Method and system for dynamic scheduling of a cleaning robot assembly line

Abstract

The application discloses a kind of dynamic scheduling methods and systems for cleaning robot assembly production line, the method comprises: according to the number of workstations between each sub-assembly line convergence and bottleneck test station Determine pipe depth, and combine the average single workstation beat time of main assembly line current Calculate pipe delay time;Collect the waiting queue length of bottleneck test station, average single piece processing time, parallel test workstation quantity and the number of in-process products in each pipe section, with the pipe delay time of each sub-assembly line as prediction window, calculate corresponding look-ahead predicted queue length;According to the comparison result of look-ahead predicted queue length and queue length upper and lower limit, respectively adjust the feeding interval of each sub-assembly line.The method can reduce the in-process product fluctuation in front of end test station, improve the running stability of assembly line.

Description

Technical Field

[0001] This application relates to the field of production scheduling and control technology, specifically to a dynamic scheduling method and system for a cleaning robot assembly line. Background Technology

[0002] Household robotic vacuum cleaners and floor scrubbers are typically assembled from multiple sub-components, including wheel assemblies, front brush assemblies, sensor assemblies, fan assemblies, dustbin assemblies, bumper assemblies, and battery assemblies. Existing assembly lines generally employ a multi-sub-assembly line structure with a main assembly line. Each sub-assembly line feeds its sub-components into the main assembly line at different workstations. The complete robot then passes through consecutive assembly stations before entering the final testing area. The final testing area typically includes multiple testing stations for functional circuit testing, depth camera calibration, TOF sensor leveling, and navigation testing. Due to variations in processing time, number of parallel workstations, and rework rates for different test items, the testing area is prone to developing periodic bottlenecks.

[0003] In actual production, although the Manufacturing Execution System (MES) can collect data such as workstation barcode scanning records, work-in-process (WIP) locations, and queue and processing data for testing stations, existing scheduling methods often adjust the feeding rhythm of each sub-assembly line based on the current queue length, the current production line cycle time, or a uniform feeding rule. Because each sub-assembly line converges at a different position on the main assembly line, the number of workstations and the transmission time required for its output to reach the bottleneck testing station vary. If all sub-assembly lines are adjusted based on the same current queue status, sub-assembly lines with earlier convergence positions often experience scheduling lag, while those with later convergence positions may adjust too quickly. This leads to fluctuations in the number of WIP items before the testing station, unstable buffer zone occupancy, and affects the utilization rate of the testing station and the stability of the overall production line cycle time. Summary of the Invention

[0004] To overcome the problems existing in related technologies, this specification provides a dynamic scheduling method and system for cleaning robot assembly lines.

[0005] According to a first aspect of the embodiments of this application, a dynamic scheduling method for a cleaning robot assembly line is provided. The assembly line includes multiple sub-assembly lines and a main assembly line, with each sub-assembly line converging at different workstations of the main assembly line, and multiple testing workstations located at the end of the main assembly line. The method includes: The pipeline depth of each sub-assembly line is determined based on the number of stations between the merging station and the bottleneck testing station on the main assembly line. The pipeline depth represents the number of stations traversed from the merging station to the bottleneck testing station. Based on the pipeline depth and the current average single-station cycle time of the main assembly line, the pipeline delay time of each sub-assembly line is obtained. Collect the current waiting queue length, average single-piece processing time, and number of parallel testing stations of the bottleneck testing station, as well as the current work-in-process quantity in the corresponding pipeline section of each sub-assembly line; For each sub-assembly line, the pipeline delay time is used as the prediction window length. The expected processing volume of the test station within the prediction window length is determined based on the number of parallel test stations and the average single-piece processing time. The forward prediction queue length of each sub-assembly line is calculated based on the current waiting queue length, the current work-in-process quantity in the corresponding pipeline segment, and the expected processing volume. Based on the comparison between the forward-looking predicted queue length of each sub-assembly line and the preset upper and lower limits of the queue length, the feeding interval of each sub-assembly line is adjusted independently.

[0006] As an optional approach, determining the bottleneck test station includes: obtaining the average single-piece processing time of each test station at the end of the main assembly line within a preset time window, and marking the test station with the longest average single-piece processing time as the bottleneck test station; when the difference between the average single-piece processing times of multiple test stations is within a preset ratio range, marking the test station with the longest current waiting queue as the bottleneck test station.

[0007] As an optional approach, the step of calculating the forward prediction queue length for each sub-assembly line includes: calculating the forward prediction queue length of the sub-assembly line based on the current waiting queue length and the current work-in-process quantity in the corresponding pipeline segment, combined with the expected processing volume of the sub-assembly line.

[0008] As an optional approach, the independent adjustment of the feeding interval for each sub-assembly line includes: increasing the feeding interval of a sub-assembly line when the forward prediction queue length of a sub-assembly line exceeds the upper limit of the queue length; shortening the feeding interval of a sub-assembly line when the forward prediction queue length of a sub-assembly line is lower than the lower limit of the queue length; and maintaining the current feeding interval of a sub-assembly line when the forward prediction queue length of a sub-assembly line is between the lower and upper limits of the queue length.

