JP7387070B1 - Scheduling device, scheduling system, scheduling method, and scheduling program - Google Patents

Abstract

The scheduling device (1A) is based on a job list (41) which is a list of a plurality of jobs to be executed by a plurality of processing machines, and a constraint condition list (42A, 42B) which is a list of constraint conditions for the processing machines. A job narrowing section (3) that creates a job list after narrowing down, which is a job list narrowed down to jobs that satisfy constraint conditions from among the jobs set in the job list, and an optimization problem based on the narrowed down job list. By solving the following, optimal job order data is created that sets the processing machine to execute the job and the order in which the jobs are executed on the processing machine so that all jobs in the job list can be completed in the shortest time. and a scheduling section (5) that creates a job execution schedule (43) in which job execution dates and times for each processing machine are set based on the job order data and the job list.

Description

The present disclosure relates to a scheduling device, a scheduling system, a scheduling method, and a scheduling program that create a job execution schedule applied to machining equipment.

In recent years, in order to efficiently execute jobs in machining equipment, it has been desired to create a job execution schedule (schedule data) that can shorten the total execution time of a plurality of jobs.

The production planning device described in Patent Document 1 obtains a processing order for obtaining a product with a specific shape by performing multiple processing operations on a material, and calculates the processing time required for each processing operation based on the processing order. It is being calculated. This production planning device uses multiple processing devices to generate production plans for producing multiple products based on each processing order, and generates multiple production plans obtained based on multiple types of processing orders. A production plan that satisfies the target conditions is extracted from

JP2021-009435A

However, in the technique disclosed in Patent Document 1, the ordering formula for determining the execution order of jobs is complicated. There was a problem in that it took a long time to create a job execution schedule because the formulation process had to be redone.

The present disclosure has been made in view of the above, and an object of the present disclosure is to provide a scheduling device that can create a job execution schedule in a short time and provide it to a user.

In order to solve the above-mentioned problems and achieve the purpose, the scheduling device of the present disclosure includes a job list that is a list of multiple jobs to be executed by multiple processing machines, and constraints that is a list of constraints on the processing machines. The present invention includes a job narrowing section that creates a job list after narrowing down, which is a job list narrowed down to jobs that satisfy constraint conditions from among the jobs set in the job list, based on the list. Furthermore, the scheduling device of the present disclosure solves an optimization problem based on the job list after narrowing down, so that all the jobs in the job list can be completed in the shortest time. , an optimization calculation unit that determines the execution order of jobs on the processing machines and creates job order data that is data indicating the processing machines to execute the jobs and the order of job execution on the processing machines; and a scheduling unit that creates a job execution schedule in which job execution dates and times for each processing machine are set based on the data and the job list. The job narrowing unit creates job list after narrowing down that satisfies the first constraint among the constraints, and the optimization calculation unit creates job order data that satisfies the first constraint. The process of adding a job to the job order data to eliminate the first constraint, and the job narrowing unit satisfying the second constraint that is the next strictest constraint after the first constraint among the constraints. A process of creating a job list after narrowing down, and an optimization calculation unit creating job order data that satisfies the second constraint condition, and adding jobs to eliminate the second constraint condition to the job order data. The process is repeated.

The scheduling device according to the present disclosure has the advantage of being able to create a job execution schedule in a short time and provide it to the user.

A diagram showing the configuration of a scheduling device according to Embodiment 1. A diagram showing an example of the configuration of a job list used by the scheduling device according to the first embodiment. A diagram showing a configuration example of a first constraint list used by the scheduling device according to the first embodiment. A diagram illustrating a configuration example of a second constraint list used by the scheduling device according to the first embodiment. Flowchart showing the procedure of scheduling processing executed by the scheduling device according to the first embodiment A diagram for explaining the first stage of the scheduling process executed by the scheduling device according to the first embodiment. A diagram for explaining the second stage of the scheduling process executed by the scheduling device according to the first embodiment. A diagram for explaining the third stage of the scheduling process executed by the scheduling device according to the first embodiment. A diagram illustrating a configuration example of a job execution schedule created by the scheduling device according to the first embodiment. A diagram showing the configuration of a scheduling system including a scheduling device according to a second embodiment. A diagram showing the configuration of a learning device included in the scheduling system according to Embodiment 2. Diagram for explaining a neural network used by the learning device according to Embodiment 2 Flowchart showing the procedure of learning processing executed by the learning device according to Embodiment 2 A diagram showing a configuration of an inference device included in a scheduling system according to a second embodiment. Flowchart showing the procedure of inference processing executed by the inference device according to Embodiment 2 Flowchart showing the procedure of scheduling processing executed by the scheduling device according to the second embodiment A diagram for explaining the scheduling process executed by the scheduling device according to the second embodiment A diagram showing the configuration of a scheduling device according to Embodiment 3 Flowchart showing the procedure of scheduling processing executed by the scheduling device according to Embodiment 3 A diagram for explaining the first stage of the scheduling process executed by the scheduling device according to the third embodiment. A diagram for explaining the second stage of the scheduling process executed by the scheduling device according to the third embodiment. A diagram showing the configuration of a job list modified by the scheduling device according to the third embodiment A diagram for explaining the third stage of the scheduling process executed by the scheduling device according to the third embodiment. A diagram showing an example of a hardware configuration for realizing a scheduling device according to Embodiment 1.

Below, a scheduling device, a scheduling system, a scheduling method, and a scheduling program according to embodiments of the present disclosure will be described in detail based on the drawings.

Embodiment 1.
FIG. 1 is a diagram showing the configuration of a scheduling device according to a first embodiment. The scheduling device 1A is a computer that creates a job execution schedule 43 for jobs to be executed in machining equipment such as an electrical discharge machine.

The job execution schedule 43 created by the scheduling device 1A is applied to a processing system that executes processing using a plurality of processing machines. In this processing system, each of a plurality of processing machines executes a plurality of types of jobs, thereby processing a plurality of workpieces and producing a plurality of products. Therefore, the job execution schedule 43 is a schedule (schedule data) for each of a plurality of processing machines to process a plurality of workpieces and produce a plurality of products by executing a plurality of types of jobs. . In the job execution schedule 43, the processing machine that executes the job, the job start time, and the job end time are set. The scheduling device 1A includes a scheduler that creates a job execution schedule 43.

The scheduling device 1A allocates all jobs to any processing machine such as a machine tool, and automatically allocates the execution order of all jobs targeted in a processing process using the processing machine. When scheduling a job using a combinatorial optimization problem or the like, the scheduling device 1A performs optimization in advance when constraints (execution restrictions) are given only to some of the processing machines and other environmental factors. Narrow down the jobs to be submitted to the solver. Thereby, the scheduling device 1A narrows down the jobs that are candidates for schedule creation, solves the optimization problem, and creates the job execution schedule 43 without switching the objective function and constraint conditions.

For example, if there is a job that cannot be executed during a part of the schedule creation target date and time, the scheduling device 1A excludes the job that cannot be executed from that time period in advance from the job list 41, and adds the job to the excluded job list 41. Determine the execution order.

The scheduling device 1A includes an input section 2, a job narrowing section 3, an optimization calculation section 4, a scheduling section 5, and an output section 6. The input unit 2 reads the job list 41 and constraint list 42A, 42B from an external device or the like.

The job list 41 is a list of jobs to be executed by the processing machine. The constraint condition lists 42A and 42B are lists of constraint conditions when assigning jobs to job order data (job order data 45 described later) used when creating the job execution schedule 43. The job order data 45 is data indicating the processing machines that execute each job and the order in which the jobs are executed by each processing machine. After creating job order data 45, the scheduling device 1A creates a job execution schedule 43 based on the job order data 45. Here, a configuration example of the job list 41 will be explained.

FIG. 2 is a diagram illustrating a configuration example of a job list used by the scheduling device according to the first embodiment. In the job list 41, job IDs (IDentifications), priorities, machining instructions, workpiece IDs, machining program IDs, and estimated times are associated with each other.

The job ID is information for identifying a job, and is illustrated in FIG. 2 as "J001", "J002", etc. The job ID is set by a job list creation device (not shown) that creates the job list 41. In the following description, jobs with job IDs "J001" to "J006" may be referred to as jobs "J001" to "J006," respectively.

Note that the scheduling device 1A may have the function of a job list creation device. In this case, the scheduling device 1A creates the job list 41 according to instructions from the user.

The priority is information on the priority of executing a job, and a job with a smaller priority number is more likely to be executed first. That is, when the constraint conditions are met, the job with the smaller priority number is assigned an earlier execution time.

Note that the priority order does not need to be set to different numerical values for all jobs, and the same numerical value may be set for different jobs. For example, the priority order may be set by the job list creation device according to an instruction from the user, or the priority order may be set automatically by the job list creation device.

The processing instruction includes information on the delivery date of the product manufactured by executing the job and the processing machine ID of the processing machine that can execute the job. The processing machine ID is information that identifies a processing machine that can be set for a job, and is illustrated in FIG. 2 as "PM001", "PM002", etc. Among jobs corresponding to job IDs, jobs whose processing machine ID is blank are jobs that may be processed by any processing machine. For example, the processing instructions may be set by the job list creation device according to instructions from the user, or may be set automatically by the job list creation device. In the following description, the processing machines with processing machine IDs "PM001" to "PM003" may be referred to as processing machines "PM001" to "PM003," respectively.

