Data Driven Optimal Control Of Cement Grinding Ball Mill System

Cement grinding process is an important process in cement production,which is to crush cement clinker,mixed materials and other materials to change their particle size and physical properties.In the process of cement grinding,the combined grinding technology consisting of roller press system and ball mill system is widely used,in which ball mill system undertakes most of the grinding tasks.The operation index of ball mill system cement particle size and ball mill load directly restrict the quality and production efficiency of cement products.The ball mill system has many complex characteristics,such as multivariable,strong coupling,non-linear,working condition change and so on.It is quite difficult to control.To realize the optimal control of the ball mill system is of great significance to stabilize the quality of cement and improve the economic benefits of the cement plant.Under the support of Shandong science and technology major project "The Key Technology Research and Application Demonstration of Intelligent Factory",this paper carried out the research on data-driven optimal control of cement combined grinding ball mill system.The main research work is as follows:(1)Aiming at the problem that ball mill load is difficult to be directly detected by instrument due to the limitation of ball mill's working characteristics,based on the analysis of influencing factors and operating data,the online detection of ball mill load is realized by using soft measurement method.The nonlinear autoregressive(NARX)model of ball mill load was established by means of self-coding and random weight neural network.The simulation results show that the model is in good agreement with the dynamic change of ball mill load,which provides data support for the subsequent research on ball mill system modeling and control.(2)Aiming at the problem that it is difficult to establish accurate mechanism model of ball mill system,the modeling method of ball mill system based on data drive is studied.On the basis of analyzing the dynamic relationship between the operating indexes(cement particle size,ball mill load)and the control parameters(the speed of powder separator,the speed of circulating fan and the speed of main exhaust fan)of the ball mill system,the data driven model was established by using the recursive neural network,and the validity of themodel was verified by the data simulation.(3)A data-driven operation optimization control method for ball mill system is designed,which includes loop set value optimization tracking control,multi-model adaptive control and material quantity control.Loop setting optimal tracking control to the cement particle size and the ball mill load control within the scope of the expectation and approximation expectations as the goal,as far as possible to(2)based on adaptive dynamic programming algorithm of data-driven model is set up as a model in the network,using the expected value of cement particle size and ball mill load known conditions,the set value of the control loop of the ball mill system is given online,and the evaluation network and the execution network in the algorithm are used to solve the optimal control law,so that the output value of the loop can track to the set value quickly;In order to deal with the condition switching of the ball mill system,a weighted adaptive dynamic programming(ADP)controller is designed based on the idea of multi model control.A number of sub models are established to cover the system condition.The matter-element extension model and evaluation function are used to monitor the system condition of the ball mill.The weight of the sub controller is determined by the design of the weighted function.As a supplement to the first two methods,the in-grinding material quantity control aims at stabilizing the amount of material entering the ball mill.An adaptive PID controller based on instant learning is designed to control the steady flow warehouse position and indirectly realize the control objective.(4)Based on Visual Studio platform,using C# language to develop the ball mill system optimization control software,based on SQL Server to establish the software supporting database,using OPC technology to achieve the data interaction between the software and the field DCS system,online granularity detection system.The optimization control software is applied to the production of cement mill in a cement factory.