| The reheating furnace is one of the key equipment in the steel industry production line.The optimal setting of the furnace temperature of the furnace will determine the quality of the furnace temperature control.Now,the existing studies discuss the furnace temperature optimization control problem in some special steady state(SS)working operations.In practical production,the reheating furnrnace will often be under the dynamic gauge change(DGC)working operation,and some working operation of delays.These phenomena are common but less studied by most researchers.In the past,when encountering such problems,most of them rely on manual experience to manually adjust the furnace temperature setting value,it is difficult to achieve dynamic optimization and adjustment of the control system.This may result in problems in production efficiency and product quality.Thus,many researchers have tried to solve the online dynamic optimization problem of the reheating furnace,thereby improving the heating quality of the heated billet,so the slabs can reach the desired temperature at the discharging time,and the energy consumption of the furnace during the production process can be reduced.In this paper,an online optimization control system for the reheating furnace based on the double-layered model predictive control method is proposed to solve the on-line optimization problem,which integrates of both the delays and the dynamic gauge change working operation.Then,we give some research achievements on problem description,modeling and online optimization algorithms,listed as below:1.The reheating process of the slabs in the reheating furnace has the characteristics of large number of variables,non-linearity and unmeasured slab temperature.Therefore,the forecast of slab temperature in the furnace is complicated,and many parameters(radiation,convection,skid and so on)need to be considered.Therefore,there is a great difficulty in constructing the temperature prediction model for slabs.Firstly,this paper analyzes and describes the heat transfer mechanism between the heating furnace and the slabs,comprehensively consider both the uneven heating and the effect of water-cooled skid mark on the heating quality of the billet,and then establish the three-dimensional temperature prediction model of slabs with the skid boundary conditions.Secondly,based on the total heat exchange factor method and the established three-dimensional temperature prediction model,the one-dimensional online temperature prediction model of slabs is proposed to meet the requirements of on-line optimization of furnace temperature for slab temperature calculation time.Thirdly,in order to achieve correction of the simplified temperature prediction model,inverse problem identification is proposed to identify the values of the total heat exchange factor in the simplified model.Based on the black box measurement data,we use the Glerkin method to discrete the forward problem,and then propose a CGM_LMA_GPM method to overcome the shortcomings of singularity of the coefficient matrix by using the conjugate gradient method(CGM).Finally,Simulations are given by Matlab language,the simulation results show that the CGM_LMA_GPM method can find the unknown total heat exchange factors in a few iterations and it overcomes oscillation manner.2.According to the characteristics of the reheating furnace under steady-state working operation,this paper establishes the optimal control problems by using the slab transient nonlinear temperature prediction model,the slab parameters and process constraints.There are some difficulties,shown as following:(1)It is difficult to find the Frechet derivative of the objective function with respect to the control variable in the process of solving the optimal control problem;(2)How to prove that the gradient of the constructed objective function is Lipschitz continuous.Thus,according to the characteristics of the optimal furnace temperature control problem of slab under steady-state working operation,the first optimization and discrete method is applied to solve the proposed optimal control problems.Dual equations for original optimization problems are obtained,and then we derive the Fre chet derivative of the objective function with respect to the control variable and prove the Lipschitz continuous.Afterwards,the gradient method and the improved conjugate gradient method are proposed to solve the optimization control problems in one-dimensional and two-dimensional,respectively.Finally,the Matlab is introduced to program the one-dimensional and two-dimensional simulations.The simulation results are given as follow:in one-dimensional case,the average calculation time is reduced from 103.88 seconds to 5.16 seconds,and the deviation between the discharging temperature and the target value is reduced from 6.78? to 0.35?;in two-dimensional case,the calculation time is reduced from 126.60 seconds to 35.10 seconds,and the uniform deviation of the discharging temperature is reduced from 57.2? to 30.1?.3.In this paper,the on-line optimization problem of furnace temperature,which integrates of both the delay problems and the dynamic gauge change working operation,is analyzed by us.Applying by the variable time domain rolling optimization method,an online optimization control system is proposed based on the double-layered model predictive controller.Firstly,in order to overcome the difficulty of establishing the objective function and the difficulty of online optimization,we additionally construct a new objective function:minimizing the difference between the actual reheating curve and the ideal reheating curve during the forecasting time.Here,the ideal reheating curves of the slabs are obtained under the steady-state working operation.Secondly,in order to overcome the shortcomings of the original method,in which the slabs have the same weighting factor,the Entropy-TOPSIS method is introduced to adjust the weighting factors of the slabs.Thirdly,we introduce the variable time domain rolling optimization scheme to solve the real-time rolling optimization problem of furnace temperature,and also give the corresponding rolling mechanism with respect to the various delays.Finally,the simulations are given to verify the proposed method.The experimental results show that the efficiency of effective heating of the slabs by proposed method is improved to 99.17%from 74.79%.4.Based on the reheating furnace simulation experiment system platform,we develop the online furnace temperature optimization setting software to complete the online furnace temperature simulation of the furnace.The simulation experiment of dynamic optimization method is verified under laboratory conditions.Firstly,in order to overcome the difficulty of large amount of computation to solve the high-dimensional model and the difficulty of online application,the numerical solution algorithm by GPU parallel framework is designed and developed.The acceleration ratio relative to the CPU approach can reach 1338.40,and the calculation time is reduced from 16.02 hours by CPU to 51.97 seconds.The information of slabs' internal temperature distribution will be used as the simulation output of the virtual reheating furnace.Secondly,combined with the former research,the online furnace temperature optimization setting software is designed and introduced.Here,the structure and function of each model in the software are introduced in detail.Finally,the interface display of the online furnace temperature optimization setting software and the simulation experiments are given.Simulation results prove the validity of the software. |
PDF Full Text Request