The work ID is information that identifies the work on which the job is executed. The machining program ID is information that identifies a machining program used when executing a job. The job list 41 does not need to include the work ID and the machining program ID.

The estimated time is the estimated time required to execute the job. The estimated time includes an estimated time for processing, an estimated time for pre-processing, and an estimated time for post-processing. The estimated time for pre-processing includes the workpiece loading time and workpiece setup time. The workpiece setup includes workpiece mounting processing, workpiece reference position measurement processing, and the like. The estimated time for post-processing includes the time for measuring the workpiece and transporting the workpiece. Note that the estimated time may be the total time of the estimated time for processing, the estimated time for pre-processing, and the estimated time for post-processing.

Next, a configuration example of the constraint list 42A, 42B will be explained. FIG. 3 is a diagram illustrating a configuration example of a first constraint list used by the scheduling device according to the first embodiment. The constraint list 42A, which is the first constraint list, includes a processing machine ID that is information that identifies a processing machine, stopped information that is information that indicates whether or not the processing machine is stopped, and unexecutable processes. , the predicted consumable life time and the maintenance execution deadline are associated with each other.

The stopped information is set to either information indicating that the processing machine is stopped or information indicating that the processing machine is not stopped. The stopped information in Fig. 3 shows a case where "0" is set as information indicating that the processing machine is not stopped, and "1" is set as information indicating that the processing machine is stopped. There is.

In the unexecutable process, a process ID of a process that cannot be executed in the processing machine is set. The process ID is information that identifies a process. For example, "0" is set in the unexecutable process when all the processes can be executed in the processing machine, and "9" is set when all the processes cannot be executed in the processing machine. Further, if "H13" is set as the process ID in the unexecutable process, the processing machine cannot execute the process corresponding to "H13".

The predicted consumable life time is the predicted time from the current time until the consumable life reaches the end of the consumable supplies used by the processing machine. Figure 3 shows a case where the predicted consumable life time is the time until the consumable is predicted to reach the end of its life, but the predicted consumable life time is the date and time when the consumable is predicted to reach the end of its life. There may be. The predicted consumable life time is set for each consumable. FIG. 3 shows a case where the consumables included in the constraint list 42A are consumables B11 and B12.

The maintenance execution deadline is the time from the current time until maintenance work is executed on the processing machine. Although FIG. 3 shows a case where the maintenance execution deadline is the time until the maintenance work is executed, the maintenance execution deadline may be the date and time when the maintenance work is executed. The maintenance execution deadline is set for each type of maintenance work. FIG. 3 shows a case where the types of maintenance work included in the constraint list 42A are maintenance H11, maintenance H12, and maintenance H13.

Note that the predicted consumable life time and the maintenance execution deadline may be set by the job list creation device according to instructions from the user, or may be set automatically by the job list creation device, for example.

FIG. 4 is a diagram illustrating a configuration example of a second constraint list used by the scheduling device according to the first embodiment. In the constraint list 42B, which is the second constraint list, process IDs are associated with whether or not manned work is required. The constraint list 42B is a list of constraints on replacement of consumables, maintenance work, and the like.

The process ID is information that identifies a work process such as replacement of consumables or maintenance work. "H11" to "H13" in the process ID indicate maintenance operations H11 to H13 explained in FIG. 3, and "B11" and "B12" indicate replacement operations for consumables B11 and B12 explained in FIG. 3. It shows.

The necessity of manned work is information indicating whether or not the process requires a worker. In other words, whether or not manned work is necessary is a constraint condition that specifies whether or not a worker is necessary. For the necessity of manned work in Figure 4, "1" is set as information indicating that the process requires a worker, and "0" is set as information indicating that the process does not require a worker. Indicates when A process for which “1” is set for whether or not manned work is required satisfies the constraint if it is possible to allocate a worker, and does not satisfy the constraint if it is not possible to allocate a worker. That will happen.

Note that the constraint condition lists 42A and 42B may include processing machines that can execute the job, availability of processing machines in the next process at the scheduled end time of the job, and the like. If "0" is set in the field of necessity for manned work, it is possible to skip cleaning jobs during times when no workers are present (unmanned times). Furthermore, if the constraint list 42A, 42B includes the availability of a processing machine in the next process at the scheduled end time of the job, the scheduling device 1A determines whether a cleaning job is set after rough machining or after finishing machining. This makes it possible to shorten the waiting time until the system is ready.

The input unit 2 inputs the job list 41 and the constraint lists 42A and 42B to the job narrowing unit 3, the optimization calculation unit 4, and the scheduling unit 5. The job narrowing down section 3 narrows down the jobs to be registered in the job order data 45 from among the jobs in the job list 41 based on the constraint lists 42A and 42B. That is, the job narrowing down section 3 narrows down the jobs to be registered in the job order data 45 from among the jobs in the job list 41 so as to satisfy the constraint list 42A, 42B. The job narrowing unit 3 sorts the jobs in the job list 41 so that jobs are not set for processing machines that are set to stop in the stopping information and for processes that are set to be unexecutable in the non-executable process. Jobs to be registered in the job order data 45 are narrowed down from among them. In the following description, a case will be described in which the scheduling device 1A creates the job execution schedule 43 mainly based on the constraint list 42A.

The job narrowing down unit 3 narrows down jobs so that consumables can be replaced by the predicted consumable life time, for example. Further, the job narrowing down unit 3 narrows down the jobs so that the maintenance work can be executed by the maintenance execution deadline, for example. The job narrowing down unit 3 sends a job list 41 in which jobs have been narrowed down to the optimization calculation unit 4. Note that in the following description, the job list 41 that has been narrowed down may be referred to as the job list 41 after narrowing down.

The optimization calculation unit 4 has an optimization solver that executes job optimization. The optimization calculation unit 4 determines a processing machine to execute each job and an execution order of the jobs in each processing machine based on the job list 41 after narrowing down and the constraint condition lists 42A and 42B. That is, the optimization calculation unit 4 sequentially assigns the jobs in the job list 41 after narrowing down to one of the processing machines so as to satisfy the constraint list 42A, 42B, thereby selecting a processing machine that executes each job. and the execution order of jobs on each processing machine.

The optimization calculation unit 4 determines the processing machine that executes each job and the order in which the jobs are executed in each processing machine by solving an optimization problem. The optimization calculation unit 4 allocates jobs to processing machines so that the higher the priority, the earlier the job is executed. Further, the optimization calculation unit 4 allocates jobs to processing machines so that the time required to complete all jobs in the job list 41 is the shortest. That is, the optimization calculation unit 4 allocates jobs to processing machines so that the total execution time of the jobs included in the job list 41 is shortened.

In the scheduling device 1A, the job narrowing section 3 and the optimization calculation section 4 create the job order data 45 while sharing the job order data 45. The job narrowing unit 3 sets constraints such as a maintenance execution deadline in the job order data 45, and extracts jobs that can be set by the maintenance execution deadline. The job narrowing unit 3 sends job order data 45 in which constraints such as maintenance execution deadlines are set, and jobs that satisfy the constraints (jobs that can be set by the maintenance execution deadline) to the optimization calculation unit 4. The optimization calculation unit 4 selects jobs that can be set in the processing machine from among the jobs that satisfy the constraint conditions, and sets them in the job order data 45. If there is no job that satisfies the constraint, the optimization calculation unit 4 sets a maintenance job or the like in the job order data 45 to eliminate the constraint. Then, the job narrowing section 3 sets the next constraint in the job order data 45 and extracts jobs that satisfy this constraint. The job narrowing section 3 and the optimization calculation section 4 set all the jobs in the job list 41 to the job order data 45 by repeating these processes. The optimization calculation unit 4 sends to the scheduling unit 5 job order data 45 that sets the processing machines that execute each job and the order in which the jobs are executed in each processing machine for all jobs in the job list 41. .

The scheduling unit 5 sets an execution date and time for each job based on the job order data 45 and the job list 41. In this case, the scheduling unit 5 refers to the processing machines that execute each job in the job order data 45 and the execution order of the jobs in each processing machine. The scheduling unit 5 also refers to the estimated time in the job list 41. The scheduling unit 5 also stores job order data 45, the time required to replace consumables for each processing machine (replacement time), and the time required for maintenance to be performed for each processing machine (maintenance time). time) and set the replacement date and time for consumables and maintenance date and time. Note that the replacement time and maintenance time may be set in the job list 41 or in the constraint list 42A, 42B.

The scheduling unit 5 creates, for each job, a job execution schedule 43 in which a processing machine to execute the job, an execution order, and an execution date and time of the job are set. The scheduling unit 5 sends the created job execution schedule 43 to the output unit 6. The output unit 6 provides the job execution schedule 43 to the user by outputting it to an external device such as a display device. Thereby, the job execution schedule 43 is displayed on the display device, and the user can refer to the job execution schedule 43.

FIG. 5 is a flowchart illustrating a procedure of scheduling processing executed by the scheduling device according to the first embodiment. The input unit 2 reads the job list 41 and the constraint lists 42A and 42B from an external device or the like (step S10). The input unit 2 inputs the job list 41 and the constraint list 42A, 42B to the job narrowing down unit 3.

The job narrowing down section 3 narrows down the jobs to be registered in the job order data 45 from among the jobs in the job list 41 based on the constraint lists 42A and 42B. That is, the job narrowing down unit 3 narrows down the jobs in the job list 41 according to the contents set as constraints in the constraint lists 42A and 42B (step S20). The job narrowing down unit 3 sends the narrowed down jobs (job list 41 after narrowing down) to the optimization calculation unit 4.

