The Research On Identifying Algorithm And The Development Of Modeling Platform Of Power Aggregate Load

Load modeling is one of the most difficult problems in the power system analysis. For a long time, the research of load modeling has lagged behind the development of power industry. Based on studying the load modeling theory comprehensively, the author validated the dynamic aggregate admittance model and analyzed its adaptability. In addition, the author identified the 3-order induction motor load model with the fault data from the experiment based on same load and the field and found that the 3-order induction motor load model has four characteristics: high ability to describe itself, high ability to extract the internal and essential property of synthetic load, high ability to describe the active power and reactive power synthetically and fine interpolation and extrapolation. Subsequently, the author qualitatively analyzed the parameter-dispersing question existing in the load modeling, and two parameters with low sensitivity or say in another way with little influence to identification results: n and a were found. At the same time, the author proposed that the question of parameter-dispersing could be reduced by improving the identification validity.Genetic algorithm is one kind of new-type optimization algorithm. It often can find the overall solution of the optimizing question with greater probability when the traditional optimizing algorithms are powerless. In recent years, because of the enormous potentiality of genetic algorithm and its successful application in many fields, the extensive concern has been received. The dissertation validated the outstanding performance of genetic algorithm compared with traditional optimization algorithm by the test data in the beginning part of the third chapter. But this algorithm itself has some defects; an outstanding problem is the contradiction between the convergence speed and the overall convergence property. In order to cushion these contradictions, this dissertation has put forward an improved comprehensive genetic algorithm. The improved genetic algorithm included three major tactics: the elitism strategy, the two-discontinuity point crossover strategy avoiding the inbreeding and the adaptive immigration strategy suppressing precocity and maintaining the population diversity. Through test function, simulation load data, test data and data coming from the field, the superiority of the improved genetic algorithm was checked up omni-directionally. The standards of measurement included algorithm convergence speed, the true degree of identification results (It is proved by simulation data.) and the parameter stability of results. The author utilized the improved genetic algorithm to model and solved the dispersed problem of identification results well, meanwhile, the modeling results showed that the identification results of the test data by the improved genetic algorithm are inosculated well with the investigation result. It illuminated not only the validity of the improved genetic algorithm but also the correctness of the investigation result. The two proved each other. Based on the deep comprehension and study to load modeling, the author established aload modeling software platform. The software platform based on parameter...