To solve the problem wherein a large amount of labors are required for adjusting a schedule, when the duty schedule of a person on duty, the workload of the day, or the like, is changed because the schedule for an accident-restoration training is managed by a human system in an electric power company, or the like, and to solve the problem wherein the training time and training content are known in advance to the employee through schedule adjustment with employee who is going to receive the accident restoration training in the schedule management by the human system and the training effect is halved.

The operation of the accident-restoration training is determined by deciding conditions, such as, the attendance status and workload of the person on duty on the day or the weather, by using a computer or the like. Accordingly, the accident-restoration training becomes more realistic for employees, and the effectiveness of the accident-restoration training can be improved.