A knowledge-data integration framework for rolling element bearing RUL prediction across its life cycle.

Prediction of Remaining Useful Life (RUL) for Rolling Element Bearings (REB) has attracted widespread attention from academia and industry. However, there are still several bottlenecks, including the effective utilization of multi-sensor data, the interpretability of prediction models, and the prediction across the entire life cycle, which limit prediction accuracy. In view of that, we propose a knowledge-based explainable life-cycle RUL prediction framework. First, considering the feature fusion of fast-changing signals, the Pearson correlation coefficient matrix and feature transformation objective function are incorporated to an Improved Graph Convolutional Autoencoder. Furthermore, to integrate the multi-source signals, a Cascaded Multi-head Self-attention Autoencoder with Characteristic Guidance is proposed to construct health indicators. Then, the whole life cycle of REB is divided into different stages based on the Continuous Gradient Recognition with Outlier Detection. With the development of Measurement-based Correction Life Formula and Bidirectional Recursive Gated Dual Attention Unit, accurate life-cycle RUL prediction is achieved. Data from self-designed test rig and PHM 2012 Prognostic challenge datasets are analyzed with the proposed framework and five existing prediction models. Compared with the strongest prediction model among the five, the proposed framework demonstrates significant improvements. For the data from self-designed test rig, there is a 1.66 % enhancement in Corrected Cumulative Relative Accuracy (CCRA) and a 49.00 % improvement in Coefficient of Determination (R2). For the PHM 2012 datasets, there is a 4.04 % increase in CCRA and a 120.72 % boost in R2.

Adaptive tacholess order tracking method based on generalized linear chirplet transform and its application for bearing fault diagnosis.

Bearing plays an important role in industrial equipment and it may operate under varying conditions. When the speed of shaft changes, whether monotonous or non-monotonous speed, common diagnostic approaches cannot effectively extract fault features. But encoders and tachometers are not always available. Therefore, tacholess order tracking methods which can directly extract the instantaneous rotating frequency (IRF) from vibration signal are very useful in bearing fault diagnosis under varying speed. Among these methods, the generalized linear chirplet transform (GLCT) can produce time-frequency representation without constructing any mathematical model, but there are two parameters must be set in advance. The parameters have great influence on the analysis result. To reduce the dependence on the prior knowledge of presetting the parameters in varying conditions, two different improved GLCT methods are proposed in this paper. To do with the situation where the trend of speed changes is monotonous, the scale-space is introduced to lift GLCT which can adaptively set a vital parameter, and the other parameter is set to default value. When faced with non-monotonous speed, the second method is proposed which the grey wolf optimizer (GWO) and Gini index are introduced to search the optimal parameters of GLCT without any prior knowledge. With the help of the proposed methods, the IRF can be extracted directly from vibration signal. Then, the raw signal can be resampled based on the IRF to eliminate the influences of speed. The morphological filtering is adopted to remove the noise and extract the fault characteristics order (FCO). Another two typical time-frequency analysis methods are used for comparisons. Three different signals are used for analysis to demonstrate the superiority of the proposed methods.