Production line scheduling method, production line system and computer readable storage medium

Abstract

The present invention provides a production line scheduling method, which is applicable to multiple tasks of an order, wherein the tasks all pass through a bottleneck site with a production machine, and the tasks correspond to multiple working conditions respectively. The production line scheduling method includes: executing multiple scheduling simulation calculations on the tasks to sequentially construct multiple scheduling simulation trees, and obtaining the work schedule and the simulated completion time of each of the tasks based on the scheduling simulation trees, and calculating the expected input time of each of the tasks at all sites based on the simulated completion time of each of the tasks, wherein each scheduling simulation tree includes at least one scheduling path, and each of the at least one scheduling path is generated by one scheduling simulation calculation, and the scheduling simulation calculation includes executing a deployment stage based on the deployment conditions and the working conditions to obtain the scheduling path.

Description

Technical Field

[0001] The present invention relates to a production line scheduling method, a production line system and a computer-readable storage medium. Background Art

[0002] Scheduling of production lines is more complicated due to various factors such as product number, expected delivery date, and available machines. In particular, it is more likely to cause resource allocation difficulties at processing sites where multiple product numbers pass through, or sites where there are insufficient machines or fixtures due to expensive machines or fixtures.

[0003] In order to make the scheduling of bottleneck sites smoother, existing technologies include Gurobi's linear programming, Monte Carlo tree search (MCTS) and variable neighborhood descent algorithm to derive scheduling data. However, no matter which method is used, the number of line changes and the delivery rate in the obtained scheduling solution are still not ideal. In addition, although the existing technology includes combining Monte Carlo tree search to shorten the total scheduling make span, because the Monte Carlo tree search in the past needs to perform simulation on all possible arrangements to obtain a more objective scheduling result, when the number of operations to be scheduled is large, it will cause a huge amount of scheduling calculations and a very long scheduling time. Therefore, it should be found that ways to improve scheduling efficiency. Summary of the invention

[0004] In view of the above, the present invention provides a production line scheduling method, a production line system and a computer-readable storage medium that can meet the above requirements.

[0005] A production line scheduling method according to an embodiment of the present invention is applicable to multiple tasks of an order, wherein the tasks all pass through a bottleneck site having at least one production machine, and the tasks correspond to multiple working conditions respectively. The production line scheduling method includes: executing a scheduling simulation calculation multiple times on the tasks to sequentially construct multiple scheduling simulation trees, and obtaining a work schedule and a simulated completion time for each of the tasks based on the scheduling simulation trees, wherein each of the scheduling simulation trees includes at least one scheduling path, and each of the at least one scheduling path is generated by one scheduling simulation calculation; and based on the simulated completion time of each of the tasks, calculating multiple expected input times of each of the tasks at multiple sites including the bottleneck site, wherein the scheduling simulation calculation includes: executing a spread point stage based on at least one spread point condition and the working conditions to obtain the scheduling path.

[0006] A production line system according to one embodiment of the present invention includes: a computing device for executing the production line scheduling method to obtain the work schedule; and a plurality of available production machines, which are communicatively connected to the computing device. After completing a current job, each of the available production machines obtains a pending job among the jobs from the computing device based on the work schedule. Each of the available production machines also outputs a processing parameter notification based on a product material number of the pending job.

[0007] A computer-readable storage medium according to an embodiment of the present invention includes a program, and a computing device executes the program to execute the production line scheduling method.

[0008] In summary, according to the production line scheduling method, production line system and computer-readable storage medium shown in one or more embodiments of the present invention, the expansion stage in the Monte Carlo tree search can be efficiently executed, and the simulation score of the Monte Carlo tree search can be calculated in advance to shorten the time to obtain the scheduling solution for each job. In addition, according to the production line scheduling method, production line system and computer-readable storage medium shown in one or more embodiments of the present invention, the expected input time of each job at each station can be back-pushed based on the completion time of each job at the bottleneck station to serve as the basis for the subsequent scheduling of the entire production line.

[0009] The above description of the content of the present invention and the following description of the implementation modes are only used to exemplarily illustrate and explain the spirit and principle of the present invention, and to further explain the scope of the patent application of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 The flowchart of the production line scheduling method according to one embodiment of the present invention is schematically shown.

[0011] Figure 2A and 2B A detailed flow chart of a production line scheduling method according to an embodiment of the present invention is schematically shown.

[0012] Figure 3A and 3B An exemplary diagram of a scheduling simulation tree according to multiple embodiments of the present invention is schematically shown.

[0013] Figure 4 An example diagram of a processing step is schematically shown.

[0014] Figure 5 A flowchart of a production line scheduling method according to another embodiment of the present invention is schematically shown.

[0015] Figure 6 A block diagram of a production line system according to an embodiment of the present invention is schematically shown.

