聚类分析的数据挖掘方法及其在机械传动故障诊断中的应用

采用网格化处理的思想,通过对基于密度的聚类分析方法进行改进,提出了一种新的聚类算法.这种算法通过对齿轮传动系统的故障信号进行测试、对故障类型进行了判定,对不同转速下齿轮传动振动信号进行谱熵计算、并采用网格划分方法将其表示在二维和三维空间分布平面内,可以较好地将正常、裂纹、磨损等类型的故障进行聚类和识别,并通过试验验证表明能够对不同工作状态的齿轮传动信号进行可靠的聚类与区分,聚类率为96%以上.说明该方法对齿轮故障进行区分与诊断是切实可行和有效的.

Analysis of data mining of clustering and its application to mechanical transmission fault diagnosis

For mechanical fault in the gear transmission system difficult to diagnosis,an improved clustering method was presented based on the DBSCAN(Density-based spatial clustering of application with noise) by using the concept of grid processing.The faults of gear transmission system were detected and the styles of gear faults were estimated by the improved clustering method,which also calculated spectral entropy of vibration signal of gear transmission at different rotate speed,and showed the spectral entropy di...