Classification And Synthesis Of Consumer Characteristics In Load Modeling And Analysis Method Of Substation Composition Characteristics

With the further development of smart grid in recent years,more and more Extra High Voltage(UHV),new energy sources and different kinds of load were connected to the power grid,which posed a certain challenge to the safe and stable operation of the power grid,and put forward higher requirements for the intelligent dispatching of the power grid.Load in power system has the characteristics of time-varying,diversity and dispersion,which makes it hard to grasp the characteristics of the composition of synthesis load in power system,resulting in a certain deviation of load modeling model parameters.This will have a negative impact on power system dispatching,operation and simulation experiments.At present,in power system simulation experiments,the general load model adopted has the characteristics of delay and error,which is hard to reflect the actual load situation.So,it is impossible to get the real-time and accurate simulation results based on the adopted model.In view of the above problems,this paper proposed a set of analytical method of synthesis load composition characteristics based on big data,which provided theoretical support for online modeling of synthesis load in substation.Firstly,aiming at the problem of low clustering quality and robustness of clustering algorithm adopted in load modeling,a similarity measurement method of the algorithm was innovated,and a load curve clustering method based on Euclidean-Dynamic Time Warping distance and Entropy Weight Method was proposed.Dynamic Time Warping(DTW)distance and Euclidean distance were combined to determine the similarity of curves comprehensively.And Entropy Weight Method was introduced to adaptively configure the weight cofficients of these three characteristics.Finally,K-means algorithm was applied to cluster the load curves.Although high-quality clustering was achieved,the efficiency of the algorithm is low.Then,aiming at the low efficiency of the algorithm,a new algorithm of load curve clustering based on Piecewise Linear Representation(PLR)and DTW was proposed.The algorithm combined the characteristics of adaptive dimension reduction of PLR with the advantage of DTW which can measure the similarity between time series of different dimension,so as to extract the feature and measure the similarity of the clustered curves.Finally,the clustering analysis was carried out by CK-means to achieve the optimal comprehensive performance.A load composition characteristic analysis method based on the Least Square Method was proposed.And the mapping relationship between the clustering results obtained by the above clustering method and the daily load curve of the substation was established to obtain the industry composition proportion of the synthesis load in substation.Based on the above theoretical research,an automatic online substation synthesis load modeling platform based on big data was realized.The model parameters of substation synthesis load of each area in certain province could be obtained by the platform automatically at fixed time,which made some breakthroughs in the accuracy and practicability of load modeling.Finally,based on the above theory,this paper built an online automatic modeling platform of substation synthesis load based on power big data,and presented the key research of the paper through the platform.