Angular position control for solar panel based on neural networks and intelligent control

Abstract

In Ciudad Juarez Chihuahua Mexico there are more than 300 sunny days a year, which means that the solar energy source is a viable form of energy transformation. Therefore, an intelligent controller proposal for the angular position of a solar panel is presented, where the controller's reference signal is obtained from the output of a multilayer neural network previously trained with sun location data. The training data was obtained from the INEGI database in Mexico and from other open access repositories. In addition, the mechanical design and its integration of the parts that make up the solar panel are shown. The closed loop controller designed makes use of a PID with the intention of becoming robust against external disturbances to the system. This allows minimizing the effect of non-modeled dynamic variables that could affect the performance of the panel-actuator system. The experimental results show an underdamped response of the controlled output at this transient and zero error at steady state. The design proposal demonstrates adequate solar reference position tracking, which can be implemented at low cost.