[0016] Description of Reference Numerals

[0017] r1: first path

[0018] r2: Second path

[0019] R: Node

[0020] J1~J5:Work

[0021] M1~M7: Available production machines

[0022] 100: Computing device DETAILED DESCRIPTION

[0023] The detailed features and advantages of the present invention are described in detail in the following embodiments, and the contents are sufficient to enable any person skilled in the art to understand the technical content of the present invention and implement it accordingly, and according to the contents disclosed in this specification, the scope of the patent application and the drawings, any person skilled in the art can easily understand the relevant purposes and advantages of the present invention. The following examples further illustrate the viewpoints of the present invention in detail, but are not intended to limit the scope of protection of the present invention.

[0024] It should be noted that the production line scheduling method of the present invention is applicable to multiple jobs of an order, wherein all of these jobs pass through a bottleneck site having at least one production machine, and these jobs correspond to multiple working conditions, that is, this bottleneck site has at least one production machine to execute the processing procedures of these jobs at the bottleneck site. For example, the products in the order can be pre-processed based on the product material number, processing procedure, processing site, available machine, available fixture and standard working hours to establish the following resource utilization tables A and B, and before establishing the resource utilization tables A and B, the work in progress (WIP) and inventory (Bank) in each site are first deducted, wherein the bottleneck site is, for example, one or more of each site in the resource utilization tables A and B.

[0025] Resource Utilization Table A

[0026]

[0027]

[0028] Resource Utilization Table B

[0029]

[0030] As shown in the following work table, resource utilization tables A and B can be further converted into work A corresponding to product material number AAA and work B corresponding to product material number BBB in the bottleneck site, and work A and work B respectively have work conditions such as product material number, processing path, and expected input time. It should be noted that although the above and subsequent contents only illustrate the situation that there is work A for producing product material number AAA and work B for producing product material number BBB, the above work A and work B are not used to limit the number of work that can be scheduled in this case, that is, the number of work that can be scheduled in this case can be two or even more.

[0031] Worksheet

[0032]

[0033] Please refer to Figure 1 , Figure 1 The flowchart of the production line scheduling method according to one embodiment of the present invention is schematically shown.

[0034] Step S10: Execute multiple scheduling simulation calculations on the tasks to sequentially construct multiple scheduling simulation trees, and obtain a task schedule and a simulated completion time of each of the tasks based on the scheduling simulation trees.

[0035] That is, each scheduling simulation tree includes at least one scheduling path, and each scheduling path is generated by one scheduling simulation calculation, and the work schedule and the simulated completion time of each work at the bottleneck site are generated based on the results of these scheduling simulation trees, wherein the scheduling simulation calculation includes executing a development point stage based on at least one development point condition and these working conditions to obtain the scheduling path. Taking the above-mentioned work A and work B as examples, since they have corresponding working conditions respectively, and these working conditions include working conditions related to time (for example, delivery date and expected input time), the computing device can perform multiple scheduling simulation calculations on work A and work B based on the working conditions such as product material number, processing path, delivery date and expected input time shown in the above work table to construct multiple scheduling simulation trees, and obtain the work schedule of work A and work B including the bottleneck site and the simulated completion time of work A and work B at the bottleneck site based on these scheduling simulation trees.

[0036] Step S20: Based on the simulated completion time of each task, the expected time required for each task at all sites is calculated.

[0037] The computing device executes step S20 to calculate the expected input time of each job at all sites including the bottleneck site. For example, when the simulated completion time of job A at the bottleneck site is estimated, the computing device can determine the expected input time of product number AAA at each of the above sites based on the simulated completion time and the required processing time of sites 1, 3, 5, 7 and 10. The detailed implementation of steps S10 and S20 will be described below.

[0038] Please refer to Figure 2A and Figure 3A ,in Figure 2A A detailed flow chart of a production line scheduling method according to an embodiment of the present invention is schematically shown, and the scheduling simulation calculation is implemented based on the Monte Carlo tree search (MCTS) architecture, for example; Figure 3A An example diagram of a scheduling simulation tree in the above-mentioned tree search process is schematically shown in an embodiment of the present invention. Figure 2A , the steps S101 to S111 of the scheduling simulation calculation are shown, wherein the scheduling simulation calculation is used to construct a scheduling simulation tree to schedule the work nodes corresponding to the multiple jobs respectively associated with the jobs in the scheduling path in the scheduling simulation tree, wherein each of the jobs corresponds to a job such as the aforementioned job A or job B, and the work nodes arranged in the scheduling simulation tree include a job and an available production machine for processing the job. For the convenience of explanation, the jobs that have not been arranged in the scheduling simulation tree will be referred to as candidate jobs.

[0039] In other words, each work node is a combination of a work and one or more available machines, and the available machines are used to process the material number corresponding to the work. For example, assuming that there are five available production machines at a bottleneck site, and both candidate work A and candidate work B will pass through the bottleneck site, the computing device can determine candidate work A and candidate work B and their corresponding production machines based on the exhibition point conditions and generate work nodes corresponding to work A and work B accordingly. Assuming that the computing device determines based on the exhibition point conditions that candidate work A is preferably scheduled before candidate work B, and determines based on the exhibition point conditions that the work node of candidate work A is a combination of work A and two of the five machines, the way in which the computing device generates the work node of candidate work B can also be based on the exhibition point conditions, determine the machine that can be used to process work B among the remaining three machines, and combine work B and the machine used to process work B into a work node, but the present invention is not limited to this.

