Research And Application Of The Expert System In Sugar Boiling Process Based On The Rough Set And Support Vector Machine

Formation and growth of sucrose crystals is closely related to the crystallization environment of the tank in sugar boiling process. It’s affected by many factors such as supersaturation of syrup, temperature and vacuum and so on. Due to the extremely complicated mechanism with many qualitative and quantitative characteristics, it’s difficult to establish a precise mathematical model. Currently there is no suitable automation controlling technology in sugar boiling process. It must be operated by people in the vast majority of sugar enterprises. In the sugar boiling process, many experienced workers of sugar enterprises observe the seed size, the sugar content and the extent of absorption in the mother liquor on glass by sampling stick extraction materials, and further analyse crystallization state and judge crystal grain is in a better crystallization state under the current sugar Brix, temperature and vacuum and so on, and finally adjust the feed rate or the amount of water.Under this background, this paper studied the sugar boiling expert system and applied it in the prediction of crystallization state based on analysis of development of measuring technology in sugar boiling process. Combining with expert’s experience, a knowledge base for the expert system was built with parameters such as Brix, temperature and vacuum as its conditions and the analysis result of crystallization state as its conclusion. And the prediction of crystallization state was drawn from the expert system.With regard to the problem in attaining the knowledge of the expert system used to predict the crystallization state in sugar boiling process, the paper studied the attaining methods and applied the method of extracting information from parameters closely related to the crystallization state through data mining, which is based on the rough set theory. However, the prediction information drew from historic data and expert’s experience is unable to solve the entire prediction problem in the online measurement of crystallization state. In order to enhance the expert system’s online prediction ability, the paper constructed the prediction model of crystallization state based on support vector machine and particle swarm optimization algorithm using the knowledge from expert system as learning example, and studied the construction method and parameter optimization algorithm of the model, and thus improved the accuracy and generalization ability of prediction model.Finally, this paper realized an expert system of prediction of crystallization state based on VC++6.0, and tested the expert system on the self-developed intelligent monitoring platform in sugar boiling process. The results showed that the forecasting accurate rate of the crystallization state of expert system reached to96.0%along with the online updating and improvement of knowledge base of expert system. Thus the expert system can be used to predict the crystallization state in sugar boiling process.