An approach, whose hardcore is a classification of charges by determining whether the charge starts at the time of disturbance occurrence, of making a re-plan was proposed to solve the problem of multi-constrained steelmaking-continuous casting re-planning. The constraints of the re-planning problem are divided into hard and soft constraints. A hybrid algorithm based on time forward inferring and genetic algorithm is designed for the processing charges, and another hybrid algorithm based on time backward inferring and genetic algorithm is designed for the un-processing charges. The feasible solution of the re-planning problem is solved with the hybrid algorithm and hard constraints, and the optimal solution is searched with the soft constraints in the feasible solution. The feasibility and effectiveness of the method were verified with the help of simulation experiments by production data in a steel plant.