粒子群优化BP神经网络的曲线光顺算法  

Curve Smoothing Algorithm for Particle Swarm Optimization BP Neural Network

作　　者：吴易泽 张旭[1] 江明阳 陈勇 WU Yi-ze;ZHANG Xu;JIANG Ming-yang;CHEN Yong(School of Mechanical and Automotive Engineering,Shanghai University of Engineering Science)

机构地区：[1]上海工程技术大学机械与汽车工程学院

出　　处：《化工自动化及仪表》2018年第12期939-942,954,共5页Control and Instruments in Chemical Industry

基　　金：国家自然科学基金项目(51205246)

摘　　要：针对曲线光顺问题,提出一种基于粒子群算法和BP神经网络两者相结合的曲线光顺算法。算法首先利用粒子群算法优化BP神经网络初始的权值和阈值,得到最优解;然后利用最优解进行BP神经网络模型的训练;最终得到光顺后的曲线。试验表明:使用笔者算法得到的曲线的光顺效果较好,在与EMD法和BP法进行比较时,该算法也有很好的优势。Aiming at the curve smoothing,a curve smoothing algorithm based on particle swarm optimization and BP neural network was proposed.Firstly,having the particle swarm optimization algorithm used to optimize both initial weights and thresholds of BP neural network so as to obtain an optimal solution;then having the optimal solution adopted to train BP neural network model and to get a smooth curve at last.The experiment results show that,the smoothing effect of the curve faired by this approach is better and it has a good advantage even compared to EMD and BP.

关 键 词：粒子群算法 BP神经网络 曲线光顺 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]