Rolling bearing transfer learning fault diagnosis method based on partial domain adversarial

Abstract

The invention discloses a rolling bearing transfer learning fault diagnosis method based on partial domain adversarial, and the method comprises the following steps: building a rolling bearing fault data sample library, and dividing source domain and target domain fault data; extracting implicit features of the fault data of the source domain and the target domain; constructing a label predictor;constructing a weighted domain classifier to obtain the probability and weight of sample features from source domain distribution; sending the weighted source domain sample features and the unweightedtarget domain sample features into another domain classifier, judging whether the sample features are from the source domain or the target domain, and constructing a gradient inversion layer; optimizing the model; inputting the test data into a feature extractor to obtain sample features, inputting the obtained sample features into a label predictor to obtain predicted labels, and calculating classification precision. According to the method, the adversarial thought is fused into a partial migration network, the strategy that domain classification is carried out after source domain sample weighting is provided, the sample domain self-adaptive capacity is improved, and the problem of carrying out unsupervised label prediction in a target domain is solved.

Description

technical field [0001] The invention relates to the field of bearing fault diagnosis, in particular to a rolling bearing migration learning fault diagnosis method based on partial domain confrontation. Background technique [0002] Rolling bearings are high-end components in various mechanized fields, such as high-speed rail, wind power, aviation, etc., with more precise transmission performance, smaller size and lighter weight. However, the maintenance of rolling bearings will also generate more costs. This is mainly because the working environment of rolling bearings is generally very harsh, especially in some important areas, rolling bearings are easily damaged. If the problematic bearings are not dealt with in time, many accidents will occur, ranging from stoppage of operation to serious safety accidents, resulting in unpredictable consequences. Therefore, a more accurate diagnosis of rolling bearing faults is of great significance. [0003] Domain adaptation theory m...

AI Technical Summary

Problems solved by technology

However, the target domain samples are generally unlabeled, and it is impossible to train a suitable weighted model through simple deep learning methods

Images