基于CS理论的24脉波整流器开路故障诊断方法

Open Circuit Fault Diagnosis Method of 24-Pulse Rectifier Based on Compressed Sensing Theory

针对24脉波整流器晶闸管开路故障待处理数据量大、诊断精度不高和诊断速度慢的缺点,提出一种基于压缩感知(CS)理论对开路故障电压信号的稀疏向量进行特征提取的分类识别方法。利用冗余字典和高斯测量矩阵对原始信号进行稀疏表示和测量,接着用正则化自适应匹配追踪算法对测量信号进行重构,得到稀疏向量;对稀疏向量进行6种特征参数的提取,将其作为BP神经网络的输入,实现对开路故障的诊断识别;选取典型的开路故障类型,进行仿真实验验证。仿真结果表明,传统方法要处理的数据长度为1000,而所提方法要处理的数据长度只有50,以很少的数据量保存了原有信号的特征信息,使得开路故障识别准确率显著提高,诊断速度加快。

In order to overcome the shortcomings of the open circuit thyristor fault of the 24-pulse rectifiers,such as large amount of data to be dealt with,low diagnostic accuracy and slow diagnostic speed,a classification recognition method based on compressed sensing(CS)theory to extract features on sparse vectors of open circuit fault voltage signals is proposed.Firstly,the original signal was sparsely represented and measured by using redundant dictionary and Gaussian measurement matrix,and then the measurement signal was reconstructed by using the regularized adaptive matching pursuit algorithm to obtain the sparse vector.Then,six feature parameters of the sparse vector were extracted and used as the input of BP neural network to realize the diagnosis of open faults.Finally,the typical open circuit fault type was selected and verified by simulation experiments.The simulation results show that the length of the data to be processed by the traditional method is 1000,and the length of the data to be processed by the proposed method is only 50.The feature information of the original signal is saved with a small amount of data,which makes the accuracy of open circuit fault recognition significantly improved,and speeds up the diagnosis.