Neural network based optimized barrier conditioned double super-twisting sliding mode controller of electric vehicle charger with grid to vehicle and vehicle to modes

The widespread adoption of electric vehicle (EV) ecosystem is tantamount to providing a reliable means of charging of vehicle battery through a critical component i.e., an EV charger. This paper presents a bidirectional EV charger system comprising of AC to DC converter, dual active bridge (DAB) and equivalent battery model has been proposed. Furthermore, a novel barrier conditioned double super-twisting sliding mode controller (BC DST-SMC) is presented in this study. The proposed approach is applied to cater the system disturbances and improve the controller's performance in terms of robustness and disturbance rejection. Furthermore, a methodical optimization approach is taken to tune the proposed controller gains instead of error-and-trial approach. The gains of the proposed controller have been tuned using meta-heuristic algorithm "genetic algorithm" (GA), "improved-grey wolf optimizer" (I-GWO) and "neural network algorithm" (NNA) with integral square error (ISE) as an objective function. The optimization results show that NNA generates the most optimized gain values. The proposed model and controller are simulated using ODE45 solver settings in Simulink, MATLAB (R) (2023a) where its comparison is done with the conventional super-twisting sliding mode controller (ST-SMC). The proposed framework has also been validated through hardware-in-loop (HIL) experimental setup using MicroLabBox DSP-based dSPACE-DS1202.