The Theory Research Of Wavelets Neural Network And Its Application In Boiler Superheated Steam Temperature Control

This paper points out merits and deficiencies of frequency domain analysis method based on transfer function, frequency characteristics and root-track and state space analysis method adopting time domain analysis by the overviews of complex system control problems. On the basis of these above discussions, possessive application value and practice feasibility of intelligent control theory and method of complex control systems, controlled objects and processes that their models are uncertain are set forth. Due to it is very difficult to obtain satisfactory performances when those complex thermal control processes with nonlinear and variability are controlled by some general control methods, an intelligent control thought and strategy based on wavelets neural network is used in the boiler superheated steam temperature control and satisfactory performances are gained. Then this does some researches and efforts to a certain extent to promote practice application of WNN in control domain.The Wavelets neural network is a new network model, which is not only provided with favorable time-frequency localization characteristics but also combined with self-study and self-organizing functions of common neural network. It preferably overcomes blindness of the structure decision and design of the previous neural network. Especially, it possesses of speediness, veracity and convergence in nonlinear function approximation. Consequently an effective select is put forward to solve those problems of complex system modeling and self-adaptive control with uncertainty, high nonlinear and large time-delay. According to the points, this paper combines merits of internal model control, such as simple design, good regulation capacity, high robustness and the ability to eliminate the unknown disturbance etc. to construct a internal model control system based on wavelets neural network, which is characterized by high robustness and quick response speed and then it can still bring good control performance when controlled objects vary in a wide range. Finally, it is triumphantly used in simulation of the 1st superheated steam temperature reduction control system of 500MW units and good performances are gained.