Performance Evaluation of PLA-SVM based Predictive Maintenance Systems

In recent years, predictive maintenance has drawn a lot of attention and emerged as an important issue in industries where dependability and efficiency are crucial. The PLA-SVM technique, based on piecewise linear programming, offers a scalable and efficient framework for handling high-dimensional industrial datasets. The study employs PLA-SVM to evaluate a predictive maintenance dataset, conducting a thorough comparison with established machine learning algorithms commonly used in the field. The evaluation study showed that the PLA-SVM algorithm achieved the highest accuracy of 97.0% and precision of 96.65%, signifying its ability to precisely identify positive cases. Furthermore, PLA-SVM performs in terms of the F1 score of 99.9%, indicating a robust equilibrium between precision and recall among the different algorithms tested, such as Random Forest (99.0%), Decision Tree (98.6%), and Support Vector Machine (97.9%). A key highlight of this study is the PLA-SVM model's rapid training time of just 2, indicating its efficiency in predicting equipment failures accurately. The study highlights PLA-SVM's superior performance and computational efficiency, positioning it as a reliable solution for predictive maintenance in industrial settings.