Scientific journal
Scientific Review. Technical science
ISSN 2500-0799
ПИ №ФС77-57440

Personal portfolio

SHORT TERM LOAD FORECASTING BY USING LEAST SQUARES SUPPORT VECTOR MACHINE THEORY

Nadtoka I.I. 1 Al-Zihery Balasim M. 1
1 South-Russia State Technical University (NPI)
Short-term forecasting of daily schedules hourly electrical load is an important basis for reliable and efficient operation of the electricity system. Accuracy of the forecast electricity consumption directly affects the quality of supervisory control and power supply reliability. Thus, the selection of an appropriate load forecasting method to improve prediction accuracy is of practical importance. The paper presents results short-term forecasting electricity consumption in the operational area of the regional supervisory control using a regression model based on the theory of support vector machine (SVM). Use a modification of the least squares support vector machine (LS-SVM). In the predictive model accounted for statistical and forecast data of air temperature and natural light, affecting the power consumption. It is shown that a significant effect on the accuracy of prediction is supported by two parameter model LS-SVM, chosen empirically.