Desarrollo de un modelo con técnicas de inteligencia artificial para el análisis y optimización de la eficiencia de colectores solares con tubos de vacío, de uso doméstico para la ciudad de Riobamba.

Abstract

In this thesis ANFIS and NNML are used to analyze and optimize the efficiency of solar collectors with vacuum tubes for domestic use in the city of Riobamba. For the purpose of designing and manufacturing solar thermal collectors, efficiency is a fundamental parameter and the input variables that differ in their performance are collector area, solar irradiation and vacuum percentage. The data generated by ANFIS simulating the behavior of the different parameters that affect the efficiency of the solar collector were normalized and classified in order to train and validate the NNML models. The NNFB, NNLR and NNRB outputs were compared with each other using RMSE, MAPE, R^2 and MSE functions. Different characteristics of the neural networks, such as number of hidden layers, number of neurons in that layer, training algorithms and activation function were modified to obtain the best architecture of the neural networks. NNRB with GDX training function, obtained better performance, with an R^2 of 0.99999; obtaining as a result that the models with artificial intelligence techniques have a very high performance to analyze and optimize the efficiency of solar collectors.