Author(s): Lena Andersson, Erik Johansson and Klara Nilsson

Abstract: The automotive manufacturing industry faces increasing pressure to optimize production systems, reduce downtime, and cut operational costs. Predictive maintenance (PdM), enabled by artificial intelligence (AI) and machine learning (ML), has emerged as a critical solution to these challenges. This paper investigates the integration of AI and ML techniques into PdM to enhance the efficiency and reliability of manufacturing systems. The research methodology involves using data collected from sensors in manufacturing equipment to train machine learning models for predicting potential failures. Techniques such as time-series forecasting using Long Short-Term Memory (LSTM) networks, anomaly detection via autoencoders, and classification using support vector machines (SVM) and random forests were employed to predict machine failures before they occur. Key findings indicate that AI-based PdM systems significantly improve predictive accuracy, reducing unplanned downtime by 30% and lowering maintenance costs by 25%. This paper introduces a novel framework for integrating real-time predictive maintenance, where AI models continuously learn and adapt to changes in manufacturing environments. Additionally, the research compares various AI techniques, assessing their effectiveness in terms of prediction accuracy and computational efficiency. The paper concludes with recommendations for future research directions, emphasizing the need for further refinement and scaling of AI-based PdM systems in automotive manufacturing.

DOI: 10.22271/27078205.2025.v6.i2a.57

Pages: 17-25 | Views: 30 | Downloads: 16

Download Full Article: Click Here