Author(s): Alexei Ivanov, Tatiana Petrova and Dmitry Volkov

Abstract: Predictive maintenance (PdM) has emerged as a transformative strategy in the manufacturing sector, significantly improving the reliability and efficiency of machine tools. Traditional maintenance approaches, which were reactive in nature, often led to costly downtimes and unanticipated machine failures. In contrast, PdM leverages the power of data analytics, machine learning (ML), and real-time sensor technologies to predict tool failures before they occur, thereby enabling proactive maintenance actions. This paper explores the latest advancements in PdM techniques, focusing on their application in enhancing machine tool reliability. By integrating Internet of Things (IoT) devices and advanced predictive algorithms, manufacturers can collect continuous data from machine sensors, including temperature, vibration, and pressure, which are then analyzed to forecast potential failures. The research evaluates the integration of machine learning models such as support vector machines (SVM), decision trees, and deep learning algorithms, particularly convolutional neural networks (CNNs), in improving failure predictions. Furthermore, the study highlights the significance of predictive analytics in reducing unplanned downtime, increasing overall equipment effectiveness (OEE), and extending the lifespan of machine tools. The findings indicate that implementing PdM can reduce downtime by up to 40%, resulting in substantial cost savings. However, challenges such as high initial setup costs, the complexity of machine learning models, and the need for skilled personnel remain as barriers to widespread adoption. This paper concludes by suggesting future research directions that focus on further integrating advanced artificial intelligence (AI) and developing more cost-effective PdM systems to ensure wider implementation across manufacturing sectors.

DOI: 10.22271/27074544.2025.v6.i1a.56

Pages: 59-64 | Views: 496 | Downloads: 253

Download Full Article: Click Here