Application of predictive maintenance systems to maintenance management system with machine learning methods in automotive industry

The maintenance approaches play a critical role in the efficiency and sustainable production of manufacturing systems. The effectiveness of equipment is an important performance indicator for sustainable production, as much as production volume and quality. While maintenance approaches traditionally faced certain limitations, today's industrial technologies, particularly under the influence of Industry 4.0, allow us to collect real-time data from machines according to specific standards, thereby enhancing maintenance effectiveness. Using the collected data, machine learning enables us to optimize maintenance processes. This way, potential failures in the production process can be predicted in advance and prevented before such issues occur. The ability to reschedule planned maintenance programs based on the predictive needs using machine learning tools can increase maintenance effectiveness and consequently improve production efficiency. In this study, equipment groups and target failures related to Pareto analysis were selected by examining the downtime, failures, and efficiency rates of equipment in an automotive factory. Various data sources were collected, categorized, and made available as algorithmic data. Artificial intelligence and machine learning tools were utilized in creating algorithms, establishing cause-and-effect relationships, and transferring the evaluation of predictions to the maintenance management system. Predictive maintenance forecasts were achieved with an 81.7% success rate using Recurrent Neural Network (RNN) machine learning methods. Keywords: Industry 4.0, Machine learning, Predictive maintenance