Abstract

The objective of the present research was to develop an intelligent photovoltaic energy capture system. A technological applied methodology was used, where the system development was carried out in three well-defined parts. The structural stage defined the components of the sensor systems, consisting of resistors and photoresistors; actuators consisting of servo motors; and the control system, which was determined as the ESP-32 due to its characteristics of connectivity for the Internet of Things. In terms of the algorithmic stage, artificial neural networks were chosen for their precision characteristics. Specifically, a neural network with 3 layers was considered: 1 input layer, 1 hidden layer, and 1 output layer consisting of 4, 10, and 2 output neurons, respectively. The inputs were trained with a learning rate of 0.3, optimized by gradient descent, resulting in minimal mean square error. Finally, in the data transmission stage for sensor control and monitoring, and for the solar cell of the prototype, the ESP-32 data transmission station connection was used, with data received by the ThingSpeak™ Internet of Things platform.