Determination of slag outflow moment during steel teeming using competitive neural network

The article describes method of competitive neural network usage to process vibration acceleration signal spectrum from a tundish surface in order to determine the moment of slag outflow from the ladle. It is proved experimentally that the neural network is able to recognize the steel teeming process state, which is prior to the slag outflow from the ladle under industrial conditions and influence of disturbances. The research results show the feasibility of the developed approach.