IoT and AI-Based Predictive Maintenance for Hybrid Vehicles in Urban Transport Systems

作者：M. S. Grace Padma, K. R, Srinivasulu Sirisala, Eswara Prasath Natarajan, Dilip Moyal, Palanivel Kandappagoundar · 发表于：2025 International Conference on Emerging Technologies and Innovation for Sustainability (EmergIN) · 年份：2025 · DOI：10.1109/emergin67762.2025.11450657

Approaches for Guaranteeing Timely and Reliable Autonomous Transport from Heterogeneous Electric Vehicle Fleets on Urban Roads with Hybrid and Non-hybrid Capabilities. The rapid integration of hybrid electric vehicles (HEVs) in urban public transport leads to new challenges related to their availability, safety, and efficient operation. In this article, a three-layered IoT and AI-based framework for predictive maintenance of hybrid vehicles is presented by combining on-board perception sensors and delay-sensitive communication protocols with cloud-based deep learning models. In this work, a stacked Long Short-Term Memory (LSTM) autoencoder is used to detect anomalies with the time series sensor data while fault categories such as battery overheating, brake system degradation, and inverter failure are identified by a Random Forest classifier. The anomaly detection trigger was also updated to a dynamic threshold using Extreme Value Theory (EVT) to be more robust and reduce false alarms. In addition, the fuzzy logic decision engine offers actionable information such as Remaining Useful Life (RUL) and maintenance urgency. Experimental validation with 50 hybrid buses and a one-year real-world dataset shows that our solution can achieve up to 37% of unplanned downtime reduction, leading to a cost saving of more than 18%, validating the scalability and applicability of the proposed approach on a large scale for intelligent urban transport systems.