Production scheduling method, device, equipment and computer program product

Abstract

The embodiment of the invention provides a production scheduling method, device and equipment and a computer program product, and the method comprises the steps: determining the loads of all technological capacity groups in a current process time chain, and taking a site corresponding to the technological capacity group with the maximum load as a bottleneck site; according to the historical operation information, determining production cycles of all stations in front of the bottleneck station in the plurality of stations; initializing a plurality of candidate water level coefficients, and determining a current water level coefficient when the cost value of a scheduling cost model is minimum and an allowable unloading quantity under the current water level coefficient based on the pre-established scheduling cost model related to the water level coefficients, the production cycle, the output capacity of the process capacity group and the time length limitation; and according to the allowable unloading quantity and the current total operation quantity of the bottleneck station and the station before the bottleneck station in the current time chain, determining the input operation quantity. By the adoption of the scheme, the risk of exceeding the time limit can be reduced, and the machining capacity is maximized.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the technical field of semiconductor manufacturing, and in particular, to a production scheduling method, device, equipment and computer program product. BACKGROUND

[0002] With the continuous reduction of the key line width of semiconductor process, the influence of Q-time on the semiconductor manufacturing process is increasing. The overall trend is that the number of Q-time is increasing, the length of Q-time is becoming shorter, and there are multiple Q-time connected together to form a chain of Q-time.

[0003] When the wafer in the Q-time exceeds the time limit, there is a risk of low yield or even scrap. Therefore, Q-time limit is crucial to semiconductor wafer manufacturing and scheduling. However, the existing scheduling scheme still has the possibility of exceeding the time limit. SUMMARY

[0004] Therefore, embodiments of the present disclosure provide a production scheduling method, device, equipment and computer program product, which can reduce the risk of exceeding the time limit and maximize the processing capacity.

[0005] To achieve the above object, the technical scheme provided by the embodiments of the present disclosure is as follows.

[0006] In a first aspect, the embodiments of the present disclosure provide a production scheduling method, comprising: determining the load of all process capacity groups in the current process time chain, and taking the station corresponding to the process capacity group with the maximum load as the bottleneck station; the current process time chain has multiple stations, each station has a corresponding time length limit, and the bottleneck station is one of the multiple stations; determining the production cycle of all stations before the bottleneck station in the multiple stations according to historical job information; initializing multiple candidate water level coefficients, and determining the current water level coefficient when the cost value of the scheduling cost model related to the water level coefficient is the smallest, and the allowed incoming quantity under the current water level coefficient based on the pre-established scheduling cost model related to the water level coefficient, the production cycle, the output capacity of the process capacity group and the time length limit; determining the amount of work that can be put into the first station according to the allowed incoming quantity, and the current total amount of work of the bottleneck station and the stations before the bottleneck station in the current time chain.

[0007] Optionally, the determination of the load of all process capacity groups in the current time chain, and the determination of the bottleneck station corresponding to the process capacity group with the maximum load, comprises: Determine the work-in-process level at each station corresponding to each process capacity group, and the real-time estimated output of available equipment within the process capacity group; Based on the work-in-process water level and the real-time estimated output, the load of each process capacity group is determined, and the station corresponding to the process capacity group with the maximum load is designated as the bottleneck station.

[0008] Optionally, the production scheduling method satisfies one or more of the following: Determining the real-time projected output of available equipment within each of the process capacity groups includes: acquiring the real-time status of each available equipment, as well as the utilization rate and hourly output of each available equipment; and determining the real-time projected output based on the real-time status of the available equipment, as well as the utilization rate and hourly output of each available equipment. When determining the real-time estimated output, the real-time status of the available equipment is converted into equipment availability data according to a preset status determination rule. The equipment availability data is used to characterize whether the equipment is available. The equipment availability data includes at least a first value indicating that the equipment is available and a second value indicating that the equipment is unavailable. When determining the load for each of the process capacity groups, a normalization process is also performed, and the maximum load is determined based on the normalized load. When there are multiple sites corresponding to the maximum load, the site with the smallest position relative to the first site is determined as the bottleneck site based on the site order in the current time chain. The expression for determining the load of each of the aforementioned process capacity groups is as follows:

[0009] Bottleneck site

[0010] in, Indicates the water level of the product, Device availability data indicating the real-time status of the available devices. This indicates the hourly output of the available equipment. This indicates the utilization rate of the available equipment. This indicates the output capacity of the process capacity group; This indicates the real-time estimated output.

[0011] Optionally, the historical operation information includes: product batches processed by all stations within the current process time chain within a preset time period, and the running time status information of each station under each product batch; The step of determining the production cycle of all stations preceding the bottleneck station among multiple stations based on historical operation information includes: Based on the running time status information of each station under each product batch, determine the average production cycle of each station in the current process time chain; The average production cycle of all stations from the first station to the bottleneck station is accumulated to obtain the production cycle.

[0012] Optionally, the running time status information includes: running time, queuing time, and suspension time; The step of determining the average production cycle of each station within the current process time chain based on the running time status information of each station under each product batch includes: Based on the running time of each site under all product batches, a first interval and a corresponding average running time are determined; a first running time that meets the first interval and a second running time that does not meet the first interval are selected from all running times, and the average running time is used to replace all the second running times; an average running time is determined based on the first running time, all the average running times, and the number of all product batches. Based on the queuing time of each station under all product batches, determine the second interval corresponding to the queuing time and the corresponding average queuing time; select the first queuing time that meets the second interval and the second queuing time that does not meet the second interval from all queuing times, and use the average queuing time to replace all the second queuing times; determine the average queuing time based on the first queuing time, all the average queuing times, and the number of all product batches. Based on the suspension time of each site under all product batches, determine the third interval corresponding to the suspension time and the corresponding suspension average value; select the first suspension time that meets the third interval and the second suspension time that does not meet the third interval from all suspension times, and use the suspension average value to replace all the second suspension times; determine the average suspension time based on the first suspension time, all the suspension average values ​​and the number of all product batches; The sum of the average running time, the average queuing time, and the average hang-up time is taken as the average production cycle of the corresponding station within the current process time chain.

[0013] Optionally, the production scheduling method satisfies one or more of the following: The step of determining the first interval based on the running time of each site under all product batches includes: sorting all the running times to obtain a first sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the first sequence respectively; and determining the first interval based on the first quartile and the third quartile corresponding to the first sequence. The step of determining the second interval based on the queuing time of each station under all product batches includes: sorting all queuing times to obtain a second sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the second sequence respectively; and determining the second interval based on the first quartile and the third quartile corresponding to the second sequence. The step of determining the third interval based on the suspension time of each site under all product batches includes: sorting all suspension times to obtain a third sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the third sequence respectively; and determining the third interval based on the first quartile and the third quartile corresponding to the third sequence.

[0014] Optionally, when the maximum load of multiple process time chains is the same, the contribution ratio of each process time chain to the production cycle is determined based on the work-in-process level of the bottleneck station in each process time chain; the production cycle from the first station to the bottleneck station in the current process time chain is determined based on the contribution ratio of each process time chain and the production cycle from the first station to the bottleneck station in each process time chain.

[0015] Optionally, the initialization of multiple candidate water level coefficients, and the determination of the current water level coefficient at which the cost value of the scheduling cost model is minimized, and the allowable shipment quantity under the current water level coefficient, based on a pre-established scheduling cost model related to the water level coefficients, and the output capacity of the production cycle and the process capacity group, includes: Based on each candidate water level coefficient, as well as the production cycle and the output capacity of the process capacity group, determine the allowable shipment quantity under each candidate water level coefficient; Based on each candidate water level coefficient and its corresponding production cycle, allowable unloading volume, and output capacity of the process capacity group, determine the operational prediction parameters of the bottleneck site under each candidate water level coefficient; The operation prediction parameters are input into the scheduling cost model to determine the cost value of the scheduling cost model under each candidate water level coefficient. The candidate water level coefficient corresponding to the minimum cost value is taken as the current water level coefficient, and the allowable shipment quantity corresponding to the candidate water level coefficient is taken as the allowable shipment quantity under the current water level coefficient.

[0016] Optionally, determining the allowable shipment quantity under each candidate water level coefficient based on each candidate water level coefficient, the production cycle, the output capacity of the process capacity group, and the time length limit includes: The total production time is determined based on the production cycle and the time limit corresponding to the bottleneck site. The estimated total production volume is determined based on the total production time and the output capacity of the process capacity group; Based on the estimated total production volume and the candidate water level coefficient, the corresponding allowable shipment volume is determined.

