基于高斯混合模型的逆变器故障诊断方法研究

Research on Fault Diagnosis Method of Inverter Based on Gaussian Mixture Model

摘要:

逆变器广泛应用于工业生产中的诸多领域。逆变器在工作过程中会出现元器件性能退化或损坏，造成经济损失甚至人员伤亡，为了提高逆变器工作的可靠性，识别出逆变器出现故障时的故障类型，提出了基于高斯混合模型的逆变器故障诊断方法。以谐振型逆变器的为例，分析了逆变器的四种典型故障，提取出逆变器在不同故障下输出电压的时域特征波峰系数和频域特征小波能量熵。使用时频特征数据训练高斯混合模型，并结合EM算法计算输入数据属于各种故障类型的概率，建立逆变器的故障诊断模型。仿真实验通过Simulink建立了并联谐振型逆变电路的模型，模拟出四种典型的故障状态，并用这些数据训练故障诊断模型。仿真结果验证了基于高斯混合模型的故障诊断方法的有效性和准确性。该故障诊断方法具有较高准确率，对于四种类型故障的总的识别率到达93.2%，可以应用于工业现场逆变器的故障诊断及其他领域。

Abstract:

Inverters are widely used in many fields of industrial production. The performance degradation or malfunction of components would occur during running, resulting in economic losses and even casualties. In order to improve the reliability of the inverter and identify the fault type when the inverter fails, a fault diagnosis method based on Gaussian mixture model is proposed. Taking the resonant inverter power supply as an example, several typical faults of the inverter are analyzed, and the time-frequency characteristics of the faults are extracted. Then the Gaussian mixture model is trained using fault characteristics, and EM algorithm is used to predict the output of the test data. The model of the parallel resonant inverter circuit was established in Simulink, and several typical power failure states were simulated. Simulation experiments verify the effectiveness and accuracy of the proposed fault diagnosis method. The fault diagnosis method is easy to implement and the predictive accuracy of the model is as high as 93.2%. It can be applied to the fault diagnosis of industrial inverters and other fields.