomers of latic acid (Tamime and Deeth, 1980) and their acidity (Osajima, 1981) could be measured; vibration a non-destructive analysis sensor, capable of measuring the gel formation of yoghurt in a cup or a container repeatedly and quickly, was the gel formation developed; was measured by its vibration frequency response compared with normal mode; a method for counting live bacteria was developed using the ATP method, and measurement of a single test was conducted within a few minutes.

Sensors for margarine manufacturing process

margarine mix. In order to improve the accuracy and stability of the sensor, a quartz glass was employed as a pipe section and the shapes of an electrode and a resonant coil were improved to increase the Q value (frequency selectivity) controlling the attenuation of microwaves. In this way, a sensor capable of measuring the moisture content over a wide range of concentrations (1530%) was developed. Calibration equations for three different kinds of products were obtained in the development programme and values of r2 obtained from the respective equations were as high as 0.988, 0.988 and 0.984, respectively. As a result, it was concluded that this sensor could be put into practical use as an on-line moisture sensor applicable in the margarine manufacturing process. Further, it was confirmed that since the physical properties of margarine varied depending on the compounding ratio and temperature, it was necessary to choose a suitable calibration equation depending on the type of margarine mix.

Sensors for confectionary baking process The process of baking biscuits and the roasting of nuts are the most important and difficult processes for product quality and are currently controlled by the experience and skill of the process operators. This research was carried out to develop sensors and systems for automatic control of the product quality. The research focused on the following three targets: ?? development

With regard to quality evaluation and analysis of margarine, remarkable progress has been made in the performance of analysing devices; however, most still depend on offline inspection. This research was carried out to develop a sensor capable of on-line inspection of the moisture content of a margarine mix, which is an important factor for the product quality. As a measuring principle, a microwave resonance method was employed to confirm the possibility that a desktop instrument could measure the moisture content of a

of an on-line sensor colour thickness for and measurements and an image processing system; ?? development of a new sensor and control system that replaces a thermometer in conventional controlling an oven; ?? development of a fuzzy logic control-applied expert system incorporating the knowledge and experience of skilled operators. For judgement of a baked product, a high rate of coincidence with results by visual judgement was obtained from the new sensor. For thickness, a measuring accuracy of 0.14mm was achieved. It was clear

Food Control 1993 Volume 4 Number 4

185

Comment

Two goals were set for the development of sensors for processing nuts. One was to develop an apparatus for detecting and removing mouldy nuts and the other was to construct an automatic assessment system for the final roasting stage. For the former objective, a sensor and sorter system using nearinfrared radiation was developed for the inspection of not only peanut surface but inspection of the entire peanut. This system made it possible to carry out nondestructive inspection of nuts and to remove mouldy nuts, including those nuts internally contaminated with moulds. As a result, it was found that mould poisons, including aflatoxin Bl, could be reduced. For the latter objective, a system capable of judging the end-point of roasting, by measuring the temperature of the various sections of the oven and calculation using regression equations, was developed and is in practical use.

quality. The conventional method for the determination of colour intensity is visual comparison under 15 watts of fluorescent light between a sample and a reference standard. Because visual measurement is used, determination of the colour number requires a skilled operator and the result can vary with different individuals. Thus, ‘a soysauce calorimeter’ was developed capable of automatic measuring, which will simplify colour determination and eliminate the need to prepare a reference liquid. A colour number for soy-sauce was obtained from the Y value calculated from a transmission factor for 31 wavelengths measured For using a spectrophotometer. miniaturization and simplification a filter type apparatus was employed and a xenon flash lamp was used as a light source. The transmitted light was received by a photodiode a colour number was then determined according to the Y value obtained and a colour number conversion table. For referencing, light passed through a glass filter, having nearly the same spectral characteristics as those corresponding to the Y value. The calorimeter was used in the measurement of colour numbers for soy-sauce and its use showed an accuracy higher than that of visual measurement plus satisfactory results in practical use. In short, the calorimeter provides easy operation, quick measurement and good portability.

Sensors for noodle manufacturing process

Sensors for oil and fat refining process

Noodles are important in Japan. However, their production lines are not highly automated and depend on control by skilled operators. In this project, a ceramic moisture sensor for the use in the production line was investigated, and a limiting current type zirconia oxygen sensor was developed. This sensor was applicable for the measurement of humidity in an atmosphere of 150°C.

