A New Energy Management Strategy Supported by Reinforcement Learning: A Case Study of a Multi-Energy Cruise Ship

Abstract

Hybrid ships offer significant advantages in energy efficiency and environmental sustainability. However, their complex structures present challenges in developing effective energy management strategies to ensure optimal power distribution and stable, efficient operation of the power system. This study establishes a mathematical model of a hybrid system for a specific ship and proposes an energy management strategy based on the deep deterministic policy gradient (DDPG) algorithm, a reinforcement learning technique. The proposed strategy’s feasibility and effectiveness are validated through comparisons with alternative energy management strategies and real-world ship data. Simulation results demonstrate that the DDPG-based strategy optimizes the diesel engine’s operating conditions and reduces total fuel consumption by 3.6% compared to a strategy based on the deep Q-network (DQN) algorithm.

Bibliographic Information