基于相空间重构和支持向量机的三电平逆变器故障诊断技术

针对三电平逆变器交叉两桥臂的两只功率管同时开路故障(非典型故障)诊断问题,提出一种基于相空间重构和支持向量机(SVM)的故障诊断方法。该方法以三相电流为检测信号,为降低特征向量的维数,对三相电流进行了Park变换,然后采用相空间重构技术,对d、q轴电流分别进行重构,得到不同形状、大小和方向的电流轨迹图形,借助图像处理技术从中提取出故障特征向量,将其作为学习样本,在SVM中训练,使分类器能够建立不同特征向量和故障类型的映射关系,实现对二极管中点箝位型(NPC)三电平逆变器的故障诊断。仿真结果表明,该方法能够准确地定位故障元,诊断精度高。

Fault Diagnosis Technology for Three-level Inverter Based on Reconstructive Phase Space and SVM

For the problem of open-circuit fault arising from two power devices on two cross bridge arms(atypical faults) in the neutral point clamped(NPC) three-level inverter, a new method of fault diagnosis is proposed based on reco-nstructive phase space and support vector machine(SVM). In this method, three-phase current is taken as measurement to classify the fault modes. Park transform is used for three-phase current to reduce the dimension of feature vectors. Then, the reconstructive phase space(RPS) method is adopted to reconstruct d-axis and q-axis current, thus the reconstructed current trajectories of different shape, size and direction are obtained. And the fault feature vectors are extracted with the help of image processing technology. Taking the feature parameters in different fault conditions as learning samples, the SVM is trained, so that the mapping relations between different feature vectors and fault types can be built for fault diagnosis of the NPC three-level inverter by the classifier. The simulation results show that the method can locate faults accurately and has high precision.