Algorithm Research of RBF Neural Network Based on Improved PSO

In view of the defect of particle swarm optimization(PSO) which easily gets into partial extremum, the paper puts out an improved particle swarm optimization(IPSO), and applies the algorithm to the selecting of parameter of RBF neural network pit function. The algorithm searches the parameter vector which has the best fitness in the whole space, according to coding mode, iterative formula, fitness function which are put out by the paper. The experiment proves that RBF neural network based on IPSO has faster convergent speed and higher precision.