The Research Of Predictive Control On The Cement Rotary Kiln Calcination System

For complex calcining system of cement kiln，because the cement kiln itself has the biglag, nonlinearity, time variation, strong coupling between variables and othercharacteristics，it is generally difficult to build a mathematical model for the rotary kilnprocess and control accuracy is naturally difficult to guarantee. With the continuousdevelopment of control technology and continuous mature of predictive controltechnology based on model predictions, rolling optimization and feedback correction, theprocess control of the cement rotary kiln embrace is embracing a new research direction.For the complex mechanism of the cement rotary kiln process, the topic proposes a modelpredictive control algorithm based on hybrid model, it uses the optimization algorithm tosolve for the calcination process variables. Specific research works are as follows:First of all, though studying the reaction mechanism of the cement rotary kiln process,analyzing the factors that affect the quality of cement clinker and energy consumption perunit of production, and further considering the present situation of Chinese cement rotarykiln control, a new model that makes oxide and oxygen content in the rotary kiln as targetvariables and makes the coal rate and high-temperature fan baffle opening as operatingvariables is put forward, to provide a basis for the prediction model of the cement rotarykiln process.Then, based on the characteristics of the cement rotary kiln process, the idea that thesteady-state model is separately identified from the dynamic model is used: thesteady-state data of the rotary kiln system is identified by BP algorithm, to obtain thenonlinear neural network steady-state model; the dynamic data of the rotary kiln system isidentified by the least squares method, to obtain the linear ARX dynamic model; theparameters of the ARX model are adjusted by the steady-state model gain in real time, toobtain the hybrid model of BP-ARX.Finally, based on the hybrid mode of BP-ARX, the predictive controller of MIMODelay System which is used for cement rotary kiln process is derived and designed. Andthe input variables of the rotary kiln system are predicted in real time though using thesequential quadratic programming optimization algorithm. Last, the data simulation of thepredictive control algorithm designed in this paper is carried out, and the implementationof the program applied in the industrial site is given out.