Research On Nonlinear Multi-model Predictive Control Technique In Gas-collector Pressure Of The Coke Oven

Coke is the main raw material in the etallurgical industry. In the coking process, large amount of by-product gas will be generated from coke ovens, and the process of recycling by-product gas is called gas collecting process. It is an important link in the iron and steel production, which not only saves energy but also reduces environmental pollution.The stability of gas-collectors’ pressure directly affects the gas quality, the life-time of ovens and the producing environment. Therefore, it is a vital significance for the iron and steel enterprise to study on the control of gas collecting process of multi-asymmetrical coke ovens.The gas collecting process of multi-asymmetrical coke ovens is a highly complex industrial process. Its characteristics of multivariable, strong coupling, asymmetry and nonlinearity, time-varying and hard in modeling are the major difficulties in the control of gas collecting process, so it is hard to achieve effective control using traditional control methods. Comprehensive consideration on the physical features and technological characteristics of coke oven gas-collector pressure system, this paper has mainly studied the mechanism modeling method and multi-model predictive control method used in the gas-collector pressure control.Firstly, with comprehensive consideration on the physical features and technological, this paper builds the mechanism model. Through the field research this paper divides the working conditions into three typical working conditions and linearizes the nonlinear model with different stable operating points, with these linear models approaching to the gas-collector nonliear process.Secondly, on the basis of models expanded in different stable working points, this paper designs predictive controllers with three models and uses a method of rebuilding the matrix of GPC algorithm as the switching strategy so as to make sure of the system switching smoothly.Finally, with the help of MATLAB this paper makes the simulation of the multi-model predictive control method and compares it with the single-model predictive control. Simuliation results demonstrate that the method used in this paper can adapt different working conditions well without the fluctuation at the switching time; with the step disturbance the simuliation shows the advantage of adapting of the method used in this paper and what’s more it also has some certain immunity, system can easily adapt to the taget pressure value; with the random perturbance, method used in this paper also shows pressure smaller deviation. All of the above results have verified the control strategy of this paper.