<![CDATA[Predictive Maintenance (PdM) has become an important application of Artificial Intelligence (AI) in modern manufacturing environments, enabling organizations to predict equipment failures, optimize maintenance schedules, and improve operational efficiency. Despite their high predictive performance, many AI-based predictive maintenance models operate as black-box systems, limiting transparency and reducing user trust in maintenance recommendations. This study aims to implement Explainable Artificial Intelligence (XAI) techniques within predictive maintenance systems to improve model interpretability and support more transparent maintenance decision-making. Industrial equipment data collected from IoT sensors, including vibration, temperature, pressure, and runtime measurements, together with historical maintenance records, were analyzed using machine learning and deep learning models, namely Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Long Short-Term Memory (LSTM). Model performance was evaluated using Accuracy, Precision, Recall, and F1-score metrics, while explanation effectiveness was assessed through interpretability analysis and expert validation involving maintenance engineers, production managers, and reliability specialists. The results demonstrate that the proposed XAI-enabled predictive maintenance framework achieves high predictive performance, with the LSTM model obtaining the highest accuracy of 95.1%, outperforming RF and XGBoost models. Furthermore, SHAP and LIME successfully identified vibration and temperature as the most influential factors contributing to equipment failure predictions and provided understandable explanations for individual maintenance decisions. These findings suggest that integrating Explainable AI into predictive maintenance systems enhances model transparency, supports effective decision-making, and promotes the practical adoption of AI technologies in industrial environments.]]>
