Research On The Modeling Of Boiler Turbine Coordination System Based On Dynamic Fuzzy Neural Network

The coordinated control system is the core unit of the power plant automation control system.Its control performance directly affects the safe and stable operation of the unit.Accurate mathematical model is the foundation of control algorithm design and performance research.Therefore,it is very necessary to establish the model of the coordination system of the machine furnace.With the wider application of thermal power units with larger capacity and higher parameters,the coupling and nonlinear characteristics of boiler turbine coordination system are stronger,which makes it difficult to establish exact mathematical models by traditional mechanism analysis.In recent years,an intelligent modeling method based on data sequence has become one of the main hot spots in the research.This paper puts forward a fuzzy neural network algorithm is a dynamic network structure with dynamic training process,based on a thorough study of the learning algorithm,according to multi parameter preset algorithm,fuzzy rule extraction is difficult to understand the shortcomings,put forward dynamic elliptic basis based on fuzzy neural network,fuzzy completeness of online distribution to avoid the random initialization parameter selection,and according to the Gauss function of input variables of each rule width on the performance of the system with the number of online correction of membership function width.The algorithm is applied to the identification of nonlinear systems.According to the MATLAB simulation results,the algorithm is superior to the dynamic fuzzy neural network in terms of performance and learning speed.Finally,based on the analysis of the boiler turbine coordinated control system static and dynamic model of the system,the nonlinear model of the simulation data and the field of industrial measurement data based on the application of improved basis functions(ellipse based EBF)dynamic fuzzy neural network algorithm for training and testing of the model.The simulation results show that the fuzzy neural network learning ability and generalization ability of strong dynamic ellipse based coordinate system established by the three in three out the non parametric model has high precision,can reflect the dynamic characteristics between the input and output of the right,to achieve the desired results.It provides an important basis for the subsequent design of model based control algorithms.