基于MOMEDA与BiLSTM的滚动轴承微弱故障识别方法

Weak Fault Identification Method of Rolling Bearing Based on MOMEDA and BiLSTM

针对传统的滚动轴承智能诊断模型计算效率低和准确率欠佳问题,课题组提出一种基于多点最优最小熵解卷积(multipoint optimal minimum entropy deconvolution adjusted,MOMEDA)和双向长短时记忆(bidirectional long short-term memory network,BiLSTM)网络相结合的滚动轴承故障诊断模型。该模型利用MOMEDA方法增强故障特征,并结合遗传算法(genetic algorithm,GA)对BiLSTM模型参数进行优化,实现滚动轴承智能、高效及鲁棒性诊断。利用该模型对经典轴承数据集以及牵引电机轴承故障数据集进行验证,平均准确率达到了99.63%,分别比传统卷积神经网络(convolutional neural network,CNN)、单层长短时记忆网络(long short-term memory network,LSTM)、双向长短时记忆网络和最新的CNN-LSTM模型高16.02%,9.98%,7.01%和5.65%,验证了该模型的有效性和优越性。

Aiming at the problems of low computational efficiency and poor accuracy of the traditional rolling bearing intelligent diagnosis model,a rlling bearing fault diagnosis model based on multipoint optimal minimum entropy deconvolution adjusted(MOMEDA)and bidirectional long and short term memory(BiSTM)network was proposed.The MOMEDA method was used in model to enhance the fault characteristics and the genetic algorithm(GA)was combined to optimize the BiLSTM model parmeters to achieve itelligent,efcient and robust diagnosis of rlling bearings.The model was validated on the classical bearing dataset and the traction motor bearing fault dataset with an average accuracy of 99.63%,which was 16.02%,9.98%,7.01%and 5.65%higher than the conventional convolutional neural network(CNN),single-layer long short-term memory network(LSTM),bi-directional long short-term memory network(LSTM)and the latest CNN-LSTM mod;l,respectively,verifying the efectivenes and superiority of the model.