일별 기온 민감도와 가중치를 적용한 지수평활법 기반 봄철 평상일 하루전 전력수요예측 알고리즘

초록

This study proposes short-term load forecasting algorithm using an improved exponential smoothing method to improve the accuracy of day-ahead load forecasting in spring. The proposed algorithm defines the range of temperature insensitivity in spring and calculates daily sensitivity of hourly temperature. Daily exponential smoothing coefficients are optimized to minimize forecasting errors. Case studies were performed on the proposed algorithm to calculate the forecast error of day-ahead load forecasting in the spring of 2022 and 2023. The proposed algorithm showed an improvement of the average prediction accuracy in spring in by 12.85%p for two years compared to the exponential smoothing and LSTM algorithm errors of the short-term load forecasting S/W of the Korea Power Exchange.

키워드