Research On Compound Modeling Of Boiler Side Core Equipment For Supercritical Units

At present,supercritical and ultra-supercritical DC boilers with large capacity,high temperature and high pressure are widely used.Correspondingly,more complicated thermal system structure and more precise process control require field personnel to have considerable knowledge and excellent operation level.The simulation machine,as a powerful platform for training operators,it is of great significance to develop a fast and effective algorithm model for thermal system.Based on the principle of establishing a simple,applicable and portable thermodynamic system model,this paper focuses on the modeling process of boiler side core equipment.The classical simulator model mostly uses mechanism model.Because this method uses conservation law,it can describe the internal mechanism and characteristics of the system in depth.But there are many parameters in the model that are difficult to determine.The neural network has strong approximation ability to the non-linear system and can determine the unknown parameters in the mechanism model.Thus,this paper establishes a composite model consisting of mechanism model and neural network model for predicting unknown parameters.Firstly,the input and output variables of each system and the main unknown parameters affecting the system characteristics are determined.Based on mass conservation,momentum conservation and energy conservation,the relevant mechanism models of coal mill,furnace combustion system,water wall and separator are established.Then,the number of input layer,hidden layer and output layer of each neural network is analyzed by the number of relevant input and unknown parameters of each system,and the structure of the neural network model is determined by using SPSA as the weight adjustment algorithm.Finally,the two are combined to establish a composite model based on the mechanism model and the neural network with unknown parameters in the prediction mechanism model.The coal mill system and the water wall system are taken as examples for detailed description.Using the historical operation data of 1000 MW supercritical unit of Guodian Jianbi Power Plant,the composite model of each system is simulated on MATLAB platform through typical working conditions such as lifting load in different time periods.The simulation results show that the trained neural network model can well predict the unknown parameters in the mechanism model,and the compound model of the system can well simulate the typical working conditions in the field.In addition,the composite model composed of mechanism model and data model has the advantages of strong universality,easy use,good expansibility and openness,so it isnecessary to promote the application research.