[0017] Optionally, determining the operational prediction parameters of the bottleneck station under each candidate water level coefficient, based on each candidate water level coefficient and its corresponding production cycle, allowable unloading volume, and output capacity of the process capacity group, includes: A multilayer perceptron neural network model is used to perform nonlinear regression on the candidate water level coefficients and their corresponding production cycles, allowable loading quantities, and process capacity groups to obtain the operating prediction parameters for each candidate water level coefficient. The multilayer perceptron neural network model is trained using each actual water level coefficient within multiple historical periods, the actual allowable loading volume, actual production cycle, output capacity of the actual process capacity group, and actual operating parameters corresponding to the actual water level coefficient.

[0018] Optionally, within a preset search space, a particle swarm optimization algorithm is used to initialize multiple candidate water level coefficients, including: randomly generating multiple initial particles within the preset search space, and setting an initial position and initial velocity for each initial particle; wherein each initial particle represents a candidate water level coefficient, and the initial velocity represents the change of the candidate water level coefficient in adjacent iterations; The step of inputting the operational prediction parameters into the scheduling cost model to determine the cost value of the scheduling cost model under each candidate water level coefficient includes: Substitute the running prediction parameters corresponding to each initial particle into the scheduling cost model to obtain the cost value of each initial particle, and record the individual optimal cost value and corresponding position of each initial particle; Compare the individual optimal cost values ​​of all the initial particles, select the particle with the smallest individual optimal cost value as the global optimal particle, and record its corresponding global optimal cost value and global optimal position. Perform at least one update iteration operation to update the velocity and position of each initial particle, and apply boundary constraints to the updated particle positions so that each updated particle is still within the search space, and recalculate the global optimal cost value and the global optimal position. The step of using the candidate water level coefficient corresponding to the minimum cost value as the current water level coefficient includes: Calculate the change in cost difference between the two global optimal cost values ​​in any two adjacent update iterations, and in response to the change in cost difference being less than a preset threshold, use the current global optimal position as the current water level coefficient.

[0019] Optionally, the production scheduling method satisfies one or more of the following: The preset search space is selected from [0.5, 2]; The preset threshold is 0.01.

[0020] Optionally, the scheduling cost model includes: a first cost function and a second cost function, wherein the first cost function has a first weight coefficient, the second cost function has a second weight coefficient, and the sum of the first weight coefficient and the second weight coefficient is a constant. Wherein, the first cost function reflects the loss cost when the candidate water level coefficient is in an over-configured state, the second cost function reflects the idle cost when the candidate water level coefficient is in an under-configured state, and the time length limit of the bottleneck station serves as the time constraint information of the first cost function.

[0021] Optionally, the operation prediction parameters include: the actual average queuing time of the bottleneck site, the output of the process capacity group corresponding to the bottleneck site, and the idle time. In response to the actual average queuing time being less than or equal to the time length limit of the bottleneck site, 0 is used as the result of the first cost function; In response to the actual average queuing time exceeding the time limit of the bottleneck station, a first cost function is determined based on the actual average queuing time. Each candidate water level coefficient has its own actual average queuing time. The process includes: weighting the deviation of the actual average queuing time from the time limit according to a preset third weighting coefficient, where the deviation is measured in squared form; dynamically adjusting the weighted result based on the ratio between the number of work-in-process exceeding the time limit and the corresponding total quantity within multiple historical periods, wherein the dynamic adjustment includes nonlinearly amplifying the ratio according to a preset proportional adjustment coefficient and an exponential adjustment parameter; and combining the weighted result with the nonlinearly amplified adjustment result to obtain the first cost function.

[0022] Optionally, a second cost function is determined based on the output capacity and idle time of the process capacity group corresponding to each of the candidate water level coefficients, including: Based on a preset fourth weighting coefficient, the deviation between the output capacity and the theoretical output capacity of the process capacity group is processed, and the deviation is nonlinearly amplified to obtain the deviation result; wherein, the output capacity of the process capacity group is determined based on the output of the process capacity group. Based on the idle time of the machines in the process capacity group and its proportion in the preset production cycle, the deviation result is adjusted to obtain the second cost function.

[0023] Optionally, the expression for the first cost function is:

[0024] The second cost function The expression is:

[0025] The expression for the scheduling cost model is:

[0026] in, Represents any of the candidate water level coefficients; This represents the actual average queuing time at the bottleneck station corresponding to the candidate water level coefficient. This indicates the time limit for the bottleneck site; This represents the third weighting coefficient; Indicates the amplification factor for historical events; This indicates the number of work-in-process items that exceed the time length limit over multiple historical periods; This indicates the total quantity of work-in-process over multiple historical periods; Indicates the order of the amplification factor for historical events; This represents the fourth weighting coefficient; This indicates the output capacity of the process capacity group corresponding to the candidate water level coefficient; This indicates the theoretical output capacity of the process capacity group; This indicates the idle time of the machines in the process capacity group within a preset production cycle; This indicates the preset production cycle; Indicates the penalty coefficient; This represents the first weighting coefficient. This represents the second weighting coefficient.

[0027] Optionally, determining the available workload from the first station based on the allowed unloading volume and the current total workload of the bottleneck station and the stations preceding the bottleneck station in the current time chain includes: In response to the allowable delivery quantity being greater than the current total operation quantity, the difference between the allowable delivery quantity and the current total operation quantity is taken as the available operation quantity; In response to the fact that the allowed unloading quantity is less than the current total operation quantity, the available operation quantity is determined to be 0.

[0028] Secondly, embodiments of this disclosure provide a production scheduling device, comprising: The first determining unit is configured to determine the load of all process capacity groups within the current process time chain, and to designate the station corresponding to the process capacity group with the largest load as the bottleneck station; the current process time chain has multiple stations, each of which has a corresponding time length limit, and the bottleneck station is one of the multiple stations. The second determining unit is configured to determine the production cycle of all stations located before the bottleneck station among the plurality of stations based on historical operation information; The processing unit is configured to initialize multiple candidate water level coefficients, and based on a pre-established scheduling cost model related to the water level coefficients, as well as the production cycle, the output capacity of the process capacity group, and the time length limit, determine the current water level coefficient when the cost value of the scheduling cost model is minimized, and the allowable unloading quantity under the current water level coefficient. The scheduling unit is configured to determine the amount of work that can be put into operation from the first station based on the allowed unloading volume and the current total workload of the bottleneck station and the stations before the bottleneck station in the current time chain.

[0029] Thirdly, embodiments of this disclosure provide an electronic device, including a memory and a processor, wherein the memory stores a computer program executable on the processor, and the processor executes the steps of the production scheduling method described in any of the foregoing embodiments when running the computer program.

[0030] Fourthly, embodiments of this disclosure provide a computer program product, including a computer program / instruction, wherein the computer program / instruction, when executed by a processor, is used to implement the production scheduling method as described in any of the foregoing embodiments.

[0031] Fifthly, embodiments of this disclosure provide a storage medium storing one or more computer instructions for implementing the production scheduling method as described in any of the foregoing embodiments.

[0032] Compared with the prior art, the technical solution of the embodiments of the present invention has the following advantages: In the production scheduling method provided in this embodiment, there are multiple stations within the current process time chain. By determining the load of all process capacity groups within the current process time chain, the bottleneck station of the current process time chain can be identified. Furthermore, based on historical operation information, the production cycle of all stations preceding the bottleneck station can be determined, reflecting the time elapsed from when a product is processed at the first station until it can enter the bottleneck station. Moreover, by initializing multiple candidate water level coefficients, based on a pre-established scheduling cost model related to the water level coefficients, and the output capacity of the production cycle and process capacity groups, the current water level coefficient with the lowest cost value of the scheduling cost model, and the allowable output quantity at the current water level coefficient can be determined, taking into account production costs. Thus, based on the allowable output quantity and the current total operation of the bottleneck station and the stations preceding the bottleneck station within the current time chain, the amount of work that can be put into operation from the first station can be determined. In the above solution, by instantly judging the workability of each process capacity group in the current process time chain, and determining the bottleneck station and production cost based on the real-time situation, the available workload from the first station in the time chain can be automatically determined. This allows the available workload to match the processing capacity and timing constraints of the bottleneck station in the current process time chain, effectively reducing the risk of exceeding time limits and maximizing processing capacity, thus having greater versatility. Attached Figure Description

[0033] Figure 1 A schematic diagram of a production process with Q-time constraints is shown. Figure 2 A flowchart of a production scheduling method according to an embodiment of this disclosure is shown; Figure 3 A flowchart illustrating the determination of a bottleneck site is shown in an embodiment of this disclosure; Figure 4 A flowchart illustrating the determination of a production cycle is shown in an embodiment of this disclosure; Figure 5 A flowchart illustrating the selection of a water level coefficient according to an embodiment of this disclosure is shown; Figure 6 An over-ratio trend graph of a work-in-process product is shown in an embodiment of this disclosure; Figure 7 A schematic diagram of the structure of a production scheduling device according to an embodiment of this disclosure is shown; Figure 8 A schematic diagram of an optional hardware structure of an electronic device provided in an embodiment of this disclosure is shown. Detailed Implementation

[0034] 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.

