| Title |
Day Ahead 24-Hours Load Forecasting Algorithm Using Latest Weather Forecasting |
| Authors |
조세원(Se-Won Jo) ; 권보성(Bo-Sung Kwon) ; 송경빈(Kyung-Bin Song) |
| DOI |
http://doi.org/10.5370/KIEE.2019.68.3.416 |
| Keywords |
Short term load forecast ; Weather forecast ; Solar photovoltaic generation ; Exponential smoothing model |
| Abstract |
Along with the spread of renewable energy, hourly load is further affected by the hourly weather. In the past, the Korea Meteorological Administration provided only weather forecasts for daily maximum and minimum temperatures and daily average amount of cloud. Currently, Dong-Nae forecasts for weather provided by Korea Meteorological Administration are provided for hourly temperature, amount of cloud, precipitation, precipitation probability, direction of the wind and wind speed by region at 3-hour intervals 8 times per day. Accordingly 24-hours load forecasting algorithm using Dong-Nae forecast is proposed to improve the performance. In the proposed algorithm, the effect of temperature to load is reflected using the weekly load sensitivity to temperature per 3-hours, while day ahead load forecasting is performed using the exponential smoothing model. In addition the effect of small solar photovoltaic generation is considered in the proposed algorithm using daytime load sensitivity to amount of cloud per 3-hours. In the case study, the load forecast is performed for the day ahead except special days in 2017. The accuracy of the proposed algorithm was improved by 27.88%, 9.25%, and 9.29% for the overall average percentage error, on Monday, weekday, and weekend, respectively, in 2017 over the overall average percentage error of the algorithm of the exponential smoothing model that reflects the effects of maximum and minimum temperatures. |