Knowledge Acquisition by Improved Fuzzy ID3 Algorithm and Stability Analysis for Jacket Tank Temperature Control
The extraction of knowledge from operation data is an important theme in an autonomous control system. An efficient method for making a decision tree for classification from data is the fuzzy ID3 algorithm using fuzzy sets. However, the definition of fuzzy sets greatly affects the generation of fuzzy trees. In this paper, we propose a new version of fuzzy ID3 algorithms to generate a fuzzy decision maximizing the expected value of transferred information by applying a random search method for determining the fuzzy set, and by using the improved fuzzy ID3 algorithm an automatic extraction of control knowledge from operational data by skilled operator of a jacket tank process. As a result, a fuzzy controller that can decide output of a control switch from both tank temperature error and differentiation of error is designed. Further, a method of analyzing the stability of the fuzzy control system by the modified fuzzy ID3 algorithm is proposed using the phase planes of tank temperature.