Computer Engineering and Applications ›› 2019, Vol. 55 ›› Issue (12): 72-76.DOI: 10.3778/j.issn.1002-8331.1809-0273

Previous Articles     Next Articles

Post-Partial Polynomial RBF Neural Network Algorithm Based on Particle Swarm Optimization

WANG Yanyan1, WANG Hongwei1,2   

  1. 1.Schoolof Electrical Engineering, Xinjiang University, Urumqi 830047, China
    2.School of Control Science and Engineering, Dalian University of Technology, Dalian, Liaoning 116024, China
  • Online:2019-06-15 Published:2019-06-13



  1. 1.新疆大学 电气工程学院,乌鲁木齐 830047
    2.大连理工大学 控制科学与工程学院,辽宁 大连 116024

Abstract: RBF(Radial Basis Function) neural network can be well applied in various fields, the key lies in the selection of network model parameter weight, network center value, base width vector and implicit layer node number. The traditional RBF neural network has the disadvantages of low accuracy, easy to fall into local optimal, slow convergence speed and so on. For these problems, the RBF neural network method is optimized by using particle swarm algorithm, that is, the weight value, network center value, and base width vector value of the RBF neural network containing the latter polynomial are optimized, and the optimal number of implicit nodes is selected. Then the PSOIRBF neural network is proposed. The effectiveness of the proposed algorithm is demonstrated by the simulation of nonlinear controlled objects such as nonlinear models and examples and the analysis of the models.

Key words: post-polynomial RBF neural network, particle swarm optimization, effectiveness

摘要: RBF(径向基函数)神经网络能在各个领域得到了很好的应用,关键在于网络模型参数权值、网络中心值、基宽向量和隐含层节点数的选取。传统的RBF神经网络存在精度不高,容易陷入局部最优,收敛速度慢等缺点。针对这些问题,提出了利用粒子群算法优化后件多项式RBF神经网络方法,即优化含有后件多项式RBF神经网络的权值、网络中心值和基宽向量值,并选取最优的隐含层节点数,进而提出了PSOIRBF(基于粒子群的后件多项式RBF)神经网络。通过对非线性模型和实例等非线性被控对象的仿真研究及对模型的分析,表明了所提出算法的有效性。

关键词: 后件多项式RBF神经网络, 粒子群优化, 有效性