基于故障演化分析的齿轮故障预测特征选择

1)南京理工大学机械工程学院,南京 210094; 2)北京特种机电技术研究所,北京 100012; 3)武警警官学院电子技术系,成都 610213

机械动力学与振动; 齿轮; 特征选择; 故障预测; 故障演化分析; 跟踪能力; 小波熵

A feature selection for fault prognosis based on fault evolution analysis
Fu Pengcheng1, 2, Wang Changming1, Duan Junbin2, and Tan Xiaodong3

Fu Pengcheng1, 2, Wang Changming1, Duan Junbin2, and Tan Xiaodong31)School of Mechanical Engineering, Nanjing University of Science and Technology, Nanjing 210094, P.R.China2)Beijing Special Electromechanical Technology Research Institute, Beijing 100012, P.R.China3)Department of Electronic Technology, Officers College of CAPF, Chengdu 610213, P.R.China

mechanical dynamics and vibration; gear; feature selection; fault prognosis; fault evolution analysis; track ability; wavelet entropy

DOI: 10.3724/SP.J.1249.2015.04434

备注

提出基于故障演化分析的齿轮故障预测特征选择方法.根据机械动力学与振动模型,计算不同故障严重程度下系统的输出响应.使用特征选择方法建立不同特征描述的齿轮故障演化曲线,计算各特征对齿轮故障演化过程的跟踪能力,选择跟踪能力最大的特征作为故障预测特征.以一阶齿轮传动系统的齿轮裂纹故障为例,论证结果表明,小波熵对齿轮裂纹故障演化过程的跟踪能力最大,采用小波熵作为齿轮裂纹故障的预测特征能有效提高故障预测的精度.

In order to improve the reliability and the effectiveness of gear fault prognosis, we propose a feature selection based on fault evolution analysis. Firstly, we build a mechanical dynamics and vibration model, which accordingly generates dynamic responses with tooth damage growth. Secondly, we obtain the trend curves of gear fault evolution for several features, and calculate the track ability for gear fault evolution of each feature. The feature with the strongest track ability is chosen as the feature for gear fault prognosis. Finally, we use a gear crack fault in a one-stage gearbox to verify the performance of the proposed method. Results show that the wavelet energy entropy with the optimal track ability for gear crack evolution among several selected features is the most suitable feature for gear crack fault prognosis.

·