SA-ANN-Based Slag Carry-Over Detection Method and the Embedded WME Platform

Slag carry-over detection technology (SCDT) is of important significance for steel continuous casting production (CCP), but has the problems with manufacture cost, service life, installation, and maintenance. Aiming at the problems, this paper brings forward a novel vibration style SCDT realization method based on simulated annealing artificial neural network (SA-ANN). According to ladle pouring process, the vibration signal of steel stream is regarded as the target signal for SCDT. Then, the time point of slag carry-over can be obtained in light of the vibration amplitude difference of pure molten steel and steel slag. Based on the fluid flow similarity principles, an embedded water model experiment (WME) platform is established. The WME platform can simulate the physical process of ladle pouring, reduce the system debugging time under formidable CCP field conditions, and improve the industrial suitability of SCDT. Using an improved SA-ANN algorithm, the status of steel stream is identified to realize automatic control for ladle pouring. WME simulated test results show that the slag detection accuracy (SDA) of this method can reach more 99%. CCP industrial field experiment proves that this method requires low cost and little rebuilding for the current CCP devices, and the practical SDA can reach more 96%.