The optimization calculation unit 4 determines the processing machine that executes each job and the order in which the jobs are executed by each processing machine based on the constraint condition lists 42A and 42B. That is, the optimization calculation unit 4 creates the processing machines for the job under the constraint conditions and the execution order for each processing machine (step S30). In other words, the optimization calculation unit 4 sets the processing machines to execute each job and the execution order of the jobs in each processing machine so as to satisfy the constraints defined in the constraint condition lists 42A and 42B. Sequence data 45 is created.

The optimization calculation unit 4 determines whether all the constraint conditions have been set in the job order data 45. That is, the optimization calculation unit 4 determines whether there is a constraint condition for which the calculation has not been completed (step S40).

If there is a constraint condition for which the calculation has not been completed (Step S40, Yes), the scheduling device 1A returns to the processing of Step S20 and repeats the processing of Steps S20 to S40.

On the other hand, if there is no constraint condition for which the calculation has not been completed (step S40, No), the optimization calculation unit 4 uses the job order data 45 that sets the determined processing machine to execute each job and the execution order of the job. It is sent to the scheduling section 5. That is, the optimization calculation unit 4 sends the execution order of jobs in each processing machine to the scheduling unit 5.

The scheduling unit 5 sets the execution date and time of each job based on the execution order of the jobs in each processing machine determined by the optimization calculation unit 4 and the estimated time of the job list 41 (step S50), and sets the execution date and time of each job in the job execution schedule 43. Create.

The output unit 6 outputs the job execution schedule 43 created by the scheduling unit 5 to an external device such as a display device (step S60). This allows the user to refer to the job execution schedule 43 in which job execution dates and times are set.

Next, a specific example of the scheduling process by the scheduling device 1A will be described using FIGS. 6 to 8. 6 to 8, a case will be described in which the scheduling device 1A executes job scheduling processing based on the job list 41 and the constraint list 42A, 42B.

FIG. 6 is a diagram for explaining the first stage of the scheduling process executed by the scheduling device according to the first embodiment. FIG. 7 is a diagram for explaining the second stage of the scheduling process executed by the scheduling device according to the first embodiment. FIG. 8 is a diagram for explaining the third stage of the scheduling process executed by the scheduling device according to the first embodiment. 6 to 8 show job order data 45 created by the scheduling device 1A.

The job narrowing down unit 3 registers the processing machine ID registered in the job list 41 in the job order data 45. The job narrowing down section 3 determines whether or not there is a stopped processing machine based on the constraint condition list 42A. The job narrowing section 3 sets the processing machine ID of the job order data 45 to "stopped" for the processing machine that is stopped. The job narrowing unit 3 here sets the processing machine "PM003" to be stopped based on the constraint condition list 42A.

The job narrowing down section 3 sets constraint conditions for allocating jobs to each processing machine in the job order data 45 according to the constraint condition list 42A. Specifically, the job narrowing unit 3 selects the process to be performed first from the consumables replacement process or the maintenance process in the constraint condition list 42A. The job narrowing unit 3 selects the maintenance H11 of the processing machine "PM002" with the closest deadline based on the constraint list 42A in FIG. Set to "PM002".

The job narrowing unit 3 narrows down the job list to jobs in the job list 41 by extracting only the list of jobs that satisfy the constraint conditions set in the job order data 45 from among the list of jobs registered in the job list 41. I do. Specifically, the job narrowing unit 3 extracts from the job list 41 jobs that can be executed by the execution deadline of maintenance H11 on the processing machine “PM002”. That is, the job narrowing unit 3 extracts only the list of jobs executable by the execution deadline of the maintenance H11 on the processing machine "PM002" from among the list of jobs in the job list 41. The job narrowing unit 3 selects from the job list 41 a list of jobs that can be executed by the execution deadline of maintenance H11, including job IDs, priorities, machining instructions, workpiece IDs, machining program IDs, and estimated times. Narrow down jobs by extracting. The job narrowing down section 3 sends the narrowed down job list 41 to the optimization calculation section 4. The job list 41 after narrowing down includes job "J002".

The optimization calculation unit 4 performs an optimization calculation to set jobs narrowed down by the job narrowing unit 3 in job order data 45 in order of priority. The optimization calculation unit 4 here sets the job "J002" to the processing machine "PM002", for example. At this point, among the jobs to be executed by the processing machine "PM001", there is no job that will be completed by the execution deadline of maintenance H11 of "PM002" (within 6 hours and 45 minutes), so the optimization calculation unit 4 Do not create a schedule for processing machine "PM001". That is, since there is no job in the processing machine "PM001" that can be executed by the maintenance H11 of the processing machine "PM002", no job is set for the processing machine "PM001".

Note that the job "J002" may be set to the processing machine "PM001", but since a job longer than 6 hours and 45 minutes can be set to the processing machine "PM001", the optimization calculation unit 4 Job "J002" is set in processing machine "PM002" which can only set jobs within 45 minutes. Thereby, the optimization calculation unit 4 can improve processing efficiency.

In order to eliminate the most severe constraint condition for the processing machine "PM002" (maintenance H11 on the processing machine "PM002"), the optimization calculation unit 4 adds maintenance H11 to the processing machine "PM002" to the job order data 45. Set. Furthermore, the optimization calculation unit 4 sets the worker (WORKER) who executes the maintenance H11 in the job order data 45 based on the constraint condition list 42B, and sets the work date and time by this worker in the job order data 45. . In FIG. 7, maintenance H11 performed by the operator on the processing machine "PM002" is illustrated as "PM002_H11". Note that if "0" is set in the requirement for manned work in the constraint condition list 42B, indicating that no worker is required, there is no need to set the worker in the job order data 45, and the work is not required. Only the date and time are set.

Next, the job narrowing unit 3 selects the process to be performed second from among the consumables replacement process or the maintenance process in the constraint condition list 42A. The job narrowing unit 3 selects the maintenance H11 of the processing machine "PM001" based on the constraint list 42A of FIG. 3, and sets the execution deadline of the maintenance H11 in the job order data 45.

The job narrowing unit 3 extracts from the job list 41 jobs that can be executed by the execution deadline of maintenance H11 on the processing machine “PM001”. That is, the job narrowing unit 3 extracts only the list of jobs executable by the execution deadline of the maintenance H11 on the processing machine "PM001" from among the list of jobs in the job list 41. The job narrowing down section 3 sends the narrowed down job list 41 to the optimization calculation section 4. The job list 41 after narrowing down includes job "J001".

The optimization calculation unit 4 performs an optimization calculation to set jobs narrowed down by the job narrowing unit 3 in job order data 45 in order of priority. The optimization calculation unit 4 here sets the job "J001" to the processing machine "PM001". As a result, the job "J001" having the first priority is set in the job order data 45.

After that, since there are no jobs that can be submitted by the next constraint deadline (maintenance H11 on processing machine "PM001"), the optimization calculation unit 4 creates a schedule for processing machines "PM001" and "PM002". do not. That is, since there are no jobs in the processing machines "PM001" and "PM002" that can be executed by the maintenance H11 of the processing machine "PM001", no jobs are set for the processing machines "PM001" and "PM002".

The optimization calculation unit 4 sets maintenance H11 for the processing machine "PM001" in the job order data 45 in order to eliminate the constraint condition for the processing machine "PM001" (maintenance H11 for the processing machine "PM001"). Further, the optimization calculation unit 4 sets the worker who will execute the maintenance H11 in the job order data 45, and sets the date and time of the work performed by this worker in the job order data 45. In FIG. 8, maintenance H11 performed by the operator on the processing machine "PM001" is illustrated as "PM001_H11".

Next, the job narrowing section 3 selects the third process to be performed from among the consumables replacement process or the maintenance process in the constraint condition list 42A. The job narrowing unit 3 selects the maintenance H12 of the processing machine "PM001" based on the constraint list 42A of FIG. 3, and sets the execution deadline of the maintenance H12 in the job order data 45.

The job narrowing unit 3 extracts from the job list 41 jobs that can be executed by the execution deadline of maintenance H12 on the processing machine “PM001”. That is, the job narrowing section 3 extracts only the list of jobs that can be executed by the execution deadline of maintenance H12 on the processing machine "PM001" from among the list of jobs in the job list 41. The job narrowing down section 3 sends the narrowed down job list 41 to the optimization calculation section 4. The job list 41 after narrowing down does not include a list of jobs.

Regarding processing machines "PM001" and "PM002", after maintenance H11 of processing machine "PM001" is added, there are no jobs that can be submitted by the execution deadline of maintenance H12, so processing machines "PM001" and "PM002" No job is set to .

The optimization calculation unit 4 sets maintenance H12 for the processing machine "PM001" in the job order data 45 in order to eliminate the constraint condition for the processing machine "PM001" (maintenance H12 on the processing machine "PM001"). That is, maintenance H12 is set for processing machine "PM001" following maintenance H11.

Further, the optimization calculation unit 4 sets the worker who will perform the maintenance H12 of the processing machine "PM001" in the job order data 45, and sets the date and time of the work by this worker in the job order data 45. In FIG. 8, maintenance H12 performed by the operator on the processing machine "PM001" is illustrated as "PM001_H12".

Thereafter, the job narrowing unit 3 selects the fourth process to be performed from among the consumables replacement process or the maintenance process in the constraint condition list 42A. The job narrowing unit 3 sets the execution deadline of the selected maintenance etc. in the job order data 45.

