CIFICEN   24414
CENTRO DE INVESTIGACIONES EN FISICA E INGENIERIA DEL CENTRO DE LA PROVINCIA DE BUENOS AIRES
Unidad Ejecutora - UE
artículos
Título:
Deep reinforcement learning approach for MPPT control of partially shaded PV systems in Smart Grids
Autor/es:
DE PAULA, MARIANO; TRIMBOLI, MAXIMILIANO; AVILA, LUIS; CARLUCHO, IGNACIO
Revista:
APPLIED SOFT COMPUTING
Editorial:
ELSEVIER SCIENCE BV
Referencias:
Año: 2020
ISSN:
1568-4946
Resumen:
Photovoltaic systems (PV) are having an increased importance in modern smart gridsresystems. Usually, in order to maximize the energy output of the PV arrays a maximumpower point tracking (MPPT) algorithm is used. However, once deployed, weatherconditions such as clouds can cause shades in the PV arrays affecting the dynamics of eachpanel differently. These conditions directly affect the available energy output of the arraysand in turn make the MPPT task extremely difficult. For these reasons, under partialshading conditions, it is necessary to have algorithms that are able to learn and adaptonline to the changing state of the system. In this work we propose the use of deepreinforcement learning (DRL) techniques to address the MPPT problem of a PV arrayunder partial shading conditions. We develop a model free RL algorithm to maximize theefficiency in MPPT control. The agent?s policy is parameterized by neural networks, whichtake the sensory information as input and directly output the control signal. Furthermore,a PV environment under shading conditions was developed in the open source OpenAIGym platform and is made available in an open repository. Several tests are performed,using the developed simulated environment, to test the robustness of the proposedcontrol strategies to different climate conditions. The obtained results show the feasibilityof our proposal with a successful performance with fast responses and stable behaviors.The best results for the presented methodology show that the maximum operating powerpoint achieved has a deviation less than 1% compared to the theoretical maximum powerpoint.