[0009] As an optional approach, the upper limit of the queue length is determined based on the physical capacity and buffer margin coefficient of the buffer zone in front of the bottleneck test station; the lower limit of the queue length is determined based on the number of parallel test stations and the safety margin.

[0010] As an optional approach, when increasing or decreasing the feeding interval, the adjustment range is determined based on the normalized deviation of the forward prediction queue length from the corresponding upper or lower limit and the adjustment gain coefficient. The normalized deviation is the ratio obtained by dividing the difference between the forward prediction queue length and the corresponding upper or lower limit by the upper or lower limit. The adjusted feeding interval is limited to the minimum feeding interval and the maximum feeding interval of the sub-assembly line.

[0011] As an optional approach, the adjustment gain coefficient is calibrated based on the capacity margin between the upper limit of the queue length and the physical capacity of the buffer zone in front of the bottleneck test station. The capacity margin is the difference between the physical capacity of the buffer zone and the upper limit of the queue length. The calibration criterion is to divide the upper limit of the queue length by the capacity margin as the calibration criterion.

[0012] As an optional approach, the method further includes: selecting a reference interval containing a preset number of workstations on the main assembly line with a preset update cycle, calculating the actual time taken by multiple work-in-process items to pass through the reference interval recently, dividing the actual time taken by each work-in-process item by the number of workstations in the reference interval and taking the average value as the updated average single-workstation cycle time, and recalculating the pipeline delay time of each sub-assembly line accordingly.

[0013] As an optional approach, the method further includes: re-comparing the average single-piece processing time of each test station at the end of the main assembly line with a preset inspection cycle; and when the bottleneck test station changes, re-determining the pipe depth and pipe delay time of each sub-assembly line according to the station number of the changed bottleneck test station.

[0014] According to a second aspect of the embodiments of this application, a dynamic scheduling system for a cleaning robot assembly line is also provided. The assembly line includes multiple sub-assembly lines and a main assembly line, with each sub-assembly line converging at different workstations of the main assembly line, and multiple testing workstations located at the end of the main assembly line. The system includes: The pipeline depth determination module is used to determine the pipeline depth of each sub-assembly line based on the number of stations between the merging station and the bottleneck test station on the main assembly line. The pipeline depth represents the number of stations traversed from the merging station to the bottleneck test station. The pipeline delay time calculation module is used to obtain the pipeline delay time of each sub-assembly line based on the pipeline depth and the current average single-station cycle time of the main assembly line. The data acquisition module is used to collect the current waiting queue length, average single-piece processing time, and number of parallel test stations of the bottleneck test station, as well as the current work-in-process quantity in the corresponding pipeline section of each sub-assembly line. The forward prediction queue length calculation module is used to determine the expected processing volume of each test station within the prediction window length for each sub-assembly line, using the pipeline delay time as the prediction window length, based on the number of parallel test stations and the average single-piece processing time. It also calculates the forward prediction queue length for each sub-assembly line based on the current waiting queue length, the current work-in-process quantity in the corresponding pipeline segment, and the expected processing volume. The feeding interval adjustment module is used to independently adjust the feeding interval of each sub-assembly line based on the comparison results between the forward-looking predicted queue length and the preset upper and lower limits of the queue length.

[0015] This application addresses the issue of bottlenecks at the end-of-line testing station in a cleaning robot assembly line, where multiple sub-assembly lines converge at different locations to the main assembly line. It determines the pipeline delay time based on the pipeline depth between each sub-assembly line's convergence point and the bottleneck testing station, using this delay time as a prediction window for the corresponding sub-assembly line. By combining the current waiting queue at the testing station, parallel testing capacity, and the amount of work-in-process (WIP) within the pipeline section, the forward-looking predicted queue length for each sub-assembly line is calculated, and the feeding interval for each sub-assembly line is adjusted accordingly. Therefore, the feeding control of each sub-assembly line no longer relies solely on the current queue state of the testing station, but rather matches the expected queuing situation when its output reaches the bottleneck testing station. This alleviates scheduling response delays or over-adjustments caused by differences in convergence points, reduces fluctuations in the amount of WIP before the testing station, lowers the risk of buffer zone congestion, and prevents the testing station from being idle due to insufficient supply, thus improving the utilization rate of the end-of-line testing station and the overall line operational stability.

[0016] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Furthermore, no embodiment in this disclosure is required to achieve all the effects described above. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this specification and, together with the description, serve to explain the principles of this specification.

[0018] Figure 1 This is a flowchart of a dynamic scheduling method for a cleaning robot assembly line according to one embodiment of this application.

[0019] Figure 2 This is a flowchart illustrating the calculation of the look-ahead prediction queue length for each sub-assembly line in one embodiment of this application.

