In most power plants, electric power is generated by rotating machinery. With the rapid growth of the unit capacity, the working conditions of bearings are becoming more and more severe. In order to increase the reliability of the units, it is important to evaluate the remaining useful life of the bearings. In this paper, a similarity-based feature extraction method is proposed. The First Prediction Time (FPT) is determined by analyzing the difference of the standard deviation of the vibration data in single window. Then, a group of time and frequency-domain features are calculated and smoothed. The 1st to 3rd order differentials of a certain window length of feature sequence are gathered to reveal the trend characteristics. The similarity between feature matrices are used as the input of the regression model. The feature set based on proposed method shows its superiority on the prediction results than traditional features.

Similarity,Feature extraction,Remaining useful life