Electromagnetic Interference resilient augmented state nonlinear observer for coupled battery converter dynamics in electric vehicles

Abstract The design and analysis of a novel EMI-aware augmented-state nonlinear observer for the electric vehicle battery-converter subsystem considering corrupted voltages and currents is presented in this study. As opposed to classical model-based observer design methodologies, which view any type of sensor imperfection as noise, this novel approach involves incorporating structured electromagnetic interference influences, sensor bias, and conversion error as part of the modeling framework. A reduced first-principles battery–converter model is developed to describe the dominant SOC, polarization-voltage, inductor-current, and DC-link-voltage dynamics required for observer design. Simulations are performed on nominal, structured EMI, reverse load, and combined stress conditions, yielding RMSE values for the state of charge of 0.05234, 0.05208, 0.05219, and 0.05204, respectively. In all cases, the RMS error for the inductor current is below 0.07, while the RMS error for the DC-link voltage ranges between 0.72869 and 0.86019. Under structured EMI and combined stress, the output reconstruction errors are low with the voltage MAE being below 0.01009 and current MAE being below 0.01836. These results confirm that the proposed observer maintains reliable state reconstruction while explicitly accounting for structured measurement corruption, thereby providing a compact and theoretically grounded estimation framework for converter-interfaced EV battery systems.