基于MPSO算法的RBF网络学习算法

RBF Neural Network Learning Algorithm Based on Modified Particle Swarm Optimization

针对标准粒子群优化(PSO)算法存在早熟收敛,易陷入局部极值的缺陷,提出了一种利用混沌优化算法确定PSO算法参数的改进粒子群优化(MPSO)算法。为了提高径向基函数(RBF)神经网络的精度和性能,提出了一种基于改进粒子群优化(MPSO)算法的RBF网络学习算法。RBF网络隐层节点个数用对手受罚的竞争学习(RPCL)算法确定后,基函数的中心矢量、方差和网络权值用MPSO算法在全局空间动态确定。采用Iris分类问题做仿真实验,并与基于标准PSO算法的方法和单纯BP网络训练进行比较。实验结果表明,该算法性能优于所比较的2种算法,并且具有良好的收敛性和模式分类能力。

Coping with the defects of prematurity and easily getting into the local optimization in the standard particle swarm optimization（PSO）,a modified particle swarm optimization（MPSO） algorithm is proposed,in which the chaos optimization algorithm is introduced to optimize parameters.In order to improve the precision and performance of radial basis function（RBF） neural network,A RBF neural network learning algorithm based on MPSO is proposed.After determination of the number of units in RBF hidden layer by using the rival penalized competitive learning（RPCL） algorithm,centers,widths of basis functions and weights of neural network are estimated dynamically in global space with MPSO.The Iris classification problem was introduced to do the simulation experiment,and compared with the standard PSO algorithm and BP algorithm.The experimental results show that,the performance of the proposed algorithm is superior to the other two algorithms with a better convergence and pattern recognition.