Intelligent Energy Scheduling in Renewable Integrated Microgrid With Bidirectional Electricity-to-Hydrogen Conversion

Abstract

The power supplying frontier in microgrids is moving from traditional fossil fuels towards clean renewable energy. Given the temporal asynchrony between intermittent renewable generation and uncertain loads, it is vital to develop an efficient energy scheduling, storing, and distributing scheme to improve renewable energy utilization (REU) and system economics. In this paper, we introduce a power-to-hydrogen (P2H) facility to convert surplus renewable energy into hydrogen through electrolysis. The conversion process is bidirectional where the hydrogen can be re-generated to electricity through a fuel cell or directly sold. Capturing the uncertainty of loads and electricity price, considering the time-coupled constraints of energy storage, we develop a deep deterministic policy gradient based hybrid energy scheduling (H-DDPG) algorithm. The H-DDPG can learn the optimal policies from historical experiences, avoid inadequate exploration by introducing decaying noise, and not rely on future information of the system. Simulations using real-world data demonstrate that 1) the developed bidirectional electricity-hydrogen conversion scenario significantly improves the system economics and REU, 2) our proposed H-DDPG algorithm outperforms the state-of-the-art Deep Q-Network algorithm.