[0020] Figure 3 This is a flowchart illustrating the adjustment of the feeding interval of each sub-assembly line in one embodiment of this application.

[0021] Figure 4 This is a schematic block diagram of a dynamic scheduling system for a cleaning robot assembly line according to one embodiment of this application.

[0022] Figure 5 This is a structural diagram of an electronic device according to one embodiment of this application. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0024] This embodiment provides a dynamic scheduling method for assembly lines of cleaning robots, applicable to assembly lines of household sweeping robots or household floor scrubbing robots. This type of production line comprises multiple sub-assembly lines and one main assembly line. Each sub-assembly line assembles different sub-components, such as wheel hub assemblies, front brush assemblies, sensor assemblies, fan assemblies, dustbin assemblies, impact plate assemblies, and battery assemblies. After each sub-assembly line completes its assembly, the sub-components are integrated into the overall assembly process at designated workstations on the main assembly line. These integration workstations are distributed across different locations on the main assembly line. The main assembly line contains dozens of continuously arranged workstations, with multiple testing stations at its end, including a Functional Circuit Test (FCT) station, a depth camera (RGBD) calibration station, a Time of Flight (TOF) sensor leveling station, and a navigation test station. The production line is equipped with a Manufacturing Execution System (MES), which can collect work-in-process positions at each workstation and process real-time data from the testing stations. This method can be executed by the scheduling calculation module within the MES system.

[0025] The implementation process of the method described in this application will be explained in detail below with reference to specific embodiments.

[0026] Please see Figure 1 , Figure 1 This is a flowchart of a dynamic scheduling method for a cleaning robot assembly line according to one embodiment of this application, such as... Figure 1 As shown, the method includes steps 101-105: In step 101: Determine the pipe depth of each sub-assembly line based on the number of stations between the merging station and the bottleneck test station on the main assembly line. The pipe depth represents the number of stations passed from the merging station to the bottleneck test station.

[0027] Pipeline depth refers to the number of stations along the main assembly line from the merging station of a sub-assembly line to the bottleneck testing station. The larger the value, the more processes the output of that sub-assembly line needs to go through from merging into the main line to reaching the testing station. Before determining the pipeline depth, the bottleneck testing station must first be determined.

[0028] According to an embodiment of this disclosure, the end-of-line testing area includes multiple testing stations, each with a different average unit processing time. The scheduling calculation module obtains the average unit processing time of each testing station at the end of the main assembly line within a preset time window. This preset time window is, for example, the past 60 minutes. The arithmetic mean of the unit processing times of all products tested within this period is used to obtain the average unit processing time of that station. After comparing the average unit processing times of each testing station, the testing station with the longest average unit processing time is marked as the bottleneck testing station.

[0029] When the average unit processing time difference among multiple test stations is within a preset proportion (e.g., the difference between two stations does not exceed 10% of the larger value), the scheduling calculation module will mark the test station with the longest current waiting queue as the bottleneck test station. When the production line has just started and there is not enough operational data, the test station with the longest nominal processing time in the process planning is used as the initial bottleneck station. After the bottleneck test station is determined, its station number on the main assembly line is recorded. .

[0030] Retrieve the entry station number of each sub-assembly line on the main assembly line from the process database. , This refers to the sub-assembly line number. The pipe depth for each sub-assembly line is also specified. Equals the bottleneck test station station number minus the inbound station number of the sub-line: ; In the formula This is the workstation number for the bottleneck testing station. For the first The entry station number of the strip assembly line is the same as the serial number in the main assembly line station sequence. It is a dimensionless positive integer.

[0031] For example, the main assembly line for a certain model of floor cleaning robot has 45 workstations (numbered 1 to 45), and the end-point bottleneck testing station is numbered... Wheel hub component sub-assembly line input station number Pipe depth Front brush component sub-assembly line import station number Pipe depth ; Collision plate component sub-assembly line input station number Pipe depth The pipe depths of the hub sub-line and the impact plate sub-line differ by more than 10 times, and this difference directly determines the classification of the subsequent prediction window length.

[0032] Optionally, the scheduling calculation module stores the numbers of all sub-assembly lines, the numbers of the incoming workstations, and the pipe depths in a mapping table for later querying. This mapping table is regenerated when production models are switched or bottleneck test stations are changed.

[0033] In step 102: Based on the pipe depth and the current average single-station cycle time of the main assembly line, the pipe delay time of each sub-assembly line is obtained.

[0034] Pipeline delay time refers to the estimated time it takes for a sub-component to travel from its entry into the main assembly line to its arrival at the bottleneck test station. Average cycle time per workstation. This represents the average time, in minutes, for a work-in-process item to pass through one station on the main assembly line. The initial value is the nominal cycle time in the process planning, and it is continuously corrected according to the dynamic update method of this embodiment. For the... Strip assembly line, pipeline delay time Calculate using the following formula: In the formula The depth of the pipe is dimensionless. The unit is minutes; The unit is minutes.