The job narrowing section 3 extracts from the job list 41 a list of jobs that can be executed by the deadline set in the job order data 45, and sends the narrowed down job list 41 to the optimization calculation section 4. The job list 41 after narrowing down includes jobs "J003" and "J005." The optimization calculation unit 4 selects jobs that can be executed by the set deadline from the narrowed-down job list 41, and sets them in the job order data 45 by performing optimization calculations.

The optimization calculation unit 4 sets job "J003" to processing machine "PM001" and sets job "J005" to processing machine "PM002." Even if job "J004" is included in the job list 41 after narrowing down, processing machine "PM001" is set for job "J004", and processing machine "PM001" is set for job "J005". PM002" is set, the optimization calculation unit 4 sets job "J005" to the processing machine "PM002."

In this way, the scheduling device 1A sets constraints such as maintenance execution deadlines in the job order data 45, and performs processing to create a narrowed-down job list 41 that extracts jobs that meet the constraints, and an optimization calculation. The process of setting the job to one of the processing machines is repeated.

In the job order data 45 created by the optimization calculation section 4, jobs to be executed by the processing machines and the order of jobs to be executed by each processing machine are set. The optimization calculation unit 4 sends the created job order data 45 to the scheduling unit 5.

The scheduling unit 5 creates a job execution schedule 43 in which job start times and end times are set based on the job order data 45 created by the optimization calculation unit 4 and the job list 41.

In this way, the scheduling device 1A solves an optimization problem under constraint conditions, sets the processing machine to execute the job and the execution order, and then selects a job (such as maintenance) that eliminates the most severe constraint at the time of setting. Add a new one to remove this constraint. Then, the scheduling device 1A solves the optimization problem under the next strictest constraint condition, sets the processing machine to execute the job and the execution order, and then executes the job (maintenance, etc.) that eliminates the most severe constraint condition at that time. Add a new one to remove this constraint. By repeating such schedule processing, the scheduling device 1A can schedule jobs that satisfy the constraint conditions.

FIG. 9 is a diagram illustrating a configuration example of a job execution schedule created by the scheduling device according to the first embodiment. In the job execution schedule 43 created by the scheduling unit 5, a job ID, a processing machine ID, a job start time, and a job end time are associated with each other.

The job execution schedule 43 includes a list of jobs "J001" to "J006" set in the job list 41 and a constraint resolution job list 50 set by the optimization calculation unit 4.

The constraint resolution job list 50 is a list of jobs newly added by the optimization calculation unit 4 in order to eliminate constraint conditions. The constraint resolution job list 50 is a list of jobs such as maintenance that are executed to resolve constraints. The constraint resolution job list 50 includes a list of jobs such as maintenance set by the optimization calculation unit 4 and a list of stopped processing machines set by the optimization calculation unit 4.

The constraint resolution job list 50 in FIG. 9 includes a list of maintenance H11 and H12 of the processing machine "PM001" and a list of maintenance H11 of the processing machine "PM002". In the job execution schedule 43 in FIG. 9, the job ID of the maintenance H11 of the processing machine "PM001" is shown as "PM001_H11", the job ID of the maintenance H12 of the processing machine "PM001" is shown as "PM001_H12", and the job ID of the maintenance H12 of the processing machine "PM001" is shown as "PM001_H12", and The job ID of maintenance H11 is shown as "PM002_H11".

Further, the job execution schedule 43 in FIG. 9 shows a case where "PM003_OUT" is set in the job ID as information indicating the stopped processing machine "PM003". The start time of the stopped processing machine "PM003" is the time when the processing machine starts stopping, and the end time is the time when the processing machine ends stopping.

Generally, when the value standards of the user of the scheduling device or the required specifications for the schedule become complicated, the job list 41 or the constraint list 42A, 42B becomes complicated. As the job list 41 or the constraint lists 42A and 42B become more complex, the policy for determining the order of job execution becomes more complex, and job scheduling becomes impossible on a rule-based basis. Furthermore, even if the scheduling device attempts to find a solution as a combinatorial optimization problem, it takes time to formulate the objective function or constraints, so if even one requirement specification differs, the formulation must be redone. .

On the other hand, in the scheduling device 1A of the first embodiment, after the job narrowing unit 3 narrows down the jobs, the optimization calculation unit 4 selects the processing machines that execute each job and the execution order of the jobs in each processing machine. It has been decided that Thereby, the scheduling device 1A can create the job execution schedule 43 in a short time because it only needs to assign a small number of jobs narrowed down according to the constraint conditions to any processing machine.

Furthermore, since the scheduling device 1A can perform scheduling without setting complicated constraints and objective functions, there is no need to review or add optimization formulas. Furthermore, the scheduling device 1A can perform scheduling without including complicated constraints in the optimization conditional expression. Furthermore, since the scheduling device 1A can simplify the optimization formula, the calculation time can be shortened.

The job narrowing section 3 uses job narrowing conditions such as priority, delivery date, processing machine that can execute the job, shipping destination of the product, information on whether or not the work is in progress, and the initial registration date and time of the job. (Date and time when the job was first registered in the job list 41) or a combination of a plurality of them may be used. That is, the job narrowing unit 3 stores the priority, delivery date, processing machines that can execute the job, shipping destination of the product, information on whether the work is in progress, and the initial registration date and time of the job. At least one of the above is used as a job narrowing condition. In this way, by applying various narrowing conditions, the job narrowing section 3 can prevent workpieces for which jobs are set from stagnation, and can control so that the number of work-in-progress items or completed work items does not increase too much. .

Further, in the scheduling device 1A, a top priority flag may be set by the user for assigning an execution date and time to a job with the highest priority, ignoring all constraints such as execution order constraints and execution date and time constraints. . Since the scheduling device 1A can set the date and time for processing the job with the highest priority flag, it can handle urgent jobs.

As described above, the scheduling device 1A of the first embodiment includes the job narrowing section 3 and the optimization calculation section 4. Then, the job narrowing down section 3 creates a job list 41 after narrowing down the jobs set in the job list 41 to jobs that satisfy the constraint list 42A, 42B. Based on the job list 41 after narrowing down, the optimization calculation unit 4 creates job order data 45, which is data indicating the processing machines that execute the jobs and the order in which the jobs are executed by the processing machines. Thereby, the scheduling device 1A can create the job execution schedule 43 in a short time and provide it to the user.

Embodiment 2.
Next, Embodiment 2 will be described using FIGS. 10 to 17. In the second embodiment, an error in job execution time is learned and scheduling is performed in consideration of the error.

FIG. 10 is a diagram showing the configuration of a scheduling system including a scheduling device according to the second embodiment. Among the components shown in FIG. 10, the components that achieve the same functions as those of the scheduling device 1A of the first embodiment shown in FIG.

The scheduling system 10 includes a scheduling device 1B, a learning device 60, a learned model storage section 65, and an inference device 70. The scheduling device 1B is a computer that creates a job execution schedule 43 similarly to the scheduling device 1A.

In the scheduling system 10, a learning device 60 and an inference device 70 are connected to a learned model storage section 65. Further, an inference device 70 is connected to the scheduling device 1B.

In the second embodiment, the learning device 60 learns the correspondence between the estimated time set in the job list 41 and the error time with respect to the estimated time. Note that the error time with respect to the estimated time is the difference between the estimated time and the actual execution time. The error time may be calculated by the learning device 60 or by a device other than the learning device 60.

The learning device 60 generates a learned model (a learned model 80 to be described later) by learning the correspondence between the estimated time and the error time. The learned model 80 is a model for inferring the error time corresponding to the estimated time. The learning device 60 generates the learned model 80, for example, based on an error list in which estimated times and error times are associated with each other. The learning device 60 transmits the learned model 80 to the learned model storage section 65.

The scheduling device 1B transmits the estimated time set in the job list 41 to the inference device 70. The inference device 70 infers the error time with respect to the estimated time set in the job list 41. The inference device 70 calculates the error time by inputting the estimated time into the trained model 80. The inference device 70 transmits the calculated error time to the optimization calculation unit 4 of the scheduling device 1B. The scheduling device 1B creates a job execution schedule 43 using the error time with respect to the estimated time.

FIG. 11 is a diagram showing the configuration of a learning device included in the scheduling system according to the second embodiment. The learning device 60 includes a data acquisition section 61 and a model generation section 62. The data acquisition unit 61 acquires the estimated time 66A and the error time 67A from outside the learning device 60.

The estimated time 66A acquired by the data acquisition unit 61 is the estimated time set in the job list 41 in the past (estimated time of a job executed in the past). When the learning device 60 generates the trained model 80, the estimated time 66A and the error time 67A are input to the data acquisition unit 61. The data acquisition unit 61 sends the estimated time 66A and the error time 67A to the model generation unit 62 as learning data.

The model generation unit 62 learns the error time 67A corresponding to the estimated time 66A based on the learning data created based on the combination of the estimated time 66A and the error time 67A sent from the data acquisition unit 61. . In other words, the model generation unit 62 learns the error time 67A corresponding to the estimated time 66A based on the learning data created based on the combination of the estimated time 66A and the error time 67A. Here, the learning data is data in which the estimated time 66A and the error time 67A are associated with each other.

Note that the data acquisition unit 61 may acquire the estimated time 66A and error time 67A for each processing machine. In this case, the model generation unit 62 learns the error time 67A corresponding to the estimated time 66A for each processing machine.

