Research On Residential Electricity Consumption Behavior Based On Load Clustering And Mid-long Term Prediction

In recent years,the country’s accelerated adjustment of the economic structure has significantly improved the living standards of residents,and the proportion of electricity load has also continued to rise.The load curve has shown some new changes.Therefore,scientific and effective analy sis of electricity consumption behavior is necessary for the construction of power grids.Power user classification is the basis of many power industry applications.Efficient user classification is of guiding significance for the formulation and application of policies such as time-sharing electricity prices and peak shifting and valley filling.The load forecast is the basic link of power grid planning,and its accuracy is also closely related to the planning and construction of the power grid and the social and economic benefits of operation.This paper studies residents’ power consumption behavior from the aspects of power user classification and medium and long-term load forecasting.Aiming at the massive data of residents’ electricity consumption,this paper adopts a clustering method based on characteristic index dimensionality reduction technology.Firstly,the data of residents’ electricity consumption is used to clean the data,and seven load characteristic indexes are extracted.Then,the principal component analysis method is used to reduce the dimension.Finally,the fuzzy C-means clustering algorithm is used to cluster the power users.This paper not only verified the method from the simulation laboratory.The feasibility of the model is verified by using two different datasets,namely the big data of residents’ electricity consumption in various cities in Shaanxi Province and the big data of specific household power users.Experimental results show that the method has high classification accuracy and speed,which greatly optimizes the operation time and accuracy of the clustering algorithm.Compared with short-term load,medium-and long-term load have complicated factors,long time span and many uncertain factors.In this paper,the identification method of the dominant factor of power load is used,and it is combined with cultural genetic algorithm for annual electricity forecast.Pearson correlation analysis was used to discriminate the correlation between the influencing factors and the power load,and the dominant factors affecting the residential power load were screened.Combined with the LMDI index decomposition method,it is possible to effectively quantify the influence weight decomposition of dominant factors.This method effectively solves the prediction model deviation caused by the subjective experience of the staff,greatly reduces the interference of irrelevant factors,and further improves the accuracy of the prediction results.After selecting the dominant factors based on the above studies,a multiple linear regression prediction model is performed,and the model is optimized based on the cultural genetic algorithm to perform medium and long-term load prediction.The cultural genetic algorithm improves the performance of the genetic algorithm from the aspects of convergence speed and convergence efficiency,and incorporating the genetic algorithm into the cultural algorithm framework overcomes the defect that the traditional genetic algorithm is prone to fall into premature convergence.Through cultural genetic algorithms,iterative optimization is performed,and finally an optimized multiple linear regression prediction model is obtained,which improves the prediction accuracy of the load prediction model.The example verifies the accuracy and effectiveness of the medium and long-term load forecasting model based on the cultural genetic algorithm after identifying the dominant factors.