[0035] Continuing with the previous example, let the current... Minutes. Wheel hub line. Minutes, front brush line Minutes, hit the board line Minutes. The output from the hub sub-line takes over an hour from merging into the main line to reaching the test station, while the impact plate sub-line takes only 6 minutes. In the tree-structured cleaning robot assembly line, the merging points of more than ten sub-assembly lines are scattered across different sections of the main line, with pipeline delay times ranging from a few minutes to sixty or seventy minutes. If the feeding rate of all sub-lines is adjusted uniformly according to the current queue status of the test station, the adjustment for sub-lines with larger pipeline depths will take a long time to transmit the effect to the test station, during which a large amount of work-in-process already in the pipeline will continue to arrive at the test area; sub-lines with smaller pipeline depths may react too quickly. To eliminate this response mismatch, subsequent steps set a prediction window length for each sub-line that matches its pipeline delay time.

[0036] Step 103: Collect the current waiting queue length, average single-piece processing time, and number of parallel testing stations of the bottleneck testing station, as well as the current work-in-process quantity in the corresponding pipeline section of each sub-assembly line.

[0037] Before setting a prediction window length that matches the pipeline delay time for each sub-line, the scheduling calculation module collects three types of real-time data from the MES system. The first type is the current waiting queue length of the bottleneck test station. The first category is the number of work-in-process units that have arrived at the testing area but have not yet been tested, reported by the programmable logic controller (PLC) of the testing station, in units of units. The second category is the average unit processing time of the bottleneck testing station. Take the average processing time of products that completed testing within the most recent time window (e.g., 30 to 60 minutes), in minutes; and simultaneously obtain the number of currently available parallel testing stations. The third category is the current work-in-process quantity within the corresponding pipeline section of each sub-assembly line. , for the Strip assembly line, To the incoming workstation To the bottleneck test station The total number of work-in-process items (WIPs) flowing along the main line segment is calculated by MES through barcode scanning records at each workstation, tracking the current workstation of each WIP item. The unit is one piece. The data collection cycle for the above data can be configured to be once every 1 to 2 minutes, and this disclosure does not limit it.

[0038] In step 104: For each sub-assembly line, the pipeline delay time is used as the prediction window length. The expected processing volume of the test station within the prediction window length is determined according to the number of parallel test stations and the average single-piece processing time. The forward prediction queue length of each sub-assembly line is calculated according to the current waiting queue length, the current work-in-process quantity in the corresponding pipeline segment, and the expected processing volume.

[0039] According to embodiments of this disclosure, the lookahead prediction queue length refers to predicting, for a given sub-assembly line, the number of work-in-process components in the waiting queue in front of the bottleneck test station when a sub-component launched from that sub-line arrives at the bottleneck test station after pipeline delay at the current moment. The prediction time span for each sub-line is equal to its respective pipeline delay time. Since the pipeline delay time is different for each sub-line, the look-ahead prediction queue length of each sub-line is generally not equal at the same time.

[0040] Specifically, Figure 2 This is a flowchart illustrating the calculation of the look-ahead prediction queue length for each sub-assembly line in one embodiment of this application, as follows: Figure 2 As shown, in step 1041, the expected processing volume of the test station within the prediction window length is determined.

[0041] For the first Strip assembly line, with its pipeline delay time The predicted window length is used as the forecast window length. Within this window period, the number of products that the bottleneck test station is expected to complete testing is the projected processing volume. : ; In the formula This refers to the pipeline delay time, expressed in minutes. The number of parallel test stations, dimensionless; The average processing time per item is in minutes. Indicates in The total amount of products that the test workstation cluster can process during the test period, and the different sub-lines Different, therefore They also differ, with the sub-line having a longer pipeline delay time corresponding to a larger expected processing capacity.

[0042] In step 1042, the forward-looking queue length is calculated based on the current waiting queue length, the number of in-process items in the pipeline segment, and the expected processing volume.

[0043] Specifically, for the first Strip assembly line, forward-looking prediction of queue length The calculation method is as follows: ; In the formula This represents the current length of the waiting queue (in pieces). This represents the current number of work-in-process items (pieces) within the corresponding pipeline segment. The estimated processing volume (pieces) is used for all three items, with each item measured in pieces. The calculation logic is as follows: the current queue inventory plus the total amount of work-in-process that will arrive in the pipeline, minus the processing volume that the test station can complete within the same time period; the difference is the estimated queue length when the output of this sub-line arrives at the test area. A negative value indicates that the test station is idle when the prediction arrives.

[0044] For example, suppose Item, , Minutes. Wheel hub line. Item, Minutes; hit the board line Item, Minutes. Estimated processing capacity of the wheel hub sub-line. The integer part is 33 pieces; Items. Estimated throughput of the impact plate sub-line. Item; Item.

