模糊神经网络系统的隶属函数初始化算法研究

Study of the Algorithm of Initializing Membership Functions in Fuzzy Neural Network System

给出一种新颖的模糊神经网络系统的隶属函数初始化算法,并分析了算法的收敛问题,与文[1]及常规自组织映照算法相比,不但每步的计算量大大减小,得到的初始化结果也更优。模拟仿真结果是令人满意的。

A novel algorithm for initializing membership functions in fuzzy neural network system is proposed and the convergence of the algorithm is analysed. Comparing with normal self-organizing mapping algorithm ,its computing cost of each step is much less and the initialized result is better. The simulation result is satisfactory.