Advanced control and conversion system for efficient PV grid integration and EV charging utilizing deep Q -learning and a hybrid dual boost converter

Aimed at strengthening the adaptability and efficiency of solar energy utilisation, this research introduces an innovative Photovoltaic (PV) grid-connected system that incorporates Electric Vehicle (EV) charging capabilities. The system features a novel Hybrid Dual Boost converter (HDBC) for effectively elevating PV voltage to a higher, more suitable level for grid integration. Multiple voltage-lift stages and a dual-boost arrangement are incorporated into the suggested converter topology. The voltage-lift stages raise the voltage level gradually in steps, and the dual-boost arrangement uses two consecutive boost sections to further increase the total gain. Unlike conventional converters that require multiple active switches and complex control, the HDBC uses a single-switch structure, thereby simplifying control circuitry and reducing switching losses. A Maximum Power Point Tracking (MPPT) technique based on Deep Q-Learning (DQL) is used to handle the nonlinear and time-varying characteristics of solar irradiation. Using a Voltage Source Inverter (VSI) with Phase-Locked Loop (PLL)-based grid synchronisation, high-gain DC output is transformed into synchronised three-phase AC. Furthermore, a Bidirectional Buck-Boost converter that is governed using a dual-loop PI controller facilitates both grid power feedback and EV charging in the instance of demand imbalance. The resilience of the suggested system is demonstrated through MATLAB simulations and hardware validations across a range of scenarios, including steady, dynamic and PV disconnection scenarios. Benchmarking against existing converters reveals the superior voltage gain of 15.625 and improved efficiency of 96%.