基于灰狼算法优化支持向量机的级联H桥五电平逆变器故障诊断

Fault Diagnosis of Cascaded H-Bridge Five-Level Inverter Based on Support Vector Machine Optimized by Grey Wolf Algorithm

摘要: 级联H桥多电平逆变器因其高效的能量转换和良好的输出谐波性能广泛应用于工业领域。逆变器开路故障会显著影响系统输出性能，故障定位的快速准确与否至关重要。本文以级联H桥五电平逆变器为研究对象，提出一种基于灰狼算法(Grey Wolf Optimizer, GWO)优化支持向量机(Support Vector Machines, SVM)的故障诊断方法。首先，利用小波包分解提取逆变器输出电压的能量熵作为特征向量，并通过概率主成分分析(Probabilistic Principal Component Analysis, PPCA)进行降维。接着，采用GWO算法对SVM中的关键参数进行优化，提升故障诊断效果。实验结果表明，所提方法在不同开路故障下均表现出优越的诊断准确率和较快的收敛速度，验证了其在复杂工况下的有效性和可行性。

Abstract: Cascaded H-bridge multilevel inverters are widely used in industrial applications due to their efficient energy conversion and excellent harmonic performance. Open-circuit faults in inverters can significantly degrade system output, making fast and accurate fault localization critical. This paper focuses on the cascaded H-bridge five-level inverter and proposes a fault diagnosis method based on Support Vector Machines (SVM) optimized by the Grey Wolf Optimizer (GWO). First, the energy entropy of the inverter output voltage is extracted as a feature vector using wavelet packet decomposition, followed by dimensionality reduction through Probabilistic Principal Component Analysis (PPCA). Subsequently, the GWO algorithm is employed to optimize the key parameters of the SVM, improving fault diagnosis performance. Experimental results demonstrate that the proposed method achieves superior diagnostic accuracy and faster convergence across various open-circuit fault conditions, confirming its effectiveness and feasibility in complex operating environments.