Intelligent System for Pasteurized Milk Quality Assessment and Prediction
Keywords:Artificial neural network, expert system, pasteurized milk, quality assessment system
AbstractPasteurized milk is one of better-processed milk products, which has nutrition composition and taste like fresh milk. The composition of milk nutrition contains essential factors needed for the human health, such as fat, protein, carbohydrate and mineral. Pasteurized milk quality needs to be assessed and controlled to ensure the factors at optimum level. Better quality assessment and control system will fulfill needs of costumer and increase the customer satisfaction and trust. This research has developed an intelligent system for pasteurized milk quality assessment and prediction. The system utilized two type of analysis: fundamental and technical analysis. These analyses were done with Expert System and Artificial Neural Network (ANN). Fundamental analysis covered raw material, working in process and end products quality factors. The technical analysis covered time and deviation of process temperature quality factors. The fundamental and technical analyses can be used to assess and predict the decrease and increase pattern in pasteurized milk quality when the related factors are change. The intelligent system for pasteurized milk quality assessment and prediction could help the quality decision maker in assessment and prediction of pasteurized milk quality from the raw material, production process until the final product packaging and storing. Information on speed and accuracy, produced by the system, increase efficiency and effectiveness in quality control and faster the product quality decision making process. Attributes used to assess raw material quality were density, temperature, composition, freshness and the microbiology of the fresh milk. Whereas, process quality attributes considered were process critical point, sanitation and characteristic of pasteurized milk quality. Some attributes for storage and packaging quality assessment were also developed. Pasteurized milk quality data was processed with entropy method. The reasoning strategy used was “Forward Chaining” and the tracing method used was “Best First Search”. Certainty Factor (CF) was used for handling uncertainty. Data Based Management System (DBMS) managed the related data. Model Based Management System (MBMS) managed the models used. The MBMS consists of five models: quality of fresh milk model, quality of production process model, quality packaging and storage model, statistical process control model and prediction model. The integrated system is implemented into computer software. Multi-layer neural network architecture was used. It consisted of one input layer, one hidden layer and one output layer. The suited activation function was Sigmoid Bipolar. Decision for fresh and pasteurized milk quality acceptance was taken based on the acceptance grade, which was practically easier and accurate. Based on the grade acceptance the management could take follow-up action and decision. This acceptance grade information, whether in grade A or B, is very important as a company bargaining position with the other processed milk industry. This system was designed, developed and implemented by using Mat lab 6.5. The system output consisted of quantitative and qualitative information and can be shown in graphic. The system was verified and validated by using real data collected from pasteurized milk and milk Products Company at West Java, Indonesia. In this company, the quality of fresh milk was at grade B, the quality of process was at grade B and the quality of packaging and storage was at grade B. This system gave suggestion for user to always improve the quality of pasteurized milk, since the current grade was still at grade B. The intelligent system was also compared with the manual system. The developed system gave more accurate result and faster time of analysis.
How to Cite
Marimin, M., Septiani, W., Sukardi, S., & Bunasor, T. K. (2007). Intelligent System for Pasteurized Milk Quality Assessment and Prediction. Proceedings of the 51st Annual Meeting of the ISSS - 2007, Tokyo, Japan, 51(2). Retrieved from https://journals.isss.org/index.php/proceedings51st/article/view/451