Forecasting Study On Quality Of Slab In Continuous Casting Slab Producting Process

The quality of continuous casting slab mainly includes macroscopical structure, central segregation, central porosity, central shrinkhole, interal crack and non-metal inclusions content, the rimhole, the skinhole and so on.The existence of central segregation, interal crack, central porosity, central shrinkhole in continuous casting slab seriously deteriorate the integrated property of steel and the rolling process of continuously casting slab, which becomes the main influence factors of the internal quality of continous casting slab. At the same time, non-metal inclusions content, the rimhole, the skinhole becomes the main influence factors of the external quality of continous casting slab, and this essay links up with the on-the-spot output datas of a large-sized steel factory, the influence of technologic and device factors on slab external and internal quality were analysed and discussed. Using the BP algorithm of man-made neural network followed, in order to develop the BP man-made neural network forecasting models of central segregation, central porosity, central shrinkhole, inter crack and small pellet impurities, the rimhole, the skinhole.The main conclusions are as follows:The samples of central segregation defect, central and trigonometry crack defect at higher level appear frequently on the conditions of high S and C content, superheat of liquid steel, cooling intensity, casting speed and low Mn/S, caster accuracy. The samples of central porosity and central shrinkhole defect at higher level appear frequently on the conditions of casting speed, the electric current of electromagnetic stirring, the frequency of electromagnetic stirring, superheat of liquid steel, C content, when these factors are changing.The samples of small pellet impurities defect at higher level appear frequently on theconditions of casting speed, the frequency of vibration, the amplitude of vibration and the inserting depth of SEN, the volume of argon to be entered into, superheat of liquid steel, when these factors are changing. The samples of the rimhole and the skinhole defect at higher level appear frequently on the conditions of the frequency of vibration, the amplitude of vibration, superheat of liquid steel, casting speed, the volume of argon to be entered into, when these factors are changing.By building up the BP man-made neural network forecasting models of these factors which are refered to before can influent the external and internal quality of slab. To combine with application of the MATLAB language. Firstly, selecting some teams of sample datas, using the half of these datas as the training sample, then the other half is used as forecasting sample, with the help of BP man-made neural network forecasting models, we can get the corresponding forecasting datas. Finally by comparing the forecasting datas with the realistic survery, we find that the results are correct under a permit of error.As a result, in the following day, to have the aid of some corresponding influented factors, we can know the quality of the continuous casting slab before they are producted.