Fuzzy system with adaptive rulebase

A fuzzy model is constructed out of recorded data in which an adaptive rulebase is introduced into fuzzy system modeling. Usually, the rulebase of a fuzzy system is generated and optimized offline once and for all. However, due to uncertainty and change of state of the system, this rulebase may not work well when the model is put into use. Hence, a certain length of recorded data from the current time instance is employed to extract rules and the rulebase is renewed by these rules. In the proposed method, the structure of the fuzzy model is determined in advance, then recorded data are used to extract fuzzy rules and construct a fuzzy model. In the application of the model, the rulebase is renewed in every computation period and this is a rolling process. This approach can deal with a system with much uncertainty since these newly generated rules reflect the most current state of the system. To demonstrate the effectiveness of the proposed method, it is used to build a model of a furnace and simulation results are satisfactory.