A bearing fault diagnosis method and apparatus based on a supervised LLE algorithm, the method comprising: acquiring training data, the training data being historical data representing bearing vibration signals, and extracting feature values of the training data and fault types corresponding to the feature values (S100); determining optimal dimensionality reduction training data of the training data and calculating the mean value and covariance matrix corresponding to each fault type in the optimal dimensionality reduction training data (S300); performing dimensionality reduction on test data received in real time to obtain dimensionality reduction test data (S400); and, on the basis of the mean values and the covariance matrices, calculating the probability value of the dimensionality reduction data in each fault type, and using the fault type with the greatest probability value as the fault type for the bearing fault diagnosis (S500). Thus, the online prediction rate of bearing fault diagnosis is improved.
展开▼
