Publication

Abstract : Hybrid Electric Vehicles involving batteries and supercapacitors are an energy-efficient alternative to conventional vehicles with internal combustion engines. However, considering the variation in the power demands, optimally allocating power between the battery and the supercapacitor in HEVs is challenging. This paper proposes an AI-based Energy Management System based on Recurrent Neural Network which can predict the required power and optimally allocate the energy required between battery and supercapacitor. Further, this method is implemented on a dual-system; a MATLAB/Simulink model and a Python-based UI with a cloud-based communication interface. The model is hence validated with real-time monitoring thereby improving the effectiveness of the RNN-based EMS over the existing conventional methods. Proposed RNN model is compared with CNN and LSTM models using Mean Square Error as performance matrix. Proposed model works superior to CNN and LSTM with an MAE of 30.59. The framework is also equipped to send the optimally allocated power values back to the Simulink model to facilitate a feedback mechanism in the real-time monitoring. The experiment helps in preventing battery degradation and optimised energy distribution in HEVs.