An Improved-WM Method Based on Optimization of Centers of Output Fuzzy Subsets for Fuzzy Rules

摘要

The generation of fuzzy rules from samples is significant for fuzzy modelling. To improve the robustness of Wang-Mendel (WM) method, an improved WM method to extract fuzzy rules from all the regularized sample data was proposed. However, the accuracy of the model with this method is degraded for the conflicting rules with small difference between support degrees. And the output subsets can only be chosen from the pre-defined ones. To solve these problems, we develop an improved-WM method based on optimization of centers of output fuzzy subsets for fuzzy rules (COiWM). This method adopts the fuzzy c-means (FCM) clustering algorithm to divide the input and output spaces, and the improved WM method which replaces the original data by regularized data is used to calculate the support degrees. Then the support degrees are used as weights to optimize the centers of output fuzzy subsets with a method of weighted averages, so as to enhance the accuracy of a fuzzy model. Experimental results of a case study on short term daily maximum electric load forecasting prove that our proposed method enhances the accuracy of a fuzzy model.