基于深度自编码网络与模糊推理相结合的矿用齿轮箱故障诊断方法

提出一种基于深度自编码网络与模糊推理相结合的矿用齿轮箱故障诊断方法。通过对完整齿轮、裂齿齿轮和缺齿齿轮3种齿轮工作状态的声信号进行小波分析并建立特征数据库,构建深度自编码网络与模糊推理系统相结合的诊断系统,实现了齿轮故障诊断与辨识。实验结果表明:这种基于声信号的故障诊断方法能够有效检测矿用齿轮箱的运行状况;与传统神经网络诊断方法以及奇异值分解诊断方法相比,该诊断方法对故障状态的辨识准确度分别提高了3.8%和8%。与传统基于振动信号的故障诊断方法相比,基于声信号的诊断方法对故障状态的辨识准确度无明显差别。表明深度自编码网络模糊推理系统同样适用于基于振动信号的矿用齿轮箱的故障特征提取与分析。

Fault Diagnosis Method of Mine Gearboxes Based on Deep Self-coding Network and Fuzzy Reasoning

A fault diagnosis method of mine gearbox based on depth self coding network and fuzzy reasoning was proposed. Through analyzing the acoustic signals of intact gear,split gear and missing gear by wavelet and establishing the characteristic databases, the diagnosis system combining depth self coding network with fuzzy reasoning system was constructed to realize the fault diagnosis and identification of the gear.The results show that the fault diagnosis method based on acoustic signal can effectively detect the operation status of mine gearbox.Compared with the traditional neural network diagnosis method and the singular value decomposition (SVD) diagnosis method, the accuracy of fault status identification of the proposed fault diagnosis method is improved by 3.8% and 8% respectively. Compared with the traditional fault diagnosis method based on vibration signal, the accuracy of fault diagnosis based on acoustic signal has no obvious difference. It shows that the depth self coding network fuzzy reasoning system is also applicable to the fault feature extraction and analysis of mine gear box based on vibration signal.