Artificial Neural Network Control In Power System Applications

In recent years, with the high-speed development of modern control theory and digital computer technology, many advanced control algorithms appears, such as generalized predictive control (GPC), internal model control (IMC), etc, they had been applied in industry. But as far as it goes, PID controller is still play a leading role in process control for its simple structure, high robust, good adaptability and can do effective control on many industrial objects. But the practical industrial production is sometimes nonlinear and time-varying, so it's difficult to establish its precise mathematic model. Consider from control, to find a kind of control theory which is high dependable, robust, good adaptability, intelligent and easy to be realized to meet the complex system's high dimension, nonlinear, uncertain, difficult to modeling is the goal of control. Thus, design a self-adaptability, self-learning intelligent neural controller is significative to improve the control effect and robust of such system.Artificial neural network had been applied in the identification of nonlinear to simulate the practical system's input/output relation because of its ability to approach any nonlinear mapping. Using the self-learning, self-adaptive and parallel process ability of neural network we can easily get an easy to be realized algorithm, and get the model of the dynamic system by training.Different from the traditional nonlinear identification method, NNI is not restricted by the nonlinear model. It get the nonlinear mapping to describe the relation of the system's input and output by learning the controlled system's input and output data. It can get an output from an input without knowing the mathematic relation of input and output. This is a noticeable new method of nonlinear system identification.This thesis researched the models and the learning algorithm of neural network, raised a main steam control system of power station based on neural network. Following is the main content.(1) Analyzed the characteristic of the neural network models and their apply range in intelligent control.(2) Analyzed the shortage of traditional PID control and raised an intelligent control system based on neural network. Chose BP network as the neural network of the NNI in the neural control system, the neural network of NNC adjust the control parameter of PID controller online according the system's status to optimize the system's capability, it take the Hebb rule as the learning algorithm.(3) Chose the main-steam temperature of power station as the control object, control the temperature using the algorithm proposed in this thesis, the simulation shows this controller has a good control ability, can solve the conflict between speediness and stability effectually.