Моделирование производительности солнечных концентраторных модулей на основе алгоритмов искусственных нейронных сетей

Abstract

Since the performance of solar power plants is quite variable, the dependence of production on weather conditions significantly increases the need for accurate forecasting. At present, the formation of a new approach to the development of predictive models of the performance of solar power plants based on artificial neural network algorithms is acquiring special relevance. The advantages of artificial neural networks in forecasting, such as the ability to learn and take into account a set of parameters that are not in a functional connection, make it possible to successfully use them in the development of models of the performance of solar power plants. The analysis of existing developments is carried out and promising areas of application of artificial intelligence algorithms in solar energy are determined. To simulate the performance of a solar concentrator module, a two-layer artificial neural network with sigmoid hidden neurons and linear output neurons has been developed. The developed performance model of a solar concentrator module based on an artificial neural network makes it possible, with a significant approximation, to determine the thermal efficiency of a solar module depending on various external conditions and operating parameters.