Aiming at the problem that a single neural network model has difficulty in accurately predicting trends of the remaining useful life of a rolling bearing, a method of predicting the remaining useful life of rolling bearings using a gated recurrent unit-deep autoregressive model (GRU-DeepAR) with an adaptive failure threshold was proposed. First, time domain and frequency domain features were extracted from the rolling bearing vibration signal. Second, its operation process was divided into a smooth operation stage and degradation stage according to the trend of the accumulated root mean square of maximum. Then, the failure threshold for different bearings were determined adaptively by the maximum of the smooth operation data. The degradation dataset of a rolling bearing was subsequently obtained. In the meantime, a GRU-DeepAR model was built to obtain predictions of the failure time and failure probability. Appropriate model parameters were determined after a large number of tests to assure the effectiveness and prediction accuracy. Finally, the trend of time series and failure times were predicted by inputting the degradation dataset into the GRU-DeepAR model. Experiments showed that the proposed method can effectively improve the accuracy of the remaining useful life prediction of a rolling bearing with good stability.

In light of the problems of a single vibration feature containing limited information on the degradation of rolling bearings, the redundant information in high-dimensional feature sets inaccurately reflecting the reliability of rolling bearings in service, and assessments of the degradation performance being disturbed by outliers and false fluctuations in the signal, this study proposes a method of assessing rolling bearings’ performance in terms of degradation using adaptive sensitive feature selection and multi-strategy optimized support vector data description (SVDD). First, a high-dimensional feature set of vibration signals from rolling bearings was extracted. Second, a method combining the Technique for Order Preference by Similarity to an Ideal Solution (TOPSIS) and K-medoids was used to comprehensively evaluate the features with multiple evaluation indicators and to adaptively select better degradation features to construct the sensitive feature set. Next, multi-strategy optimization of the SVDD model was carried out by introducing the autocorrelation kernel regression (AAKR) and a multi-kernel function to improve the ability of the evaluation model to overcome outliers and false fluctuations. Through validation, it could be seen that the method in this study uses samples of rolling bearings in the healthy early stage to establish the evaluation model, which can adaptively determine the starting point of the bearing’s degradation. The stability and accuracy of the model were effectively improved.

Aiming at the existing life prediction methods for rolling bearing degradation information mining is not sufficient, the critical time step information degree is insufficient, resulting in the loss of key degradation information, model prediction accuracy and model generalization ability is insufficient, this paper proposes a novel deep multiscale feature fusion network with dual attention for rolling bearing remaining useful life (RUL) prediction. First, multi-domain feature sets of rolling bearing vibration signals are acquired. Subsequently, it is proposed to use Squeeze-and-Excitation (SE) attention mechanism to calibrate and weigh temporal significance of feature sequences, thereby capturing critical temporal information. Then, a multi-scale feature extraction and fusion module with deep network is constructed, consisting of multiple identical multi-scale residual pyramid layers connected in series to further delve into the state information of rolling bearings. Additionally, a relative position encoding method suitable for time series prediction is introduced within the multi-head attention mechanism, a network architecture based on a dual attention-enhanced Transformer encoding layer is established. This enhancement significantly improves model prediction accuracy, generalization capability, and sequence data comprehension ability. Finally, the high-level features output from the feed-forward layer are mapped through a regression layer to obtain the final prediction results. Experimental results demonstrate the superior performance of the proposed method in terms of both prediction accuracy and generalization capability for rolling bearing life prediction. A more robust and accurate framework for RUL prediction in rolling bearings is provided.

Rolling element bearings are critical components in rotating machinery. To tackle the problem of difficult to accurately characterize the operating state of rolling bearings caused by irrelevance and varying sensitivity of multiple features to performance degradation, and introduction of subjective errors in determination of hyperparameters of deep learning models, which can affect the accuracy and efficiency of remaining useful life (RUL) prediction. To address these challenges, this paper proposed a novel RUL prediction method for rolling bearings with secondary feature selection and Bayesian optimization of self-attention mechanisms for bidirectional long short-term memory (BSBiLSTM). Firstly, multi-domain features are extracted from noise-reduced vibration signals. Then, a three-criterion constraint-based feature selection algorithm is used and a secondary selection algorithm with Pearson correlation coefficient is proposed to improve data quality. Next, the 3σ criterion is integrated to determine the first prediction time for rolling bearings and to divide the degradation stage. Subsequently, the BiLSTM model with Bayesian optimization and self-attention mechanism is proposed to predict the RUL of rolling bearings to further improve the algorithm efficiency. Finally, experimental validation is carried out based on the PRONOSTIA platform dataset and the XJTU-SY rolling bearing dataset, and the results show that the method proposed in this paper is better than many mainstream life prediction methods for rolling bearings at present, and the prediction accuracy is higher.

Important state parameters, such as torque and angle obtained from the servo control and drive system, can reflect the operating condition of the equipment. However, there are two problems with the information obtained through the network from the control and drive system: the low sampling rate, which does not meet the sampling theorem and the nonuniformity of the sampling points. By combing equivalent sampling and nonuniform signal reconstruction theory, this paper proposes a reconstruction method for signal obtained from servo system in periodic reciprocating motion. Equivalent sampling combines the low rate and nonuniform samples from multiple periods into one single period, so that the equivalent sampling rate is far increased. Then, the nonuniform samples with high density are further resampled to meet the reconstruction conditions. This step can avoid the amplitude error in the reconstructed signal and increase the possibility of successful reconstruction. Finally, the reconstruction formula derived from basis theory is applied to recover the signal. This method has been successfully verified by the simulation signal of the robot swing process and the actual current signal collected on the robot arm testbed.