POWER DEMAND PREDICTION METHOD AND POWER DEMAND PREDICTION PROGRAM

PROBLEM TO BE SOLVED: To realize a power demand prediction method that can predict power demand with a high degree of accuracy in a short time not by means of intuition and experience.SOLUTION: In a power demand prediction method that predicts power demand for a predetermined time unit by multiple regression analysis, the power demand prediction method makes an objective variable be a power load, uses at least temperature, square of the temperature, a day of week, a busy period, a latest weeks load average value, and a latest weeks temperature average value as explanatory variables, divides an objective variable=a+b*X1+c*X2+d*X3+e*X4+f*X5+g*X6 as a basic arithmetic equation according to a weekday and non-weekday condition, a temperature boundary condition and a time zone condition to create a plurality of prediction formulas, inputs actual values for past several years into the objective variables of the plurality of prediction formulas and the explanatory variables X1-X6, and after obtaining values of a constant a, and coefficient b, c, d, e, f, g corresponding to each prediction formula, assigns the prediction target explanatory variables X1-X6 in the corresponding prediction formula to calculate a prediction value of the power demand.SELECTED DRAWING: Figure 2 【課題】人の勘と経験によらず短時間で高精度に電力需要を予測することが可能な電力需要予測方法を実現する。【解決手段】重回帰分析によって所定時間単位で電力の需要を予測する電力需要予測方法において、目的変数を電力負荷とするとともに、説明変数として、少なくとも、気温と、気温の２乗、曜日、繁忙期、直近数週間負荷平均値、直近数週間気温平均値を使用し、基本演算式として、目的変数＝ａ＋ｂ*Ｘ1＋ｃ*Ｘ2+ｄ*Ｘ3＋ｅ*Ｘ4＋ｆ*Ｘ5＋ｇ*Ｘ6を、平日と平日以外の条件と、温度境界条件と、時間帯条件とに応じて分割して複数の予測式を作成し、該複数の予測式の前記目的変数と前記説明変数Ｘ1〜Ｘ6に、過去数年分の実績値を入れて、各予測式に対応した定数ａおよび係数ｂ，ｃ，ｄ，ｅ，ｆ，ｇの値を得た後、予測対象の説明変数Ｘ1〜Ｘ6を対応する予測式に代入して電力需要の予測値を算出するようにした。【選択図】図２