[0035] As described in the background section, Q-time constraints are critical for semiconductor wafer manufacturing and scheduling. Currently, most factories with multiple production lines and technology nodes require complex manufacturing and operational systems to address multi-step Q-time constraints.

[0036] See Figure 1 The diagram shown illustrates a production process with Q-time constraints, as follows: Figure 1 As shown, the current production process includes: Station A, Station B, Station C, Station D, and Station E, with a corresponding Q-time limit of 12 hours. That is, the time limit for the processed product (e.g., wafer) from the end of Station A (corresponding to the start point of Q-time calculation) to the end of Station E (corresponding to the end point of Q-time calculation) is 12 hours.

[0037] Each station has a corresponding processing time. For example, the processing time of station A is 1 hour, that of station B is 1 hour, that of station C is 2 hours, that of station D is 1 hour, and that of station E is 1 hour. Furthermore, the processing times of stations B through D are considered part of the Q-time.

[0038] Furthermore, there is a permitted waiting time between adjacent stations. For example, the waiting time between the first station A and the second station B is 2 hours, the waiting time between the second station B and the third station C is 2 hours, the waiting time between the third station C and the fourth station D is X hours, and the waiting time between the fourth station D and the fifth station D is Y hours, where X + Y is less than 4.

[0039] In other words, Q-time includes the process time corresponding to a station and the waiting time between adjacent stations. Each process time chain is affected by both process time and waiting time.

[0040] To reduce the impact of Q-time on semiconductor manufacturing, current methods for time chain control in semiconductor production include: The first method is at the time chain starting point (corresponding to...) Figure 1 In the first station A), a virtual station, namely a Dummy station, is added. The number of work-in-process products that can be put into the time chain is calculated manually every day based on the work-in-process level and equipment capacity in the time chain. Work-in-process products that can be put into the time chain are skipped by the Dummy station, or the number of work-in-process products that can be put into the time chain in each Dummy station is summarized and a CSV document is generated and uploaded to the Real-Time Dispatch (RTD) system, and the RTD will complete the station skipping in a unified manner.

[0041] Because it requires manual monitoring of all equipment status within the time chain and calculation of the number of products that can be added to the time chain based on the number of available equipment, hourly output of equipment, and existing work-in-process inventory, the process is cumbersome and computationally intensive. The calculation results are highly subjective, and the accuracy is strongly correlated with the experience of the calculation personnel, making it unsuitable for dynamic production processes like semiconductor manufacturing.

[0042] The second approach involves predicting the risk of processes exceeding Q-time and selecting the optimal scheduling action while maximizing capacity and minimizing the risk of exceeding Q-time. This primarily involves the rational scheduling of work-in-process (WIP) already at Q-time stations to reduce the risk of exceeding Q-time and balance capacity.

[0043] However, in the second approach, when equipment malfunctions within the time chain, leading to reduced production capacity or even line stoppage, the order quantity at the dummy station cannot be adjusted immediately. Manual intervention is required to abnormally suspend work-in-process within the time chain, which can easily result in low yields or scrapped products due to exceeding time limits.

[0044] The third approach is to predict the availability of equipment to determine whether to schedule wafers. This includes wafer scheduling and management before planned equipment downtime, as well as the advance delivery time and quantity of wafers before planned equipment restart. Scheduling is carried out according to a fixed value of the number of workable wafers to reduce the risk of over Q-time caused by downtime.

[0045] However, the third approach only assesses and schedules the work-in-process (WIP) level within the current Q-time loop, without considering the impact of equipment downtime or other factors that could lead to timely responses such as line shutdowns. Therefore, it cannot fundamentally control the Q-time WIP value to address the risk of over-Q-time.

[0046] Therefore, it is evident that current Q-time management solutions are unable to dynamically adjust the amount of goods that can be shipped based on real-time manufacturing and production conditions.

[0047] Based on this, this disclosure provides a production scheduling method. The current process time chain has multiple stations. By determining the load of all process capacity groups within the current process time chain, the bottleneck station of the current process time chain can be identified. Furthermore, based on historical operation information, the production cycle of all stations preceding the bottleneck station can be determined, reflecting the time elapsed from product processing at the first station to its entry into the bottleneck station. Moreover, by initializing multiple candidate water level coefficients, based on a pre-established scheduling cost model related to the water level coefficients, and considering the production cycle and the output capacity of the process capacity groups, the current water level coefficient with the lowest cost value of the scheduling cost model, as well as the allowable output quantity at the current water level coefficient, can be determined, taking into account manufacturing costs. Thus, based on the allowable output quantity and the current total operation of the bottleneck station and the stations preceding the bottleneck station within the current time chain, the available operational volume from the first station can be determined. In the above solution, by instantly judging the workability of each process capacity group in the current process time chain, and determining the bottleneck station and production cost based on the real-time situation, the available workload from the first station in the time chain can be automatically determined. This allows the available workload to match the processing capacity and timing constraints of the bottleneck station in the current process time chain, effectively reducing the risk of exceeding time limits and maximizing processing capacity, thus having greater versatility.

[0048] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be described by way of example below with reference to the accompanying drawings.

[0049] See Figure 2 The flowchart of a production scheduling method in an embodiment of this disclosure is shown below. Figure 2 As shown, the following scheduling process can be executed: S201, determine the load of all process capacity groups in the current process time chain, and take the station corresponding to the process capacity group with the largest load as the bottleneck station; the current process time chain has multiple stations, each of which has a corresponding time length limit, and the bottleneck station is one of the multiple stations.

[0050] In some embodiments, each process time chain may include multiple stations (e.g. Figure 1 The diagram shows five stations, from station A to station E (though there may be more or fewer stations). Each station corresponds to at least one process capacity group, which processes the operations performed at the current station.

[0051] Considering that the real-time equipment conditions within the current process time chain vary, causing bottleneck equipment to continuously shift, it is necessary to determine the real-time bottleneck equipment based on the load of the process capacity group, and then identify the bottleneck site. The bottleneck site can be understood as the site that first reaches load saturation and constrains the maximum available workload. This dynamic flow consideration overcomes the limitations of traditional methods that determine bottleneck equipment solely based on the static capacity value of the capacity group.

[0052] Specifically, within the time chain of the current process, load calculations are performed on all process capacity groups included in that time chain to determine the production load that each process capacity group needs to bear within the time chain. The load of a process capacity group can be determined based on the processing demand, work cycle time, or resource occupancy of the corresponding station per unit time.

[0053] After determining the load of each process capacity group, the loads of all process capacity groups are compared, the process capacity group with the largest load value is selected, and the station corresponding to the process capacity group is determined as the bottleneck station.

[0054] In some embodiments, each station corresponds to a different process or processing step, and each station has a pre-set time limit to constrain the maximum time that the product to be processed can stay at the current station after the processing is completed.

[0055] In some embodiments, a site corresponds to a process, and the process can refer to a primary process step such as photolithography or etching, or it can refer to a specific operation step or sub-process within the photolithography process, such as: coating, soft baking, alignment, exposure, post-baking, development, etc.

[0056] In this embodiment, the process refers to processes such as photolithography and etching.

[0057] In some embodiments, a bottleneck site refers to any site other than the first site among a plurality of sites. This is because the first site is the first to receive the products to be processed, and the quantity disposed of during unloading is sufficient to meet the processing capacity of the first site.

[0058] In some embodiments, a process capacity group refers to a capacity management unit formed by logically aggregating at least one site based on process type or production capacity similarity, used to characterize the overall production capacity available for performing at least one process.

[0059] A site refers to a specific production resource unit that performs a process, including but not limited to a single piece of equipment, a workstation, or a combination thereof, used to carry out the actual processing or handling operations of the process.

[0060] A process refers to a standardized processing or handling step performed on work-in-process during the production process. It is not limited to a specific execution site and can be performed by one or multiple sites in parallel.

[0061] The MaxQ-time time limit for each station refers to the maximum allowed dwell time within the current time chain, from the completion of product processing at the previous station until entry into the current station. For example, see... Figure 1 The time limit MaxQ-time for the second station B refers to the maximum dwell time after the first station A completes processing the product and before entering its own station.

[0062] S202, Based on historical work information, determine the production cycle of all stations among the multiple stations that are located before the bottleneck station.

[0063] In some embodiments, the processing time varies at different times when the same station processes products to be processed (e.g., the same product to be processed or different products to be processed). Historical operation information is used to characterize the operation status of each station in the corresponding process during the historical production process. Therefore, based on the historical operation information, the time information corresponding to each station can be determined.

[0064] In this way, after identifying the bottleneck site, the first site can be used as the starting site. The time consumed by the product or work object from entering the first site, passing through each intermediate site in sequence, until it can reach the bottleneck site for operation can be counted. This includes the actual operation time of each site as well as the waiting time and transfer time between sites. Thus, the production cycle of all sites from the first site to the bottleneck site can be determined.

