Abstract
A method to distribute the temporal emission in a thermal power plant, which is expected to improve the accuracy of the air pollution forecasting system, has been developed. A linear multiple regression formula, which estimates the hourly electric supply and demand energy in the future and which is based on the month, day of the week, clock time and discomfort index as explanation variables, was modeled. The result of the cross-validation during the testing period, not the same as the learning period to build it, indicated that the model is sufficiently reproducible. The hourly electric supply and demand energy in a gas-fired power plant and oil thermal power plant were estimated according to the energy ratio of the generation method, and its temporal distribution coefficient was calculated. Comparing the regression coefficients, the outstanding differences are estimated each clock time in the weekday and holidays, month, day of the week and discomfort index. Comparing the regression coefficients for 9 major electric power companies, the coefficients indicated that each area of the electric power company has its regional characteristic corresponding to its operating district area. Comparing the regression coefficients of the years 2008, 2010, 2012 and 2014 of TEPCO, the coefficients indicated that each year has its annual characteristic and the influence of power savings after the Great East Japan Earthquake in 2011. The results indicated the necessity to consider regional and annual characteristics while distributing the temporal emissions.
