Research On Load Modeling Based On Power Quality Monitoring Data

In order to provide data support for the automatic monitoring and control of the smart grid,the establishment of a complete power quality information platform,and the achievement of analysis for deep information and effective management for massive monitoring data have become the trend of development and research.The power quality disturbances carry valid information describing the operating status of the grid and reflecting the characteristics of the load,which can provide data supplement for the measurement-based load modeling,in order to fully utilize existing resources.Because of the characteristics of random time-varying and complexity,the load needs long-term monitoring data for modeling,and the data scale is large.In order to alleviate the problem of collection,compression,storage and transmission of massive monitoring data,compressed sensing theory overcomes various drawbacks of traditional compression modes,reduces the hardware implementation difficulty of monitoring terminals,and makes more rational use of system configuration.The distributed compressed sensing theory solves the compressed sensing problem for multiple measurement vectors with shared support set as the object.The independent measurement and joint reconstruction processing mode is used to fully exploit the correlation between signals and reduce the number of measurements required for accurate reconstruction.This paper proposes the idea of load modeling based on the power quality monitoring system applied compressed sensing theory to deal with disturbance data and carries out research.Firstly,based on the requirements of sample data and the performance of grid monitoring systems,the feasibility and practicability of load modeling based on various types of system monitoring data are compared.Through analysis,the advantages of load modeling based on the power quality monitoring system are obtained,and the pre-processing process for transient power quality monitoring data modeling is summarized.The single-sample modeling of load based on power quality disturbance compressed sampling data is discussed.Based on PSCAD,an IEEE-14-node standard test system was built to obtain power quality disturbance event records by setting up faults.On the basis of analyzing the characteristics of power quality disturbance data as modeling samples,in order to improve the data processing efficiency,a program of piecewise compression sampling and reconstruction based on an improved SP method are designed,which completes data acquisition and processing before modeling.And the performance of the data processing scheme was analyzed through simulation.Based on the reconstructed data,a load modeling data sample was formed and the parameters of the model were identified.The effectiveness of the load model based on the reconstructed data is verified by examining the ability to describe load characteristics.It shows that the accuracy of the reconstruction method satisfies the load modeling requirements,and proves the validity of the data processing scheme.It also illustrates the feasibility of load modeling based on compressed sampling values of power quality disturbance signals.Comprehensive load modeling based on power quality compressed sampling data is discussed.Considering the time-varying of load,the idea of applying distributed compressed sensing theory to solve the comprehensive problem of load characteristics is proposed for the classification and comprehensive of load characteristics program.For data with similar load characteristics,the mixed-support set joint sparse model was used to describe the data correlation,and a distributed compressed sensing program which includes piecewise compression sampling and based on thresholds backtracking mechanism for the joint reconstruction was designed.The reconstruction scheme does not require signal sparsity as a priori condition and reduces the number of measurements required.Through simulation,the performance of joint reconstruction of the load data of the same type is analyzed.Using the data reconstructed by the algorithm,based on the measured response,the load characteristics are integrated and a generic model of the class is established.By checking the load model,it has generalization to a certain extent,that is,the time-varying problem of the load is solved,and the correctness of the proposed model and method for dealing with load comprehensive data is proved.