Machine maintenance and breakdown in steel plants affect the execution of production schedule. In response to the uncertainty of machine in the steel plant, this study investigates the predictive-reactive scheduling problem of steel plants considering machine maintenance. In the predictive scheduling stage, based on the availability of machine, a predictive scheduling model considering machine preventive maintenance is established, and an adaptive iterative greedy algorithm is designed to solve the model, forming a predictive schedule with maintenance time windows. In the reactive scheduling stage, a reactive scheduling model for machine breakdown is established with scheduling efficiency and stability as optimization objectives. A multi-objective evolution algorithm based on preference information is designed to optimize and adjust the predicted schedule. Finally, comparative experiments are conducted based on production performance to verify the effectiveness and superiority of the proposed method.