BFA优化SVM的刀具故障诊断

Tool Fault Diagnosis Based on BFA Optimized SVM

支持向量机(SVM)对机床刀具磨损阶段监测的诊断能力与其参数惩罚因子C和核函数参数g紧密相关,SVM参数的优化对其诊断精度影响很大。为解决人工选取支持向量机参数效率低、准确率不高的问题,提出一种细菌觅食算法(BFA)优化SVM参数的刀具故障诊断方法。将SVM的诊断准确率作为细菌觅食算法的优化目标,利用细菌觅食算法对SVM参数全局寻优,得到最优参数组合。实验结果分析表明,相对于传统的SVM,优化参数后的SVM对刀具磨损阶段的监测准确率至少提高了5%,验证了此方法的可靠性。

The diagnostic ability of support vector machine(SVM)for machine tool wear stage monitoring is closely related to its parameter penalty factor C and kernel function parameter g. The optimization of SVM parameters has a great influence on its diagnostic accuracy. In order to solve the problem of low efficiency and low accuracy of manual selection of support vector machine parameters,a tool fault diagnosis method based on bacterial foraging algorithm(BFA)to optimize SVM parameters was proposed. The diagnostic accuracy of SVM was used as the optimization target of the bacterial foraging algorithm. The bacterial foraging algorithm was used to optimize the SVM parameters globally,and the optimal parameter combination was obtained. Compared with the traditional SVM,the experimental results show that the SVM with optimized parameters improve the monitoring accuracy of the tool wear stage by at least 5%,which verified the reliability of this method.