Particle swarm optimization-neural network algorithm and its application in the genericarameter of microstrip line

Abstract

To solve the general model for S-parameter of microstrip line quickly, this paper proposes Particle Swarm Optimization-Neural Network (PSO-NN) algorithm, which is based on the research of Particle Swarm Optimization (PSO) algorithm and neural network algorithm. By testing and analyzing PSO-NN, PSO and BP neural network algorithm respectively with the performance check function, we find PSO-NN the best performance. Finally, PSO-NN algorithm is applied to the general model for S-parameter of microstrip line which has made use of CST software to get the training data and validation data of the S-parameter of microstrip line. By training and validating PSO-NN, PSO and BP neural network algorithm, we prove that PSO-NN algorithm has the minimum average error and standard deviation in acceptable time. Compared with CST software, the PSO-NN algorithm has shorter simulation time at the same precision level .Therefore, this paper is of great value to the research of PCB board. © 2013 Springer-Verlag.