BEARING FAULT DIAGNOSIS METHOD FOR DYNAMIC JOINT DISTRIBUTION ALIGNMENT NETWORK UNDER VARIABLE WORKING CONDITIONS

摘要

A bearing fault diagnosis method for a dynamic joint distribution alignment network under variable working conditions, comprising the following steps: collecting bearing vibration data under different working conditions to obtain a source domain sample and a target domain sample (S1); constructing a deep convolutional neural network model for dynamic joint distribution alignment (S2); simultaneously feeding the source domain sample and the target domain sample into a parameter-initialized deep convolutional neural network model, so that a feature extractor extracts high-level features of the source domain sample and the target domain sample, and calculates a marginal distribution distance and a conditional distribution distance (S3); obtaining a joint distribution distance according to the marginal distribution distance and the conditional distribution distance, and combining the joint distribution distance with label loss to obtain an objective function (S4); optimizing the objective function by using a stochastic gradient descent method, and training the deep convolutional neural network model to obtain an optimized deep convolutional neural network model (S5); and inputting the target domain sample into the optimized deep convolutional neural network model to obtain a predicted label of a target domain, and comparing the predicted label of the target domain with a true label of the target domain to obtain diagnosis accuracy (S6). The method can reduce the influence of domain drift, so that a deep learning model can well complete a fault diagnosis task under variable working conditions; moreover, the speed is high, and the calculation amount is small.