融合能量熵编码和分类模型的牵引电机故障诊断

Fault Diagnosis of Traction Motor Based on Fusion of Energy Entropy Coding and Classification Model

针对牵引电机故障特征不明显、识别定位困难等问题,提出一种融合能量熵编码与分类模型的故障特征量化诊断方法。结合故障机理特性,对故障严重程度进行建模,用微弱电流信号重构对故障敏感的电磁转矩信号,建立基于经验模态分解能量熵和故障属性知识编码的故障特征矩阵;为消除牵引电机故障样本少、非线性模式识别对精确诊断的影响,提出一种改进的灰狼优化算法(IGWO)对支持向量机分类SVM模型参数进行辨识,通过对多类故障准确识别率寻优实现对牵引电机状态预测。在高速列车牵引系统半实物仿真平台进行优化模型对比试验,通过对故障诊断指标分析可知,能量熵编码与IGWO-SVM融合方案可以很好地识别牵引电机故障。

To address the difficulty in detecting and isolating the obscure symptoms of incipient faults in traction motors,a quantitative fault feature diagnosis method combining the energy entropy coding and the classification model was proposed.Based on the characteristics of fault mechanisms,the severity of faults was modelled,and the electromagnetic torque signal sensitive to the faults was reconstructed using weak current signals.A fault feature matrix was established based on empirical mode decomposition energy entropy coding and fault attribute knowledge encoding.To eliminate the impact of few fault samples and nonlinear pattern recognition on accurate diagnosis of traction motor faults,an improved grey wolf optimization(IGWO)algorithm was proposed to identify the parameters of the support vector machine classification model.The prediction of the status of traction motors was realized by optimizing the accurate recognition rate of multiple types of faults.An optimization model comparison experiment was conducted on the hardware-in-the-loop simulation platform of high-speed train traction system.The analysis of fault diagnosis indicators shows that the fusion scheme of energy entropy coding and IGWO-SVM can effectively identify traction motor faults.