Research On Temperature Control Algorithm Based On Improved Predictive Control For Coke Oven

The process of coking oven heating,is an important work before the coking oven is put into production.The quality of the heating process determines the service life and the production quality after the coking oven is put into operation.We desire the average temperature among all the coking ovens increasing gradually according to the designed heating curve and the specific temperature of each combustion chamber changing uniformly,which can protect the coking oven and extend its service life.Therefore,it is significant to research temperature control of the coking oven heating process.It is difficult to establish an accurate temperature model for the coking oven because there are too many factors to interfere it.So we cannot utilize the most of advantage control strategies which need to feed an accurate mathematical model.At the same time,the traditional PID control strategy is difficult to tune reasonable parameters in real-time for this nonlinear controlled object.Fortunately,the Generalized Predictive Control(GPC)based on inaccurate model is suitable for our controlled object and requirements.GPC algorithm does not require high accuracy of the model and adopts a multi-step predictive optimization strategy.GPC is widely used in practical industrial control because of its good performance,anti-interference and robustness.In this paper,we first study the process of coke ovens,including the basic principles of ovens,gas flow models and traditional coke oven temperature mathematical models.A feedback correction method based on ridge regression is proposed to solve the problem of without solution when the matrix is singular in least squares feedback correction.In order to reduce the computational complexity,we propose a step-by-step scrolling optimization combing with multiple constraints,which follows the actual control flow.Through theoretical derivation and simulation experiments,it is proved that the ridge regression prediction method can effectively solve the collinearity problem in the coke oven temperature prediction problem.In the unconstrained rolling optimization problem,the constraint of valve increment and the temperature heating rate are added to make the rolling optimization meet the practical application problem,avoiding the high-frequency component of the control quantity due to repeated bidirectional changes,thus avoiding the actuator damage.Then the whole oven temperature control system is designed.According to the multivariate generalized prediction method,the coupling problem encountered in the actual control is solved.The modular multivariable generalized predictive control strategy is proposed.The simulation experiment and the field test have obtained good control effects.At the end of this paper,the temperature control system platform of the coke oven in practical application is briefly introduced,including hardware configuration and software functions.The system has been successfully applied to many ovens and achieved excellent control effects.