The Research On Spatial Load Forecasting Method Based On CEEMD-PE And Cellular Load Interaction

Spatial Load Forecasting(SLF)is an important part of power grid planning.Accurate and reliable load forecasting are very important to the planning and design of large power network.Compared with previous load forecasting methods,SLF results include not only load size,but also load spatial distribution.The high-precision prediction results of SLF can guide the design and operation of substations and determine the capacity and optimal location of power supply facilities.This paper elaborates the research status of data preprocessing and spatial load forecasting methods at home and abroad,and classifies and summarizes the current SLF methods according to different ways.The characteristic analysis of power load is the basis of SLF.The classification of spatial power load,the influencing factors and the growth rule of load change are studied successively.This paper analyzes the causes of space power load forecasting error and introduces the evaluation method of SLF error analysis.In order to better study SLF,Geographic Information System(GIS)is introduced into the spatial load prediction and the electric power Geographic Information System is established.In view of the problem that singular data exist in the historical load data measured directly,which reduces the precision of the spatial load forecasting results,a method to determine the reasonable maximum value of cell load based on CEEMD and permutation entropy is proposed.Firstly,CEEMD technology is used to decompose the load sequences of each class I cell generated by the power supply range of the 10 k V feeder in the predicted area.Secondly,the permutation entropy algorithm is used to distinguish and identify the high frequency components of the decomposed IMF components,and then eliminate them.Then,the low frequency component and residual component are reconstructed to generate a new principal component,and the maximum annual load of the reconstructed principal component of each class I cell load is determined as the reasonable maximum of the cell load,based on which the spatial load prediction is realized.Finally,the rationality and effectiveness of the data preprocessing method proposed in this paper are verified through the analysis of an engineering example.In view of the fact that the existing spatial power load forecasting methods do not consider the load interaction relationship among the power supply cells in the area to be predicted,a spatial power load forecasting method which considers the acceptability of the standard cell and the influence of the load of the adjacent cells is proposed,and it is mainly applicable to the annual load forecasting of urban power grid planning.Firstly,a quantitative model is established by using the coefficient of variation to reflect the acceptability of the standard cell to the load of adjacent cells.Secondly,the spatial composite weight matrix is generated by the distance factor between each cell and the power load similarity factor in the region to be predicted,and the quantitative model of the load influence of adjacent cells is established by using the spatial convolution to reflect the load influence of adjacent cells on the standard cell.Then,the spatial power load forecasting model is constructed by considering the acceptability of the standard cell to accept load and the influence of the adjacent cells to generate load.Finally,the correctness and effectiveness of the proposed method and model are verified by an engineering example.