针对神经网络的一些缺陷,研究神经网络基于粒子群优化的学习算法,将粒子群优化算法用于RBF神经网络的学习训练。提出了一种基于粒子群优化(PSO)算法的径向基(RBF)网络参数优化算法,首先利用减聚类算法确定网络径向基函数中心的个数,再用PSO算法优化径向基函数的中心及宽度,最后用PSO算法训练隐含层到输出层的网络权值,找到神经网络权值的最优解,以达到优化神经网络学习的目的。最后,通过一个实验与最小二乘法优化的神经网络进行了比较,验证了算法的有效性。
A training algorithm for neural network based on particle swarm optimization was investigated. Introduced a parameter optimization method of radial basis function(RBF) neural network algorithm based on particle swarm optimization(PSO) algorithm. First, it used subtractive clustering method to determine unit's number in RBF layer. Second, it optimized central position and directional width used PSO algorithm. Third, it optimized connection weights between the RBF layer and the output layer used PSO algorithm. Through the comparison of least square method, the result shows that it is effective.