[0045] Faced with the same current waiting queue length, the hub sub-line, due to its greater pipe depth and larger backlog of work-in-process, has a forward-looking predicted queue length of 13 pieces; the impact plate sub-line, with its shorter pipe and smaller inventory, has only 8 pieces. This difference reflects the varying pipe delays between each sub-line and the bottleneck testing station. Under traditional scheduling methods, all sub-lines adjust their feed based on the same current queue status, failing to reflect this difference. However, by setting differentiated prediction window lengths for each sub-line, each sub-line can make feed decisions based on the predicted queue status when its actual output reaches the bottleneck station.

[0046] In step 105: Based on the comparison results between the forward-looking predicted queue length of each sub-assembly line and the preset upper and lower limits of the queue length, the feeding interval of each sub-assembly line is adjusted independently.

[0047] The feeding interval refers to the time interval between two consecutive material feedings on a sub-assembly line, measured in minutes. A longer feeding interval results in a lower output rate. This step independently adjusts the feeding interval for each sub-line based on the forward-looking prediction queue lengths calculated in the preceding steps.

[0048] Specifically, Figure 3 This is a flowchart illustrating the adjustment of the material feeding interval of each sub-assembly line in one embodiment of this application, as follows: Figure 3 As shown, in step 1051, the upper limit and lower limit of the queue length are set.

[0049] Queue length limit To predict the threshold for increasing the feeding interval when the queue length exceeds this value, it is determined by multiplying the physical capacity of the buffer zone in front of the bottleneck test station by the buffer margin coefficient. It is assumed that the buffer zone can accommodate a maximum of... Work-in-process, buffer margin factor is Given a value ranging from 0.60 to 0.85, then: ; In the formula The physical capacity (units) of the buffer. This is a dimensionless coefficient. The value of the buffer margin coefficient is determined based on the buffer space situation; specifically, it can be taken from the past 24 hours. maximum value With the physical capacity of the buffer The ratio as The reference value, i.e. And limit the results to Within the range; when the production line is started for the first time, the initial value can be manually selected according to the sufficiency of the buffer space. For example, when the space is sufficient, take 0.80 to 0.85, and when the space is tight, take 0.60 to 0.70. After multiplying by this coefficient, leave a margin between the physical capacity and the upper limit of the queue to avoid the actual queue overflowing the buffer due to prediction deviation.

[0050] Queue length lower limit To predict the threshold for triggering a shortened feeding interval when the queue length falls below this value, the number of parallel test stations is used. The sum of the safety margin and the safety margin is determined. When multiple test stations are running simultaneously, at least [number] are required. One product is in a waiting state, with a safety margin of, for example, 1 to 2 pieces. or .

[0051] As an example, if Item, ,but Item; if If the safety margin is taken as 1 piece, then Item.

[0052] In step 1052, the feeding interval of each sub-assembly line is adjusted according to different situations. Let the first... The current feeding interval of the strip assembly line is (minutes). Based on forward-looking prediction of queue length. With the upper limit of queue length and lower limit of queue length The comparison results are handled in three cases.

[0053] Specifically, in the first case... Exceed This sub-line needs to increase the feeding interval; the adjusted feeding interval is... Calculate as follows: First, calculate the normalized bias of the look-ahead prediction queue length deviating from the upper limit of the queue length, i.e. This value is a dimensionless ratio; then multiplied by the adjustment gain coefficient. Add 1 to this value as an adjustment factor, and multiply it by the current feeding interval to obtain the adjustment value: ; In the formula The method for determining the gain coefficient (dimensionless) is described in S4.3; This is the maximum allowable feeding interval (in minutes) for this sub-line, taken as 2.5 to 3 times the normal feeding interval, determined by minimum output requirements and personnel scheduling constraints. The operation limits the results to what is physically feasible.

[0054] The second scenario, Below The feeding interval needs to be shortened: ; In the formula This is the minimum allowable feeding interval (in minutes) for this sub-line, which is limited by the maximum processing rate of the sub-line equipment.

[0055] The third scenario, In and In between, maintain the current feeding interval. .

[0056] Since the look-ahead prediction queue lengths of each sub-line are generally unequal, each sub-line may fall into different situations within the same scheduling period. Continuing with the example data mentioned above, let's assume... Item, Item, The current feeding interval for each sub-line is 1.5 minutes. (Wheel hub sub-line) Items, exceeding Normalization bias is Adjustment factor , Minutes, feeding pace slows down; impact plate line The item falls within the designated area and remains unchanged for 1.5 minutes. At the same time, the hub line slows down due to the greater depth and pressure in the conduit, while the impact line requires no adjustment.

[0057] Optionally, the scheduling calculation module sends the adjusted feeding intervals of each sub-line to the workstation terminals of the corresponding sub-assembly lines, and the sub-lines execute material feeding according to the new intervals until the next scheduling cycle is recalculated and updated.

