A Deep Reinforcement Learning-Based MPPT Control for PV Systems under Partial Shading Condition

Abstract

On the issues of global environment protection, the renewable energy systems have been widely considered. The photovoltaic (PV) system converts solar power into electricity and significantly reduces the consumption of fossil fuels from environment pollution. Besides introducing new materials for the solar cells to improve the energy conversion efficiency, the maximum power point tracking (MPPT) algorithms have been developed to ensure the efficient operation of PV systems at the maximum power point (MPP) under various weather conditions. The integration of reinforcement learning and deep learning, named deep reinforcement learning (DRL), is proposed in this paper as a future tool to deal with the optimization control problems. Following the success of deep reinforcement learning (DRL) in several fields, the deep Q network (DQN) and deep deterministic policy gradient (DDPG) are proposed to harvest the MPP in PV systems, especially under a partial shading condition (PSC). Different from the reinforcement learning (RL)-based method, which is only operated with discrete state and action spaces, the methods adopted in this paper are used to deal with continuous state spaces. In this study, DQN solves the problem with discrete action spaces, while DDPG handles the continuous action spaces. The proposed methods are simulated in MATLAB/Simulink for feasibility analysis. Further tests under various input conditions with comparisons to the classical Perturb and observe (P&O) MPPT method are carried out for validation. Based on the simulation results in this study, the performance of the proposed methods is outstanding and efficient, showing its potential for further applications.

Investigation on Maximum Power Point Tracking Algorithm (INC and FLC) for Effective Utilization of Power Under Partial Shaded Photovoltaic System

“…Novel FLC mppt is implemented with adding one more input variable beta which may reduce the dependency of user knowledge and complicated rules. Because of this reason the proposed FLC (5.83 %) provides better tracking efficiency compared to conventional FLC (3.21%) methods (Sundararaj et al 2020;Phan, Lai, and Lin 2020).…”

“…Hybrid algorithm is combination of Cauchy preferential crossover (CC) with the flower pollination algorithm (FPA). The performance is analysed under partial shaded condition which provides better efficiency in the proposed system (99.6 %) over FLC method (94%) (Sundararaj et al 2020;Phan, Lai, and Lin 2020)). Novel FLC mppt is implemented with adding one more input variable beta which may reduce the dependency of user knowledge and complicated rules.…”

Photovoltaic System MPPT Evaluation Using Classical, Meta-Heuristics, and Reinforcement Learning-Based Controllers: A Comparative Study

“…Deep RL [31], [32], [33], which is seen as an advancement of reinforcement learning, has attained great success in robotics, language processing, and many other areas of application [34]. By integrating reinforcement learning with deep learning, deep RL has the advantage of using deep neural networks to efficiently approximate policy functions or value functions without using tables to store large state-action pairs [29].…”

Comparative Analysis of Robustness and Tracking Efficiency of Maximum Power Point in Photovoltaic Generators, Using Estimation of the Maximum Power Point Resistance by Irradiance Measurement Processing

“…It is well known that the MPP is not a static point, but is constantly fluctuating throughout the day, depending on the temperature and irradiance received by the PVG [ 1 , 2 , 3 , 4 ], hence the need to use effective MPPT. Compared to traditional tracking techniques [ 5 , 6 , 7 , 8 ], new ones have been developed, such as model-based techniques [ 9 , 10 , 11 , 12 , 13 , 14 , 15 ], and techniques based on artificial intelligence and bioinspired methods [ 2 , 3 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 ]. In general, these techniques seek to read the maximum power point without considering, at least a priori, the possibility that the installation works at a nonglobal maximum.…”

“…In general, these techniques seek to read the maximum power point without considering, at least a priori, the possibility that the installation works at a nonglobal maximum. For this reason, some techniques focus specifically on the search for the global maximum [ 4 , 16 , 17 , 19 , 21 , 23 , 24 , 25 , 26 , 27 ].…”