[0065] For example, the stations in the current timeline are sequential, with each station occupying a position within that sequence. Thus, when a bottleneck station is identified, the production cycles of all stations preceding it can be determined. See also... Figure 1 Assuming the bottleneck site is the third site C, the total production cycle of the first site A and the second site B is determined based on historical operation information.

[0066] S203, initialize multiple candidate water level coefficients, and based on the pre-established scheduling cost model related to the water level coefficients, as well as the production cycle, the output capacity of the process capacity group, and the time length limit, determine the current water level coefficient when the cost value of the scheduling cost model is minimized, and the allowable unloading quantity under the current water level coefficient.

[0067] In some embodiments, steps S201 and S202 are used to determine the bottleneck site and some time information and processing capacity information corresponding to the bottleneck site. Considering the uncertainty of the processing process, and taking into account both maximizing the capacity of the bottleneck equipment and minimizing over-Q-time losses, based on the determination of the bottleneck site, the possible allowable loading quantity is further determined, including the minimum allowable loading quantity and the maximum allowable loading quantity.

[0068] Specifically, after identifying the bottleneck site, multiple candidate water level coefficients are initialized (these multiple candidate water level coefficients can be grouped together, and subsequent candidate water level coefficients can be modified based on this group), and based on the already determined production cycle and the output capacity of the process capacity group, the allowable shipment quantity corresponding to each candidate water level coefficient can be determined.

[0069] Next, based on the allowable shipment volume, production cycle, and output capacity of the process capacity group corresponding to each candidate water level coefficient, the scheduling cost corresponding to each candidate water level coefficient can be determined using the scheduling cost model.

[0070] By comparison, the minimum scheduling cost can be determined, and the candidate water level coefficient corresponding to the minimum scheduling cost can be used as the current water level coefficient. At the same time, the allowable unloading quantity corresponding to the minimum scheduling cost can be used as the allowable unloading quantity of the current water level coefficient.

[0071] Thus, adopting a scheme based on polling and scheduling cost models can significantly reduce the risk of products exceeding time limits and avoid bottleneck equipment idleness caused by low work-in-process levels within the time chain control process.

[0072] S204. Based on the allowed unloading volume and the current total workload of the bottleneck station and the stations before the bottleneck station in the current time chain, determine the amount of work that can be put into operation from the first station.

[0073] In some embodiments, by performing the above steps, the allowable unloading volume within the current time chain can be determined. Since some stations within the current time chain may already be in the processing stage, the allowable unloading volume can be compared with the current total workload to determine the available workload from the first station within the current time chain.

[0074] In other words, calculate the total WIP value from the Q-time starting point to the bottleneck process in each time chain, which is the total WIP value of the bottleneck equipment operations that have entered the time chain and are yet to be used. .

[0075] For example, given a timeline containing 1, 2, ..., m processes, corresponding to m stations, and assuming the bottleneck station is the (m-2)th process, then its... Assuming the bottleneck process is the m-th process, then .

[0076] Accordingly, in response to the allowable delivery quantity being greater than the current total operation quantity, the difference between the allowable delivery quantity and the current total operation quantity is taken as the available operation quantity.

[0077] Specifically, if the allowed unloading volume exceeds the current total workload, it indicates that the current workload is not saturated and there is still capacity for further product processing. Therefore, the difference between the allowed unloading volume and the current total workload is taken as the available workload, and the unloading volume is gradually increased starting from the first station.

[0078] Correspondingly, in response to the allowable unloading quantity being less than the current total operation quantity, the available operation quantity is determined to be 0.

[0079] Specifically, allowing the unloading volume to be less than the current total workload indicates that the current workload is already saturated and no new workload should be added.

[0080] For example, suppose a time chain is established, and the determined allowable delivery quantity is... The current workload within this time chain is Calculate the amount of available tasks for the available time chain. ,in:

[0081] like The amount of work that can be put into operation is Value; if If so, the amount of work that can be put into operation is 0.

[0082] It should be noted that, firstly, if The available workload is then fed back to the real-time dispatch system, and work is carried out according to the available workload. The first method is to distribute goods based on value; the second method is to control the quantity of goods that can be distributed by controlling the dummy site of the first site.

[0083] In some embodiments, the real-time bottleneck equipment is defined by combining the quantity of work-in-process in the buffer zone within the production capacity group and the real-time machine status of the equipment to calculate the load of the machine group. In this way, the process module with the largest machine group load in the time chain is defined as the current bottleneck process module.

[0084] See Figure 3 The flowchart shown in this embodiment of the present disclosure illustrates a method for determining bottleneck sites, such as... Figure 3 As shown, it includes: S301, determine the work-in-process level at each station corresponding to each process capacity group, and the real-time estimated output of available equipment within the process capacity group.

[0085] In some embodiments, each process capacity group includes at least one processing equipment. Based on the MES and EAP equipment, the work-in-process level at the current station can be determined in real time, that is, the number of work-in-process items in the process capacity group when the corresponding process at this station is executed, which reflects the load status of this process capacity group.

[0086] Furthermore, the processing is carried out by the processing equipment within the process capacity group, thus allowing for a further determination of the real-time projected output of available equipment.

[0087] Available equipment can refer to equipment that can be normally arranged for production / operation at the current moment or within the planned period.

[0088] In one embodiment, determining the real-time projected output of available equipment within each of the process capacity groups includes: Obtain the real-time status of each available device, as well as the utilization rate and hourly output of each available device; determine the real-time projected output based on the real-time status of the available devices, as well as the utilization rate and hourly output of each available device.

[0089] Specifically, considering the uncertainties in the production process, it is usually necessary to obtain the real-time status (available or unavailable) of each available device, as well as the utilization parameters of the available devices themselves, in order to determine the real-time expected output.

[0090] In short, based on the number of available equipment for each process m... Each process m corresponds to the available equipment i with an hourly output ( , indicating the process number; , indicating the process module equipment number.

[0091] In some embodiments, the initially acquired real-time equipment status is unstructured data, such as Wait_ENG, Hold_ENG, RUN, IDLE, etc. Therefore, when determining the real-time estimated output, the real-time equipment status of the available equipment is converted into equipment availability data according to a preset status determination rule. The device availability data Used to characterize whether a device is available; wherein, the device availability data It includes at least: a first value indicating that the device is available and a second value indicating that the device is not available.

[0092] In one embodiment, when the device is available, device availability data... The value is 1; when the device is unavailable, the device availability data is... The value of is 0.

[0093] Specifically, when the real-time status of the device is Wait_ENG or Hold_ENG, the corresponding device availability data is... The value is 0; when the device's real-time status is RUN or IDLE, the corresponding device availability data is... The value is 1.

[0094] It should be noted that the real-time device status listed above is used to indicate the status of available devices and does not limit other statuses of available devices.

[0095] Therefore, the real-time projected output can be determined using the following expression: .

[0096] in, Device availability data indicating the real-time status of the available devices. This indicates the hourly output of the available equipment, for example, the hourly wafer output of a bottleneck site. This indicates the utilization rate of the available equipment. This indicates the capacity of the process capacity group.

[0097] S302, based on the work-in-process water level and the real-time estimated output, determine the load of each process capacity group, and designate the station corresponding to the process capacity group with the maximum load as the bottleneck station.

[0098] In some embodiments, the work-in-process level reflects the quantity of products being processed, while the total output characterizes the output capacity of the current process capacity group. Based on the work-in-process level and the total output, the load of each process capacity group can be determined, and the station corresponding to the process capacity group with the highest load can be designated as the bottleneck station.

[0099] In one example, the ratio between the finished product water level and the total output is used as the load of the process capacity group. That is, the expression for determining the load of each process capacity group is:

[0100] Correspondingly, bottleneck sites

[0101] in, Indicates the water level of the product, Device availability data indicating the real-time status of the available devices. This represents the hourly output of the available equipment (e.g., all available equipment). This indicates the utilization rate of the available equipment. This indicates the output capacity of the process capacity group; This indicates the real-time estimated output.

[0102] Considering the differences in dimensions and orders of magnitude among the outputs involved in the production plan, directly comparing the load of each process capacity group would be insufficient. This can easily lead to an unreasonable dominant role.

[0103] Therefore, when determining the load of each of the process capacity groups, a normalization process is also performed, and the maximum load is determined based on the normalized load.

[0104] For example, and based on Determine the maximum load.