[0058] In one embodiment, the method further includes determining an adjustment gain coefficient, wherein the adjustment gain coefficient is... Control the response amplitude of the feeding interval as the deviation changes. Based on the upper limit of queue length Physical capacity of the front buffer zone of the bottleneck test station The capacity margin is determined by calibration. The difference between the physical capacity of the buffer and the upper limit of the queue length is... The unit is pieces. The calibration criterion is: when the forward-looking prediction of the queue length deviating from the upper limit of the queue length reaches the capacity margin, the feeding interval is increased to a preset multiple of the current value (e.g., 2 times). Substituting this condition into the adjustment formula for the first case, the normalized deviation is... Let the adjustment factor equal to 2, then we get: ; In the formula and All are in units of pieces. It is dimensionless. As an example, Item, Item, Item, The smaller the buffer space, the better. The larger the value, the more sensitive the system is to deviations. The calculation is performed using the formula above during the initial operation of the production line. After running for several shifts, adjustments can be made based on actual queue fluctuations. The percentage of times the current waiting queue length exceeds the upper and lower limits within multiple consecutive scheduling cycles should be statistically analyzed. If the exceedance rate is too high, the limits should be appropriately increased. If the feeding interval shows alternating increases and decreases, then the interval should be appropriately reduced. .

[0059] In one embodiment, the average cycle time per station on the main assembly line Fluctuations in equipment status and personnel rotation, coupled with prolonged periods without updates, can cause pipeline delays to deviate from actual values. The scheduling calculation module recalculates at a preset update interval (e.g., 10 to 20 minutes). .

[0060] Specifically, a reference interval containing a preset number of workstations is selected on the main assembly line, for example, 10 workstations from workstation 10 to workstation 20; the MES records the times when multiple work-in-process items (e.g., 20 items) enter and leave this reference interval, and the actual time taken for each item to pass through the reference interval is divided by the number of workstations in the reference interval and then averaged. ; In the formula For the number of samples, and The first The time (in minutes) when a product leaves and enters the reference range. This refers to the number of workstations in the reference range. (Updated) Substitution Retrieve the pipeline delay time for each sub-line. Optionally, remove extreme values ​​before calculation, for example, remove the two largest and two smallest samples before taking the average.

[0061] In some embodiments, the bottleneck station in the end-of-line testing area changes with the production process. The scheduling calculation module re-compares the average single-piece processing time of each testing station at a preset inspection cycle (e.g., every 2 hours); when the bottleneck testing station changes, the pipe depth and pipe delay time of each sub-assembly line are re-determined based on the station number of the changed bottleneck testing station, and the updated mapping table takes effect from the next scheduling cycle.

[0062] In some embodiments, data acquisition, forward-looking prediction queue length calculation, and material feeding interval adjustment are executed cyclically according to a preset scheduling cycle. The scheduling cycle is no greater than half of the shortest pipeline delay time in each sub-assembly line. For example, if the shortest pipeline delay time for the impact plate sub-line is 6 minutes, the scheduling cycle is 3 minutes. Optionally, when different models of cleaning robots are produced alternately on the same assembly line, the scheduling calculation module reads the process route information of the new model and re-executes it when switching models to generate the corresponding pipeline depth mapping table.

[0063] Using the above method, the material feeding rate adjustment for each sub-assembly line is based on the future predicted queue state calculated according to its respective pipeline delay time, rather than the current queue state of the end-of-line testing station. Sub-lines with larger pipeline depths have a longer prediction window, enabling them to adjust their material feeding rates in advance, ensuring that the scheduling instructions take effect precisely when they reach the bottleneck station after a long pipeline journey. Sub-lines with smaller pipeline depths only predict and adjust within a short period, responding quickly without excessive fluctuations. This differentiated prediction and adjustment mechanism improves the scheduling response mismatch caused by different pipeline delays, reduces the fluctuation range of work-in-process before the end-of-line testing station, and improves the stability of the testing station utilization rate.

[0064] Please see Figure 4 , Figure 4 This is a schematic block diagram of a dynamic scheduling system for a cleaning robot assembly line according to this application. The assembly line includes multiple sub-assembly lines and one main assembly line. The sub-assembly lines converge at different workstations on the main assembly line, and multiple testing workstations are located at the end of the main assembly line. Figure 4 As shown, the system includes: The pipeline depth determination module 4001 is used to determine the pipeline depth of each sub-assembly line based on the number of stations between the merging station and the bottleneck test station on the main assembly line. The pipeline depth represents the number of stations passed from the merging station to the bottleneck test station. The pipeline delay time calculation module 4002 is used to obtain the pipeline delay time of each sub-assembly line based on the pipeline depth and the current average single-station cycle time of the main assembly line. The data acquisition module 4003 is used to collect the current waiting queue length, average single-piece processing time and number of parallel test stations of the bottleneck test station, as well as the current work-in-process quantity in the corresponding pipeline section of each sub-assembly line. The forward prediction queue length calculation module 4004 is used to determine the expected processing volume of the test station within the prediction window length for each sub-assembly line, using the pipeline delay time as the prediction window length, based on the number of parallel test stations and the average single-piece processing time, and to calculate the forward prediction queue length of each sub-assembly line based on the current waiting queue length, the current work-in-process quantity in the corresponding pipeline segment, and the expected processing volume. The feeding interval adjustment module 4005 is used to independently adjust the feeding interval of each sub-assembly line based on the comparison results between the forward-looking predicted queue length of each sub-assembly line and the preset upper and lower limits of the queue length.

