基于SVD-ITD和支持向量机的潜水磨碎泵故障诊断

Fault diagnosis of submersible grinding pump based on SVD-ITD and support vector machine

针对潜水磨碎泵流道堵塞的故障诊断,提出了一种基于奇异值分解(SVD)和固有时间尺度分解(ITD)的电流信号分析和构建支持向量机故障识别模型相结合的方法。搭建了潜水磨碎泵堵塞实验台,采集了潜水磨碎泵在不同堵塞程度下的电流信号,利用SVD滤除电流工频信号,采用ITD分解得到各固有旋转分量(PCRs),再将PCRs的均方根值(RMS)与相对应的堵塞程度标签组成数据特征集,最后输入到支持向量机确定故障状态。实验结果表明提出的方法能够准确地识别出四种不同堵塞等级,准确率高达92.3%。因此所提出的方法在少量样本情况下对堵塞程度有较高的识别能力,具有很高的实际应用价值。

For the fault diagnosis of submerged grinding pump flow channel clogging,a method combining current signal analysis based on singular value decomposition(SVD)and inherent time scale decomposition(ITD)and building a support vector machine fault recognition model was proposed.A submersible grinding pump clogging test bench was built.The current signals of the submersible grinding pump under different clogging degrees were collected.The current power frequency signal was filtered using SVD.The inherent rotation components(PCRs)were obtained by ITD decomposition.The root mean square value(RMS)and the corresponding clogging degree label form a data feature set,and finally input the data set into the support vector machine to determine the fault state.Experimental results show that the proposed method can accurately identify four different clogging levels with an accuracy rate of up to 92.3%.Therefore,the method proposed in herein has a high ability to identify the degree of clogging in the case of a small number of samples and has a high practical application value.