[0105] By adopting the above scheme, the static capacity assessment is transformed into a dynamic load analysis based on the on-site operating status by determining the work-in-process level and the real-time estimated output of the available equipment within each process capacity group at the corresponding site. This allows the load of the process capacity group to be updated in real time as the work-in-process level and equipment availability change. Based on this, by comprehensively calculating the work-in-process level and the real-time estimated output, the actual carrying capacity of each process capacity group can be dynamically reflected. This enables timely and accurate identification of the bottleneck site corresponding to the process capacity group with the highest load, avoiding bottleneck judgment lag caused by fixed parameters or historical data, and improving the real-time performance, accuracy, and adaptability to production fluctuations in bottleneck identification.

[0106] In actual processing scenarios, a process capacity group may correspond to multiple stations within the same process time chain. When the load of this process capacity group is at its maximum, there are multiple bottleneck stations. At this point, one of the bottleneck stations is selected based on its position.

[0107] For example, when there are multiple sites corresponding to the maximum load, the site with the smallest position relative to the first site is determined as the bottleneck site based on the site order in the current time chain.

[0108] Specifically, the products to be processed are processed sequentially. After being processed at the first station, they are transferred to the second station, and so on, until they are processed by the last station in the current process time chain. In other words, there is a sequence among the multiple stations. Thus, selecting the station with the smallest position relative to the first station as the bottleneck station can expose the restricted location earlier, thereby improving the timeliness and effectiveness of system scheduling and control.

[0109] For example, see Figure 1 If the sites corresponding to the maximum load include the second site B and the third site C, then the second site B will be regarded as the bottleneck site.

[0110] When identifying a bottleneck site, it is possible to further determine the production cycle of all sites from the first site up to the bottleneck site.

[0111] In some embodiments, historical job information includes: product batches processed by all stations within the current process time chain within a preset time period, and the running time status information of each station under each product batch.

[0112] Specifically, the preset time period can be configured according to production management needs, such as a specific shift, a specific production day, or a specific statistical period.

[0113] A product batch refers to a production unit that enters and completes the current process time chain processing flow within the preset time period. Each product batch is processed or handled sequentially through multiple stations according to the process sequence. The current process time chain is used to characterize the order in which the product passes through the stations in the current process and their corresponding time relationships.

[0114] Furthermore, for each product batch, the historical operation information also includes the running time status information of that product batch at each site. The running time status information includes at least one or more of the following at the corresponding site: start processing time, end processing time, and processing duration. It may also include time-dimensional information such as waiting time, idle time, processing time, or abnormal stoppage time at the site.

[0115] By collecting and storing the aforementioned historical operation information, the actual operation status of each product batch at different stations in the current process time chain can be comprehensively reflected within the preset time period, thereby providing data support for subsequent production cycle analysis, bottleneck station identification, process optimization, and production plan adjustment.

[0116] For example, as shown in Table 1, this is a historical process record table for the historical operation cycle. The specific data content includes: Lot ID (production batch identifier, used to track the processing flow of the same batch of products), Stage (process stage), StepNo (step number, corresponding to the station in this solution), Run Time (running time), Que Time (queue time), and Hold Time (suspend / wait / hold time).

[0117] Specifically, Table 1 shows S1 to S... n The historical production cycle of each stage is illustrated using Step 1 in S1 as an example.

[0118] Referring to Table 1, for Step 1 in S1, the corresponding Lot IDs are Lot1 to Lot. N And their respective QT1, RT1, and HT1, as well as QTN RT N and HT N .

[0119]

[0120] Table 1 Historical Assignment Information For example, as shown in Table 1, if 10 lots will pass through stations A to E, then each station will have 10 lots with Run Time, Que Time, and Hold Time.

[0121] Correspondingly, see Figure 4 The flowchart illustrating a production cycle determination in this embodiment of the present disclosure is shown below. Figure 4 As shown, the following steps can be performed: S401, based on the running time status information of each station under each product batch, determine the average production cycle of each station in the current process time chain.

[0122] In some embodiments, when processing products, each batch of products will sequentially pass through multiple stations for production and processing in its corresponding process time chain.

[0123] For any given site, the runtime status information for different product batches (as a non-limiting example, including processing start time, processing end time, waiting time, idle time, and abnormal downtime, etc.) can be obtained.

[0124] In this step, based on the running time status information of the same site under multiple different product batches, the actual processing time of the site can be statistically analyzed, thereby calculating the average production cycle of the site within the current process time chain.

[0125] As a non-limiting example, the average production cycle can be the arithmetic mean of the actual processing time of the site across multiple product batches, or it can be a weighted average; this embodiment of the invention does not impose any limitation on this.

[0126] It should be noted that by introducing the average production cycle, the impact of a single product batch or abnormal processing on the assessment of the site's production capacity can be eliminated, so that the obtained site production cycle can more accurately reflect the overall operating level of the site in the current process time chain.

[0127] In one embodiment, runtime status information includes runtime, queuing time, and hang time. That is, the average production cycle is actually the average of these three runtime status information: runtime, queuing time, and hang time.

[0128] In this case, step S401 may include: S4011, based on the running time of each of the stations under all the product batches, determine the first interval and the corresponding average running time; select the first running time that meets the first interval and the second running time that does not meet the first interval from all the running times, and use the average running time to replace all the second running times; determine the average running time based on the first running time, all the average running times and the number of all product batches.

[0129] In some embodiments, considering the fluctuations in runtime in historical job information, some outliers may occur. Therefore, after obtaining multiple runtimes, further anomaly removal processing is performed, dividing all runtimes into a first runtime that meets a first interval and a second runtime that does not meet the first interval. Then, based on the first runtime, the second runtime, and the corresponding batch number, the average runtime is obtained.

[0130] For example, if the number of running times is consistent with the number of batches, by using the method of determining the first interval, the number of second running times outside the first interval can be determined. Then, the average running time is used to replace these second running times, and the final average running time is recalculated.

[0131] In one example, suppose we now have 11 running times [6, 7, 14, 36, 39, 40, 41, 42, 43, 47, 49]. After determining the first interval and the average of 33, assuming that 6, 7, and 14 can be identified as outliers, we replace 6, 7, and 14 with the average of 33, and then use [33, 33, 33, 36, 39, 40, 41, 42, 43, 47, 49] to determine the average running time.

[0132] In one example, the quartile method is used to determine the first interval.

[0133] Specifically, all running times are sorted to obtain a first sequence arranged in ascending order; the first quartile and the third quartile corresponding to the first sequence are calculated respectively; the first interval is determined based on the first quartile and the third quartile corresponding to the first sequence.

[0134] For example, the lower limit of the first interval is:

[0135] The upper limit of the first interval is:

[0136] Thus, after determining the upper and lower limits of the first interval, the running time outside this interval is called the second running time, and the average value of the second running time is used to replace these second running times.

[0137] That is, for A certain data in or Then, the average of all values ​​is used to replace these outliers:

[0138] in, Indicates the number of times the second run takes; This represents any one of the second running times.

[0139] S4012, based on the queuing time of each station under all product batches, determine the second interval corresponding to the queuing time and the corresponding average queuing time; select the first queuing time that meets the second interval and the second queuing time that does not meet the second interval from all queuing times, and use the average queuing time to replace all the second queuing times; determine the average queuing time based on the first queuing time, all the average queuing times and the number of all product batches.

[0140] In some embodiments, considering the fluctuations in queuing times in historical job information, some outliers may occur. Therefore, after obtaining multiple queuing times, further anomaly removal processing is performed, dividing all running times into a first queuing time that meets the second interval and a second queuing time that does not meet the second interval. Then, based on the first queuing time, the second queuing time, and the corresponding batch number, the average queuing time is obtained.

[0141] In one example, the quartile method is used to determine the second interval.

[0142] For example, determining the second interval based on the queuing time of each of the stations under all the product batches includes: sorting all the queuing times to obtain a second sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the second sequence respectively; and determining the second interval based on the first quartile and the third quartile corresponding to the second sequence.

[0143] The process for determining the second interval can be found in the description of the first interval; the only difference between the two is the object being processed.

[0144] S4013, based on the suspension time of each station under all product batches, determine the third interval corresponding to the suspension time and the corresponding suspension average value; select the first suspension time that meets the third interval and the second suspension time that does not meet the third interval from all suspension times, and use the suspension average value to replace all the second suspension times; determine the average suspension time based on the first suspension time, all the suspension average values ​​and the number of all product batches.

[0145] In some embodiments, considering the fluctuations in suspension times in historical job information, some outliers may occur. Therefore, after obtaining multiple suspension times, further anomaly removal processing is performed, dividing all runtimes into a first suspension time that meets the third interval and a second suspension time that does not meet the third interval. Then, based on the first suspension time, the second suspension time, and the corresponding batch number, the average suspension time is obtained.

[0146] In one example, the quartile method is used to determine the third interval.

[0147] For example, determining the second interval based on the queuing time of each of the stations under all the product batches includes: sorting all the queuing times to obtain a third sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the third sequence respectively; and determining the third interval based on the first quartile and the third quartile corresponding to the third sequence.

