基于改进 PSO 优化 SVM 网络的数控机床故障分类器设计

摘要

为实现小样本下数控机床故障分类器的设计，通过分析故障分类器构造的基本原理，采用支持向量机（SVM）的神经网络实现，然后用改进的粒子群算法（PSO）对 SVM 的参数进行优化，改进的 PSO 算法主要采取了团体互助优化策略，系统分别采用了训练样本和测试样本，并用 BP 神经网络算法和 PSOSVM 神经网络算法进行测试，通过对比测试说明改进 PSO 算法的优越性。%For small sample of the design of the NC machine fault classifier,the basic principle of fault classifier structure,was illustrated and using the neural network implementation of support vector machine (SVM),and used the improved particle swarm (PSO)algorithm to opti-mize parameters of SVM,the improved PSO algorithm mainly adopted association cooperation optimization strategy,system adopted the training sample and test sample respectively,and the BP neural network algorithm and PSOSVM neural network algorithm,through the con-trast test shows that improved PSO algorithm is good.