Key points are not available for this paper at this time.
Accurate reliability predictions for state systems are essential for industries such as manufacturing, energy, and transportation and utilizing artificial intelligence (AI) considerably enhances this process. However, the several features that affect the target variable and their complicated relationships make this process challenging. This research overcomes these obstacles by applying feature engineering and selecting features using the Boruta method, which guarantees that the analysis is informed solely by the most relevant features. Applying three machine learning models such as Naive Bayes, SVM, and artificial neural networks (ANN) to this refined dataset showed that the artificial neural networks outperformed the competing models with an accuracy of 0.84, largely due to its capacity to represent complicated, non-linear relationships. With an accuracy of 0.79, SVM proved its validity despite feature dependencies, whereas Naive Bayes attained 0.76. These results demonstrate that ANN has great promise as a tool for data-driven decision-making in important industries, with the ability to improve operational efficiency, reduce system failures, and strengthen maintenance plans.
Guo et al. (Sat,) studied this question.