Further, the data acquisition unit 61 may acquire the estimated time 66A and error time 67A for each process. In this case, the model generation unit 62 learns the error time 67A corresponding to the estimated time 66A for each process.

Further, the data acquisition unit 61 may acquire the estimated time 66A and error time 67A for each job. In this case, the model generation unit 62 learns the error time 67A corresponding to the estimated time 66A for each job.

Further, the data acquisition unit 61 may acquire the estimated time 66A and error time 67A for each work. In this case, the model generation unit 62 learns the error time 67A corresponding to the estimated time 66A for each work.

Further, the data acquisition unit 61 may acquire the estimated time 66A and error time 67A for each combination of at least two of the processing machine, process, job, and workpiece. In this case, the model generation unit 62 learns the error time 67A corresponding to the estimated time 66A for each combination of at least two of the processing machine, process, job, and workpiece.

The model generation unit 62 can use known algorithms such as supervised learning, unsupervised learning, and reinforcement learning as learning algorithms. As an example, a case will be described in which a neural network is applied to the learning algorithm used by the model generation unit 62.

The model generation unit 62 learns an appropriate error time 67A corresponding to the estimated time 66A, for example, by so-called supervised learning according to a neural network model. Here, supervised learning means that by giving a data set (learning data) of input and result (label) to the learning device 60, it learns the features included in these learning data and extracts the results from the input. A method for inferring

A neural network is composed of an input layer consisting of a plurality of neurons, an intermediate layer (hidden layer) consisting of a plurality of neurons, and an output layer consisting of a plurality of neurons. The intermediate layer may be one layer or two or more layers.

The learned model storage unit 65 stores the learned model 80 generated by the learning device 60. The trained model 80 stored in the trained model storage unit 65 is read out by the inference device 70 when the inference device 70 infers an error time (an error time 67B to be described later).

Note that at least one of the learning device 60, the inference device 70, and the learned model storage unit 65 may be a separate device from the scheduling system 10, which is connected to the scheduling system 10 via a network, for example. good. Furthermore, at least one of the learning device 60, the inference device 70, and the learned model storage unit 65 may be built in the scheduling device 1B. Furthermore, at least one of the learning device 60, the inference device 70, and the learned model storage unit 65 may exist on a cloud server. Further, the learning device 60 and the inference device 70 may be realized by different computers, or the learning device 60 and the inference device 70 may be realized by one computer.

FIG. 12 is a diagram for explaining a neural network used by the learning device according to the second embodiment. For example, in a three-layer neural network as shown in Figure 12, when multiple inputs are input to the input layer (X1 to X3), the values are multiplied by weights W1 (w11 to w16) and the middle layer (Y1 to Y2). Then, the result is further multiplied by weight W2 (w21 to w26) and output from the output layer (Z1 to Z3). This output result changes depending on the values of weight W1 and weight W2.

The neural network used by the learning device 60 of FIG. The error time 67A corresponding to is learned. In other words, the neural network used by the learning device 60 in FIG. 67A, an error time 67A corresponding to the estimated time 66A is learned by so-called supervised learning.

In other words, the neural network learns by adjusting the weights W1 and W2 so that the result output from the output layer by inputting the first input, the estimated time 66A, approaches the second input (correct answer). do.

In this way, the neural network learns by inputting the estimated time 66A into the input layer and adjusting the weights W1 and W2 so that the result output from the output layer approaches the error time 67A. By learning the correspondence between the estimated time 66A and the error time 67A, the neural network generates a trained model 80 that can output an appropriate error time 67A when the estimated time 66A is input. In this way, the learning device 60 learns the trained model 80 that can output the correct error time 67A when the estimated time 66A is input.

The model generation unit 62 generates and outputs the learned model 80 by performing the above learning. The trained model storage unit 65 stores the trained model 80 output from the model generation unit 62.

Next, a processing procedure for the learning device 60 to learn the trained model 80 will be described using FIG. 13. FIG. 13 is a flowchart showing the procedure of learning processing executed by the learning device according to the second embodiment.

The data acquisition unit 61 acquires learning data used for learning (step S110). Specifically, the data acquisition unit 61 acquires an estimated time 66A and an error time 67A. It is assumed that the data acquisition unit 61 acquires the estimated time 66A and the error time 67A at the same time, but the estimated time 66A and the error time 67A may be input in association with each other. Therefore, the data acquisition unit 61 may acquire the estimated time 66A and the error time 67A at different timings. The data acquisition unit 61 sends the estimated time 66A and error time 67A to the model generation unit 62.

The model generation unit 62 executes a learning process using the estimated time 66A and the error time 67A (step S120). Specifically, the model generation unit 62 generates the estimated time 66A by so-called supervised learning according to the learning data created based on the combination of the estimated time 66A acquired by the data acquisition unit 61 and the error time 67A. The error time 67A corresponding to is learned, and a learned model 80 is generated.

After generating the trained model 80, the model generation unit 62 outputs the trained model 80 to the trained model storage unit 65 (step S130). The trained model storage unit 65 stores the trained model 80 generated by the model generation unit 62.

FIG. 14 is a diagram showing the configuration of an inference device included in the scheduling system according to the second embodiment. The inference device 70 includes a data acquisition section 71 and an inference section 72. The data acquisition unit 71 acquires the estimated time 66B from outside the inference device 70 (scheduling device 1B in the second embodiment). The estimated time 66B is the same information as the estimated time 66A. That is, the estimated time 66B is the estimated time set in the job list 41.

The data acquisition unit 71 acquires the estimated time 66B using the same method as the data acquisition unit 61. The data acquisition unit 71 sends the acquired estimated time 66B to the inference unit 72. In the scheduling system 10, the data acquisition section 71 is a first data acquisition section, and the data acquisition section 61 is a second data acquisition section.

The inference unit 72 receives the estimated time 66B sent from the data acquisition unit 71. Further, the inference unit 72 reads out the trained model 80 from the trained model storage unit 65. The inference unit 72 uses the trained model 80 to infer an error time 67B corresponding to the estimated time 66B. That is, by inputting the estimated time 66B acquired by the data acquisition unit 71 into the trained model 80, the inference unit 72 can output the error time 67B inferred from the estimated time 66B. The inference unit 72 transmits the inferred error time 67B to the scheduling device 1B.

In the second embodiment, a case has been described in which the inference device 70 outputs the appropriate error time 67B using the learned model 80 learned by the model generation unit 62 of the scheduling system 10. The trained model 80 may be acquired from an external source such as a scheduling system. In this case, the inference device 70 outputs an appropriate error time 67B based on the learned model 80 obtained from another scheduling system or the like.

Next, with reference to FIG. 15, a processing procedure in which the inference device 70 infers the error time 67B using the learned model 80 will be described. FIG. 15 is a flowchart showing a procedure of inference processing executed by the inference device according to the second embodiment.

The data acquisition unit 71 acquires inference data used for inference of the error time 67B (step S210). Specifically, the data acquisition unit 71 acquires the estimated time 66B. The data acquisition unit 71 sends the estimated time 66B to the inference unit 72. The inference unit 72 acquires the estimated time 66B from the data acquisition unit 71 and acquires the learned model 80 from the learned model storage unit 65.

The inference unit 72 inputs the estimated time 66B to the trained model 80 (step S220) and obtains an appropriate error time 67B.

The inference unit 72 outputs data inferred using the learned model 80 and the estimated time 66B (step S230). Specifically, the inference unit 72 transmits the appropriate error time 67B obtained by the learned model 80 to the scheduling device 1B.

The scheduling unit 5 of the scheduling device 1B creates the job execution schedule 43 using the error time 67B corresponding to the estimated time 66B.

FIG. 16 is a flowchart illustrating a procedure of scheduling processing executed by the scheduling device according to the second embodiment. Note that among the processes shown in FIG. 16, the same steps as those described in FIG. 5 are given the same step numbers, and redundant explanations will be omitted.

The processing of steps S10 and S20 executed by the scheduling device 1B is similar to the processing of steps S10 and S20 executed by the scheduling device 1A. The optimization calculation unit 4 sends the estimated time set in the job order data 45 (the job execution time set in the job execution schedule 43) to the inference device 70. Thereby, the inference device 70 infers the error time corresponding to the estimated time and sends it to the optimization calculation unit 4. The optimization calculation unit 4 acquires the error time corresponding to the estimated time from the inference device 70 (step S25).

The optimization calculation unit 4 determines the processing machine that executes each job and the order in which the jobs are executed in each processing machine based on the constraint list 42A, 42B and the error time. That is, the optimization calculation unit 4 creates the processing machines for the job under the constraint conditions and the execution order in each processing machine using the error time (step S31). In other words, the optimization calculation unit 4 sets the processing machines to execute each job and the execution order of the jobs in each processing machine so as to satisfy the constraints defined in the constraint condition lists 42A and 42B. Sequence data 45 is created using error time.

The optimization calculation unit 4 determines whether there is a constraint condition for which the calculation has not been completed (step S40), and if there is a constraint condition for which the calculation has not been completed (step S40, Yes), the scheduling device 1B performs the step Returning to the process of S20, the processes of steps S20 to S40 are repeated.

On the other hand, if there is no constraint condition for which the calculation has not been completed (step S40, No), the scheduling unit 5 calculates the execution order of the jobs in each processing machine determined by the optimization calculation unit 4 and the estimated time in the job list 41. , and the error time, the execution date and time of each job is set (step S52), and a job execution schedule 43 is created.