In the edible oil industry, crude oil extracted from vegetable seed, fruits and animals is refined through deoxidation, decoloration and deodorization and made into edible oil. In these processes, colour and iodine value (I.V.) are used as important indicators for quality control. The conventional method for colour measurement is visual observation by operators and, for measurement of I.V., a system utilizing a manual titration requiring a complicated manual operation is used. An on-line measurement technique using an optical method was developed to attain stable and continuous measurements.

that the newly developed heat flux control method was superior to a conventional temperature control method in that the former could maintain the quality of biscuits at a constant level regardless of changes in the wind velocity in an oven. Moreover, a system was developed capable of diagnosing the state of operation of a manufacturing line on the basis of data or signals received from operators or a sensor. Sensors for nut processing

Sensors for soy-sauce manufacturing process The colour of soy-sauce is a major property for evaluation of its

166 Food Control 1993 Volume 4 Number 4

For colour measurement, spectral transmission was observed and used to make corrective calculations of X, Y, Z and x, y, respectively. A computer colour matching method was then used for the generation of a Lovibond Surface Color value. In order to reduce deflection of an optical axis due to the flow of oil or a change in temperature, improvements to the optical system and correction of the software calculations were made. It was found that the use of correlation equations for a value by a sensor and a value by visual measurement could result in colour measurement with an accuracy equal to or higher than that of visual measurement. For I.V. measurements, near/ infrared spectrophotometric analysis was used (Iwamoto, 1987). The problem of broad absorption was solved by computer processing using a differential method and multiple regression analysis. An online system was constructed as follows. After filtration, the oil was passed through a deaeration vessel to remove air bubbles and then equilibrated at the proper temperature for measurement in a flow cell. After several improvements, accuracy in the on-line system increased to a level similar to that obtained at the laboratory. The correlation coefficient between values obtained by manual analysis and values from the sensor was 0.993 and the standard error of calibration (SEC) was 0.476. Sensors for soup concentrate manufacturing process In many food manufacturing processes, human sensory evaluation is important for quality control. The objective was to develop an on-line sensing system for taste of consomme type soups (broths). From the data obtained from an omission test, it was found that the taste of the soup was greatly influenced by glutamic acid, inosinic acid, lactic acid and glucose, as well as by aromatic volatile compounds. Based on six factors consisting of the above-mentioned five indices plus a Brix index, multiple regression analysis was made for measured values from soup samples and organoleptic test values to obtain a correlative equation.

Comment

A newly developed taste-sensing system comprised four parts: a taste component measuring unit using an enzymatic FIA system; an automatic measuring unit using an oxide semiconductor device as a gas sensor for indicating a relative intensity of aroma; a Brix flow meter; an automatic control and evaluated value computing system using a personal computer. This taste-sensing system was used for sensory evaluation of 39 real samples with 15 skilled panelists. The correlative analysis showed a correlation as high as 0.806 in terms of the multiple regressive contribution rate.

Sensors for the inspection of packaged meat products In recent packaging processes, gaspackaging methods sealed (modified atmosphere packaging, MAP, where mainly N2 gas is used) and vacuum packaging methods are used. Pinholes or defective sealing of a packaging film causes decomposition, discolouration and fat oxidation of the product, resulting in the loss of the product’s commercial value (Gilchrist et al., 1985; Lampi, 1980). The objective of this research was to develop sensors for performing continuous inspection of every product on the line for pinholes and defective sealing. Gas-sealed packaged products are sealed with nitrogen gas (or a specially formulated gas mix containing mainly N2) instead of air in order to prevent oxidation. The principle employed was to use a gas mixture composed of 95% N2 and 5% He for the pakaging gas. The helium gas leakage from pinholes or defective sealing was detected by a detector. A packaged product sealed with the above-mentioned gas is automatically fed into a chamber and then depressurized. With this apparatus pinholes of more than 25p.m diameter were detectable. If defective sealing and pinholes exist in a vacuum packaged product, air comes inside the package to form an air layer. When an ultrasonic wave pulse is transmitted into such a defective product, there is a difference in time for the

reflected wave pulse to appear and the amplitude of the reflected wave is increased. A sensor was developed capable of detecting such differences using electric signals to perform continuous inspection. Sliced ham packs were used for the experiments. It was confirmed that any product having an ingressed amount of air of more than 0.3ml could be detected as a defective package.

Sensors for inspection of external appearance of packaging for milk products In the milk product manufacturing process, a number of visual inspections occur. The objective was to automate such inspections. An image-processing apparatus was developed as a new inspection technique and used in the data printing process of a yoghurt paper package by an ink-jet data printer and the address printing process of a gabletop paper package. For the development of a new image processor, it was necessary to know what kind of algorithms should be used in the inspection. Thus, an interactive imageprocessing system was constructed on a multi-use UNIX minicomputer. Subsequently, hardware for implementing the algorithms was developed. The hardware was modularized by function, to provide the ability to share resources and multi-usability of the system. Another feature was that the system was provided with specifications suitable to the manufacturing site, unique mechanical structures and an input/output processing function like a sequencer. The system was used on an actual production line as described above. The processing time was about 300 ms.

