Industrial tube-furnaces are of characteristics of strong coupling, multivariable and large time delay. Motivated by distributed model predictive control, this thesis studies the solution of furnace temperature balance, and proposes a comprehensive optimization scheme for furnaces. The main research work is presented as follows.Firstly, based on the Infinitesimal method and Bejionkob method, we build the first-principles models for convection chambers and radiation chambers. The influence of the inlet flow on the outlet temperature, the chamber temperature, and the flue gas temperature are analyzed by means of dynamic model simulations, which offers ameaningful guidance for furnace optimization control.Secondly, traditional control strategies of furnaces are insusceptible to eliminate the difference in branch-temperature due to their strong coupling. In response, a distributed MPC control scheme based on conjugate gradient transform is introduced in this thesis, where each subsystem can be independent inachieving the optimal input of each subsystem with a conjugate transformation to solve the coupling problems. Compared with methods of centre MPC and PID, the proposed approach is verified having good performances of resisting disturbances.Finally, conventional optimization methods for furnace susually less concern the holistic performance. In response, we propose a closed-loop solution based on distributed model predictive control. The simulations are carried out to compare with centre MPC and PID methods in terms of response time, optimization results and economic performance, demonstrating the effectiveness of the proposed method. |