Machine Learning-Driven PdMin Software Engineering: Models, Methods, and Applications
Abstract: Machine Learning (ML) has been increasingly applied in the domain of Software Engineering (SE) for improving Predictive Maintenance (PdM) approaches. In this paper, we discuss ML-based PdMin software systems, with an aim to develop models, methodologies, and applied practices. PdM approach, which was initially used in manufacturing and industrial system applications, has great potential in the software domain to predict software failures, performance, and system vulnerabilities before potential failures happen. This paper provides an overview of various ML algorithms, that is, supervised, unsupervised, and reinforcement learning, used for software fault diagnosis, anomaly detection, and system performance optimization. Moreover, we address the problem and discuss the emerging problems, including the lack of data, features, and interpretation of the models. Applying the ML-based PdM methods to several real-world scenarios, this paper illustrates the potential of the ML-driven PdM for the downtime and reliability of software systems, and the operational costs of the software industry. Finally, the paper outlines future work on how to extend the models and optimize their scalability and applicability in a real-world software system.
Keywords: ML, predictive maintenance, SE, performance, optimization, software failure
How to Cite?: Dr. Ashish Jolly, "Machine Learning-Driven PdMin Software Engineering: Models, Methods, and Applications", Volume 15 Issue 5, May 2026, International Journal of Science and Research (IJSR), Pages: 1095-1101, https://www.ijsr.net/getabstract.php?paperid=SR26517131831, DOI: https://dx.doi.org/10.21275/SR26517131831