基于粒子群优化算法的LS-SVM电动机断条故障诊断  

Failure Diagnosis of Motor Broken Bar LS-SVM Based on Particle Swarm Optimization Algorithm

作　　者：许智颖 许允之[2] 郑辞晏 沈佳[2] 仝年 

机构地区：[1]大连市旅顺口区第57中学,辽宁大连116041 [2]中国矿业大学动力与工程学院,江苏徐州221116

出　　处：《煤矿机电》2016年第6期62-66,共5页Colliery Mechanical & Electrical Technology

摘　　要：支持向量机(Support Vector Machine,SVM)算法是基于统计学习理论的一种新的学习方法,应用于故障诊断技术中,具有训练所需样本少、诊断率高等优点。最小二乘支持向量机(Least Squares Support Vector Machines,LS-SVM)算法是标准支持向量机的一种扩展,能在保证精度的同时大大降低计算机的复杂性,加快求解速度。该算法的超参数对支持向量机的性能有着重要的作用。因此,采用粒子群优化(Particle Swarm Optimization,PSO)算法对LS-SVM算法寻找最优超参数,进一步提高LS-SVM对电动机断条故障诊断的效率和准确率有着重要作用。实验结果表明,综合PSO与LS-SVM两种算法的优点,可有效减少故障诊断中误判、漏判的发生。Support vector machine （SVM） is a new kind of learning method based on statistical learning theory. This method can be applied in fault diagnosis with advantages of requiring fewer training samples and having higher diagnosis rates. Least squares support vector machines （LS-SVM） are an expansion of standard SVM which can truly lower the complexity of computer programs and accelerate the speed of solving mathematical problems without damaging the accuracy. The hyper parameters are of great importance to the performance of SVM. Accordingly, uses particle swarm optimization （PSO） to discover the best hyper parameter in order to enhance the efficiency and accuracy of fault diagnosis in terms of broken bars in motors. Finally, it can learn from the experimental results that it can synthesizes the advantages of both PSO and LS-SVM algorithm. As a result, the occurrence rates can be cut down due to the decrease of wrong judgment and missing judgment.

关 键 词：粒子群优化算法 最小二乘支持向量机 电动机 故障诊断 

分 类 号：TM343.3[电气工程—电机] TP206.3[自动化与计算机技术—检测技术与自动化装置]