Akademik Veri Yönetim Sistemi
ICAIS'25 INTERNATIONAL CONFERENCE ON ARTIFICIAL INTELLIGENCE STUDIES, İstanbul, Türkiye, 8 - 11 Kasım 2025, (Yayınlanmadı)
The rapid digitalization of industrial systems, driven and accelerated by the Industry 4.0 paradigm, has fundamentally reshaped the way production processes and maintenance operations are managed. Within this new technological ecosystem, where machines, sensors, and systems are interconnected through the Internet of Things (IoT), vast amounts of real-time data are continuously generated and analyzed. Traditional maintenance strategies—such as reactive maintenance, which addresses failures only after they occur, and preventive maintenance, which relies on scheduled interventions regardless of actual equipment conditions—often lead to inefficient resource use, unplanned production downtime, and increased operational costs. In contrast, predictive maintenance approaches, which integrate IoT-based monitoring with advanced artificial intelligence and machine learning algorithms, enable early detection of potential failures, optimized maintenance scheduling, and data-driven decision-making. As a result, predictive maintenance has emerged as a forward-looking, efficient, and sustainable solution that enhances equipment reliability, reduces environmental impact, and supports the overall goals of smart manufacturing within the Industry 4.0 framework. This study analyzes the potential of machine learning–based predictive maintenance models in forecasting failures within production lines. A dataset containing 1,500 machine operation records, including parameters such as air temperature, process temperature, rotational speed, torque, tool wear, and machine type, was used for model development. The target variable was defined as machine failure. Data preprocessing involved converting categorical variables into numerical form and addressing class imbalance through the Synthetic Minority Oversampling Technique (SMOTE). The Random Forest algorithm was applied to train the model, yielding 95% accuracy and an F1-score of 0.75. The application of SMOTE significantly improved the model’s ability to predict rare failure cases, demonstrating the efficiency of oversampling in imbalanced industrial datasets. The findings indicate that IoT-enabled predictive maintenance models can play a crucial role in reducing downtime, lowering maintenance costs, and improving overall operational efficiency in manufacturing industries. The purpose of this article is to evaluate the impact of predictive maintenance using IoT data on industrial operations, focusing on the benefits of proactive strategies, challenges in implementation, and their broader implications for digital transformation and sustainable manufacturing.