[0148] For the process of determining the third interval, please refer to the descriptions of the first and / or second intervals; the only difference between the two is the object being processed.

[0149] S4014, the sum of the average running time, the average queuing time, and the average hang-up time is taken as the average production cycle of the corresponding station within the current process time chain.

[0150] For example,

[0151] In this way, a production cycle configuration table can be constructed for the current process time chain. As shown in Table 2, the Q-timetraget between two adjacent stations (Q-time from step-Q-time to step) in the current process time chain, and the corresponding CT, are illustrated.

[0152]

[0153] Table 2 Production Cycle Configuration Table For example, when Q-time from step is 1 and Q-time to step is 2, the corresponding Q-time target is Max. Q-time1 (Determined based on the characteristics of the product in production), CT is CT1.

[0154] S402, the average production cycle of all stations before the bottleneck station is accumulated and used as the production cycle.

[0155] Specifically, after completing step S401, each station in the current process time chain corresponds to its average production cycle, and the positions of the first station and the bottleneck station can be predetermined in the process time chain. Among them, the bottleneck station is the station that has the greatest impact on the overall production cycle.

[0156] In this step, the average production cycle of all stations from the first station to the bottleneck station can be accumulated sequentially according to the order of the stations in the process time chain, so as to obtain the overall production cycle of all stations from the first station to the bottleneck station.

[0157] This overall production cycle is used to characterize the total production time required for a product to move from the current process timeline to the bottleneck site.

[0158] For example, .

[0159] in, Used to indicate the first station in the process time chain (e.g., Figure 1 The first station shown is A). Used to represent the station before the bottleneck station within a process time chain (e.g., Figure 1 The bottleneck station shown is the third station C. (For the second station B) This indicates the average production cycle for each site.

[0160] For example, referring to Table 2, assuming site 3 is the bottleneck site, then .

[0161] It should be noted that by calculating the production cycle of the stations from the first station to the bottleneck station, the impact of the processes before the bottleneck station on the overall production line cycle time can be assessed more accurately, providing a reliable data basis for subsequent production line balancing, cycle time optimization, or bottleneck elimination.

[0162] Furthermore, the inventors discovered that if multiple different process capacity groups with the same maximum load exist within multiple time chains, then the production cycle from the first station to the bottleneck station within the current process time chain can be determined based on the production cycle corresponding to the bottleneck station within the multiple time chains.

[0163] For example, when the maximum load of multiple process time chains is the same, the contribution ratio of each process time chain to the production cycle is determined based on the work-in-process level of the bottleneck station in each process time chain; the production cycle from the first station to the bottleneck station in the current process time chain is determined based on the contribution ratio of each process time chain and the production cycle from the first station in each process time chain to the bottleneck station.

[0164] Specifically, when analyzing a production system, multiple processes covering the entire production flow can be constructed into multiple process time chains according to their temporal sequence.

[0165] When multiple process time chains have their maximum load concentrated in the same process capacity group within the same production cycle, it indicates that the process capacity group constitutes a constrained resource in multiple process time chains.

[0166] In this case, bottleneck stations located in the process capacity group can be further identified in each process time chain, and the current work-in-process level information corresponding to the bottleneck station can be obtained. The work-in-process level is used to represent the number of products that have not yet been processed before the bottleneck station.

[0167] As a non-limiting example, if the maximum load of process time chains Q1, Q2 and Q3 is concentrated in process capacity group G1, and the corresponding bottleneck stations in G1 are M1, M2 and M3 respectively, then the work-in-process levels W1, W2 and W3 at M1, M2 and M3 can be determined respectively.

[0168] Thus, when multiple process time chains share the same maximum load process capacity group, the impact of different process time chains on the overall production cycle is not entirely the same, and this impact can be quantified by the work-in-process level at their respective bottleneck stations.

[0169] In this embodiment, the work-in-process level at the bottleneck station corresponding to each process time chain can be normalized to determine the contribution ratio of each process time chain in the current production cycle.

[0170] Among them, the higher the work-in-process level in a process time chain, the greater its constraint on the production cycle, and the higher its corresponding contribution ratio.

[0171] As a non-limiting example, if the work-in-process levels at the bottleneck stations of process time chains Q1, Q2, and Q3 are 10, 20, and 30 respectively, their contribution ratios can be determined to be 1 / 6, 1 / 3, and 1 / 2 respectively according to the preset calculation rules.

[0172] It should be noted that the contribution ratio can be calculated linearly or adjusted by combining empirical weights or correction coefficients. This embodiment of the invention does not impose specific limitations on this.

[0173] Furthermore, given the known contribution ratio of each process time chain to the overall production cycle, the production cycle of key segments in each process time chain can be dynamically allocated based on the contribution ratio.

[0174] By breaking down the overall production cycle according to the contribution ratio of each process time chain, and combining this with the basic production rhythm of the process time chain itself, the target production cycle from the first station to the bottleneck station in the process time chain can be determined.

[0175] By employing the above methods, when multiple process time chains share the same maximum load process capacity group, it is possible to avoid excessive occupation of bottleneck resources by a single process time chain, thereby achieving load balancing among the process time chains. Alternatively, it can be considered that when the production cycle of each process time chain has been reasonably allocated according to its contribution ratio, the overall production system has reached a stable state, and the space for further optimization is limited. At this point, the determined production cycle can be regarded as a production cycle configuration that approximates the global optimum.

[0176] By adopting the above method, when multiple process time chains compete for the same bottleneck resources, the method comprehensively considers the work-in-process level at the bottleneck station and the actual contribution of each process time chain to the production cycle. Compared with scheduling methods based solely on fixed beats or single priorities, it can allocate the production cycle more rationally, improve the utilization rate of bottleneck resources, and thus enhance the overall production efficiency and stability of the production line.

[0177] Similarly, bottleneck sites also affect time length limits. When the maximum load of the process capacity group is the same when multiple process time chains are determined, the actual time length limit of the bottleneck site can also be determined in the same way as the production cycle.

[0178] In short, if the bottleneck device group is the same across multiple time chains, then it is determined according to the time chain. Together with CT, determine the current event chain based on proportions. And CT scan.

[0179] In some embodiments, based on the determined production scheduling parameters, a water level coefficient for adjusting the amount of work that can be put into operation is further introduced. The value of the water level coefficient is determined by minimizing the scheduling cost model, so that the scheduling cost model can adaptively weigh the importance of different cost factors. This optimizes the overall operating cost while meeting production constraints, avoids the problem of scheduling results deviating from actual production needs due to unreasonable fixed weight settings, and improves the economy, flexibility and applicability of the scheduling scheme.

[0180] In one example, see Figure 5 The flowchart shown in this embodiment of the present disclosure illustrates a method for selecting a water level coefficient, as follows: Figure 5 As shown, it includes: S501, based on each of the candidate water level coefficients, the production cycle, and the output capacity of the process capacity group, determine the allowable shipment quantity under each candidate water level coefficient.

[0181] In some embodiments, the candidate water level coefficient is a parameter used to measure the production load level. Different candidate water level coefficients correspond to different capacity utilization levels. The output capacity of the process capacity group refers to the maximum production quantity that the process capacity group can complete under ideal or limited conditions within the production cycle.

[0182] Accordingly, the allowable output quantity refers to the number of products that the process capacity group is allowed to put into production within the corresponding production cycle under the condition of meeting the candidate water level coefficient constraints. It is used to characterize the production load level that can be carried under the candidate water level coefficient.

[0183] For example, when the candidate water level coefficient is small, it indicates that the load level of the process capacity group is low, and the allowable shipment quantity calculated based on the production cycle and the capacity of the process capacity group is relatively small; when the candidate water level coefficient is large, it indicates that the load level of the process capacity group is high, and the corresponding allowable shipment quantity is relatively large.

[0184] Based on this, the allowable shipment quantity under each candidate water level coefficient can be calculated according to the output capacity, production cycle and the corresponding relationship between the process capacity group and the candidate water level coefficient, thereby providing a basis for subsequent production planning and shipment decisions.

[0185] It should be noted that the candidate water level coefficient is the only variable when determining the allowable discharge volume. That is, different candidate water level coefficients correspond to different allowable discharge volumes.

[0186] In one example, step S501 may include: determining the total production time based on the production cycle and the time length limit corresponding to the bottleneck site; determining the estimated total production volume based on the total production time and the output capacity of the process capacity group; and determining the corresponding allowable shipment quantity based on the estimated total production volume and the candidate water level coefficient.

[0187] Specifically, the production cycle is used to limit the production time from the first station to the bottleneck station. The time limit restricts the maximum dwell time of the product after it leaves the previous station and before it enters the bottleneck station. Together, these reflect the total time corresponding to the bottleneck station. Based on this, and using the known capacity of the process capacity group, the estimated total production volume can be determined, that is, the processing volume that the current production line can expect to complete.

