A novel fuzzy neural network for the control of complex systems
Jian-Nan Lin, Shin‐Min Song · 2002
Various fuzzy neural networks (FNNs) were proposed to enhance the performance of neural networks (NNs) for complex, uncertain systems with highly nonlinearities. In this paper, we propose a novel FNN with the following features: a simple structure of three layers with different types of fuzzy neurons; a straightforward method for generating suitable FNN rule base connection structure; a simple learning algorithm and a method for obtaining a good guess of the initial weights of the proposed FNN. The design of the fuzzy neurons and network structure of this FNN are presented. This FNN is then evaluated by a simulation study of inverse kinematics of a two degrees of freedom manipulator. It is shown that the proposed FNN outperforms conventional feedforward multilayer neural networks and is simpler than existing FNN proposed by Lin and Lee (1991). The potential applications of this FNN are discussed.>