Abstract:

In order to solve the problem that the RUL prediction accuracy of rolling bearings was not high due to the distortion of degradation feature distribution under the condition of limited full-life monitoring data of rolling bearings, a prediction method of the RUL of rolling bearings based on EWM and SVR was proposed. Firstly, the time-domain and frequency-domain features of the vibration signal were extracted, and the logarithmic transformation was performed on the features. Then, the index weights were determined by EWM to realize the feature selection. Finally, SSA was used to optimize the SVR model, and the low-dimensional features after dimensionality reduction by principal component analysis(PCA) were used as the input of the optimized SVR model, and the RUL percentage was used as the output, so as to realize the prediction of the RUL of the bearing. The results show that under the condition of limited monitoring data, compared with other methods, the proposed method not only has a more stable prediction performance, but also has an average reduction of 19.51% in absolute error and 17.73% in mean square error(MSE).

Key words: entropy weight method(EWM), support vector regression(SVR), rolling bearing, remaining useful life(RUL)prediction, sparrow search algorithm(SSA)