Research On Prediction Model And Processing Parameters Analysis Of Breakout In Slab Continuous Casting

Continuous casting is an important link to produce qualified steel products,and high quality,high efficiency and low energy consumption are the goals pursued by continuous casting technology.High drawing speed is an important means to improve the production efficiency and reduce energy consumption in steel enterprises,however,the increase of drawing speed will increase the chance of steel leakage in continuous casting,which will bring huge economic loss to steel mills in case of steel leakage accident.In order to avoid the occurrence of steel leakage accident,that is,we can optimize the field casting conditions from the perspective of continuous casting process parameters to avoid the steel leakage accident from the root,however,from the production practice,only from the process parameters optimization,it is impossible to completely avoid the steel leakage accident,thus,the development of an accurate steel leakage prediction model can effectively prevent the steel leakage accident.Therefore,this paper investigates the following four aspects,as follows.First,the formation mechanism of cohesive steel leakage is analyzed,and the formation and evolution process of billet cohesion is described.At the same time,the temperature characteristics of continuous casting process are analyzed based on the pouring data of a domestic steel plant,and the temperature characteristics and laws of stable pouring,open pouring and sticking steel leakage are summarized in terms of pouring speed,pouring temperature and liquid level fluctuation.Secondly,the main influencing factors of 44 cases of cohesive steel leakage were statistically analyzed in terms of steel grade,thickness,width,time and operation.In terms of steel grade,the number of sticking is higher for low alloy steel than for low carbon steel;in terms of billet thickness,the number of sticking per thousand casting heats is 2.5,0.5 and 0.6 for 220,260 and320 mm thickness billets,respectively,and the sticking is significantly higher for 220 mm thick billets with higher pulling speed than for 260 and 320 mm thick billets;in terms of billet width,the sticking increases with width due to the deterioration of slag protection.The adhesion increases gradually with the increase of width.Again,by calculating the transverse and longitudinal propagation rates of the adhesion leakage in the crystallizer,the crack propagation behavior of the adhesion leakage was analyzed,and the "T" type temperature rate characteristic reconstruction method was proposed by combining the temperature and temperature rate characteristics of the adhesion leakage steel.Based on the temperature rates of four thermocouples in the upper and lower rows,the temperature rates of three adjacent thermocouples in the first row for 30 seconds were summed up and then connected with the temperature rates of the middle thermocouple in the second row,and the reconstructed data could capture both transverse and longitudinal propagation behavior of the bonded leaky steel.After that,the data reconstruction characteristics of 40 cases of true and 40 cases of pseudo-sticky junctions were statistically analyzed,and the main differences between true and pseudo-sticky junctions of the reconstructed data were summarized.Finally,based on the above true and pseudo-sticky knot leakage steel reconstruction data,three leakage steel forecasting models,SVM,GS-SVM and DE-SVM,were established respectively,and the three models were tested after training.The results showed that there was one case of leakage in SVM with 90% reporting rate,and the reporting rate of both GS-SVM and DE-SVM was 100%,and the alarm of GS-SVM and DE-SVM models The accuracy rates were96.2% and 92.3%,respectively,and GS-SVM had a higher alarm accuracy rate.