[0188] Furthermore, taking into account changes in production line status, a candidate water level coefficient was introduced based on the estimated total production volume. The candidate water level coefficient is also subject to cost constraints, ensuring that the determined allowable output quantity simultaneously meets both time and cost constraints.

[0189] In one example, the allowed order quantity The expression is:

[0190] in, This represents the candidate water level coefficient (it should be noted that when the candidate water level coefficient meets the selection requirements, the candidate water level coefficient is the current water level coefficient). Indicates a time limit; Indicates the production cycle; This indicates the output capacity of a process capacity group, i.e., the processing capacity per hour.

[0191] S502, based on each candidate water level coefficient and its corresponding production cycle, allowable unloading quantity, and output capacity of the process capacity group, determine the operating prediction parameters of the bottleneck station under each candidate water level coefficient.

[0192] In one embodiment, after determining the allowable delivery quantity, the operating prediction parameters under each candidate water level coefficient are further determined based on the candidate water level coefficient, the production cycle, allowable delivery quantity, and output capacity of the process capacity group corresponding to the candidate water level coefficient. This indicates the operating parameter information of the equipment under the determined production cycle, allowable delivery quantity, and output capacity of the process capacity group, as well as the selected candidate water level coefficient.

[0193] In this way, by determining the operating forecast parameters again based on the determined operating forecast parameters before generating the final allowable quantity, the final scheduling result is generated within the feasible range limited by the operating forecast parameters, thereby avoiding the generation of unexecutable scheduling schemes that do not meet the actual production conditions and improving the executability and stability of the production scheduling results.

[0194] In some embodiments, a multilayer perceptron neural network model is used to perform nonlinear regression on the candidate water level coefficients and their corresponding production cycles, allowable unloading quantities, and output capacity of process capacity groups to obtain the operating prediction parameters under each candidate water level coefficient.

[0195] Specifically, the candidate water level coefficients, their corresponding production cycles, allowable unloading quantities, and output capacity of the process capacity groups are used as inputs. These input parameters are then processed by a multilayer perceptron neural network model to determine the operational prediction parameters for each candidate water level coefficient.

[0196] The multilayer perceptron (MLP) model consists of an input layer, one or more hidden layers, and an output layer.

[0197] In this embodiment, the multilayer perceptron model includes one input layer, two hidden layers, and one output layer, wherein: Input layer: Responsible for receiving raw data. The input layer receives a fixed-length feature vector. The number of neurons is fixed at 4, corresponding to 4 input feature vectors. In this model, the input vector is the input vector. .

[0198] Hidden layers: Located between the input and output layers, they are the core of the model's feature abstraction and transformation. They perform preliminary processing and combination of the input data to discover hidden patterns within it.

[0199] Output layer: Produces the final prediction result. Output dimension: The number of neurons in the output layer corresponds to the number of values ​​to be predicted.

[0200] In this embodiment, the multilayer perceptron model outputs three values, so there are three neurons: Q_time, ... _ and _ .

[0201] It should be noted that the specific structure of the multilayer perceptron model in this scheme can be set according to actual needs. Only one example is given at present, and it should not be used to limit this application.

[0202] Furthermore, the multilayer perceptron neural network model is trained using each actual water level coefficient within multiple historical periods, the actual allowable loading volume corresponding to the actual water level coefficient, the actual production cycle, the output capacity of the actual process capacity group, and the actual operating parameters.

[0203] Specifically, semiconductor manufacturing is a dynamic process that is updated at regular intervals. This allows the acquisition of multiple historical actual water level coefficients and the actual performance of the production line under those actual water level coefficients. This enables the initial multilayer perceptron neural network model to be trained to obtain the corresponding multilayer perceptron neural network model, which is then used to predict the operating parameters corresponding to the candidate water level coefficients in each current state.

[0204] In one example, nonlinear regression is performed using a multilayer perceptron neural network model (MPL) based on data from multiple historical periods to determine the running prediction parameters.

[0205] For example, input vector ,in, Indicates the actual water level coefficient within a historical period. The average production cycle from the latest WIP (Work In Progress) to the target site at the time of the update.

[0206] Accordingly, the output vector (i.e., the running prediction parameters) These are the actual average queuing times of the bottleneck stations within one cycle. The output (i.e., the number of wafers produced) of the process capacity group corresponding to the bottleneck site within one cycle. Bottleneck equipment idle time within one cycle .

[0207] S503, the operation prediction parameters are input into the scheduling cost model to determine the cost value of the scheduling cost model under each candidate water level coefficient.

[0208] In some embodiments, when determining the running forecast parameters, the running forecast parameters can be input into the scheduling cost model to determine the cost value under each candidate water level coefficient.

[0209] In one example, the scheduling cost model may include: a first cost function and a second cost function, the first cost function having a first weight coefficient, the second cost function having a second weight coefficient, and the sum of the first weight coefficient and the second weight coefficient being a constant.

[0210] The first cost function reflects the loss cost when the candidate water level coefficient is in an over-configured state, the second cost function reflects the idle cost when the candidate water level coefficient is in an under-configured state, and the time length limit of the bottleneck station is used as the time constraint information of the first cost function.

[0211] In other words, when determining the cost value, both excessively large and excessively small candidate water level coefficients are considered, making the determined cost value closer to the actual cost and thus more closely reflecting the actual production process. Furthermore, by using the time constraint of the bottleneck site as the time constraint information of the first cost function, the loss cost under over-configuration conditions can be better evaluated, ensuring that a water level coefficient that meets the actual production and manufacturing requirements is selected.

[0212] Furthermore, the sum of the first and second weighting coefficients is a fixed value. This constraint gives the weights a clear meaning of cost structure proportion, effectively avoiding the problems of cost scale distortion and double pricing, and improving the comparability between schemes under different candidate water level coefficients.

[0213] In one embodiment, the first weighting coefficient It could refer to the yield loss weighting coefficient, or the cost per unit of overwork, rework, or scrap loss (unit price / pcs).

[0214] Second weighting coefficient This could refer to the idle capacity weighting coefficient or the cost of capacity loss from bottleneck equipment (unit price / hrs).

[0215] and The sum of is 1.

[0216] In some embodiments, the operational prediction parameters include: the actual average queuing time of the bottleneck site, the output of the process capacity group corresponding to the bottleneck site, and the idle time.

[0217] Accordingly, in response to the actual average queuing time being less than or equal to the time length limit of the bottleneck station, 0 is used as the result of the first cost function.

[0218] Specifically, if the average queuing time is less than or equal to the time limit of the bottleneck station, it indicates that the currently determined candidate water level coefficient can meet the production and manufacturing requirements, and there will be no overload due to an excessively large candidate water level coefficient. The loss is such that the value of the first cost function is 0.

[0219] In response to the actual average queuing time being greater than the time length limit of the bottleneck station, a first cost function is determined based on the average queuing time, wherein each of the candidate water level coefficients has its own average queuing time.

[0220] Specifically, if the actual average queuing time is greater than the time limit of the bottleneck station, it indicates that the currently determined candidate water level coefficient is beyond expectations, and there is a loss due to the excessively large candidate water level coefficient. Therefore, it is necessary to consider the impact of the first cost function on production scheduling.

[0221] In some embodiments, the first cost function is determined in the following manner: According to a preset third weighting coefficient, the deviation of the actual average queuing time from the time length limit is weighted, wherein the deviation is measured in square form; the weighted result is dynamically adjusted according to the ratio between the number of work-in-process exceeding the time length limit and the corresponding total number in multiple historical periods, wherein the dynamic adjustment includes nonlinearly amplifying the ratio according to a preset proportional adjustment coefficient and an exponential adjustment parameter; the first cost function is obtained by combining the weighted result with the nonlinearly amplified adjustment result.

[0222] That is, after determining the impact of the actual average queuing time itself, the first cost function is also modified according to the production scheduling of historical cycles to adapt to the operating status of the production line under different conditions.

[0223] In some embodiments, the expression for the first cost function is:

[0224] in, Represents any of the candidate water level coefficients; This represents the actual average queuing time corresponding to the candidate water level coefficient; This indicates the time limit for the bottleneck site; This represents the third weighting factor, such as the basic unit overtime cost factor; This represents the amplification factor for historical events, where it increases if there are consecutive over-the-top events. That is, to amplify the punishment; This indicates the number of work-in-process items that exceed the time length limit over multiple historical periods (e.g., the number of work-in-process items over Q_time over the last q periods). It represents the total number of work-in-process items over multiple historical periods (e.g., the total number of work-in-process items over the last q periods). This indicates the order of the amplification of historical events.

[0225] It should be noted that, for a set of candidate water level coefficients, the first cost function It is static. First cost function. The dynamics are reflected in the first cost function determined for different groups of candidate water level coefficients. There are differences.