The output unit 6 outputs the job execution schedule 43 created by the scheduling unit 5 to an external device such as a display device (step S60). This allows the user to refer to the job execution schedule 43 in which job execution dates and times are set.

Note that the learning device 60 may calculate the number of sets of learning data used by the model generation unit 62 when creating the trained model 80, and may send the number of sets of learning data to the inference device 70. In this case, when inferring the error time corresponding to the job execution time, the inference device 70 calculates the reliability of the inferred error time. The inference device 70 calculates higher reliability as the number of sets of learning data increases. The reliability calculated by the inference device 70 is displayed on a display device or the like. This allows the user to refer to the reliability of the error time.

Further, the job narrowing section 3 may transmit the estimated time to the inference device 70. In this case, the inference device 70 transmits the inferred error time to the job narrowing down section 3. The job narrowing down section 3 narrows down jobs based on the estimated time and error time. The optimization calculation unit 4 sets the job in the job order data 45 based on the estimated time and the error time.

In this way, in the scheduling system 10, the learning device 60 learns the error time corresponding to the estimated time based on the estimated time and error time of jobs executed in the past, and generates the learned model 80. The inference device 70 inputs the estimated time set in the job list 41 into the trained model 80 to infer the error time corresponding to the estimated time. The scheduling device 1B creates a job execution schedule 43 based on the estimated time set in the job list 41 and the estimated error time so that the job satisfies the constraint conditions. Since the scheduling system 10 learns and makes inferences about estimation errors, it is possible to reduce the risk of a job exceeding a constraint condition with an execution deadline.

FIG. 17 is a diagram for explaining the scheduling process executed by the scheduling device according to the second embodiment. In FIG. 17, the optimization calculation unit 4 selects the processing machines that execute each job and the execution order of the jobs in each processing machine based on the job list 41 after narrowing down, the constraint condition lists 42A and 42B, and the error time. A case will be explained in which the following is determined. FIG. 17 shows a state in which the job “J002” and the error time Tx of the job “J002” are set in the job order data 45.

The optimization calculation unit 4 converts the job and the error time into job order data 45 so that the total time of the estimated job time and the error time satisfies the constraint condition (execution deadline for maintenance H11 of the processing machine "PM002"). Set. That is, even if the job satisfies the constraint, if the total time of the job's estimated time and error time does not satisfy the constraint, the optimization calculation unit 4 adds the job and the error time to the job order data 45. Not set.

As described above, the scheduling system 10 of the second embodiment uses the learning device 60 that creates the trained model 80 for inferring the error time from the estimated time of the job, and the trained model 80 to calculate the error time from the estimated time of the job. The inference device 70 infers the error time. Then, the scheduling device 1B creates a job execution schedule 43 using the error time inferred using the learned model 80. Thereby, the scheduling system 10 can reduce the risk of a job exceeding a constraint condition with an execution deadline.

Embodiment 3.
Next, Embodiment 3 will be described using FIGS. 18 to 23. In the third embodiment, after the optimization operation is performed, it is detected whether there is a violation of the constraint conditions (constraint violation), and if there is a violation of the constraint, the optimization operation is performed again.

The scheduling device of Embodiment 3 determines the execution order of jobs including jobs with constraint conditions, and selects jobs that violate constraints (conditions in which jobs with execution restrictions cannot be executed) based on job order data 45. The job order data 45 is temporarily excluded from the job order data 45 according to the constraint conditions.

FIG. 18 is a diagram showing the configuration of a scheduling device according to the third embodiment. Among the components in FIG. 18, components that achieve the same functions as the scheduling device 1A of the first embodiment shown in FIG. 1 are denoted by the same reference numerals, and redundant explanation will be omitted.

The scheduling device 1C is a computer that creates a job execution schedule 43, similar to the scheduling device 1A. The scheduling device 1C includes an input section 2, an optimization calculation section 4, a scheduling section 5, a constraint violation detection section 7, and an output section 6.

In the scheduling device 1C, the input unit 2 is connected to the optimization calculation unit 4, the scheduling unit 5, and the constraint violation detection unit 7, and the optimization calculation unit 4 is connected to the scheduling unit 5. Furthermore, in the scheduling device 1C, the scheduling section 5 is connected to the constraint violation detection section 7. A constraint violation detection section 7 is connected to the output section 6 and the optimization calculation section 4.

The input unit 2 of the scheduling device 1C inputs the job list 41 to the optimization calculation unit 4, the scheduling unit 5, and the constraint violation detection unit 7. The input unit 2 also inputs the constraint list 42A, 42B to the constraint violation detection unit 7.

The optimization calculation unit 4 of the scheduling device 1C determines the processing machine that executes each job and the order in which the jobs are executed by each processing machine, based on the job list 41 before job narrowing down. The optimization calculation unit 4 allocates jobs to processing machines so that the time required to complete all jobs in the job list 41 is shortened.

The scheduling unit 5 creates a job execution schedule 43 for each job, in which a processing machine to execute the job, an execution order, and an execution date and time are set. The scheduling unit 5 sends the created job execution schedule 43 to the constraint violation detection unit 7.

The constraint violation detection unit 7 detects a job that violates a constraint in the job execution schedule 43 or the job order data 45 based on the job execution schedule 43 and the constraint list 42A, 42B. In the third embodiment, the optimization calculation unit 4 determines the processing machines that execute each job and the order in which the jobs are executed in each processing machine, without considering the constraints set in the constraint condition lists 42A and 42B. is determined, the job execution schedule 43 and job order data 45 may include jobs that violate the constraint conditions (jobs that do not satisfy the constraint conditions). Therefore, the constraint violation detection unit 7 detects a job that violates the constraint condition within the job execution schedule 43 or within the job order data 45.

When the constraint violation detection unit 7 detects a job that violates the constraint conditions, it deletes the job that violates the constraint conditions from the job execution schedule 43 and the job order data 45. Furthermore, when the constraint violation detection unit 7 detects a job that violates the constraint conditions, it corrects the job list 41 by deleting from the job list 41 jobs that do not violate the constraint conditions. That is, the constraint violation detection unit 7 creates a modified job list 41 in which jobs that do not violate the constraint conditions are deleted from the job list 41. In this way, the constraint violation detection unit 7 determines the execution schedule of a job that does not violate any constraint conditions, and deletes this job from the job list 41. As a result, jobs that violate the constraint conditions remain in the job list 41.

Further, the constraint violation detection unit 7 adds a job corresponding to the constraint condition (a job for resolving the constraint condition) to the job list 41. The constraint violation detection unit 7 adds, to the job list 41, a job such as a maintenance job that is executed to eliminate a constraint condition, for example.

The constraint violation detection unit 7 sends to the optimization calculation unit 4 a job list 41 in which jobs that do not violate the constraint conditions are deleted and jobs such as maintenance jobs are added. Further, the constraint violation detection unit 7 sends job order data 45 in which jobs violating the constraint conditions have been deleted to the optimization calculation unit 4. Further, the constraint violation detection unit 7 sends the job execution schedule 43 in which the job violating the constraint has been deleted to the optimization calculation unit 4.

When the constraint violation detection unit 7 does not detect a job that violates the constraint conditions, it sends the job execution schedule 43 to the output unit 6. The output unit 6 outputs the job execution schedule 43 to an external device such as a display device.

When the optimization calculation unit 4 receives the corrected job list 41 from the constraint violation detection unit 7, the optimization calculation unit 4 determines the processing machine that executes each job and the job performance of each processing machine based on this job list 41. Determine the execution order. That is, the optimization calculation unit 4 selects the processing machine that executes each job and the processing machine in each processing machine based on the job list 41 in which jobs that do not violate the constraint conditions have been deleted and jobs such as maintenance have been added. Determine the job execution order. Here again, the optimization calculation unit 4 allocates jobs to processing machines so that the time required to complete all jobs in the job list 41 is shortened. The optimization calculation unit 4 adds the determined processing machines that execute each job and the execution order of the jobs in each processing machine to the modified job list 41. Then, the optimization calculation unit 4 leaves jobs that do not violate the constraint conditions in the job order data 45, and deletes jobs that violate the constraint conditions from the job order data 45.

In the scheduling device 1C, the process executed by the optimization calculation unit 4, the process executed by the scheduling unit 5, and the process executed by the constraint violation detection unit 7 are repeated. That is, the optimization calculation unit 4 determines the execution order of jobs in each processing machine, the scheduling unit 5 creates the job execution schedule 43, and the constraint violation detection unit 7 determines the job execution order in the job execution schedule 43. The process of detecting jobs that violate the constraint conditions is repeated.

FIG. 19 is a flowchart illustrating the procedure of scheduling processing executed by the scheduling device according to the third embodiment. Note that among the processes shown in FIG. 19, the same steps as those described in FIG. 5 are given the same step numbers, and redundant explanations will be omitted.

The processing of steps S10 and S60 executed by the scheduling device 1C is similar to the processing of steps S10 and S60 executed by the scheduling device 1A. The input unit 2 reads the job list 41 and the constraint lists 42A and 42B from an external device or the like (step S10). The input unit 2 inputs the job list 41 to the optimization calculation unit 4, the scheduling unit 5, and the constraint violation detection unit 7. The input unit 2 also inputs the constraint list 42A, 42B to the constraint violation detection unit 7.

Based on the job list 41, the optimization calculation unit 4 determines the processing machine that executes each job and the order in which the jobs are executed by each processing machine. That is, the optimization calculation unit 4 creates a processing machine and an execution order for the job (step S30a).

