Publication

Abstract : As recently observed, the implementation authenticity of emerging photovoltaic (PV) array reconfiguration methods is regularly under suspicion as it may fail to relocate the PV modules during varying irradiance on a real-time platform. To address this problem, a self-learning Internet of Things (IoT) framework is designed and implemented in the article to enhance the power generation under the variable and nonuniformly shaded 6×6 total-cross-tied (TCT) connected PV arrays. To make the IoT framework self-learning, it allows each PV module to harvest PV array information, trace I-V and P-V curves, alter the configuration within the PV array using the Advantage Actor-Critic (A2C) algorithm, and perform the closed-loop operation. The designed closed-loop and self-learning IoT framework employs message queuing telemetry transport (MQTT) communication protocol, enabling communication with different PV modules of the array, user, and IoT components during nonuniform shading conditions. Finally, the designed method provides optimal electrical locations of PV modules based on the information received from the IoT cloud. In this view, it has innovative competencies e.g., precise real-time monitoring and control over varying irradiance with a higher rate for each PV module and local processing of congregated data within the array. A comparative analysis among the various reconfigurations is also assessed to show the satisfactory performance of the proposed IoT framework. Furthermore, to validate the effectiveness of the designed framework, a hardware prototype is developed and performance is found satisfactory.