ANovelCloudTheoryBasedTime-series Predictive MethodforMiddle-term Electric LoadForecasting

Themiddle-term electric loadforecasting isan existing difficult workandoftenhasa largeerror.To address theproblem, thispaperproposes a novelcloud theory basedtime-series predictive methodformiddle-term electric loadforecasting. Inthismethod, thetimeseries of daily maximumloadispartitioned into twoparts, historical dataset and currenttendency dataset, backwardcloud algorithm isapplied tothetwo datasets toformthe historical cloudand the currentcloud,and the corresponding rulesets aremined. Thenthehistorical cloud andcurrent cloudisintegrated tocreated predictive cloud through synthesized cloud. Finally, viacloudreasoning, the forecast result canbeobtained. Thispredictive method effectively integrates quasi-periodical regularity andcurrent tendency oftime-series data, andhasa simple computing model.Thecasestudyshowsthattheproposed methodis accurate andpractical.