Enhanced hybrid MPPT approach strategy in PV–fuel cell–battery–supercapacitor systems powered electric vehicle applications under partial shading conditions

This work introduces an innovative method for intelligent control and optimization of the electric drive system, utilizing an integrated combination of photovoltaic (PV), fuel cell (FC), battery, and supercapacitor sources. The system is designed to provide and regulate power for a brushless DC (BLDC) motor in an Electric Vehicle (EV), aiming for optimal operational efficiency under fluctuating irradiance, particularly in Partial Shading Conditions (PSC). This paper presents an adaptive hybrid maximum power point tracking (MPPT) methodology that integrates a pretrained neural network (NN) with the perturbation and observation (P&O) method. The system in development employs an intelligent decision engine and a MATLAB function block to manage the dynamic exchange of solar irradiance variations from irradiation difference and the output power levels from associated plants/sites. This technique attains superior tracking precision with little oscillation by enabling context-sensitive, fluid transitions between P&O and NN control operations. The hybrid energy system of the proposed vehicle consists of four power sources connected to a DC bus through the DC-DC power converters. A DC bus is further connected to a DC-AC inverter used to run the electric motor of a vehicle. A Fractional Order PI (FOPI) controller, tuned with the Chaotic JAYA algorithm, is compared to a Proportional-Integral (PI) controller optimized through Particle Swarm Optimization (PSO) for motor speed control. The FOPI controller displays enhanced stability and transient response. The simulation results across diverse PSC scenarios illustrate that the proposed hybrid MPPT and energy control architecture exhibits resilience. The proposed method achieved superior MPPT performance, with an average efficiency under partial shading. It significantly reduces and improves convergence speed compared to conventional methods, yielding performance metrics such as MeanSquare Error (MSE) of = 0.22. These results demonstrate the method’s high accuracy, stability, dependable motor performance and practical viability for real-world EV applications.