Abstract:

In the framework of the study, a melty forecast model developed using ARIMA and artificial nerve networks for short-term energy forecast processes that have an important place in energy management systems has been revealed. In the framework of this model, the reliability of the system has been tested with real-life data obtained from a company that produces the tobacco industry. According to the results obtained from the ARIMA method, some of the factors affecting electricity consumption have been added as input data to the artificial nerve network model. Considering the correlation between solar energy production, working hours, production amounts and past electricity consumption data in the data sets, the best system for the desired company is designed to be planned by testing the various number of neurons and different training algorithms. Based on the information obtained, the melting forecast model provides an improvement of 39.9% compared to the forecast data made only with the ARIMA model.

Keywords: