FAULT DIAGNOSIS METHOD BASED ON ADAPTIVE MANIFOLD EMBEDDED DYNAMIC DISTRIBUTION ALIGNMENT

摘要

A fault diagnosis method based on adaptive manifold embedded dynamic distribution alignment. The optimal subspace dimension is automatically calculated, and the geodesic flow kernel and transformed manifold feature representation are calculated, thereby preventing feature distortion of data in an original Euclidean space. The use of a similarity measure A-distance to define an adaptive factor and dynamical adjustment of relative weights of conditional distribution and edge distribution of sample data reduce the distribution difference between source domain and target domain samples and improve the accuracy and effectiveness of rolling bearing fault diagnosis under variable working conditions. The present method has high interpretability and low requirements on computer hardware resources, can achieve fast execution speed and excellent diagnostic accuracy, algorithm convergence, and parameter robustness, is especially suitable for multi-scenario and multi-fault bearing fault diagnosis under variable working conditions, and can be widely applied to fault diagnosis tasks under variable working conditions of complex systems such as machinery, electric power, chemical industry, and aviation.