Ant Colony Optimal Design Of Generator Excitation Controller And Model Identification In Neural Network

Synchronous generator excitation control system is one of the important parts of power system control. In the power system normal movement or the accident movement, the synchronous generator excitation control system can reduce the voltage'fluctuation, balance the distribution of inactive power, increase anti-interference and steady operation of the system. The fine excitation control system may not only guarantee the generator operation safely and provide the quality of electrical energy, but also effectively enhance the performance index of the system.Therefore, the optimization of synchronous generator excitation control system has decisive effect to overall power system and the new research has strong use value. On the base of research of mathematic model of idle operation and one machine - infinity bus system operation, the paper firstly established the system state equations, then obtain a set of difference equation of state equation in using Runge-Kutta means, by which the better result can be gotten. This state equation and its solution of one machine - infinity bus system has practical application value in the emulation research and is the basis of the dynamic emulation experiments of excitation control system.To the defect of traditional PID excitation controller, the paper adopt the Ant Colony Algorithm, which has the advantage of the distributed computing and avoiding local optimum, to select the best parameter in the excitation controller, that can improve the performance of controller and increase the operation stability of system. This excitation controller has the well control ability to operating mode of deviating design operating point when optimizing parameter by Ant Colony Algorithm. So it can ensure the safe and stable operation of generator and the system and possess extensive application worth.Because the power system is a huge system and its safety is very important, it cannot carry on the experiment in the actual system, but adopt the simulation experiment to research and design excitation controller and the other devices. In the simulation experiment, it often uses the state equation to substitute the generator operating condition, so it has certain approximate. To this disadvantage, this paper design Neural Net model with sequence structure to identify the actual generator, utilizing the advantage of compulsory simulation and non-linearity of Neural Net. So it can overcome the defect of the state equation and make the simulation effect approach actual operation.A series of simulations experiment indicate, the excitation controller based on Ant Colony Algorithm in this paper has a stronger robustness, compatibility and the better dynamic quality. The Neural Net model with sequence structure has the better identification effect to the actual generator, that may substitute the state equation in the simulation experiment. So that provides possible way for reducing the approximation in the experiment, has practical application value.