Application Of Decoupling Internal Model Control In Grinding Process

In the actual grinding classification process,the overflow concentration of classifier is an important process index.However,the overflow concentration of classifier not only has a complex strong coupling and nonlinear relationship with the ore supply and water supply,but also suffers from random noise,measurement noise,model parameter perturbation and uncertainty disturbance in the transmission process.The influence of the grinding classification process is characterized by multivariable,strong coupling and large time lag,which leads to low control precision and low robustness of the grinding classification process,which makes the overflow concentration of classifier difficult to control.Therefore,this thesis takes the grinding classification process as the research object,and the overflow concentration stability control of classifier as the control target.The mathematical model of the grinding classification process is obtained by identification method,and two improved multivariable internal model decoupling control methods are introduced to realize the decoupling control of the grinding classification process and the set value of the overflow concentration stability tracking of the classifier.The concrete contents are as follows:Firstly,the overflow concentration of classifier and recurrent sand amount are taken as the output,the classifier adds water and the ore supply amount as the process variable,the mathematical model of the grinding classification process is established.Parameters of the model are identified by the forgetting factor recursive least squares method.The coupling degree analysis of the system is performed by the relative gain matrix.Secondly,the improved V norm three degree of freedom internal model decoupling control method is proposed to decoupling the grinding classification process.By introducing filters,the three degree of freedom internal model structure is constructed to separate the set point response from the disturbance response.Meanwhile,the dual-port control is introduced to improve the system's anti-interference ability.Taking the comprehensive performance index of ITAE combined with robust performance as the objective function of tuning filter parameters.The chaos local enhancement operator improved seeker optimization algorithm(CLEOISOA)based on chaos search and local enhancement operator search is proposed to optimize the parameters of filter.The simulation results show that under the perturbation systems,compared with other controlmethods,the improved V norm three degree of freedom internal model decoupling control method makes the grinding classification system have better decoupling and robustness.Furthermore,the Improved inverse decoupling active disturbance rejection internal model control method is proposed to decoupling the grinding classification process.The inverse decoupling method is used to realize the decoupling of the grinding classification system,and the improved internal model control and the linear active disturbance rejection control are adopted for the decoupled subsystem.By introducing internal model compensator and gain to compensate the time delay,the dependence of the system on the model is reduced.The model mismatch,external disturbance and uncertainties are suppressed by adjusting the parameters of the linear active disturbance rejection controller,the internal model compensator and gain.The simulation results show that compared with the improved V norm three degree of freedom internal model decoupling control method and the inverse decoupling internal model control method,the Improved inverse decoupling active disturbance rejection internal model control method has better decoupling performance,tracking performance and robust performance.It is proved that this method is effective.Finally,two improved multivariable internal model decoupling control methods are verified on the semi-physical simulation platform of grinding.The results of the semi-physical simulation platform show that the two improved multivariable internal model decoupling control methods are effective.