Desarrollo de un modelo con técnicas de inteligencia artificial para medir la eficiencia de un colector solar de tubos al vacío bajo la irradiancia de la ciudad de Riobamba.

Abstract

In this work, the multilayer perceptron (MLP) artificial intelligence models and the neuro-diffuse adaptive inference system (ANFIS) are used to evaluate the instantaneous efficiency of a stable-state vacuum tube solar collector, which has been installed in the city of Riobamba. Efficiency plays a crucial role in the design of solar thermal collectors and the factors that influence their performance are several, among them we have solar irradiation, ambient temperature, outlet temperature and hot water consumption, these experimental and meteorological data They are obtained taking into consideration the climatic conditions of the city of Riobamba, all the information is normalized and classified for the training and validation of the MLP and ANFIS models. The results of the multilayer perceptron model (MLP) were compared with the results obtained by the neuro-fuzzy adaptive inference system (ANFIS) using the integral criterion of the square of the error, the root mean square error and the coefficient of determination. Various types of training algorithms such as backpropagation, Levenberg-Marquardt (LM) and hybrid are used by the models. The values predicted by these models are closely related to the experimental data obtained. The results show that the MLP model has a higher accuracy compared to the real data calculated R2 = 0.99215, RMSE = 0.0101 and MSE = 0.00019 with respect to ANFIS with R2 = 0.7742, RMSE = 0.0702 and MSE = 0.00492. It was evidenced in the results that these models are an effective tool for the study of solar thermal energy systems and allow us to optimally evaluate the performance of a vacuum tube solar collector.