Smart electric vehicle multi-phase energy management using modulated network and predictive battery health modeling

Current energy management systems for electric vehicles face significant challenges such as inefficient handling of transient loads, battery stress leading to capacity loss of up to 20% over vehicle lifespan, and suboptimal energy utilization from hybrid sources. Conventional and heuristic control strategies cannot frequently adapt in real-time to changing driving conditions, thus they may be responsible for about 15% higher energy loss and less battery life. To overcome such problems, the present study is the implementation of a smart multi-phase energy management framework that combines the solar photovoltaic, battery, and supercapacitor sources with an adaptive dual-phase control approach as a solution. Phase one aims at controlling supercapacitor reference voltage for stabilizing rapidly changing loads over 10 kW, while phase two is engaged in total power flow optimization, which led to a 30% reduction of battery current fluctuations and a 12% increase in the overall system efficiency. The Modulated Deformable Coyote and Badger Graph Convolution Network is a one-of-a-kind network that represents complex spatiotemporal dependencies of vehicle speed (0–600 RPM), state of charge (SOC: 5%–100%), current (0–75 A), and acceleration. It is trained and optimized by a Symbiotic Mechanism-Based Coyote and Badger Optimization Algorithm that achieves rapid convergence within 1000 epochs. Simulation results confirm a battery degradation rate reduction of up to 25% and transient response enhancement concerning the proposed method as compared to the existing ones. This framework offers a robust, intelligent, and efficient solution for sustainable energy management in next-generation electric vehicles.