Improvement And Research Of Fuzzy Model Algorithm

The T-S fuzzy model is linear rule consequent, which is used abroad in math models application. Fuzzy model identification is the chief means of fuzzy modeling. Optimizing structure of fuzzy systems is the key to fuzzy model identification. Among the methods of fuzzy model identification, fuzzy clustering algorithm is better than others with many samples, which will improve the efficiency of fuzzy identification. In order to optimize the fuzzy model structure, the number of rules is the key. Therefore the paper introduces the following issues with above questions:Firstly, In order to improve the efficiency of fuzzy identification, firstly research the method that fuzzy clustering algorithm based on objective function. FCPM find the cluster prototypes effectively, collaborative fuzzy clustering make the membership of the entity more exact. An improved fuzzy clustering algorithm is proposed based on the combination of FCPM and Collaborative Fuzzy Clustering. CFCPM clustering algorithm can be obtain which is deduced from FCPM and Collaborative Fuzzy Clustering. The membership and cluster prototypes are improved. The cluster effect of the data set is better. The experimental results obtained on the wine set show the effectiveness of the proposed method.Secondly, Based on CFCPM, a new approach to build fuzzy model is proposed. The approach is composed of two phases: the first one is to remove redundant information by feature selection approach using feature similarity. The second one is to identify the initial fuzzy system using the collaborative fuzzy clustering algorithm. The antecedent and consequent parameters of fuzzy model can be optimized. The collaborative fuzzy clustering is applied to extracted features to improve the parameters and efficiency of the fuzzy model. The results of experiments show the effectiveness of the proposed method for fuzzy modeling.Thirdly, Based on fuzzy modeling with collaborative G-K fuzzy clustering algorithm, in order to optimize the fuzzy model structure, method of T-S fuzzy model identification with growing and pruning rules is proposed. Based on the maximum absolute error (MAE) index, the fuzzy rules are extracted from real-time data. Prune the rule by the impact degree of one local model. The entire algorithm presents a completely online identification of the T-S model and gains a structural and parameter adaptability. The optimized fuzzy model structure can be gained as the accuracy of the T-S model is guaranteed.