Parametric Optimization of DC–DC Boost Converter Using LSTM–HHO for Renewable Energy Systems Under Bounded Time-Varying Loads

The integration of renewable energy sources such as photovoltaic (PV) and fuel cell systems demands highly efficient and adaptive DC–DC power conversion. However, existing DC–DC boost converters face several challenges, such as poor adaptability to time-varying loads, inefficient parameter tuning, and increased switching stress during fluctuating operating conditions. Traditional control strategies including, proportional-integral (PI) controllers and basic neural networks, are unable to maintain performance under dynamic conditions. This paper presents a long short-term memory and Harris Hawks optimization (LSTM–HHO) framework for parametric optimization of DC–DC boost converter architectures. A detailed mathematical model of the boost converter under time-varying load conditions is developed, aiming to describe the key problems and their theoretical solutions. The model uses LSTM networks to predict time-varying load behaviors and adjusts control parameters through the HHO algorithm. The framework is validated across multiple converter topologies, including conventional boost, flyback, and interleaved configurations, using advanced semiconductor devices such as Silicon Carbide metal-oxide-semiconductor field-effect transistor (SiC-MOSFET) and Gallium Nitride high-electron-mobility transistor (GaN-HEMT). Simulation results under bounded time-varying load conditions and source perturbation scenarios show that the proposed framework achieves improved voltage regulation, reduced output ripple, lower switching stress, and faster convergence than conventional PI, fuzzy-logic, deep reinforcement learning-enhanced model predictive control (DRL–MPC), and RBF-FLC approaches. Validation with nonideal simulations and hardware-in-the-loop (HIL)/controller-in-the-loop (CIL) testing under bounded time-varying loads R(t) ∈ [R_min, R_max] confirms reduced ripple and switch stress at embedded-feasible compute cost.