The scheduling unit 5 sets the execution date and time of each job based on the execution order of the jobs in each processing machine determined by the optimization calculation unit 4 and the estimated time of the job list 41 (step S50a), and sets the execution date and time of each job in the job execution schedule 43. Create.

The constraint violation detection unit 7 determines whether or not there is a job that violates the constraint conditions in the job execution schedule 43, based on the job execution schedule 43 and the constraint condition lists 42A and 42B (step S55).

If there is a job that violates the constraint condition in the job execution schedule 43 (step S56, Yes), the constraint violation detection unit 7 deletes the job that does not violate the constraint condition from the job list 41, and deletes the job that does not violate the constraint condition. The violating job is left in the job list 41. Further, the constraint violation detection unit 7 adds a job for resolving the constraint condition to the job list 41. In this manner, the constraint violation detection unit 7 deletes jobs that do not violate the constraint conditions from the job list 41, and adds jobs for resolving the constraint conditions to the job list 41 (step S57). The job list 41 in which jobs that do not violate the constraint conditions are deleted and jobs for resolving the constraint conditions are added is a job list 41X, which will be described later.

The constraint violation detection unit 7 sends to the optimization calculation unit 4 a job list 41X in which jobs violating the constraint conditions have been deleted and jobs for resolving the constraint conditions have been added. The scheduling device 1C returns to the process of step S30a and repeats the processes of steps S30a to S56.

If there is no job that violates the constraint conditions in the job execution schedule 43 (step S56, No), the output unit 6 outputs the job execution schedule 43 created by the scheduling unit 5 to an external device such as a display device ( Step S60). This allows the user to refer to the job execution schedule 43 in which job execution dates and times are set.

In this way, the scheduling device 1C determines the processing machines that execute each job and the order in which the jobs are executed by each processing machine, without using the constraint list 42A, 42B. That is, the scheduling device 1C determines the execution order of jobs including jobs with constraint conditions. If a constraint condition that prevents the job from being executed is met, the scheduling device 1C automatically excludes the job from the job execution schedule 43 and does not allocate execution time to the job. As a result, the scheduling device 1C can exclude only jobs that do not satisfy the constraint conditions from the job execution schedule 43 after executing scheduling on the premise that all jobs will be executed without excluding jobs from the beginning. The rules for excluding jobs that do not meet the criteria can be simplified. Thereby, the scheduling device 1C can create the job execution schedule 43 that satisfies the constraint conditions with fewer calculations and shorter calculation time than the scheduling device 1A.

FIG. 20 is a diagram for explaining the first stage of the scheduling process executed by the scheduling device according to the third embodiment. FIG. 21 is a diagram for explaining the second stage of the scheduling process executed by the scheduling device according to the third embodiment.

Based on the job list 41, the optimization calculation unit 4 allocates each job to a processing machine in order of priority. For example, the optimization calculation unit 4 sets job "J001" to processing machine "PM001", sets job "J002" to processing machine "PM003", and sets job "J003" to processing machine "PM003". Set to "PM001". Further, the optimization calculation unit 4 sets the job "J004" (not shown in FIGS. 20 and 21) to the processing machine "PM001", and sets the job "J005" to the processing machine "PM002". Then, job "J006" is set to processing machine "PM003". Note that the job "J002" may be set to the processing machine "PM001".

The scheduling unit 5 creates a job execution schedule 43 based on the job execution order in each processing machine determined by the optimization calculation unit 4 and the estimated time of the job list 41.

The constraint violation detection unit 7 determines whether or not there is a job that violates the constraint conditions in the job execution schedule 43, based on the job execution schedule 43 and the constraint condition lists 42A and 42B.

For example, when the job execution schedule 43 corresponds to the job order data 45 shown in FIG. , it is determined that the constraint condition is violated.

In addition, the constraint violation detection unit 7 detects that the job "J005" in which the maintenance H11 of "PM002" is not set by the deadline and the execution deadline of the maintenance H11 of "PM002" has been set is It is determined that the constraint condition is violated.

In addition, the constraint violation detection unit 7 detects that a job " J003" and "J004" are determined to violate the constraint condition.

FIG. 21 shows a state in which the constraint violation detection unit 7 has extracted a job that violates the constraints. The constraint violation detection unit 7 leaves the job that violates the constraint in the job list 41 and deletes the job "J001" that does not violate the constraint from the job list 41.

FIG. 22 is a diagram showing the structure of a job list modified by the scheduling device according to the third embodiment. As described above, the constraint violation detection unit 7 of the scheduling device 1C leaves the jobs "J002" to "J006" that violate the constraints in the job list 41, and leaves the job "J001" that does not violate the constraints in the job list 41. Delete from 41. Thereby, the constraint violation detection unit 7 generates a job list 41X by modifying the job list 41.

Furthermore, the constraint violation detection unit 7 adds a job for resolving the constraint condition to the job list 41X. FIG. 22 shows a case where "PM001_H11", "PM001_H12", "PM002_H11", and "PM003_OUT" are added to the job list 41X as job IDs of jobs for resolving constraint conditions.

As described above, "PM003_OUT" indicates that the processing machine "PM003" is stopped. Further, "PM001_H11" and "PM001_H12" respectively indicate maintenance H11 and H12 to the processing machine "PM001", and "PM002_H11" indicates maintenance H12 to the processing machine "PM002". Note that the delivery date of the job for resolving the constraint is the job execution deadline.

The constraint violation detection unit 7 sets the highest priority to the stopped processing machine. The constraint violation detection unit 7 here sets the first priority to "PM003_OUT". In addition, the constraint violation detection unit 7 sets a higher priority for jobs that have closer execution deadlines (delivery dates) among jobs for resolving constraint conditions. Furthermore, the constraint violation detection unit 7 lowers the priority of the original job set in the job list 41 in the job list 41X.

The optimization calculation unit 4 determines the processing machine that executes each job and the order in which the jobs are executed by each processing machine, based on the job list 41X. FIG. 23 is a diagram for explaining the third stage of the scheduling process executed by the scheduling device according to the third embodiment.

The optimization calculation unit 4 allocates each job to a processing machine in order of priority based on the job list 41X. In this case, if there is a job that can be executed before maintenance, the optimization calculation unit 4 sets the job before maintenance, even if the job has a lower priority than maintenance.

The optimization calculation unit 4 here sets "PM003_OUT" indicating that the processing machine "PM003" is stopped, based on the job order data 45. In addition, the optimization calculation unit 4 sets job "J002" that can be executed before maintenance H11 of processing machine "PM002" to processing machine "PM002", and then performs maintenance H11 on processing machine "PM002". Set.

In addition, the optimization calculation unit 4 sets job "J001" that can be executed before maintenance H11 of processing machine "PM001" to processing machine "PM001", and then sets maintenance H11 to processing machine "PM001". do. Further, similarly to the first embodiment, the optimization calculation unit 4 sets maintenance H12 to the processing machine "PM001", and then sets job "J003" to the processing machine "PM001". Further, the optimization calculation unit 4 sets job "J005" to processing machine "PM002".

In the scheduling device 1C, the process executed by the optimization calculation unit 4, the process executed by the scheduling unit 5, and the process executed by the constraint violation detection unit 7 are repeated. If the job execution schedule 43 created by the scheduling unit 5 does not include a job that violates the constraint conditions, the constraint violation detection unit 7 sets a maintenance job in the job execution schedule 43 based on the constraint list 42B. .

This allows the scheduling device 1C to create the same job execution schedule 43 as the scheduling device 1A. Note that the constraint violation detection unit 7 detects the job execution schedule 43 at any timing as long as there is no constraint violation in a job that eliminates a constraint condition such as maintenance, and the setting of this job to the job execution schedule 43 is confirmed. You can also set up maintenance jobs.

Note that the scheduling device 1C may be applied to the scheduling system 10 described in the second embodiment. In this case, the scheduling system 10 includes a scheduling device 1C, a learning device 60, a learned model storage section 65, and an inference device 70. In this scheduling system 10 as well, the inference device 70 is connected to the scheduling section 5 and sends the inferred error time to the scheduling section 5.

As described above, the scheduling device 1C of the third embodiment includes the optimization calculation section 4 and the constraint violation detection section 7. Then, the optimization calculation section 4 creates job order data 45 without considering the constraint conditions, and the constraint violation detection section 7 deletes jobs that do not satisfy the constraint conditions from the job execution schedule 43 and the job order data 45. A modified job list 41 is created by deleting jobs that meet the constraint conditions and adding jobs to the job list 41 to eliminate the constraint conditions. The optimization calculation unit 4 solves an optimization problem based on the corrected job list 41 to execute the jobs so that all the jobs in the corrected job list 41 are completed in the shortest time. The processing machine and the execution order of jobs on the processing machine are determined and added to the job order data 45. Thereby, the scheduling device 1C can simplify the rules for excluding jobs that do not satisfy the constraint conditions. Therefore, the scheduling device 1C can create a job execution schedule 43 that satisfies the constraint conditions in a short time.

Here, the hardware configuration of the scheduling devices 1A to 1C will be explained. Note that since the scheduling devices 1A to 1C have similar hardware configurations, the hardware configuration of the scheduling device 1A will be described here.

FIG. 24 is a diagram illustrating an example of a hardware configuration for realizing the scheduling device according to the first embodiment. Scheduling device 1A can be realized by input device 300, processor 100, memory 200, and output device 400. An example of the processor 100 is a CPU (Central Processing Unit, also referred to as a central processing unit, processing unit, arithmetic unit, microprocessor, microcomputer, or DSP (Digital Signal Processor)) or a system LSI (Large Scale Integration). Examples of the memory 200 are RAM (Random Access Memory) and ROM (Read Only Memory).

