CMAC Neural Network Model Based on Compose Particle Swarm Optimization

In order to improve the training precision of traditional CMAC model, this paper suggests a new CMAC model, whose weights are trained by composite particle swarm optimization. Traditional model´s weights are trained by LMS algorithm, which can´t learn approaching function´s reciprocal and unfitted nonlinear hyperplane. The new method makes full use of the disadvantages of swarm intelligence, and improves above disadvantages effectively.