Daily electrical load forecasting in power supply systems on the basis of fuzzy neural networks

In this article the approach of daily electrical load forecasting, based on fuzzy neural networks is proposed. This approach models behavior of load on those areas where it is primarily a function of temperature. Forecasting was carried out separately for the working days and for the weekend days. As the initial information daily archival records of developed capacity, temperatures of air and an overflow of the electric power have been used. And these data were set as the minimal and maximal value. The created programs have been tested and determined mistakes of forecasting which have made no more than 2.5% for the working days and 2% for the weekend days.