The degradation starting time is an important variable affecting the accuracy of degradation path prediction, but little work has been considered in existing studies. This article investigates the problem of predicting the performance of rolling element bearings based on early degradation analysis. Based on an improved dual linear structural support vector machine with envelope spectrum algorithm and $\mu +4\sigma$ criteria, a new health indicator is proposed to detect the degradation starting time. As well the detected time is sensitive to early anomalies. In addition, according to the degradation starting time, a convolutional neural network prediction model is established to predict the degradation path. Experiments show the effectiveness and superiority of the proposed method.

As a critical component of the engine, the failure of aviation fuel pumps can lead to serious safety accidents, necessitating the development of effective maintenance programs. Fault Tree Analysis (FTA) has a clear structure and strong interpretability in maintenance decision making. However, it heavily relies on expert knowledge, which is subject to uncertainty and incoherence. Therefore, this paper proposes the NOA (Nutcracker Optimization Algorithm)–GNN (Graph Neural Network) model to enhance the accuracy and robustness of FTA by mitigating the uncertainty and inconsistency in expert knowledge. The NOA algorithm efficiently searches the solution space to identify globally optimal solutions. An FTA-TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) maintenance decision-making framework has also been developed. By integrating FTA with TOPSIS, the proposed method provides a comprehensive and systematic approach that combines qualitative and quantitative analyses, thereby improving the effectiveness and reliability of maintenance decision making.

Evidence reasoning (ER) rule can effectively integrate qualitative knowledge and quantitative data and simultaneously flexibly deal with various types of uncertain information. Although ER rule has been applied in many engineering fields, many challenges still need to be addressed. For example, due to the lack of labeled data samples and the inherent fuzziness in information transformation of semantic reference values in health assessment of complex electromechanical systems, the accuracy of the assessment results is difficult to be measured quantitatively, which reduces the credibility of the ER model. To address this issue, a new ER model considering the credibility of results is proposed in this article, denoted as the ER-c model. First, a method for calculating the assessment results credibility is designed by establishing the relationship between the internal and external measurement information of a complex electromechanical system. Second, the objective function for the ER-c model optimization process is established according to the credibility and optimized using the covariance matrix adaptation evolution strategy (CMA-ES) algorithm. Finally, in the fields of instrument measurement in practical engineering, the problem that the hidden information such as health status of complex electromechanical system cannot be accurately measured is improved in this article. A case study of a body-in-white welding robot is conducted, and the results demonstrate the efficacy of the proposed method and provide a reliable foundation for measuring the credibility of assessment results.

Real-time prediction of remaining useful life (RUL) is one of the most essential works in prognostics and health management (PHM) of the micro-switches. In this paper, a linear degradation model based on an inverse Kalman filter to imitate the stochastic deterioration process is proposed. First, Bayesian posterior estimation and expectation maximization (EM) algorithm are used to estimate the stochastic parameters. Second, an inverse Kalman filter is delivered to solve the errors in the initial parameters. In order to improve the accuracy of estimating nonlinear data, the strong tracking filtering (STF) method is used on the basis of Bayesian updating Third, the effectiveness of the proposed approach is validated on an experimental data relating to micro-switches for the rail vehicle. Additionally, it proposes another two methods for comparison to illustrate the effectiveness of the method with an inverse Kalman filter in this paper. In conclusion, a linear degradation model based on an inverse Kalman filter shall deal with errors in RUL estimation of the micro-switches excellently.

Feature distributions shift under variable-speed gearbox operation, which can cause a model trained only on healthy samples to misclassify normal operating changes as anomalies. A semi-supervised anomaly detection method is proposed in this study. Using healthy data, the method first fits speed-dependent trends for the selected time- and frequency-domain statistics, and the deviations from these trends form the statistical residuals. Computed order tracking then converts the vibration signal to the angular domain, where five mechanism features describe meshing energy, harmonic structure, sideband modulation, and order-spectrum entropy. Removing the corresponding healthy speed trends yields the mechanism residuals. Robust Bounded Health-Consistency Weighting (RB-HCW) weights these residuals according to their variability in healthy data before they are fused with the statistical residuals and modeled by Deep Support Vector Data Description (DeepSVDD). The Sequential Bayesian Queue-Based Alarm (SBQA) module then confirms whether abnormal decisions persist across successive windows. Across the four fault types under the two separately modeled load conditions, the proposed method achieved macro-averaged true positive rate (TPR), accuracy (ACC), and F1-score values of 93.31%, 92.58%, and 94.02%, respectively.