Font Size: a A A

Study Of Dynamical Characteristic Of Load Modeling Based On WAMS

Posted on:2009-02-22Degree:MasterType:Thesis
Country:ChinaCandidate:L NiuFull Text:PDF
GTID:2132360245496353Subject:Power system and its automation
Abstract/Summary:PDF Full Text Request
With the extended scale and increased complexity of power grid, digital simulation of power system become the major means for layout, planning, running, and analysis, using different models would lead to different simulation results. Especially the huge changes occurred in modern system's load structure, modeling work becomes much more difficult, so the models face the flinty challenge. In order to meet the simulation needs, new load model must be found.At present, the measurement technology and computer technology has rapid development. Aim at the actuality of load modeling, this paper present a new method which uses wide-area measurements and support vector machine (SVM) for load modelling, and found a sort of new model. The three main contents of this paper are as follow:Firstly, paper makes a summary of load modeling on development and actuality, and introduces the wide-area measurement system (WAMS) and SVM theory which are used for modeling load. Based on WAMS actual time dynamic monitoring and SVM, paper presents a new method of load modeling and a new sort of load model.Secondly, modeling load with SVM and identifying the parameters of load model. Paper use SVM for non-linear regression analysis. By using Gauss Radial Basis Kernel Function, SVM mapping input space to a eigenspace, then the non-linear issue of input space changes to linear issue in eigenspace. Through linear regressing in hyper-dimension space, the non-linear function is get. Through its three levels, Bayesian Framework can infer the value of four parameters respectively, and then a kind of nonmechanical model is founded.Finally, paper makes simulation by using the actual measurements of a 220kV transformer substation in Shandong Province. In the simulation, paper selects voltage and frequency as input variable; active power and reactive power as output variable. By changing the structure and parameters of model, paper proves the validity of this method in choosing model structure and optimizing parameters, moreover, analyze the generalization ability, flexibility of structure, calculate speed and so on.The simulation results verify the validity of this method, and this model can describe the actual load more exactly.
Keywords/Search Tags:Load modelling, Support Vector Machine (SVM), Wide-area Measurement System (WAMS), Non-linear systems identification, Bayesian Evidence Framework
PDF Full Text Request
Related items