摘要
为实现异步电机在运行状态下快速准确诊断故障类型,构建了优化的最小二乘支持向量机(least squares support vector machine,简称LSSVM)智能故障诊断模型。首先,LSSVM将不等式约束改为等式约束,收敛速度更快;其次,改进粒子群优化算法(particle swarm optimization algorithm,简称PSO)在迭代过程设置1个粒子变异过程,随机放置的粒子可带领种群摆脱局部最优的束缚,找到全局最优点;最后,采用智能诊断模型对西储大学轴承数据及异步电机等7类故障诊断实验数据进行诊断。结果表明:LSSVM的诊断时间仅不到传统支持向量机(support vector machine,简称SVM)的30%,对轴承和异步电机的诊断精度分别为100%和94.3%,相较于传统SVM,LSSVM具有更快的收敛速度和更高的诊断精度。
To enable the rapid and accurate diagnosis of fault types in asynchronous motors during operation,an optimized intelligent fault diagnosis model based on the least squares support vector machine(LSSVM)is constructed.First,by replacing inequality constraints with equality constraints,LSSVM achieves a faster convergence speed.Second,an improved particle swarm optimization algorithm is proposed,which incorporates a particle mutation process during iteration.This process allows randomly placed particles to guide the population in escaping local optima and locating the global optimum.Finally,the intelligent diagnosis model is applied to diagnose both bearing data from Case Western Reserve University and experimental data from 7 types of faults in asynchronous motors.The results show that the diagnosis time of LSSVM is less than 30%of that required by traditional support vector machine(SVM),the diagnostic accuracy for bearings and asynchronous motors reaches 100%,and 94.3%respectively.Compared with the traditional SVM,LSSVM demonstrates faster convergence speed and higher diagnostic accuracy.
作者
王保建
张小丽
王延启
尹昱东
WANG Baojian;ZHANG Xiaoli;WANG Yanqi;YIN Yudong(National Experimental Teaching Demonstration Center of Mechanical Foundation,Xi'an Jiaotong University Xi'an,710049,China;Key Laboratory of Road Construction Technology and Equipment of Ministry of Education,Chang'an University Xi'an,710064,China)
出处
《振动.测试与诊断》
北大核心
2025年第6期1247-1253,1281,共8页
Journal of Vibration,Measurement & Diagnosis
基金
国家自然科学基金重大研究计划集成资助项目(92060302)。
关键词
异步电机
最小二乘支持向量机
粒子群优化算法
故障诊断
asynchronous motors
least squares support vector machine
particle swarm optimization algorithm
fault diagnosis
