BEARING FAULT DIAGNOSIS METHOD BASED ON ADAPTIVE MANIFOLD PROBABILITY DISTRIBUTION

摘要

A bearing fault diagnosis method based on adaptive manifold probability distribution, comprising the following steps: constructing a plurality of migratable domains and migration tasks; then converting a data sample in each migration task into frequency domain data using a Fourier transform, inputting the frequency domain data into a GFK algorithm model, and calculating, using the GFK algorithm model, a manifold feature representation matrix in each migration task associated with a bearing fau according to the manifold feature representation matrix, calculating a cosine distance between the center of a target domain and the center of a source domain in each migration task, and defining an objective function for intra-domain classifier learning; then solving the objective function to obtain a probability distribution matrix of the target domain; selecting, from the probability distribution matrix, a label corresponding to the maximum probability value corresponding to each data sample in the target domain as a prediction label of the data sample of the target domain to complete bearing fault diagnosis. The diagnosis method improves the bearing fault diagnosis accuracy and diagnosis efficiency.