[0040] Step S101: Select the current node.

[0041] In this step S101, the computing device specifies a current node from the nodes listed in the scheduling simulation tree as the basis for the subsequent serial connection of the pre-selected node. If this scheduling simulation tree is connected to a selected node, and there are no other nodes in this scheduling simulation tree, the selected node is designated as the current node, which is the initial node of this scheduling simulation tree; and if there are no nodes in this scheduling simulation tree, a random node or an empty node can be used as the initial node of this scheduling simulation tree. In addition, the current node selected in this step S101 can also be a pre-selected node in the scheduling path that already exists in the scheduling simulation tree. The relevant descriptions of the current node, pre-selected node, selected node and scheduling path described herein will be further described below.

[0042] For example, assuming that the total number of jobs is 100 and the schedules of 5 of the jobs have been selected, the computing device may select the combination of the job node (selected node) of the fifth job selected in the schedule and the available production machine as the current node (initial node) of the scheduling simulation tree, i.e., the root node (root, such as Figure 3A The node R shown in the figure); the other 95 jobs that have not been selected can be regarded as candidate jobs. In addition, if the schedules of the 100 jobs are not selected, the initial node can be a null node or a combination of a randomly selected job and an available production machine as a job node.

[0043] The above step S101 is the selection phase (Selection) of the Monte Carlo tree search of this case, and then the expansion phase (Expansion) of the Monte Carlo tree search of this case is performed in the following steps S103 to S107, and the simulation phase (Simulation) of the Monte Carlo tree search of this case is performed in step S109 to obtain the simulation scores of each pre-selected node directly or indirectly derived from the current node in the expansion phase, and then the backpropagation phase (Backpropagation) of the Monte Carlo tree search of this case is completed in step S111 to update the simulation scores of each pre-selected node connected to the current node, and the selected node that can be used as the next current node is determined from these pre-selected nodes based on the simulation scores. It should be explained that the expansion phase is used to schedule the candidate tasks in the scheduling path respectively, wherein the expansion phase includes multiple selection procedures, and each selection procedure includes the following steps S103 to S105 (first selection subroutine) and step S107 (second selection subroutine). The above steps will be further explained below.

[0044] Step S103: determining whether at least one of the tasks meets at least one of the plurality of deployment conditions.

[0045] Since each job corresponds to one or more job conditions, the computing device can select a node suitable for connecting to the current node from the candidate jobs based on the job conditions and the exhibition point conditions, wherein each of the exhibition point conditions preferably includes at least one of a material number condition, a fixture handling condition, and a production start time condition, but is not limited thereto. For example, the material number condition can be that the product material number in the job condition is the same as the current material number of the current node, or that the expected investment time in the job condition falls within a first threshold time and the product material number is the same as the current material number of the current node, wherein the first threshold time is, for example, two days; the fixture handling condition can be that a machine fixture in the job condition is the same as the current machine fixture of the current node; the production start time condition can be that the available machine time in the job condition is the earliest one that can start operation among the job conditions, or that the expected investment time falls within a second threshold time, and this second threshold time is shorter than the aforementioned first threshold time. In addition, the exhibition point conditions can also include factors such as daily production capacity conditions and line change times, and the present invention does not limit the exhibition point conditions. In other words, by using the exhibition point conditions as the basis for selecting suitable candidate tasks as work nodes, the construction of the scheduling simulation tree can be made more efficient.

[0046] Step S104: Randomly select a candidate task as a pre-selected node. When the computing device determines that the task conditions do not meet the deployment point conditions, the computing device may randomly select one of the candidate tasks as a pre-selected node that follows the current node.

[0047] Step S105: Select candidate jobs whose working conditions meet the deployment point conditions as pre-selected nodes. When the computing device determines that one of the working conditions of the candidate jobs meets at least one of the deployment point conditions, the combination of the candidate jobs meeting the deployment point conditions and the available production machines is selected based on the current node (e.g. Figure 3A The node R shown in FIG. 1 is used as a preselected node following the current node (eg Figure 3A In addition, when multiple of the candidate jobs meet one or more deployment conditions, the computing device may use the combination of the candidate job with the earliest required input time and the available production machine as the pre-selected node.

[0048] After determining the pre-selected nodes in step S104 or S105 , the computing device may then execute step S107 : determining whether the working time of each production machine reaches a third threshold time, or whether all candidate tasks have been pre-selected nodes.

[0049] Specifically, since each site corresponds to one or more production machines, the computing device can determine whether each working time corresponding to each production machine reaches the third threshold time, for example, two days, based on the pre-selected node and at least one current node connected in series to the pre-selected node, wherein the at least one current node connected in series to the pre-selected node includes the aforementioned current node. In other words, by determining whether the working time of each production machine reaches the third threshold time, it can be ensured that each production machine is assigned an equal or similar workload, so that the product number can enter the next production machine as soon as possible after the previous production machine completes processing, so as to shorten the line change time of the product number. The third threshold time is preferably the same as the first threshold time, but the present invention is not limited thereto.

