一种基于离散粒子群的自适应径向基网络模型

提出了一种基于离散粒子群的自适应RBF网络模型。即融合两个二进制编码粒子群,通过对最近邻聚类中心选择法的改进,RBF网络模型能自适应地优化隐节点数、中心向量与宽度,且在保证网络性能的前提下,使网络的结构相对简单(较少的隐层节点数)。同时为进一步提高网络性能,采用梯度下降法与递推最小二乘法混合学习策略,分别对基函数参数(中心与宽度)和输出层线性权值进行学习。仿真实验证明了该方法模型的有效性。

Adaptive radial basis function neural network model based on discrete particle swarm optimization algorithms

A adaptive radial basis function(RBF) neural network model based on discrete particle swarm optimization(PSO) algorithms is proposed.The RBF network model can adaptively optimize hidden nodes,the center vector and the spread parameter of basis function via fusing two binary coded particles and improving the nearest-neighbor clustering center selection method.The model can also make the structure of the network be relatively simple(fewer hidden nodes) under the premise of the guarantee network performance.Furthermore,the gradient descent and recursion least squares are used to optimize the parameters of the basis function and the linear weight of the output layer,respectively.The simulation experiment proves the validity of the proposed model.