Optimising regenerative braking performance in EVs through advanced deep learning models

Purpose The system for regenerative braking in the electric car functions for deceleration and energy recovery during braking. However, insufficient studies on how the energy diffusion process within electronic braking might occur prevent its complete energy recovery. Therefore, this work aims to address the energy recovery problem of a regenerative braking system using independent motor drives in both front and rear axles. Design/methodology/approach The innovative Bat-based LeNet Braking System (BbLBS) has been developed to optimise braking settings, thus maximising energy recovery and enhancing brake performance. A regenerative brake system for an electric vehicle (EV), incorporating vehicle dynamics, an electric motor, a battery and a brake subsystem, is designed and simulated within a MATLAB/Simulink environment. After the block design, a recovery analysis of energy is conducted by simulating various driving conditions, including urban traffic, highway cruising and downhill driving profiles, in this phase. The bat-optimised fitness feature regulates the regenerative braking approach to enhance energy recovery by adjusting motor torque generation. After the energy recovery, the continuous motor operates in generator mode during braking, which can cause overheating. The optimised LeNet algorithm mechanism effectively controls heat and enhances braking performance. Ultimately, the outcomes of the developed approach are calculated and compared to those of other conventional methods in terms of speed, state of charge, torque, temperature and energy consumption. Findings The proposed method yields the following outcomes: speed of 2.56 km/h, state of charge of 90.38%, torque of 3.1 Nm, energy recovery efficiency of 95.0% and energy consumption of 59.60 W/km. Originality/value The effectiveness of the proposed system demonstrates that the proposed method is well-suited for this study.