基于DPSO优化支持向量机的水轮机组振动故障诊断  

Hydraulic Generating Vibration Faults Diagnosis by Support Vector Machine Based on Particle Swarm Optimization

作　　者：张欣伟[1] 陈文献[2] 张锋利 

机构地区：[1]西安理工大学水利水电学院,陕西西安710048 [2]安康供电局,陕西安康725000 [3]陕西地方电力设计有限公司,陕西西安710075

出　　处：《西安理工大学学报》2013年第2期172-175,共4页Journal of Xi'an University of Technology

基　　金：国家自然科学基金资助项目(51279161)

摘　　要：针对PSO算法在寻找支持向量机最优参数的后期容易陷入局部最佳,影响支持向量机在水轮机振动故障诊断中的正确率这一问题,选用动态粒子群算法(DPSO)来优化支持向量机,将水轮机组的故障特征向量输入优化后的支持向量机进行故障类型诊断。仿真结果表明,DPSO优化的支持向量机有较好的分类识别准确率和较高的诊断精度,可以寻找到全局最优解。According to the basic PSO algorithm, searching for optimum parameters of support vector ma- chine in the late stage is easy to fall into local optimum, and further affects support vector machine in hy- draulic turbine vibration fault diagnosis correct rate. With an aim at this problem, the dynamic particle swarm algorithm （DPSO） is selected to optimize the support vector machine. The hydraulic turbine fault feature vector is input into the optimized support vector machine fault diagnosis. The simulation results show that DPSO optimized SVM can find the global optimal solution, thereby having good classification accuracy. In the hydraulic turbine vibration fault diagnosis compared to the traditional PSO optimized support vector machine has higher diagnostic accuracy.

关 键 词：水轮机 振动故障诊断 动态粒子群算法 支持向量机 

分 类 号：TK738[交通运输工程—轮机工程]