Study On Modeling And Control Method In The Grinding Classification Process

The grinding fineness is an important technological index in the grinding classification process,which has a strong-coupling and nonlinear relationship with the ore feeding quantity,the adding water and the classifier overflow concentration.It is difficult to establish an accurate mathematical model and realize the ideal control for grinding fineness by using these present control methods.However,the change of the classifier overflow concentration whose detection methods are simpler and detection devices are cheaper can reflect the variation of the classifier overflow fineness.Therefore,this thesis chooses the primary grinding closed-loop as the research object and the classifier overflow concentration is served as the control objective.It establishes a mathematical model between the classifier overflow concentration and the classifier adding water through decoupling,and introduces the model reference adaptive method based on the improved differential evolution into the system,which realizes the adaptive control for the classifier overflow concentration.The main contents of this thesis are as follows:First of all,for establishing the mathematical model,the classifier overflow concentration and the return sand were taken as output variables while the classifier adding water and the ore feeding quantity were taken as input variables.The established multivariable model parameters were identified by the improved least square method.It adopted diagonal matrix decoupling control to eliminate multivariable coupling and obtained the mathematical model between the classifier overflow concentration and the classifier adding water.Thus,the classifier overflow concentration was only under the control of the classifier adding water.Then,in order to overcome the effect of the parameter variations and kinds of interference to system performance in the grinding operation process,the model reference adaptive control was introduced into the control loop of classifier overflow concentration for adjusting the controller's variable gain adaptively,whose adaptive control algorithm was realized respectively by the gradient method,differential evolution algorithm and improved differential evolution algorithm.The simulation result showed that the model reference adaptive control based on the improved differential evolution performed a smaller overshoot,quicker responsivity and better robustness than the others,which realized the adaptive control for the classifier overflow concentration.Finally,in order to verify the application of the proposed theories and methods,an experiment was carried out on the grinding synthesize automation platform through loading the modular algorithm program into the grinding virtual operation equipments.The result showed that the model reference adaptive control method based on the improved differential evolution got a better control performance for the classifier overflow concentration.