Abstract: In steelmaking-continuous casting ladle matching， the ladle properties are numerous and the targets are difficult to meet at the same time. The performance indicators are established by first-order rule-learning to maximize the ladle temperature and ladle use times， minimize the ladle material grade and nozzle number. The ladle temperature， life， material， nozzle， skateboard and frame use times are constraints. The selection rules of ladle are given by minimal generalization method. The method of heuristic ladle selection based on rule priority is proposed. The actual data simulation and applying results show that the proposed method can improve the production efficiency and the number of online ladle and ladle maintenance per day can be reduced.

Key words: steelmaking-continuous casting, ladle matching, first-order rule-learning, minimal generalization, simulation