AI Powered Maintenance Forecasting in Mechanical Systems: A Data Driven Approach

This research proposes an AI-based predictive maintenance (PdM) system with real-time operation in mechanical systems. It offers a neuromorphic computing, machine learning, and explainability from large language models-based hybrid architecture to reduce downtime by over 70% and increase diagnostic readability and trustworthiness. The system is running on edge devices like Raspberry Pi and Intel Loihi with sub-5ms latency and power consumption less than 50mW, making it possible for real-time monitoring without cloud dependency. Domain testing by domain experts confirms both its human usability and technical performance and, therefore, is the ideal solution for the industrial environment today.