[pp.37-42]
Daisuke KATO, Kanako TANAKA, Michinori KIMURA, Tomoko IWATA, Ryuji MATSUHASHI, and Koichi YAMADA, (Japan Science and Technology Agency / The University of Tokyo)
Abstract:
We have developed a forecasting model for power consumption in the area supplied by the Tokyo Electric Power Company. The model has been constructed based on power consumption and meteorological data from 2011. The model has forecasted power consumption based on weather forecasting data. The forecast information based on the model is available to citizens through the “Networking for Prevention of Blackouts”. For this system, “Center for Low Carbon Society Strategy (LCS)” has collaborated with local governments that can directly network with citizens living in the forecast area. The standard error between estimated and real power consumption during daytime weekdays is 0.85 GW, which corresponds to 2.25 % of real power consumption. In 76.1 % of all weekdays from July 1 to September 9, the differences between estimated and real peak values are within 1 GW. Comparison of parameters between 2010 and 2011 indicates that the component of power consumption depending on temperature decreases about 0.3 GW per Celsius degree, and the component not depending on temperature about 5 GW. The model includes a negative offset for Thursday and Friday of about 0.3 GW. This is caused by the holiday shift by the automobile companies from Saturday and Sunday to Thursday and Friday from July to September of 2011.
Key Word:
power consumption, forecasting model, saving electricity