Advanced optimization strategies for resilient and cost-efficient hydrogen-based AC/DC microgrids with integrated RES and BESS

Abstract

Currently, renewable energy sources (RESs) are crucial for advancing sustainable and resilient energy systems owing to their integration into alternating current and direct current microgrids. However, the current approaches face limitations when it comes to suboptimal topology designs, inefficient resource sizing, and limited adaptability under dynamic conditions. These constraints limit cost efficiency, energy efficiency, and resiliency, particularly when coupled with emerging sectors such as hydrogen-powered transportation. To address these challenges, this framework develops an innovative, multifaceted approach based on advanced artificial intelligence and optimization techniques. The first is the use of a graph convolutional neural network (GCNN) with stochastic optimization to obtain robust and efficient layouts that integrate RES, battery energy storage system (BESS), and hydrogen electrolyzers, allowing transportation sector coupling. It ensures the optimal sizing of hydrogen electrolyzers and BESS, based on the criteria of balancing capital expenditure, operational expenditure, and energy efficiency, using a multi-objective mixed integer linear programming approach. Second, it proposes a hierarchical model predictive control strategy that utilizes hybrid long short-term memory–Gaussian process forecasts for real-time, multi-time-scale operational optimization of energy dispatch, along with hydrogen production. Finally, a deep reinforcement learning model with proximal policy optimization ensures adaptive and fault-tolerant energy management by incorporating resiliency metrics into the learning process. The proposed framework brings about the following significant advancements: an approximately 25% reduction in power losses, around 30% faster outage recovery, hydrogen production costs of 2–3 $·kg⁻¹, and renewable utilization rates near 95%. It further supports cost and energy efficiency and creates a more resilient blueprint that sets up a promising integration of RES, BESS, and hydrogen technologies into the future towards sustainability.