The Research On Identifying Algorithm And Model Structure

An appropriately accurate representation of load is necessary for power system stability and dynamic analysis.This article first points out that choosing reasonable model structure is the only identifying condition. Based on the analyzing load model structure, author puts an improved model structure; it has clear physics meaning and better insert extrapolate characteristic, moreover, it can improve parameters' stability in a certain extent.Based on studying the load modeling theory comprehensively, first points common traditional optimization algorithms that are often used in load modeling causing large parameter-dispersing, so the author choose genetic algorithms which have the overall situation search ability as the identification algorithm, it often can find the overall solution of the optimizing question with greater probability when the traditional optimization algorithms are powerless. Aiming at the shortcoming of basic genetic algorithm as slow rapidity of convergence and easy to precocity, the author obtain well-chosen random-elite strategy in colony choice, avoiding close relative propagation in two-point crossover and self-adaptive modulating strategy of crossover and mutation probability, then present a synthetically improved genetic algorithm and apply it to power system aggregate load modeling. The practical modeling based on the field measured data from power substation proves that this improved genetic algorithm can effectively ameliorate colony multiformity and overcome precocity in evolution course. It has great effect in shortening convergence time, improving model precision, conquering the decentralization of parameters, and is an excellent optimum arithmetic and fairly fits power system load modeling.On this basis, the authors systemically investigate the operation mechanism of the genetic algorithm, analyze the different search ability of genetic operator; point out that the key factor of GA performance is population's variety, obtain the related restraint between population's variety and algorithm parameter; provide the settlement rule of the genetic parameter on analyzing from the theory, thoroughly research the population scale, crossover and variation and the control strategy, as well as the initial parameter range etc genetic algorithm key operation parameter how influence the algorithm performance rule, put forward reasonable population scale, initial parameter range and a self-adaptive modulating strategy of crossover and variation probability. The result of study indicates that it is the key to excavating GA enormous latent energy that the rational parameter is made up, can improve...