Abstract—The world wide resource crisis led scientists and engineers to search for renewable energy sources. Photovoltaic systems are one of the most important renewable energy sources. In this paper we propose an intelligent solution for solving the maximum power point tracking problem in photovoltaic systems. The proposed controller is based on reinforcement learning techniques. The algorithm performance far exceeds the performance of traditional maximum power point tracking techniques. The algorithm not only reaches the optimum power it learns also from the environment without any prior knowledge or offline learning. The proposed control algorithm solves the problem of maximum power point tracking under different environment conditions and partial shading conditions. The simulations results show satisfactory dynamic and static response and superior performance over famous perturb and observe algorithm.
Index Terms—Photovoltaic, maximum power point tracking, reinforcement learning.
Ayman Youssef and Mohamed El. Telbany are with the Electronics Research Institute, Giza, Egypt (e-mail: aymanmahgoub@eri.sci.eg, telbany@eri.sci.eg).
Abdelhalim Zekry is with the Electronics and Communication Department, Ain Shams University, Egypt (e-mail: author@nrim.go.jp).
[PDF]
Cite:Ayman Youssef, Mohamed El. Telbany, and Abdelhalim Zekry, "Reinforcement Learning for Online Maximum Power Point Tracking Control," Journal of Clean Energy Technologies vol. 4, no. 4, pp. 245-248, 2016.