Optimization of Charging Prediction in Electric Vehicle Fleets: An Integrated Approach with EdgeAI and Data Flow Pipelines

In this study, a system is proposed to optimize the charging prediction process of electric vehicle fleets and is integrated with a data flow pipeline constructed along with an EdgeAI-based charge prediction approach. The system processes the data collected from the vehicles on the EdgeAI unit, predicts the current state of charge, transfers it to database through the Fiware middleware, enabling its visualization with data analytics tools. Proposed for the effective management of battery charge states in electric vehicle fleets, this system aims to make a more effective contribution to environmental sustainability and energy efficiency goals by optimizing the data flow. The system was prototyped and tested using raw data collected through the battery management system of an actual electric vehicle. The results show that the proposed system is reliable and can be easily scaled in fleet scenarios. In the future, it is planned to scale and implement the system for fleet vehicles in real time and conduct stress tests.