The structure of an intelligent control system for a converter, which includes a subsystem for recognizing and evaluating the slopping of a gas-slag-metallic emulsion in real time based on computer vision by explicit features and a subsystem for predicting slopping by indirect signs, which are most often used to monitor the oxygen-converter process, is proposed. Procedures for forecasting and detecting slopping based on artificial neural networks and a precedent approach are developed. The morphological signs of slopping and the factors influencing their occurrence are described.