Parameter Identification And Optimization Of Regenerator Based On Surrogate Model

Horseshoe flame glass furnace is a kind of thermal equipment for glass production.As a typical energy recovery equipment,regenerator plays an important role in thermal field.Energy efficiency optimization of regenerator can save cost and protect environment.How to build and optimize the regenerator model is the key to energy efficiency optimization.At present,the modeling methods of regenerator mainly include experimental method,analytical method and numerical simulation method.Because of the cost of material and labor,the experimental method is insufficient.The analytical method is not only difficult to express the physical process of regenerator accurately,but also insufficient.The numerical simulation method only needs to be carried out on the computer and is easy to realize,which is suitable for building the regenerator model.Aiming at the problems of long modeling time and accuracy of regenerator and energy efficiency optimization of regenerator,this paper proposes a set of process to solve these problems.Firstly,the computational fluid dynamics(CFD)model of regenerator is established.Secondly,aiming at the problem of uncertainty of thermophysical parameters of regenerator,the identification based on Kriging surrogate model is carried out.Finally,aiming at the shortcomings of single agent model,a new combined agent model is proposed.Based on the combined agent model,multi-objective optimization is carried out to obtain the optimal process and structural parameters.The main contents of this paper are as follows:(1)Aiming at the construction of regenerator temperature model,the actual operation of regenerator is simulated based on computational fluid dynamics model.The three-dimensional model of regenerator was established by Solid Works,and then it was imported into ANSYS for meshing and setting boundary conditions,model parameters and material parameters.Finally,the preliminary simulation results are obtained.(2)Aiming at the uncertainty of thermal parameters in the computational fluid dynamics(CFD)model of regenerator,an identification method of thermal physical parameters of regenerator based on Kriging tlbo algorithm is proposed.Firstly,Sobol sensitivity analysis method is used to select the parameters to be identified to reduce the identification complexity.At the same time,Kriging model is used as the alternative model of parameter identification,and the maximum expectation plus point criterion is used to improve the accuracy of the model.Finally,the thermophysical parameters of the regenerator are identified by tlbo intelligent optimization algorithm.(3)Aiming at the problem of energy efficiency optimization of regenerator,considering the deficiency of single agent model and the research status of combined model,an adaptive ensemble surrogate model method is proposed.In the case of evaluating the weighted average coefficient,each model is selected first,and the model with poor effect will be excluded.The selected model uses the weighted average method proposed by acar to calculate the coefficients.In the evaluation of the point by point weighting coefficient,the point by point weighting criterion method considering the local cross validation error and distance factors is used.This method is proved to be effective after testing function.(4)The multi-objective optimization of regenerator regenerative period was carried out.Based on ANSYS fluent simulation platform and combined with the design requirements of regenerator,a combined surrogate model was established with the design variables of heat storage time,inlet velocity,porosity and equivalent diameter,and the performance indexes of heat storage capacity,heat storage rate and heat storage efficiency.Spea-ii multi-objective algorithm is used to optimize the combined agent model of regenerator,and the optimal solution of regenerator process and structural parameters is obtained.