Studies Of Load Model And Identification System Of Load Characteristics Based On Virtual Instrumentation

An appropriately accurate representation of load is necessary for power system stability and dynamic Analysis. This thesis studies the important problems in electric load modeling of measurement-based data. They are static electric load based on spline function, the synthesis of dynamic load models based on neural network, and identification system of power load characteristics based on virtual instrumentation. Main contents are as following. 1. Spline function model of static electric loadThe static load model expresses the active and reactive powers at any instant of times as functions of voltage and frequency at slow change, and the static load model is often used in power system computing, analysis and control. For example, power flow, reactive power compensation, voltage stability, frequency stability, and analysis of long terms dynamic process. Comparing to make dynamic load model, it is easier to build static load model.With the development of power system, the wide application of all kinds of power electric equipments and new non-linear load, as a result power load appear considerable changes. With new technology promoting power system development and improvement, the new problems occurred in power system. As the traditional static load model already had great difficulty in describing the static characteristic of modern load, it requires that a new model will be introduced to descript load characteristic better.To facilitate the work in load static characteristic analysis and description, a new idea of static electric load characteristic based on spline function is introduced in this paper. Based on the analysis of the structure for traditional model, the reason of a higher error is pointed in this paper. According to the theory of spline function, this paper presents methods and arithmetic of interpolation of solving the equation or fitting method of optimization based on application occasion and data features.The interpolation method can be obtained by solving the equations group with terminal condition, and coefficient matrix is a tridiagonal matrix. It is convenient to solve model parameter by special tridiagonal matrix method. Use the functionapproximation to solve the problem of fitting method. After transforming the general spline function expression into spline function containing the step polynomial, the spline function containing the step polynomial is considered as fitting function. In solving model parameters, the parameters of model are obtained using the generalized least-square method. In unallocated the node, a layered optimal strategy is proposed. The inside arithmetic is for optimal polynomial coefficient of curve fitting with given the node applying the generalized least-square method. The outside arithmetic is for the optimal the node by using one of nonlinear optimization methods-direct optimization. Therefore, it can realize not only optimize the node but also the approximation function.Test results on typical load reveal that the load model can describe the static characteristic of complex load, this research provide a new effective method to identify load characteristic of unknown structure. 2. The synthesis of dynamic load models based on neural networkIn recent years, more and more researchers have realized the importance of dynamic load models for the power system dynamic performance analysis. As for the model identification is concerned, some researchers suggested the utility of neural network in power load dynamic property modeling, and some valuable achievements have been received. The reason of ANN can be used in dynamic characteristics modeling is that it can approach the object data limitlessly. As for the loads with variable and complicated dynamic characteristics, it's easy for ANN to approach the single record data very well.As the accumulation of the data recorded by load characteristics recorded system, it is inevitable to integrate many kinds of load models into one synthetical load model so as to drive the load modeling form the stage of studying to...