基于DSmT与小波网络的齿轮箱早期故障融合诊断  

Gearbox incipient fault fusion diagnosis based on DSmT and wavelet neural network

作　　者：陈法法[1] 汤宝平[1] 姚金宝[1] 

机构地区：[1]重庆大学机械传动国家重点实验室,重庆400030

出　　处：《振动与冲击》2013年第9期40-45,共6页Journal of Vibration and Shock

基　　金：国家自然科学基金项目(51275546);重庆市自然科学杰出青年基金(CQ cstc2011jjjq70001)

摘　　要：针对齿轮箱早期故障特征十分微弱难以有效辨识问题,提出基于DSmT理论与小波神经网络的齿轮箱早期故障融合诊断模型。利用多个振动传感器合理布置在齿轮箱的多个关键部位采集多源振动信息并进行特征提取;利用多个并联小波神经网络实现齿轮箱早期故障的初级诊断获得彼此独立的多个证据;利用DSmT理论对多个独立证据进行融合决策得出齿轮箱的最终诊断结论。DSmT理论克服了传统DST证据理论的局限性,小波神经网络实现多源证据信度分配的客观化。诊断实验结果表明,该方法能有效提高齿轮箱早期故障特征的辨识精度、降低诊断的不确定性。Aiming at that gearbox incipient fault signals are usually very weak and their features are difficult to be distinguished, a diagnosis model based on DSmT and wavelet neural network for gearbox incipient faults was proposed. Multiple vibration sensors were reasonably arranged at critical positions of a gearbox to collect multi-source vibration informations for feature extraction. Several shunt-wound wavelet networks were used to carry out the primary fault diagnosis and acquire evidences independent of each other. Then, DSmT was used to combine different independent evidences and get the final decision result. The DSmT overcomes the shortcoming of the traditional DST theory, and the wavelet neural network realizes the objectivity of multi-source evidence belief assignment. The diagnostic tests show that the method can effectively improve the identification accuracy of gearbox incipient fault features and reduce diagnostic uncertainty.

关 键 词：DEZERT-SMARANDACHE理论 信息融合 小波神经网络 齿轮箱 故障诊断 

分 类 号：TH132[机械工程—机械制造及自动化]