Sensors for inspection of external appearance of bottles Development of sensors for inspecting the mouth on a bottle was made to automate visual inspections of a whisky or brandy bottling line. In the experimental laboratory, an method was first inspection screened which offered a greater possibility of practical use than the inspection methods currently available. The biggest factor in screening

of an inspection method using an image-processing technique is whether or not variation in a characteristic quantity for a defective bottle is large and whether or not variation in a characteristic quantity for a normal bottle is small. As a result of screening, it was determined that a special lighting method, known as the multi-spot method for radiating a number of beams having high directivity to the bottle mouth section, was the most practical. Subsequently, a prototype system which incorporated this lighting method was fabricated and installed in an experimental line to be subjected to a test using a great number of bottles. Evaluation of its practical use was carried out for both phases of detecting for defective bottles and yield for normal bottles. In the inspection of the bottle mouth section, this sensor could obtain a permissible range of yield (99.97%) with detectivity equal to that of the visual inspection currently carried out.

Sensors for detecting metallic foreign materials Research was carried out to develop a highly reliable sensor detecting having an enhanced capability for detecting metallic foreign materials such as stainless steel etc. in food products, plus one that is also capable of responding to production lines of different kinds and small amounts of food. The research began with investigation of the present situation at the manufacturing site and determined that a product effect (PE; a phenomena in which a product to be inspected produces the same signal as that produced by a piece of metal) was caused by lower preset sensitivity. Subsequently, analyses of PE revealed that the product temperature was correlated with PE and the relationship between PE and a signal from a piece of metal varied according to the exciting frequency of a sensor section. In view of this, an automatic inspection system was developed metallic foreign for detecting materials with high sensitivity by measuring the product temperature with an infrared radiation thermometer set to detect a metal section

Food Control 1993 Volume 4 Number 4

187

Comment

at the optimum condition according to the measured product temperature. This trial system has improved sensitivity by 20-40% for a metallic ball having the detectable diameter comparable with that of a conventional system.

References Biggs, D.A. (1978) Instrumental infrared estimation of fat, protein, and lactose of milk: collaborative study. J. Assoc. Off. Anal. Chem. 61, 1015-1034 Biggs, D.A. (1979) Infrared estimation of fat, protein, and lactose in milk: evaluation of multispec instrument. J. Assoc. Off. Anal. Chem. 62, 1202-1210 Edwards, R.A., Dainty, R.H. and Hibhard, C.M. (1983) The relationship of bacterial numbers and types to diamine concentration in fresh and aerobically stored beef,

pork and lamb. J. Food Technol. l&$777788 Gilchrist, J.E., Rhea, U.S., Dickerson, R.W. and Campbell, J.E. (1985) Helium leak

test for micron-sized holes in canned foods. J. Food Prof. 48, 856-860 Iwamoto, M (1987) Near-infrared technology in Asia. Near-Infrared Technology in the Agricultural and Food Industries (Ed. P. Williams and K. Norris) American Association of Central Chemists, Inc., Minnesota, USA Retort pouch: the development of a basic packaging concept in today’s high technology era. J.

Lampi,

R.A.

(1980)

Food Process Eng. 4, l-18 Y. (1981) Electrochemical measurement of organic acid and salt content in food. Nippon Shokuhin Kogyo

Osajima,

Gakkaishi 28, 608-613 Sato, T., Yoshino, M., Furukawa, S., Someya, Y., Yano, N., Uozumi, J. and Iwamoto, M. (1987) Analysis of milk

180 Food Control 1993 Volume 4 Number 4

constituents by the near infrared spectrophotometric method. Jpn J. Zootech. Sci. 58, 698-706 Sato, T., Yoshino, M. Iwamoto,

M., Cho, R.K. and (1988) Near-infrared

reflectance spectroscopic analysis for butter constituents. Jpn J. Zootech. Sci. 59,806-813 Tamime, A.Y. and Deeth, H.C. (1980) Yoghurt: technology and biochemistry. J. Food Prot. 43, 939-977 F.R. (1980) Evaluation of Milkoscan 104 infrared milk analyzer. J.

Van de Voort,

Assoc. Off. Anal. Chem. 63, 973-980 Wortberg,

B. and Weller,

R. (1982)

Zue

Qualitaet und Frische von Fleisch und Fleischwaren im Hinblick auf ihren Gehalt an biogenen Aminen. Fleischwirt schaft 62, 14.57-1460,

1463

Young, R.S. (1978) Calibration and standar-

dization of the infrared milk analyzer. The California experience. J. Dairy Sci. 61, 1279-1283