The Research Of The Computer Excitation Intelligent Controll System Of Asynchronized Synchronous Turbogenerators

Asynchronized synchronous generator is a new type of generator whose performances are significantly better than those of conventional generator. Its appearance shows a new way to resolve the problem of continuous over-voltage on extra-high voltage electric network and the stability of power system with long transmission lines.On the grounds of theory of asynchronized turbogenerator (ASTG), based on the practice of the analog asynchronized generator, this dissertation designs and debugs the computer-controlled system, and investigates detailedly the intelligent excitation control method.This dissertation firstly expounds the basic principle and electric structure of ASTG, and analyzes the principle of excitation control. Because in actual system the excitation current is always the control target, the mathematical model with excitation-current controlled, which can be referenced by simulation of the same type system, is put forward.For the measurement of analog signals such as voltage and current of ASTG, a filter card for pre-disposing is designed, in order to improve real-time performance, algorithms for AC analog sampling is simplified, and a compensation algorithms for phase error due to sequence-sampling is brought forward. For the measurement of impulse-width signals such as rotor-speed and rotor-position-angle, an intelligent interface card based on ISA bus is designed. All programs for measure and control based on C++ are compiled and debugged and the flow for system debugging is summarized. The running-states indicate that the system's performance is basically met the demand of design.Multiple-targets control strategy for active power channel, which extended the item-amount of controller's formula and help to enhance the controller's performance, is brought forward, after analyzing thoroughly the effect of controller's parameter, an approach on how to debug controller's parameter is concluded. The running of both simulation and practice system indicate these strategy and method are feasible. Based on the multiple-targets control strategy, utilizing the properties of self-adapting and simpleness of nerve cell, multiple-targets neural control strategy is brought forward, which facilitates the on-line regulation of controller's parameter. The simulation indicates this control strategy has remarkable self-adaptability and robustness character.　A strategy of on-line optimization based on genetic algorithms (GA), which utilizesthe global convergence of GA, is put forward to overcome the difficulty of specifying the neural controller's parameter, avoiding the problem such as local optimum when using BP algorithms. The optimum selection scheme is applied as GA operation, which enhances the convergence speed, the course's stability and result's dependability by comparing with conventional scheme.