[0050] When the working time of not every production machine reaches the third threshold time, the computing device can define the pre-selected node as another current node and execute another point selection procedure based on the other current node. Figure 3A For example, the computing device defines the aforementioned pre-selected node (J1, M3) as another current node, and executes another point selection procedure based on the other current node (J1, M3) (i.e., starts again from step S103), and then selects another candidate job whose working conditions meet the deployment point conditions from the remaining unselected candidate jobs, or randomly selects another candidate job and an available production machine as a pre-selected node following the other current node (e.g., Figure 3A Node (J3, M3) shown). In detail, when executing step S107 for the first time, the computing device makes a judgment based on the first current node (i.e., the initial node) and the pre-selected nodes connected thereto; however, when executing step S107 subsequently, the computing device makes a judgment based on all current nodes (i.e., including the initial node and all nodes defined as current nodes by the pre-selected nodes) and the pre-selected nodes connected thereto. In other words, step S107 judges whether the working time of each production machine of the bottleneck node has reached the third threshold time based on the pre-selected nodes and at least one current node connected in series to the pre-selected nodes.

[0051] When the computing device determines in step S107 that the working time of each production machine has reached the third threshold time (or all candidate tasks have been selected as nodes), it means that the point-spreading stage of the Monte Carlo tree search of this embodiment has been completed, so the computing device can stop executing the point-spreading stage. At this time, after stopping the point-spreading stage, the computing device can continue to execute the simulation stage, that is, execute step S109: calculate the simulation score of the task schedule including the pre-selected node and at least one current node connected in series to the pre-selected node.

[0052] For example, the computing device may be configured to calculate a simulation score based on an allowed number of line changes using a formula, calculate a simulation score based on an expected output quantity using an expected number of production capacity formula, and calculate a simulation score for a work schedule including a preselected node and at least one current node connected in series to the preselected node using an allowed number of late submissions formula based on an allowed number of late submissions.

[0053] The formula for the allowed number of line changes is, for example, the following formula (1), wherein score_setuptime in the formula for the allowed number of line changes is the simulation score; thresh_time is preferably the same threshold time as the third threshold time mentioned above, but thresh_time can also be other threshold times, which are used to evaluate the relative relationship between the number of line changes within the threshold time and the allowed number of line changes; allowed_swi is the allowed number of line changes; and total_swi is the total number of line changes within the threshold time.

[0054]

[0055] The expected capacity formula is, for example, the following formula (2), wherein score_qty is the simulation score; generated_qty is an output quantity within the threshold time; expected_qty is the expected output quantity; thresh_time is preferably the same threshold time as the third threshold time mentioned above, but thresh_time can also be other threshold times, which are used to evaluate the relative relationship between the output quantity within the threshold time and the expected output quantity.

[0056]

[0057] The formula for the allowed number of late submissions is, for example, the following formula (3), wherein score_delay is the simulation score; allowed_delay is the allowed number of late submissions; total_delay is the total number of late submissions within the threshold time; thresh_time is preferably the same threshold time as the third threshold time mentioned above, but thresh_time can also be other threshold times, which are used to evaluate the relative relationship between the number of late submissions within the threshold time and the allowed number of late submissions.

[0058]

[0059] In addition, the method of obtaining the simulation score can also be to use the above-mentioned simulation score score_setuptime, simulation score score_qty and simulation score score_delay as reference simulation scores and multiply them by corresponding weight values ​​respectively, and then the sum of the reference simulation scores multiplied by the weight values ​​is the final simulation score. That is, according to the use requirements, the simulation score can be any one of the simulation scores calculated by the above three formulas, or it can be the sum of the simulation scores calculated by the above three formulas multiplied by the corresponding weight values. The present invention does not limit the actual calculation parameters of the simulation score.

[0060] After the simulation phase is completed, the back-pushing phase may be performed, namely, step S111: adding the simulation score to the total score of the pre-selected node. In addition, the back-pushing phase may also be to add the simulation score to the first pre-selected node (i.e., Figure 3A The first initial node (J1, M3) of the first path r1 or the first initial node (J5, M2) of the second path r2).

[0061] After calculating the simulation score, the computing device can add the simulation score to the total score of each node on the scheduling path, that is, the computing device can add the simulation score to the total score of all pre-selected nodes determined in the deployment phase. In short, after executing a scheduling simulation calculation, a corresponding scheduling path will be obtained, and the scheduling path has a corresponding simulation score. The computing device performs the back-pushing phase (step S111) to add the simulation score to the total score of all nodes on the scheduling path.

[0062] by Figure 3A For example, assuming that the initial node selected in the selection phase is node R, and the pre-selected nodes generated in the deployment phase are nodes (J1, M3) and (J3, M3), the simulation score calculated in the simulation phase is the score of the scheduling path r1 including nodes R, (J1, M3) and (J3, M3). Then, in the back-pushing phase, the computing device can add the simulation score to the pre-selected nodes (J1, M3) and (J3, M3), which is to complete a scheduling simulation calculation, and the scheduling path r1 is a scheduling path generated by this scheduling simulation calculation.

