Sistema Ciberfísico para Estimação e Controle de Posição Aplicado à Sistemas Fotovoltaicos

Abstract

This dissertation proposes the systematic development of a Cyber-Physical System (CPS). The methodology involves the integration of intelligent sensors based on Artificial Neural Networks (ANNs) and an optimal controller of the Discrete-time Linear Quadratic Tracking (DLQT) type, tuned through reinforcement learning, for positioning a dual-axis structure considering circular motion in two degrees of freedom. The motivation for applying the CPS stems from the increasing demand for renewable energy sources, combined with the low efficiency of fixed photovoltaic modules, which require technological solutions capable of optimizing solar irradiance harvesting. The system is divided into two modules: a Smart Solar Incidence Tracking Sensor, developed in accordance with applicable technical standards and dedicated to estimating the coordinates (elevation and azimuth angles) of maximum solar irradiance, and a position-control module for photovoltaic panels, which operates along the trajectory estimated by SIRIS. The designed module is not restricted to the proposed application, being compatible with other implementations such as aerospace engineering, military systems, and telecommunications. The use of IoT-based techniques characterizes the cyber-architecture of the CPS. Experimental results and simulations demonstrate the effectiveness of the proposed approach in maximizing the energy generated while reducing control effort, even under parametric variations. The findings highlight the potential of the proposed CPS as an innovative and efficient solution for intelligent solar tracking in photovoltaic systems.