[0065] The implementation process of the functions and roles of each module in the above system is detailed in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0066] Figure 5 This is a schematic diagram of an electronic device according to one embodiment of this application. Exemplarily, this electronic device can be used to deploy a MES system, such as... Figure 5 As shown, at the hardware level, the electronic device includes a processor 1102, an internal bus 1104, a network interface 1106, memory 1108, and non-volatile memory 1110, and may also include other hardware required for business operations. One or more embodiments of this specification can be implemented in software, for example, the processor 1102 reads the corresponding computer program from the non-volatile memory 1110 into the memory 1108 and then runs it. Of course, in addition to software implementation, one or more embodiments of this specification do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to each logic module, but can also be hardware or logic devices.

[0067] Based on the same concept as the above-described method, this application also provides a machine-readable storage medium storing a plurality of computer instructions. When executed by a processor, these computer instructions can implement the dynamic scheduling method for cleaning robot assembly lines disclosed in the above examples of this application. The machine-readable storage medium can be any electronic, magnetic, optical, or other physical storage device, and can contain or store information such as executable instructions, data, etc.

[0068] Based on the same application concept as the above method, this application embodiment also provides a computer program product, which includes a computer program that, when executed by a processor, implements the dynamic scheduling method for cleaning robot assembly line disclosed in the above examples of this application.

[0069] In the above embodiments, the descriptions of each embodiment have different focuses. Parts not described in detail in a certain embodiment can be referred to in the relevant descriptions of other embodiments. The above descriptions are merely preferred embodiments of this application and explanations of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solutions formed by specific combinations of the above technical features, but should also cover other technical solutions formed by arbitrary combinations of the above technical features or their equivalent features without departing from the inventive concept.

Claims

1. A dynamic scheduling method for a cleaning robot assembly line, characterized in that, The assembly production line includes multiple sub-assembly lines and one main assembly line. The sub-assembly lines converge at different workstations of the main assembly line, and multiple testing stations are located at the end of the main assembly line. The method includes: Based on the number of stations between the merging station and the bottleneck testing station on the main assembly line of each sub-assembly line, the pipeline depth of each sub-assembly line is determined. The pipeline depth represents the number of stations traversed on the main assembly line from the merging station of a certain sub-assembly line to the bottleneck testing station. Determining the bottleneck testing station includes: obtaining the average single-piece processing time of each testing station at the end of the main assembly line within a preset time window, and marking the testing station with the longest average single-piece processing time as the bottleneck testing station; when the difference between the average single-piece processing times of multiple testing stations is within a preset ratio range, marking the testing station with the longest current waiting queue as the bottleneck testing station. Based on the pipe depth and the current average cycle time per station of the main assembly line, the pipe delay time of each sub-assembly line is obtained. The specific calculation formula is as follows: ,in For pipe depth, This represents the average time it takes for a work-in-process item to pass through one station on the main assembly line. The current waiting queue length, average single-piece processing time, and number of parallel testing stations of the bottleneck testing station are collected, as well as the current work-in-process quantity in the corresponding pipeline segment of each sub-assembly line. The current work-in-process quantity in the corresponding pipeline segment of each sub-assembly line is the total number of work-in-process products flowing on the main line segment from the merging station to the bottleneck testing station. When the bottleneck testing station is changed, the pipeline depth and pipeline delay time of each sub-assembly line are re-determined according to the station number of the changed bottleneck testing station. For each sub-assembly line, the pipeline delay time is used as the prediction window length. The expected throughput of the test stations within this prediction window length is determined based on the number of parallel test stations and the average single-piece processing time. The specific calculation formula is as follows: ； in, For pipeline delay time, This represents the number of parallel test stations. The average single-item processing time is calculated; and the forward-looking predicted queue length of each sub-assembly line is calculated based on the current waiting queue length, the current work-in-process quantity in the corresponding pipeline segment, and the expected processing volume. Based on the comparison between the forward-looking predicted queue length of each sub-assembly line and the preset upper and lower limits of the queue length, the feeding interval of each sub-assembly line is adjusted independently.

2. The method according to claim 1, characterized in that, The step of calculating the forward prediction queue length for each sub-assembly line includes: calculating the forward prediction queue length of the sub-assembly line based on the current waiting queue length and the current work-in-process quantity in the corresponding pipeline segment, combined with the expected processing volume of the sub-assembly line.