The scheduling device 1A is realized by the processor 100 reading and executing a computer-executable scheduling program stored in the memory 200 for executing the operations of the scheduling device 1A. The scheduling program, which is a program for executing the operations of the scheduling device 1A, can also be said to cause a computer to execute the procedure or method of the scheduling device 1A.

The scheduling program executed by the scheduling device 1A has a module configuration including a job narrowing section 3, an optimization computing section 4, and a scheduling section 5. is loaded onto the main storage, and the job narrowing section 3, optimization calculation section 4, and scheduling section 5 are generated on the main storage.

The input device 300 receives the job list 41 and the constraint lists 42A and 42B and sends them to the processor 100.

Memory 200 stores scheduling programs and the like. Furthermore, the memory 200 is used as a temporary memory when the processor 100 executes various processes. The output device 400 outputs a job execution schedule 43.

The scheduling program may be provided as a computer program product by being stored in a computer readable storage medium in an installable or executable file format. Further, the scheduling program may be provided to the scheduling device 1A via a network such as the Internet. Note that some of the functions of the scheduling device 1A may be realized by dedicated hardware such as a dedicated circuit, and some may be realized by software or firmware.

Note that the hardware configuration of the job narrowing section 3 may be the hardware configuration shown in FIG. 24. Further, the hardware configuration of the optimization calculation section 4 may be the hardware configuration shown in FIG. 24. Further, the hardware configuration of the scheduling section 5 may be the hardware configuration shown in FIG. 24. Further, the hardware configuration of the learning device 60 may be the hardware configuration shown in FIG. 24. Further, the hardware configuration of the inference device 70 may be the hardware configuration shown in FIG. 24.

The configurations shown in the embodiments above are merely examples, and can be combined with other known techniques, or can be combined with other embodiments, within the scope of the gist. It is also possible to omit or change part of the configuration.

1A to 1C scheduling device, 2 input unit, 3 job narrowing unit, 4 optimization calculation unit, 5 scheduling unit, 6 output unit, 7 constraint violation detection unit, 10 scheduling system, 41, 41X job list, 42A, 42B constraint conditions List, 43 Job execution schedule, 45 Job order data, 50 Constraint resolution job list, 60 Learning device, 61, 71 Data acquisition unit, 62 Model generation unit, 65 Learned model storage unit, 66A, 66B Estimated time, 67A, 67B error time, 70 inference device, 72 inference unit, 80 learned model, 100 processor, 200 memory, 300 input device, 400 output device.

Claims (12)

Based on a job list that is a list of a plurality of jobs executed by a plurality of processing machines and a constraint list that is a list of constraints for the processing machines, the constraints are selected from among the jobs set in the job list. a job narrowing unit that creates a job list after narrowing down, which is a job list narrowed down to jobs that meet conditions;
By solving an optimization problem based on the job list after narrowing down, the processing machine that executes the job and the processing machine in the processing machine are selected so that the time to complete all jobs in the job list is the shortest. an optimization calculation unit that determines a job execution order and creates job order data that is data indicating a processing machine that executes the job and the job execution order in the processing machine;
a scheduling unit that creates a job execution schedule in which execution dates and times of the jobs for each of the processing machines are set based on the job order data and the job list;
Equipped with
a process in which the job narrowing unit creates the narrowed-down job list that satisfies a first constraint among the constraint conditions;
a process in which the optimization calculation unit creates the job order data that satisfies the first constraint condition, and adds a job to the job order data to eliminate the first constraint condition;
a process in which the job narrowing unit creates the narrowed-down job list that satisfies a second constraint that is the next strictest constraint after the first constraint among the constraints;
a process in which the optimization calculation unit creates the job order data that satisfies the second constraint condition, and adds a job to the job order data to eliminate the second constraint condition;
is repeated,
A scheduling device characterized by: The optimization calculation unit sets a processing machine to execute the job and an execution order of the job in the processing machine so as to satisfy the constraint.
The scheduling device according to claim 1, characterized in that: The job that satisfies the constraint condition is a job that can be completed by the deadline for performing maintenance work on the processing machine.
The scheduling device according to claim 1, characterized in that: A job that satisfies the constraint is a job that can be completed within the lifespan of the consumables of the processing machine.
The scheduling device according to claim 1, characterized in that: A job that satisfies the constraint is a job to which a worker can be assigned;
The scheduling device according to claim 1, characterized in that: The optimization calculation unit sets a processing machine to execute the job and a job execution order in the processing machine in the job order data, and then selects a job to eliminate the most severe constraint condition at the time of setting. Lifting the most severe constraint by adding it to the job order data;
The scheduling device according to claim 1, characterized in that: The job narrowing unit includes a priority order for executing the job, a delivery date of a product manufactured by executing the job, a processing machine that can execute the job, a shipping destination of the product, and the job. using at least one of information as to whether the work to be executed is in progress and the date and time when the job was first registered in the job list as job narrowing conditions;
The scheduling device according to claim 1, characterized in that: By solving an optimization problem based on a job list, which is a list of multiple jobs to be executed by multiple processing machines, the jobs are arranged so that all the jobs in the job list complete in the shortest time. Determine the processing machine to be executed and the order in which the jobs are executed by the processing machines, and create job order data that is data indicating the processing machine to be executed and the order in which the jobs are executed by the processing machines. an optimization calculation section;
a scheduling unit that creates a job execution schedule in which execution dates and times of the jobs for each of the processing machines are set based on the job order data and the job list;
When a job that does not satisfy the constraint conditions for the processing machine is detected among the jobs set in the job execution schedule or the job order data, the job that does not satisfy the constraint conditions is deleted from the job execution schedule and the job order data. and a constraint violation detection unit that creates a modified job list by deleting a job that satisfies the constraint condition from the job list and adding a job to eliminate the constraint condition to the job list;
Equipped with
The optimization calculation unit creates the job order data without considering the constraint conditions,
The optimization calculation unit executes the jobs so that all jobs in the revised job list can be completed in the shortest time by solving an optimization problem based on the revised job list. determining a processing machine to be executed and a job execution order in the processing machine, and adding data indicating a processing machine to execute the job and a job execution order in the processing machine to the job order data;
A scheduling device characterized by: A scheduling device according to any one of claims 1 to 8,
an inference device that infers an error time that is an error in the estimated time required to execute the job;
has
The inference device includes:
a first data acquisition unit that acquires the estimated time;
an inference unit that infers the error time from the estimated time using a trained model for inferring the error time corresponding to the estimated time;
Equipped with
The optimization calculation unit creates the job order data using the error time.
A scheduling system characterized by: further comprising a learning device that generates the learned model based on the estimated time and the error time,
The learning device includes:
a second data acquisition unit that acquires the estimated time and the error time as learning data;
a model generation unit that generates the learned model using the learning data;
Equipped with
10. The scheduling system according to claim 9. A scheduling device selects jobs set in the job list based on a job list that is a list of a plurality of jobs to be executed by a plurality of processing machines and a constraint list that is a list of constraints for the processing machines. a job narrowing step of creating a job list after narrowing down, which is a job list narrowed down to jobs that satisfy the constraint conditions;
a processing machine in which the scheduling device solves an optimization problem based on the narrowed-down job list to execute the jobs so that all jobs in the job list are completed in the shortest time; an optimization calculation step of determining a job execution order in the processing machine and creating job order data that is data indicating a processing machine to execute the job and the job execution order in the processing machine;
a scheduling step in which the scheduling device creates a job execution schedule in which the execution date and time of the job for each processing machine is set based on the job order data and the job list;
including;
The scheduling device includes:
a process of creating the job list after narrowing down that satisfies a first constraint condition among the constraint conditions;
creating the job order data that satisfies the first constraint condition, and adding a job to the job order data to eliminate the first constraint condition;
A process of creating the narrowed-down job list that satisfies a second constraint that is the next strictest constraint after the first constraint among the constraints;
creating the job order data that satisfies the second constraint condition, and adding a job to the job order data to eliminate the second constraint condition;
repeat,
A scheduling method characterized by: Based on a job list that is a list of a plurality of jobs executed by a plurality of processing machines and a constraint list that is a list of constraints for the processing machines, the constraints are selected from among the jobs set in the job list. a job narrowing step for creating a job list after narrowing down, which is a job list narrowed down to jobs that meet conditions;
By solving an optimization problem based on the job list after narrowing down, the processing machine that executes the job and the processing machine in the processing machine are selected so that the time to complete all jobs in the job list is the shortest. an optimization calculation step of determining a job execution order and creating job order data that is data indicating a processing machine to execute the job and the job execution order in the processing machine;
a scheduling step of creating a job execution schedule in which execution dates and times of the jobs for each of the processing machines are set based on the job order data and the job list;
make the computer run
a process of creating the job list after narrowing down that satisfies a first constraint condition among the constraint conditions;
creating the job order data that satisfies the first constraint condition, and adding a job to the job order data to eliminate the first constraint condition;
A process of creating the narrowed-down job list that satisfies a second constraint that is the next strictest constraint after the first constraint among the constraints;
creating the job order data that satisfies the second constraint condition, and adding a job to the job order data to eliminate the second constraint condition;
is repeated,
A scheduling program characterized by:

Images