摘要
研究六相永磁容错电机的故障诊断问题,有利于飞行器电力作动系统稳定运行。由于永磁容错电机属于强耦合高阶非线性系统,故障信号样本时间短且具有时变性,导致永磁容错电机故障诊断困难。传统的机器学习理论方法易陷入局部极值点,准确性较低。为提高永磁容错电机的故障准确率,采用支持向量机联合小波分析算法进行故障诊断。首先采集电机故障状态下的输出转矩信号,进行小波分解提取能量特征向量,采用最小二乘支持向量机构造诊断模型,克服局部最优问题,实现准确判别电机故障。对比实验表明,提出故障诊断方法明显优于传统支持向量机、BP神经网络模型方法,诊断准确率达到90%以上。
This paper studies fault diagnosis of six-phase permanent magnet fault-tolerant motor, which is condu- cive to the stable operation of aircraft electric power system. The six-phase PMSM belongs to the strong coupling and high order nonlinear system, and the fault signal sample time is short, so the fault diagnosis of six-phase PMSM is difficult. The accuracy of the traditional theory of machine learning method is low. In order to improve the accuracy, this paper combines wavelet analysis algorithm and SVM ( support vector machine) for fault diagnosis. We collecte the torque samples of fault tolerant permanent magnet motor, and then the torque signals are decomposed by wavelet. Moreover, the diagnosis model is built by the LS-SVM (least square support vector machine), to overcome the local optimal problem and realize accurate diagnosis. Simulation results show that the proposed method is superior to tradi- tional support vector machine and BP neural network model of diagnostic methods. The diagnostic accuracy is more than 90%.
出处
《计算机仿真》
CSCD
北大核心
2015年第9期109-113,共5页
Computer Simulation
基金
陕西省自然科学基金项目(2012JM8016)