There is a production floor with W workstations and N jobs with M operations( different processing times per operation ). A job is completed only if its M Operations are completed. Objective is to reduce Idle time amongst Workstations

I am trying to use Deep Q Learning to solve this to come up with an optimized schedule but Need inputs on what you feel would the right State for RL in this case, please.


  • $\begingroup$ As stated, it appears we could just sort Jobs by total duration, and schedule them in descending order. There must be other aspects of the task that you want to share above. Maybe a given Workstation can only do a certain subset of Operations? Are completion times constant? Maybe there's a cost when a Station switches among Jobs? Or a setup cost when it switches Operation types? Or a transportation cost to move materials between Stations? Also, has your Reinforcement Learner seen good + bad schedules in the training data? What heuristic rules has it picked up on so far? $\endgroup$
    – J_H
    Commented Dec 24, 2022 at 21:45
  • $\begingroup$ Yes , there are a few qualifications we need to take care off - like certain workstations can work with certain operations $\endgroup$
    – ArchanaR
    Commented Dec 26, 2022 at 0:47
  • $\begingroup$ Processing times are different for the operations and currently transportation time is not considered . I am looking for a State for my reinforcement learning and I think you mean the regular constraint programming when you say sorting the jobs and calculating . Any idea how the state should look like ? $\endgroup$
    – ArchanaR
    Commented Dec 26, 2022 at 0:49


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.