Abstract: The longitudinal facial crack is a kind of slab defect with high incidence. Online judgment of the influence factors of longitudinal facial cracks is of great importance to realize the intelligence and green of continuous casting production.According to the formation mechanism and influential factors of longitudinal facial cracks, an nonlinear relationship model of influence factors and longitudinal facial crack was established by BP neural network. Then, the model was optimized by the slab data and the LM algorithm. Furthermore, the weight of each influential factor was calculated using the MIV algorithm. Based on the above processes, the online judgment in the intelligence of the weight of influential factors of longitudinal facial cracks could be realized. When the longitudinal facial crack was occurred, the occurrence probability of longitudinal facial cracks was decreased by adjusting the process parameters online based on the judgment results. And the formation cause of longitudinal facial cracks could be promptly found based on the judgment resultsAfter the cast.

Key words: longitudinal facial crack, influence factors, intelligence, online judgment, LMBP-MIV