Energy Optimization of Airborne Wind Energy System via Deep Reinforcement Learning

Abstract

This paper addresses the energy optimization problem of the airborne wind energy system. Note that it is difficult to build the airborne wind energy system model. A deep deterministic policy gradient (DDPG) algorithm is first designed to solve the optimization problem of continuous action space in this paper. To achieve convergence with fewer training episodes, a proximal policy optimization (PPO) algorithm is designed. The experimental results show that the energy harvesting performance of reinforcement learning algorithms is 46% higher than traditional algorithms. Meanwhile, the convergence performance of the PPO algorithm is faster than it of the DDPG algorithm.