[0063] Please refer to Figure 2B , Figure 2B The detailed flowchart of the production line scheduling method of one embodiment of the present invention is schematically shown. After the back-pushing stage is completed, the aforementioned multiple executions of the scheduling simulation calculations on the tasks to sequentially construct the scheduling simulation trees, and obtaining the task schedule and the simulated completion time of each of the tasks based on the scheduling simulation trees may also include the following steps S113, S115 and S117.

[0064] Step S113: Determine whether the scheduling simulation tree including the initial node reaches the termination condition.

[0065] The computing device determines whether the scheduling simulation tree constructed by executing multiple scheduling simulation calculations has reached a termination condition, where the termination condition is, for example, that the number of iterations of the scheduling simulation calculation has reached an upper limit, or that the total running time of the scheduling simulation calculation has reached an upper limit, etc. The present invention does not limit the setting of the termination condition.

[0066] When the computing device determines in step S113 that the scheduling simulation calculation has not reached the termination condition, the computing device may perform another scheduling simulation calculation, that is, the computing device may perform step S101 again to further select another candidate task (such as Figure 3A The combination of work J4 or work J5) and available production machines as a node connected in series to the current node (such as Figure 3A Node (J4, M1 / M7) or node (J5, M2) shown). Taking the second path r2 of node R connected to node (J5, M2) as an example, this is another scheduling path formed by executing the scheduling simulation calculation again. Figure 3A From the example scheduling simulation tree, the scheduling simulation tree with node R as the initial node includes a first path r1 from node R via node (J1, M3) to node (J3, M3), a second path r2 from node R to node (J5, M2), and a third path r3 from node R via node (J1, M3) to node (J4, M1 / M7). In addition, when there is only one candidate task that has not yet become a selected node, and thus only a single scheduling simulation calculation needs to be performed, a scheduling simulation tree may also include only one scheduling path, and the present invention does not limit the number of scheduling paths in the scheduling simulation tree.

[0067] On the other hand, when the computing device determines in step S113 that the scheduling simulation calculation has reached the termination condition, the computing device may execute step S115: selecting one from a plurality of first pre-selected nodes connected to the initial node as a selected node corresponding to the scheduling simulation tree, wherein the total score of the selected node is higher than the total score of the first pre-selected nodes other than the selected node.

[0068] For example, if the scheduling simulation reaches the termination condition, the scheduling path obtained is Figure 3AAs shown in the first path r1 and the second path r2, the computing device can compare the total scores of the pre-selected node (J1, M3) in the first path r1 and the pre-selected node (J5, M2) in the second path r2, wherein the pre-selected node (J1, M3) and the pre-selected node (J5, M2) are both the first pre-selected nodes connected in series after the initial node R. When the computing device determines that the total score of the pre-selected node (J1, M3) is higher than the total score of the pre-selected node (J5, M2), the computing device can select the pre-selected node (J1, M3) as a selected node.

[0069] Next, the computing device executes step S117: determining whether all the candidate tasks have been selected as nodes. That is, the computing device determines whether all the candidate tasks have been selected as corresponding to multiple selected nodes in the scheduling simulation trees. When the computing device determines that at least one candidate task has not been selected as a selected node among the candidate tasks, the computing device can perform the scheduling simulation calculation again (i.e., execute step S101); and when the computing device determines that all the candidate tasks have been selected as corresponding to the selected nodes of the scheduling simulation trees, the computing device can obtain the simulated completion time of each task including the task schedule of the selected nodes, and execute step S20 accordingly.

[0070] by Figure 3A and 3B For example, if all candidate tasks to be scheduled into the task schedule are tasks J1 to J5, then step S117 is to determine whether all candidate tasks J1 to J5 have been listed as selected nodes. When the computing device determines in step S117 that not all candidate tasks J1 to J5 have been selected nodes in the task schedule, the computing device re-executes step S101 and uses the most recently selected selected node as the current node, and can update the first to third threshold times accordingly (for example, the first threshold time is updated from the original two days to three days).

[0071] Otherwise, please refer to Figure 3B ,in Figure 3B The example diagram of the scheduling simulation tree of another embodiment of the present invention is schematically shown. When the computing device determines in step S117 that all candidate jobs J1-J5 have been selected nodes and arranged into the job schedule (i.e., all job nodes J1-J5 have been assigned to corresponding available production machines and are selected nodes in the scheduling path r1), the computing device can output the following: Figure 3B The work schedule shown includes all the work nodes (J1, M3) ~ (J5, M2) and calculates the simulation completion time of the work nodes (J1, M3) ~ (J5, M2) at the bottleneck site. It should be noted that Figure 3A and 3B The node R shown in the figure may be a null node or a working node, and the present invention does not limit the setting of the node R.

