基于EEMD模糊熵和PSO-KELM的NPC三电平逆变器故障诊断

功率开关器件是逆变器的核心部件，但其易发生开路故障，故对其进行故障诊断方法研究很有必要。针对中点钳位型（Neutral Point Clamped,NPC）三电平逆变器功率开关管器件的开路故障，提出一种基于总体经验模态分解（Ensemble Empirical Mode Decomposition,EEMD）模糊熵和粒子群算法（Particle Swarm Optimization,PSO）优化的核函数极限学习机（Kernal Extreme Learning Machine,KELM）的故障诊断方法。首先采样功率开关器件的桥臂输出端的三相电压作为故障信号以区分各种故障类型，然后利用EEMD模糊熵提取故障特征向量，最后将其划分为训练集和测试集送入PSO-KELM中，识别故障类型并输出诊断结果。经MATLAB平台仿真实验得到该方法的故障诊断率超过98%，通过与其他方法的对比实验分析，该方法的有效性与优势得到验证。

Fault Diagnosis of NPC Three-Level Inverter Based on EEMD Fuzzy Entropy and PSO-KELM

Power switching device is the core component of inverter, but it is prone to open circuit fault, so it is necessary to study the fault diagnosis method.Aiming to the open-circuit fault of power electronic devices in Neutral-point Clamped(NPC) three-level inverter,a diagnosis approach based on Ensemble Empirical Mode Decomposition(EEMD) with fuzzy entropy and Kernel Extreme Learning Machine(KELM) with Particle Swarm Optimization(PSO).Firstly,the bridge three-phase voltage of power switch devices are sampled as the characteristic signals to classify different fault types.Afterwards,the feature parameters of the fault diagnosis is extracted by EEMD with fuzzy entropy.Finally,taking the feature parameters in various fault situations as training samples and testing samples,the PSO-KELM algorithm is utilized to identify different kinds of faults and output the consequences of fault diagnosis.Through MATLAB simulation experiments,the capability of the proposed method is more than 98%.Through comparing with other method,the effectiveness and superiority of the proposed method is proved.