| In recent years,China’s power load peak is growing rapidly,especially in the east part of China,the load peaks continue to hit new peaks,which results in the imbalance between power supply and demand at the peak load period.Accurate load peak prediction can provide support for the grid to improve the power supply service capacity,and at the same time,improve economic efficiency.In response to the above problems,this thesis proposes a novel power load peak prediction method based on bayesian network and analyze the correlation relationship between the total load and the load of the station area during peak load period.In this thesis,we analyze the daily and weekly load peak characteristics of the total load in Pudong district,Shanghai,and study the peak prediction model.A two-stage method based on the forward stepwise selection of regularization is proposed to select the model factors,and six optimized variables are selected by example verification,the feasibility and effectiveness of the method is verified.A power load peak prediction model based on bayesian network is proposed.Two peak prediction modes,bayesian network load peak prediction based on peak load period and bayesian network load peak prediction based on k-means clustering are constructed.The two models are used to predict the daily and weekly load peaks by Shanghai Pudong power load data.The prediction results of the two models have great advantages in peak time prediction accuracy and model training time.The feasibility and effectiveness of the Bayesian network-based peak prediction method are verified.Considering the similarity between the load of the station(including the common station area and the special transformer area)and the total load during the peak period,and the contribution rate to the total load.A load correlation analysis method based on quadratic clustering between peak load and station area is proposed.The load data of 8954 stations in Pudong district of Shanghai is taken as an training data to test the performance of the algorithm.According to the order of the typical load curve after the second clustering,it can be considered as a kind of peak-shift scheduling priority.The feasibility and effectiveness of the proposed method are verified. |
