基于优化算法的风机振动故障模拟与诊断

摘要

In order to solve the problems of the vibration fault data shortage and low accuracy of fault diagnosis rate in traditional algorithm for direct drive permanent magnet synchronous wind power generator ( DDPMG) vibration fault, according to the trait of traditional synchronous generator vibration characteristic is similar to DDPMG,adopts synchronous generator to simulate the vibration faults of DDPMG, acquires the vibration data of simulating DDPMG under the different operation conditions, using genetic algorithm (GA) hierarchically to optimize the parameters of least square support vector machine (LS-SVM)makes fault diagnosis. Compare with the no optimization LS-SVM method, the diagnosis result shows that:the accuracy of fault diagnosis by the optimized algorithm is much higher,it can provide reference for later DDPMG vibration fault monitoring and diagnosis technology research.%为解决直驱永磁同步风力发电机（Direct Drive Permanent Magnet synchronous wind power Generator，DDPMG）振动故障数据匮乏和传统算法故障诊断准确率低的问题，根据传统同步发电机振动特性与DDPMG相似的特点，采用同步发电机对DDPMG进行振动故障模拟，获取DDPMG模拟振动在不同运行工况下的数据；并利用遗传算法（Genetic Algorithm，GA）分层优化最小二乘支持向量机（Least square Support Vector Machine，LS-SVM）参数的算法进行故障诊断。与未优化的LS-SVM诊断对比结果表明：采用优化算法的故障诊断准确率更高，可为后续DDPMG振动监测与故障诊断技术研究提供参考。