기상에 대한 동네예보를 활용한 제주도의 특수일 전력수요예측 알고리즘

초록

It is necessary to increase the accuracy of the load forecast by reflecting the characteristics of the load in Jeju and the effect of temperature on load when performing load forecast for Jeju special day. In order to the accuracy of load forecast, Jeju special day load forecasting algorithm that reflects the effect of temperature on load by using the Dong-Nae forecast provided by Korea Meteorological Administration is proposed. The proposed algorithm uses a fuzzy linear regression model to perform 24-hour special day load forecasting. Then, the load sensitivity for temperature per 3-hour for the special day is calculated using the load on the special day normalized by the basic load and temperature. Using load sensitivity for temperature, the effect of temperature on load is adjusted by the difference between the four days of weekday before special day and special day of the past three years and by the difference between the four days of weekday before the forecast day and forecast day. Case studies were performed for the proposed algorithm from 2014 to 2016. The accuracy of the proposed algorithm is improved by 41.51% from 5.42% to 3.17% on MAPE.

키워드