[0072] In step S20 of the present embodiment, after calculating the simulated completion time, the computing device can further deduct the flow factor (Flow Factor) from the simulated completion time of each job and multiply it by the required processing time of each job, thereby obtaining the expected input time of each job at the bottleneck site (expected input time = simulated completion time - flow factor × required processing time), wherein the flow factor is the value obtained by dividing the product delivery time of each of these jobs by the actual processing time, and the product delivery time is the time from the start of processing to the completion of processing of the product, and the actual processing time is the time from the product delivery time to the waiting time. Similarly, the expected input time corresponding to each job at the bottleneck site at other sites on the production line can be calculated in the same estimation method, and finally the computing device can output the expected input time of all jobs of the production line at each site. Accordingly, the job with the smallest expected input time, the shortest line change time and / or the earliest completion of the previous process can be prioritized into the actual production line schedule, but the present invention is not limited to this.

[0073] In addition, in an embodiment where the number of bottleneck sites is multiple, Figure 4 For example, assuming the production process is as follows Figure 4 As shown in the example of stations 1 to 10, where stations 3 and 10 are both bottleneck stations, the computing device can first calculate the expected input time of each job at station 10 (the last bottleneck station) based on the required delivery date of each job using the flow factor (i.e., expected input time = required delivery date completion time - flow factor × actual processing time), and then calculate the simulated completion time of each job at station 10 using the scheduling simulation calculation method, and determine the delivery date of each job at station 10 using the calculated simulated completion time. Then, the computing device back-calculates the expected input time of each job at station 3 using the flow factor based on the delivery date of each job at station 10, and then calculates the simulated completion time of each job at station 3 using the scheduling simulation calculation method, and determines the delivery date of each job at station 3 using the simulated completion time calculated this time. Therefore, the computing device can calculate the expected input time of each job at each station using the flow factor based on the delivery date of each job at station 3.

[0074] In another embodiment where the number of bottleneck sites is multiple, and also taking site 3 and site 10 as both bottleneck sites, the computing device can first calculate the expected input time of each job at site 3 and site 10 based on the required delivery date of each job with the flow factor, and calculate the simulated completion time of these jobs at site 3 and site 10 respectively in a similar manner as described above, and determine the delivery dates of these jobs at site 3 and site 10 respectively according to the two simulated completion times. Next, the computing device calculates the expected input time of these jobs at each site based on the delivery dates of these jobs at site 3 and site 10 respectively with the flow factor, and the computing device can calculate the average of the expected input time calculated based on the delivery dates of these jobs at site 3 and site 10 respectively, and use the calculated average as the actual input time of these jobs at each site.

[0075] Please refer to Figure 2A and Figure 5 ,in Figure 5 The flowchart of the production line scheduling method according to another embodiment of the present invention is schematically shown. Figure 2A The detailed implementation of step S103 of determining whether the working conditions meet the conditions of the exhibition point can be as follows: Figure 5 The sub-steps shown.

[0076] Step S1031: Determine whether at least one of the working conditions meets the material number condition. When the computing device determines that none of the working conditions meets the material number condition, step S1033 can be executed to determine whether at least one of the working conditions meets the jig handling condition. When the computing device determines that none of the working conditions meets the jig handling condition, step S1035 can be executed to determine whether at least one of the working conditions meets the production start time condition.

[0077] On the contrary, when it is determined in step S1031 that at least one of the working conditions meets the material number condition; in step S1033, it is determined that at least one of the working conditions meets the fixture handling condition; or in step S1035, it is determined that at least one of the working conditions meets the production start time condition, the computing device may execute Figure 2A Step S105.

[0078] It should be noted that the material number condition described in step S1031 may also include that the sum of the working time of all pre-selected nodes and the current node connected to the pre-selected node obtained in a site deployment phase is less than a continuous operation time, where the continuous operation time is, for example, two days. In other words, since the current node and the pre-selected node are usually the same material number and correspond to the same machine, the mechanism of setting the continuous operation time can make the working time of each machine more even (e.g. Figure 3AThe node (J1, M3) shown and the node (J3, M3) connected in series thereto).

[0079] In the above description, Figure 4 The order of steps S1031, S1033 and S1035 shown is only an example. The judgment priority of these exhibition point conditions can be adjusted according to usage requirements, and exhibition point conditions can be added or deleted according to usage requirements. The present invention is not limited to this.

[0080] Please refer to Figure 6 , Figure 6 The block diagram of a production line system according to an embodiment of the present invention is schematically shown. The above-mentioned production line scheduling method and scheduling simulation calculation thereof are preferably executed by the computing device 100 of the production line scheduling system, and are suitable for implementation in a production line system including this computing device 100 and a plurality of available production machines M1 to M7. The computing device 100 is, for example, a device with computing capabilities such as a central processing unit. The present invention does not limit the type of computing device 100; the available production machines M1 to M7 are communicatively connected to this computing device 100. That is, the computing device 100 is used to execute the above-mentioned production line scheduling method to obtain the expected input time of each job at the bottleneck site, so that each available machine M1 to M7 can obtain a pending job from the remaining jobs based on the job scheduling after completing a current job. Figure 3B For example, when machine M3 completes job J1, it can be determined based on the job schedule that job J3 among the remaining jobs (J2-J5) also needs to be processed by machine M3, so machine M3 can adjust the processing parameters according to the job J3 to be processed.

