Method and apparatus for automatically scheduling jobs in computer numerical control machines using machine learning approaches

A method of automatically scheduling jobs in computer numerical control machines using machine learning, the method comprising the steps of:

• collecting a schedule job list from a database;
• generating a plurality of schedules for a schedule job to be processed with respect to the schedule job list;
• calculating an evaluation index for the plurality of generated schedules;
• determining whether the calculated evaluation index for the plurality of schedules has reached a target evaluation index;
• and
• selecting two schedules corresponding to evaluation indices when the calculated evaluation index does not reach the target evaluation index and generating two new schedules using a genetic algorithm,
• wherein the method further comprises the steps of:
• returning to the calculating the evaluation index for the plurality of generated schedules after the selecting the two schedules and the generating the two new schedules are repeatedly performed until new schedules corresponding to the number of the plurality of schedules are generated,
• setting selection probabilities to the plurality of generated schedules such that a schedule having a higher evaluation index has a higher selection probability, and returning the set selection probabilities to a user, in response that the calculated evaluation index reaches the target evaluation index,
• in response that one of the plurality of schedules is selected, incorporating the selected schedule into job assignment information for each machine,
• displaying the job assignment information for each machine,
• redistributing jobs in accordance with the schedule incorporated into the job assignment information for each machine, and
• visualizing and displaying the redistributed results on a screen,
• wherein the calculating the evaluation index for the plurality of generated schedules comprises, for each of the plurality of generated schedules, calculating the evaluation index based on the number of jobs having violated deadlines, a total of delay hours for jobs over due, and an end time of the last job processed, and
• wherein the selecting two schedules and the generating the two new schedules comprise:
• grouping two schedules into a pair among the plurality of schedules, and generating and updating two new schedules through crossover operation and mutation operation for each pair of two schedules,
• establishing an advanced genetic algorithm, in order to satisfy the target evaluation index, by repeatedly performing the generating and updating until new schedules corresponding to the number of the plurality of schedules are generated, digitizing degree that constraints regarding the evaluation index are violated, and gradually lowering the probability that schedules not reaching the target evaluation are selected by applying a penalty to the evaluation index, and
• generating optimized schedules by using the advanced genetic algorithm.