A New ANFIS Model based on Multi-Input Hamacher T-norm andSubtract Clustering

This paper proposed a novel adaptive neuro-fuzzy inference system (ANFIS), which combines subtract cluster- ing, employs adaptive Hamacher T-norm and improves the prediction ability of ANFIS. The expression of multi-input Hamacher T-norm and its relative feather has been originally given, which supports the operation of the proposed system. Empirical study has testified that the proposed model overweighs early work in the aspect of benchmark Box-Jenkins dataset and may provide a practical way to measure the importance of each rule.