Research On Mid-long Term Load Forecasting Based On Load Characteristic Analysis

Mid-lo ng term load forecasting res ults are the basis of power syste m p lanning,and are the premise of carrying o ut suc h work as power bala nce, and power structureadjustme nt. As the guarantee o f re liab ility a nd ratio na lity o f power grid operation,promoting mid-long term load forecasting is he lp ful to optima l resource a llocation,fue l pla n, and arrange me nt of operation. And it also has important theoretica lsignificance and practical value for the power grid planning department.As an important work in the power syste m, load characteristic ana lys is he lpsforecasters accurate ly grasp the c hange rule of load curve to ga in ideal forecastresult. Based on the statistica l results of mid-long term load characteristicparameters in a n area, the re lations hip between load characteristic index andinflue nce factor was ana lyzed in this paper. And the n the regularity of loadcharacteristic index was s ummarized, which prepares for mid-lo ng term loadforecasting work.Compared with s hort term load, mid-long term load has more co mp licatedinflue nce factors, longer time span a nd more uncerta int ies. This paper proposed adominant factor identification method based on the research o f mid-long term loadcharacteristic ana lys is, whic h was comb ined with par tia l least squares and applied tomid-lo ng term load forecasting. Fro m the a ngle o f load characteristic index, themethod chose benc hmark by the result o f load ana lys is. And the corre lation betwee ninflue nce factor and benc hmark was ca lc ulated by Pearson c orrelatio n a na lys is toide ntify the lead ing factor in the power load. By mea ns o f calc ulatio n exa mp le, wecan come to the conc lus io n that this me thod can red uce the error caused bysubjective experie nce of forecaster and enha nce the practicability o f mode l. At thesame time, the mode l witho ut irre le va nt factors had more adva nta ge in e xtractingmain component and obtaining predicted result.An electric power de mand forecasting and load characteristic ana lys is platfor mwas deve loped for the long term deve lop ment o f electric power enterprise. Takingthe techno logy of power pred iction a nd data mining as the core, the syste mdeveloped with Java platform a nd Oracle database was c losely co mbined withcomputer network techno logy, communicatio n tec hno lo gy, informa tio n sec uritytechno lo gy and inte llige nt tec hno logy. In add itio n, the syste m assoc iated with the actua l needs of grid work ca n effective ly impro ve the accurac y o f load forecastingand provide technical support for power grid planning.