An entropy-based fuzzy membership partition method used in operator functional state prediction

In this paper, an entropy-based adaptive fuzzy membership partition method is proposed. The method is based on the definition of entropy with an expectation of balancing the total entropy of the training data under certain partition setting. Without any prior knowledge of the data, the method can adaptively find out how many partitions are suitable for each variable. Firstly, the method is tested in the Mackey-Glass time series and shows good performance. Secondly, it is adopted in a fuzzy model which is constructed by using Wang-Mendel method for operator functional state prediction. The prediction result shows the proposed method is quite useful and can be used in the future fuzzy system construction work.