基于符号动力学信息熵与SVM的液压泵故障诊断

发布时间：2018-12-15 07:46

【摘要】：针对泵车液压泵早期故障特征信号微弱、故障特征难以提取的问题,提出了一种基于符号动力学信息熵与支持向量机(support vector machine,简称SVM)的泵车液压泵故障诊断方法。分别模拟了液压泵9种故障状态,测取了各状态下多测点的振动信号样本值。利用时间序列的符号动力学信息熵,计算各振动信号的符号动力学信息熵Hk,确定了各状态下相应的信息熵特征向量。建立了不同状态特征向量训练集,再结合支持向量机对液压泵故障模式进行诊断与识别,测试结果准确率为98.71%。将该方法与改进的BP(back propagation,简称BP)神经网络诊断结果进行了对比,结果表明该方法的识别率更高,诊断时间更短,适用于现场液压泵故障的在线诊断。
[Abstract]:In order to solve the problem that the early fault characteristic signal of hydraulic pump is weak and it is difficult to extract the fault feature, a fault diagnosis method based on symbolic dynamic information entropy and support vector machine (SVM) for hydraulic pump is proposed in this paper. Nine fault states of hydraulic pump were simulated, and the vibration signal samples of multiple measuring points were measured. The symbolic dynamics information entropy of each vibration signal is calculated by using the symbolic dynamics information entropy of the time series, and the corresponding information entropy eigenvector under each state is determined by using the symbol dynamics information entropy of the time series. The training sets of different state characteristic vectors are established, and the fault modes of hydraulic pump are diagnosed and identified with support vector machine. The accuracy of the test results is 98.71. The method is compared with the improved BP (back propagation, (BP) neural network diagnosis results. The results show that the method has higher recognition rate and shorter diagnosis time and is suitable for on-line diagnosis of field hydraulic pump faults.
【作者单位】：武汉科技大学机械自动化学院;
【基金】：国家科技支撑计划资助项目(2012BAF02B01,2011BAF11B01)
【分类号】：TU646

【相似文献】