Year of Graduation
Development of Neural Network System for Prognosis of a Thermal Regime for the Spacecraft "Spektr-R"
In the paper, the problem of predicting the temperature parameters of the "Spektr-R" spacecraft is considered. To solve this problem, it was decided to use the apparatus of neural networks. As a result of the research, a model of a multilayer perceptron of direct propagation is constructed. Numerous studies have been carried out to find learning parameters to reduce the resulting error. The method of back propagation of the error was used as the learning algorithm. As a result, software based on this model was implemented, tests were carried out on real data. An analysis of the results obtained showed that more research is needed to improve the accuracy of the model.