Method to Detect Slag Flow in Course of Steel Casting Based on Machine Learning Algorithms

The paper considers the possibility of implementing a system capable of identifying the ingress of slag into the intermediate ladle at an earlier stage of casting. To do this., it is proposed to analyze the vibration acceleration signal measured on the steel filling unit by a neural network with an autoencoder architecture to determine the moment when the slag starts to flow out of the steel filling ladle and transfer it to the control system