Improved Parameter Identification Method Based on Moving Rate

TL;DR

This paper introduces an improved fuzzy inference and parameter identification method for fuzzy neural networks that reduces computational complexity and enhances effectiveness, demonstrated through precipitation and security prediction tasks.

Contribution

The paper proposes a novel fuzzy inference algorithm based on production terms that simplifies calculations and improves parameter identification in fuzzy neural networks.

Findings

Reduced calculation complexity compared to existing methods

Improved accuracy and effectiveness in parameter identification

Lower learning time and error in practical applications

Abstract

To improve the problem that the parameter identification for fuzzy neural network has many time complexities in calculating, an improved T-S fuzzy inference method and an parameter identification method for fuzzy neural network are proposed. It mainly includes three parts. First, improved fuzzy inference method based on production term for T-S Fuzzy model is explained. Then, compared with existing Sugeno fuzzy inference based on Compositional rules and type-distance fuzzy inference method, the proposed fuzzy inference algorithm has a less amount of complexity in calculating and the calculating process is simple. Next, a parameter identification method for FNN based on production inference is proposed. Finally, the proposed method is applied for the precipitation forecast and security situation prediction. Test results showed that the proposed method significantly improved the effectiveness of identification, reduced the learning order, time complexity and learning error.