基于多分辨率分析的模糊系统结构辨识算法

Structure Identification in Fuzzy System Based on Multi-resolution Analysis

分析了尺度函数多分辨率逼近性以及与T-S模糊系统结构的相似性,提出基于多分辨率分析的模糊系统(MAFS)。在此基础上,利用B-样条尺度函数作为模糊隶属函数具有紧支撑性、多分辨率逼近的特性,从时-频域局域化角度,采用WTMM技术,给出了模糊系统结构辨识算法。仿真结果验证了这种模糊系统在不影响辨识精度情况下,模糊隶属函数具有模糊语义的完备性和可解释性,同时表明该结构辨识算法的正确性和有效性。

The multi-resolution approximation performance of scale function and its similarity with T-S fuzzy system are analyzed. Then, the T-S fuzzy system based on multi-resolution analysis (MAFS) is constructed. According to the compactness and multi-resolution characteristics of B-spline scale function as fuzzy membership, a novel fuzzy structure identification algorithm is proposed by adopting WTMM technology. Simulations are given to demonstrate the completeness and interpretability of fuzzy variables without decreasing approximation performance and the effectiveness of the novel structure identification algorithm.