基于深度学习的齿轮箱故障模糊诊断方法

Faults fuzzy diagnosis method for gearbox based on deep learning

风电机组齿轮箱作为传动系统重要组成部分,其运行状态关乎整个风电机组的健康运行。由于齿轮箱振动信号具有非线性、非平稳等特性,传统时频分析方法分解故障信号和提取故障特征的能力有限。因此,文章提出将深度学习应用于齿轮箱故障诊断中,通过构建一维卷积神经网络模型对齿轮箱不同状态下的特征向量进行高效提取、重构。同时,将模糊理论应用于分类器,构建一个模糊多分类器(FMC)对故障进行识别,提出了以平均隶属度作为故障等级判断标准。实验结果表明,文章所提方法在确保齿轮箱故障诊断高准确率的同时,提升了故障分类的精度。

The gearbox is an important part of the wind turbine transmission system,and its operating state is related to the wind turbine unit healthy operation.Because the gearbox vibration signals have nonlinear and non-stationary characteristics,the traditional time-frequency analysis method has limited ability in decomposing the fault signal and extracting the fault feature.Therefore,the deep learning method was proposed to be applied to the gearbox fault diagnosis.Using a one-dimensional convolutional neural network,the feature vectors in different states of the gearbox were extracted and reconstructed efficiently.At the same time,a fuzzy multi-classifier(FMC)using fuzzy theory is constructed to identify the fault and the average membership degree was used as the fault classification standard.The results showed that the proposed method improves the accuracy of fault classification and ensure the high accuracy of gearbox fault diagnosis.