Hierarchical Fuzzy Modeling Of Precalcining System Of New Dry Cement Production Line

As an important work of the simulation training system of the cement production line, it is necessary to model every part. Precalcining of new dry-process cement system is a kind of object with multiple input variables, large delay, large inertia and uncertainty characteristics, the scholars found it difficult to achieve a precise mathematical model through the long-term research. This also resulted in that the traditional control theory is difficult to obtain a more excellent control effect. Fuzzy inference system already has many successful applications because of its "universal approximation". But in the face of the new dry-process cement precalcining system with the multi-variable characteristics, it is inevitably necessary to solve the "dimension disaster" problem. Base on the technical features and control features of the new dry process cement precalcining system, a kind of modeling approach with the hierarchical fuzzy system is proposed in this paper. Two kinds of hierarchical fuzzy modeling approach were proposed in the modeling of the precalcining system pressure and temperature, which solved the phenomenon of "dimension disaster" problem in the fuzzy system modeling effectively, established the hierarchical fuzzy model with good approximation and dynamic nature of the performance successfully.1. In the modeling of the precalcining system pressure, a kind of "one plus" hierarchical fuzzy model was proposed, also gave the complete program of steps to achieve it. First, the grey relationship degree method was used in the correlation analysis of the input variables and output variables to obtain the importance order of input variables, the static hierarchical fuzzy model of the system was obtained. Then applied the grey relationship degree method again to make the correlation analysis between the historical value of the output variables of the system and the current output value, and introduced the historical information properly into the static hierarchical fuzzy model to obtain the dynamic characteristics of a hierarchical fuzzy model. This program can reduce the model error from the static error 158.103Pa to the 57.8285Pa. And the model is dynamic because of the introductions of the historical information. The test error rate of the model is about 5%, so this model has good generalization ability.2. In the modeling of the precalcining system temperature, for the shortcomings of the modeling program in the precalcining system pressure modeling, a new attempt was made in the hierarchical fuzzy modeling. Main attempt direction is the introduction of a number of compliance expertise, technology features of input combinations. Through a large number of experiments, a "hybrid" structure of the hierarchical fuzzy model was ultimately obtained. In this model, not only to consider the analysis of the relationship between the variables, but also take the actual combination operation existed in the system into account. Some new attempts were made to correct the neglect of the relations between the input variables and the process connection in manual experience. The final results of model training and model test show that the program has achieved satisfactory results. The training error of this hierarchical fuzzy model for the TSD decomposition furnace temperature reduced to 4.66℃, the test error rate of about 3%. Still with dynamic characteristics, in line with the needs of the new dry cement production process simulation and operator training system modeling.