At present, monthly preparing job assignments in maintenance depots for work such as electric equipment/track maintenance are human-intensive, which is very time-consuming. An automatic job assignment system for crews has been studied previously, but little research has been carried out on a similar system for maintenance workers.
Accordingly, the RTRI built a mathematical programming model in consideration of the work patterns of maintenance workers to help planners in preparing job assignments, and developed a job assignment system based on the model.
After a user inputs data such as names, qualifications, jobs (date and time, work details, number of people required, and necessary qualifications), designated holidays, leave of absence desired, etc., and selects relevant constraints, the system generates a job assignment candidate. If it does not satisfy the constraints, the system indicates the causes. Users can modify the output, if necessary, and have it checked again (Fig. 1). These functions make it possible to flexibly generate an assignment that is adapted to local rules of each group. The system can also cope flexibly with job schedules that vary greatly by month and special work patterns for maintenance workers.
We conducted computational experiment on a commodity PC using actual job schedule data. This experiment revealed that for a group of about ten people can be created within dozens of seconds. An on-site trial demonstrated that a job assignment previously taking a few hours to a dozen hours can be performed within dozens of minutes using the system.