The Design And Realization Of Prediction Model In Expert System For Coal Blending

In recent years, due to the constant updation of technology of blast-furnaces production, there is a increasing demand for the quality and stability of coke. The coal blending is mixing different nature of the single kind of coal by a certain rate. On this basis, coking coal blending is proceeding a series of phisical and chemical reactions of coke oven to obtain the coke. At present, scholars at home and abroad is devoted to the study of the coke quality prediction model, disturbed by different factors of manufacturing technique, coal resources and coal quality, leading to the lack of a unified prediction model that can be commonly used in all coking plants so far.Currently, there are a number of small and medium coking plants that still adopt the experience of blending expert to set specific ratio for the single kind of coal to guide the production, which is liable to make the produced coke quality unstable.According to the actual situation of coking coal blending of D coking plant,establishing prediction model for fired coal quality and coke quality respectively.Among of those, using the method of polynomial fit for prediction model of fired coal quality, and using the method of multiple linear regression analysis for coke quality.The paper has a reasearch on dynamical management of model, making the model dynamically revisable. By comparing the actual value with the predicted value, and the error of them is within the admitted scope, explaining the fact that the established prediction model has the pretty good prediction result, so it can realize the stable production of coke.