三电平逆变器开路故障诊断研究

Research on open-circuit fault diagnosis of three-level inverter

针对传统三电平逆变器开路故障诊断方法存在计算复杂、准确率低等问题,提出了一种基于小波分析与粒子群优化支持向量机的三电平逆变器开路故障诊断方法(WT-PSO-SVM)。在分析三电平逆变器三相电流信号特征的基础上,利用三层小波对电流信号进行分解,提取各频带能量作为故障特征;小波变换提取到故障能量特征后,部分故障下所提取的能量十分接近,无法有效区分,进而引入正半周比例系数作为辅助特征;将归一化的能量和正半周比例系数作为特征向量输入支持向量机进行分类训练,同时利用粒子群算法优化支持向量机的参数以达到最好的分类效果,最终实现故障诊断。实验结果表明:WT-PSO-SVM方法可以有效识别三电平逆变器的开路故障,较其他故障诊断方法有更高的诊断精度和速度,在变负载和有噪声影响情况下仍有较高的故障识别准确率,准确率达到97.3%。

In view of problems of complicated calculation and low accuracy existed in traditional open-circuit fault diagnosis methods of three-level inverter,an open-circuit fault diagnosis method of three-level inverter based on wavelet analysis and particle swarm optimization support vector machine(WT-PSO-SVM)was proposed.On the basis of analyzing the characteristics of the three-phase current signal of the three-level inverter,the current signal is decomposed by using the three-layer wavelet,and the energy of each frequency band is extracted as the fault feature.After the energy was extracted by wavelet transform,the extracted energy under partial faults is very close and cannot be distinguished effectively,and then the positive half-cycle proportional coefficient is introduced as auxiliary feature.The normalized energy and the positive half-cycle proportional coefficient are used as feature vectors to input support vector machines for classification training,and the parameters of support vector machine are optimized by particle swarm optimization algorithm to achieve the best classification effect,so as to realize fault diagnosis.The experimental results show that the WT-PSO-SVM method can effectively identify open-circuit faults of the three-level inverter,which has higher diagnostic accuracy and speed than other fault diagnosis methods,and still has a higher fault identification accuracy of 97.3%in the case of variable load and noise.