Forecasting of energy consumption with artificial neural networks using CUDA technology

The existing methods of forecasting energy consumption are analyzed. The choice of the method of forecasting with the help of artificial neural network is explained. The application of parallel calculations on GPU is described. The paper considers a new algorithm of forecasting energy consumption based on the theory of artificial neural networks using CUDA technology to optimize forecasting energy consumption at an industrial enterprise. For practical usage of the new algorithm a software for Windows is designed, it is programed by C++. The experiment is made; on its bases the analysis of software operation and calculation of the consumed resources are carried out. In accordance with the results of the experiments, the developed parallel algorithm has reached the required accuracy of forecasting for a short period of time. The application of the algorithm will allow the enterprises to get more accurate forecasting and reduce the costs concerning electricity payments.