[0081] It can be understood that after obtaining the pending work, each available production machine M1~M7 can also output a corresponding processing parameter notification based on the product material number of the pending work, and after the processing parameters of the available production machine have been set to correspond to the pending work, start processing the pending work. Specifically, after obtaining the pending work, each available production machine can determine whether the product material number of the pending work is the same as the material number of the previous work (i.e., another product material number of the current work). When the available production machine determines that the product material number of the pending work is the same as the material number of the previous work, it can output a processing parameter notification including a parameter continuation notification; conversely, when the available production machine determines that the product material number of the pending work is different from the material number of the previous work, it can output a processing parameter notification including a parameter change notification. Accordingly, when the product material number changes, this available production machine can load the corresponding processing parameters, or notify the factory personnel to adjust the processing parameters.

[0082] In addition, the present invention also includes a computer-readable storage medium, which includes a program, and the program is executed by a computing device to execute the above-mentioned production line scheduling method and its scheduling simulation calculation, wherein the computer-readable storage medium can be a non-transitory computer-readable storage medium (non-transitory computer-readable medium), such as a CD, a hard disk or a random access memory (RAM). The present invention does not limit the type of computer-readable storage medium.

[0083] In summary, according to the production line scheduling method, production line system and computer-readable storage medium shown in one or more embodiments of the present invention, the expansion stage in the Monte Carlo tree search can be efficiently executed, and the simulation score of the Monte Carlo tree search can be calculated in advance to shorten the time to obtain the scheduling solution for each job. In addition, according to the production line scheduling method, production line system and computer-readable storage medium shown in one or more embodiments of the present invention, the expected input time of each job at each station can be back-pushed based on the completion time of each job at the bottleneck station to serve as the basis for the subsequent scheduling of the entire production line.

Claims

1. A production line scheduling method, applicable to multiple tasks of an order, characterized in that: The tasks all pass through a bottleneck site having at least one production machine, and the tasks correspond to a plurality of working conditions respectively. The production line scheduling method includes: Executing a scheduling simulation calculation multiple times on the tasks to sequentially construct a plurality of scheduling simulation trees, and obtaining a task schedule and a simulated completion time of each of the tasks based on the scheduling simulation trees, wherein each of the scheduling simulation trees includes at least one scheduling path, and each of the at least one scheduling path is generated by one execution of the scheduling simulation calculation; and Based on the simulated completion time of each of the tasks, multiple expected input times of each of the tasks at multiple sites including the bottleneck site are calculated. The scheduling simulation calculation includes: executing a deployment phase based on at least one deployment condition and the working conditions to obtain the scheduling path; The method of executing the scheduling simulation calculations for the tasks multiple times to sequentially construct the scheduling simulation trees, and obtaining the task schedule and the simulated completion time of each of the tasks based on the scheduling simulation trees includes: Determining whether the tasks have all been selected as a plurality of selected nodes respectively corresponding to the scheduling simulation trees; When it is determined that at least one of the tasks has not been selected as the selected node, executing the scheduling simulation calculation again; and When it is determined that the tasks have all been selected as the selected nodes respectively corresponding to the scheduling simulation trees, the simulated completion time of each of the tasks is obtained using the task schedule including the selected nodes; The at least one deployment point condition is a plurality of deployment point conditions, the deployment point stage of the scheduling simulation calculation is used to schedule the tasks in the scheduling path, the deployment point stage includes a plurality of point selection procedures, and each of the point selection procedures includes a point selection sub-program, wherein the point selection sub-program includes: Determining whether at least one of the working conditions meets at least one of the exhibition point conditions; When it is determined that one of the working conditions meets at least one of the deployment point conditions, the work whose working condition meets the at least one deployment point condition and a corresponding available production machine are selected based on a current node as a pre-selected node following the current node; as well as When it is determined that the working conditions do not meet the deployment point conditions, one of the workings and another corresponding available production machine are randomly selected as the pre-selected node following the current node; The point selection subprogram is a first point selection subprogram, and each of the point selection subprograms further includes a second point selection subprogram, which is executed after the first point selection subprogram, wherein the second point selection subprogram includes: Determining whether at least one working time corresponding to the at least one production machine has reached a threshold time based on the pre-selected node and at least one current node connected in series to the pre-selected node, wherein the at least one current node connected in series to the pre-selected node includes the aforementioned current node; When it is determined that the working time of one of the at least one production machine does not reach the threshold time, defining the pre-selected node as another current node, and executing another point selection procedure based on the another current node; and When it is determined that the at least one working time has reached the threshold time, the point selection procedures are stopped.

2. The production line scheduling method according to claim 1, characterized in that: The at least one exhibition point condition includes at least one of a material number condition, a fixture transportation condition, and a production start time condition.

3. The production line scheduling method according to claim 1, characterized in that: The scheduling simulation calculation further includes a simulation phase, which is executed after the display point phase is stopped. The simulation phase includes: calculating a simulation score of the scheduling path corresponding to the scheduling simulation calculation.

4. The production line scheduling method according to claim 3, characterized in that: In the point selection procedure executed for the first time in the point development stage, the current node is an initial node, the pre-selected node is a first pre-selected node, and the scheduling simulation calculation also includes a back-pushing stage, which is executed after the simulation stage. The back-pushing stage includes: superimposing the simulation score to a total score of the first pre-selected node.