3. The method according to claim 1, characterized in that, The method of independently adjusting the feeding interval of each sub-assembly line includes: increasing the feeding interval of a sub-assembly line when the forward prediction queue length of a sub-assembly line exceeds the upper limit of the queue length; shortening the feeding interval of a sub-assembly line when the forward prediction queue length of a sub-assembly line is lower than the lower limit of the queue length; and maintaining the current feeding interval of a sub-assembly line when the forward prediction queue length of a sub-assembly line is between the lower and upper limits of the queue length.

4. The method according to claim 3, characterized in that, The upper limit of the queue length is determined based on the physical capacity and buffer margin coefficient of the buffer zone in front of the bottleneck test station; the lower limit of the queue length is determined based on the number of parallel test stations and the safety margin.

5. The method according to claim 3, characterized in that, When increasing or decreasing the feeding interval, the adjustment range is determined by the normalized deviation of the forward prediction queue length from the corresponding upper or lower limit and the adjustment gain coefficient. The normalized deviation is the ratio obtained by dividing the difference between the forward prediction queue length and the corresponding upper or lower limit by the upper or lower limit. The adjusted feeding interval is limited to the minimum feeding interval and the maximum feeding interval of the sub-assembly line.

6. The method according to claim 5, characterized in that, The adjustment gain coefficient is obtained by calibrating the capacity margin between the upper limit of the queue length and the physical capacity of the buffer zone in front of the bottleneck test station. The capacity margin is the difference between the physical capacity of the buffer zone and the upper limit of the queue length. The calibration criterion is to divide the upper limit of the queue length by the capacity margin as the calibration criterion.

7. The method according to claim 1, characterized in that, The method further includes: selecting a reference interval containing a preset number of workstations on the main assembly line with a preset update cycle, calculating the actual time taken by multiple work-in-process items to pass through the reference interval recently, dividing the actual time taken by each work-in-process item by the number of workstations in the reference interval and taking the average value as the updated average single-workstation cycle time, and recalculating the pipeline delay time of each sub-assembly line accordingly.

8. The method according to claim 1, characterized in that, The method further includes: re-comparing the average single-piece processing time of each test station at the end of the main assembly line with a preset inspection cycle; when the bottleneck test station changes, re-determining the pipe depth and pipe delay time of each sub-assembly line according to the station number of the changed bottleneck test station.

9. A dynamic scheduling system for a cleaning robot assembly line, characterized in that, The assembly production line includes multiple sub-assembly lines and one main assembly line. The sub-assembly lines converge at different workstations on the main assembly line, and multiple testing stations are located at the end of the main assembly line. The system includes: The pipeline depth determination module is used to determine the pipeline depth of each sub-assembly line based on the number of stations between the merging station and the bottleneck testing station on the main assembly line. The pipeline depth represents the number of stations traversed on the main assembly line from the merging station of a sub-assembly line to the bottleneck testing station. Determining the bottleneck testing station includes: obtaining the average single-piece processing time of each testing station at the end of the main assembly line within a preset time window, and marking the testing station with the longest average single-piece processing time as the bottleneck testing station; when the difference between the average single-piece processing times of multiple testing stations is within a preset ratio range, marking the testing station with the longest current waiting queue as the bottleneck testing station. The pipeline delay time calculation module is used to obtain the pipeline delay time of each sub-assembly line based on the pipeline depth and the current average single-station cycle time of the main assembly line. The specific calculation formula is as follows: ,in For pipe depth, This represents the average time it takes for a work-in-process item to pass through one station on the main assembly line. The data acquisition module is used to collect the current waiting queue length, average single-piece processing time, and number of parallel testing stations of the bottleneck testing station, as well as the current work-in-process quantity in the corresponding pipeline segment of each sub-assembly line. The current work-in-process quantity in the corresponding pipeline segment of each sub-assembly line is the total number of work-in-process products flowing on the main line segment from the merging station to the bottleneck testing station. When the bottleneck testing station is changed, the pipeline depth and pipeline delay time of each sub-assembly line are re-determined according to the station number of the changed bottleneck testing station. The forward-looking queue length calculation module is used to determine the expected processing volume of each test station within the prediction window length for each sub-assembly line, using its respective pipeline delay time as the prediction window length, based on the number of parallel test stations and the average single-piece processing time. The specific calculation formula is as follows: ； in, For pipeline delay time, This represents the number of parallel test stations. The average single-item processing time is calculated; and the forward-looking predicted queue length of each sub-assembly line is calculated based on the current waiting queue length, the current work-in-process quantity in the corresponding pipeline segment, and the expected processing volume. The feeding interval adjustment module is used to independently adjust the feeding interval of each sub-assembly line based on the comparison results between the forward-looking predicted queue length and the preset upper and lower limits of the queue length.

Images