5. The production line scheduling method according to claim 4, characterized in that: Executing the scheduling simulation calculations on the tasks multiple times to sequentially construct the scheduling simulation trees includes: Determining whether the scheduling simulation tree including the initial node has reached a termination condition; When it is determined that the scheduling simulation tree has not reached the termination condition, executing another scheduling simulation calculation; as well as When it is determined that the scheduling simulation tree has reached the termination condition, one is selected from a plurality of first pre-selected nodes connected to the initial node as a selected node corresponding to the scheduling simulation tree, wherein the total score of the selected node is higher than the total score of the first pre-selected nodes other than the selected node.

6. The method according to claim 2, characterized in that The material number condition includes: a product material number in the working condition is the same as a current material number of the current node.

7. The method according to claim 6, characterized in that The material number condition further includes: an estimated input time in the work condition falls within the threshold time.

8. The method according to claim 2, characterized in that The jig transport condition includes: a machine jig in the working condition is the same as a current machine jig of the current node.

9. The method according to claim 2, characterized in that The production start time condition includes: an available machine time in the working condition is the earliest one among the working conditions.

10. The method according to claim 1, characterized in that Determining whether at least one of the working conditions meets at least one of the exhibition point conditions includes: Determining whether at least one of the working conditions meets a material number condition; and When it is determined that none of the working conditions meet the material number condition, determining whether at least one of the working conditions meets a fixture transport condition; When it is determined that none of the working conditions meet the jig transport condition, determining whether at least one of the working conditions meets a production start time condition; as well as When it is determined that none of the working conditions meet the production start time condition, another one of the working conditions is randomly selected as the pre-selected node.

11. The method according to claim 1, characterized in that When multiple of the working conditions meet the deployment point condition, the scheduling simulation calculation further includes: The work corresponding to the work condition having the earliest required input time among the work conditions is used as the pre-selected node.

12. The method according to claim 1, characterized in that The exhibition point conditions include a material number condition, the material number condition includes: the sum of a working time of the pre-selected node and the at least one current node serially connected to the pre-selected node obtained in the exhibition point stage is less than a continuous operation time, and judging whether at least one of the working conditions meets at least one of the exhibition point conditions includes: Determine based on the working conditions whether the sum of the working time of the pre-selected node and the at least one current node serially connected to the pre-selected node obtained in the deployment phase is less than the continuous operation time; as well as When it is determined that the sum of the working time is not less than the continuous operation time, it is determined whether at least one of the working conditions meets another one of the exhibition point conditions.

13. The method according to claim 3, characterized in that Calculating the simulation score of the scheduling path corresponding to the scheduling simulation calculation includes: The simulation score is calculated based on an allowed number of line changes, an expected output quantity, and an allowed number of late deliveries.

14. The method according to claim 13, characterized in that The simulation score is calculated based on the allowed number of line changes: the simulation score is calculated based on the threshold time, the allowed number of line changes within the threshold time, and a total number of line changes within the threshold time.

15. The method according to claim 13, characterized in that Calculating the simulation score based on the expected output quantity includes: calculating the simulation score based on the threshold time, an output quantity within the threshold time, and the expected output quantity within the threshold time.

16. The method according to claim 13, characterized in that The simulation score is calculated based on the allowed late submission number: based on the threshold time, the allowed late submission number within the threshold time, and a total late submission number within the threshold time.

17. The method according to claim 13, characterized in that Calculating the simulation score based on the allowed number of line changes, the expected output quantity, and the allowed number of late deliveries includes: Calculate a reference simulation score corresponding to the allowed number of line changes, the expected output quantity and the allowed number of late deliveries respectively; multiplying the reference simulation scores corresponding to the allowed number of line changes, the expected output quantity, and the allowed number of late deliveries by corresponding weight values; and The simulation score is obtained by multiplying the sum of the reference simulation scores by the corresponding weight values.

18. The method of claim 1, wherein: Calculating one of these expected time investments involves: Dividing a product delivery time of one of the tasks by an actual processing time to obtain a flow factor; and The expected input time is obtained by deducting the flow factor from the simulated completion time of one of the tasks and multiplying the actual processing time.

19. A production line system, characterized in that: Include: a computing device for executing the production line scheduling method of claim 1 to obtain the work schedule; and A plurality of available production machines are communicatively connected to the computing device. After completing a current job, each of the available production machines obtains a pending job from the computing device based on the job schedule. Each of the available production machines also outputs a processing parameter notification based on a product material number of the pending job.

20. The production line system according to claim 19, characterized in that: Each of the available production machines outputs the processing parameter notification based on the product number of the pending job, including: When each of the available production machines determines that the product material number is different from another product material number currently being processed, the processing parameter notification outputted includes a parameter change notification. When each of the available production machines determines that the product material number is the same as the material number of the other product currently being processed, the processing parameter notification outputted by each of the available production machines includes a parameter adoption notification.

21. A computer-readable storage medium comprising a program, wherein a computing device executes the program to execute the production line scheduling method according to claim 1.

Images