[0226] Accordingly, a second cost function is determined based on the output capacity and idle time of the process capacity group corresponding to each candidate water level coefficient.

[0227] Specifically, based on a preset fourth weighting coefficient, the deviation between the output capacity and the theoretical output capacity of the process capacity group is processed, and the deviation is nonlinearly amplified to obtain the deviation result; wherein, the output capacity of the process capacity group is determined based on the output of the process capacity group; based on the idle time of the machines in the process capacity group and its proportion in the preset production cycle, the deviation result is adjusted to obtain the second cost function.

[0228] The second cost function The expression is: .

[0229] in, This refers to the fourth weighting coefficient, for example: the equipment idle penalty baseline coefficient; This indicates the output capacity of the process capacity group corresponding to the candidate water level coefficient, i.e., the actual output processing capacity (pieces / h). The theoretical output capacity (pieces / hour) of the process capacity group is determined by the characteristics of the equipment itself. This indicates the idle time of the machines in the process capacity group within a preset production cycle; This indicates the preset production cycle; This represents the penalty coefficient, for example: idle duration penalty coefficient.

[0230] In this embodiment, the output capacity of the process capacity group is determined based on the output of the process capacity group, which means: .

[0231] Based on this, the expression for the scheduling cost model is:

[0232] This represents the first weighting coefficient. This represents the second weighting coefficient.

[0233] S504, the candidate water level coefficient corresponding to the minimum cost value is taken as the current water level coefficient, and the allowable delivery quantity corresponding to the candidate water level coefficient is taken as the allowable delivery quantity under the current water level coefficient.

[0234] In some embodiments, after determining the cost value corresponding to all current candidate water level coefficients, all cost values ​​are compared, and the candidate water level coefficient with the smallest cost value is selected as the current water level coefficient.

[0235] In some embodiments, within a preset search space, a particle swarm optimization algorithm is used to initialize multiple candidate water level coefficients, and the candidate water level coefficients obtained from the initial initialization are iteratively updated until a water level coefficient that satisfies the convergence state is determined.

[0236] The preset search space limits the range of values ​​for the candidate water level coefficients.

[0237] More specifically, within the preset search space, multiple initial particles are randomly generated, and an initial position and initial velocity are set for each initial particle; wherein each initial particle represents a candidate water level coefficient, and the initial velocity represents the amount of change of the candidate water level coefficient in adjacent iterations.

[0238] By adopting a water level coefficient selection method based on particle swarm optimization, efficient optimization of scheduling parameters can be achieved in the highly complex and multi-constrained production environment of semiconductor manufacturing. This improves production line throughput while reducing production risks, enhances the scheduling system's adaptability to dynamic disturbances, and improves the stability and reliability of the overall manufacturing process.

[0239] For example, semiconductor manufacturing processes typically include multiple key steps such as photolithography, etching, thin film deposition, ion implantation, and chemical mechanical polishing. These steps have strict sequential dependencies, equipment-specific constraints, and re-entry process characteristics, which makes the production scheduling problem exhibit high-dimensionality, nonlinearity, and strong coupling.

[0240] Particle swarm optimization (PSO) can perform a global search of complex scheduling spaces through a group cooperative search approach. It can effectively handle complex scheduling problems introduced by changes in equipment combinations, process path differences, and dynamic constraints in semiconductor manufacturing, and avoid the problem that traditional rule or linear optimization methods are prone to getting trapped in local optima under complex constraints.

[0241] In some embodiments, the preset search space is selected from [0.5, 2], that is, the value of the water level coefficient is between [0.5, 2]. Here, the interval [0.5, 2] is not a numerical selection, but a technical constraint on the allowable variation range of the original scheduling quantity. Its purpose is to achieve scheduling correction within the executable range of the manufacturing system without destroying the original scheduling structure.

[0242] Specifically, by limiting the water level coefficient to a multiple range of 0.5 to 2 relative to the original scheduling quantity, the adjustment range of the scheduling quantity is constrained. This allows for the effective amplification or compression of the original scheduling quantity to accelerate the scheduling correction process when scheduling deviations are large or production status changes. Furthermore, as scheduling gradually stabilizes, excessive fluctuations in the scheduling quantity are suppressed, ensuring that the scheduling results remain within a range that matches the manufacturing resource execution capacity. This achieves a balance between rapid response and stable execution of production scheduling without disrupting the original scheduling structure.

[0243] Accordingly, an iterative update method is used to determine the cost value for each candidate water level coefficient.

[0244] In one example, the running prediction parameters corresponding to each initial particle are substituted into the scheduling cost model to obtain the cost value of each initial particle, and the individual optimal cost value and corresponding position of each initial particle are recorded; the individual optimal cost values ​​of all initial particles are compared, and the particle with the smallest individual optimal cost value is selected as the global optimal particle, and its corresponding global optimal cost value and global optimal position are recorded; at least one update iteration operation is performed to update the velocity and position of each initial particle, and boundary constraint processing is applied to the updated particle position so that each updated particle is still within the search space, and the global optimal cost value and global optimal position are calculated again.

[0245] Specifically, for all initial particles in the current group, the running prediction parameters corresponding to each initial particle are substituted into the scheduling cost model to obtain the cost value corresponding to each initial particle, thereby determining the global optimal cost value for this operation.

[0246] Furthermore, each initial particle has a corresponding individual historical optimal cost value, thus, based on the current process and the historical processes already performed, the global optimal cost value and the global optimal position can be determined.

[0247] Based on this, each initial particle updates itself according to the initial velocity configured for it, thereby obtaining the initial particle for the next round and recording its corresponding position, while satisfying that the position of each updated initial particle is located in the search space, thereby determining the global optimal cost value and the global optimal position.

[0248] In other words, each substitution operation can determine a globally optimal cost value and the corresponding globally optimal position. Thus, based on any two adjacent iterations, the current water level coefficient can be determined.

[0249] Specifically, the change in cost difference between the two global optimal cost values ​​is calculated in any two adjacent update iterations, and in response to the change in cost difference being less than a preset threshold, the current global optimal position is taken as the current water level coefficient.

[0250] For example, suppose that in two consecutive update iterations, the cost value determined in the first iteration is Cost1, and the cost value determined in the second iteration is Cost2. If the water level is less than the preset threshold, the corresponding current global optimal position will be used as the current water level coefficient.

[0251] It should be noted that when determining multiple candidate water level coefficients in each iteration, it is necessary to calculate the corresponding operational prediction parameters and determine the current water level coefficient based on these operational prediction parameters.

[0252] In one embodiment, the preset threshold is 0.01. This indicates that in two adjacent update iterations, the cost value corresponding to the global optimal position is basically the same, and the cost gradually converges, which can basically meet the constraints of cost and time at the same time.

[0253] In one example, the water level coefficient can be determined in the following ways: Initialize the particle swarm. Within the search space. Y initial particles are randomly generated; each particle represents a candidate water level coefficient. Value, and record its initial position. and speed .

[0254] Determine the runtime prediction parameters for each particle in the particle swarm. The method for determining the runtime prediction parameters can be found in the previous example.

[0255] Assess individual fitness. Input each particle into a bi-objective optimization model: Record the individual optimal fitness value of the particles. and corresponding positions .

[0256] Evaluate the global optimal fitness. Compare the fitness of all particles. Select the minimum value: Record the corresponding optimal position .

[0257] Update particle velocity and position. Simultaneously ensure the updated position remains within the search space.

[0258] Determine the convergence condition. In two consecutive iterations... If the change is less than the threshold of 0.001, the search stops and the optimal solution is output. (Corresponding to the water level coefficient obtained from the last convergence), which is the current global optimal position. As the target water level coefficient output, the allowable cargo volume corresponding to the target water level coefficient can be determined.

[0259] In this way, by automatically and in real-time calculating the quantity of work-in-process (WIP) available for each dummy station in the time chain based on equipment status, WIP level, time limit length, and production cycle, the system can control WIP delivery in real-time when there are downtime processes within the time chain. In non-downtime situations, it dynamically identifies bottleneck equipment based on equipment downtime within the time chain. The dummy delivery quantity is determined based on the objectives of maximizing bottleneck equipment capacity and minimizing over-Q-time losses, providing a basis for the real-time dispatch system. This significantly reduces the risk of products exceeding time limits and avoids bottleneck equipment idleness caused by low WIP levels within the time chain control processes. After adopting a semiconductor production scheduling system with multi-process time chain control, the proportion of products exceeding time chain limits can be effectively improved.

[0260] See Figure 6 The over-ratio trend chart of a work-in-process product shown in this embodiment of the present disclosure is as follows: Figure 6 As shown, using the production scheduling scheme provided in this embodiment, during the period from Week 1 to Week 16, driven by the increase in the number of work-in-process (CNT), the over-time ratio and the over double ratio gradually decrease.

[0261] Furthermore, focusing on the monthly period (i.e., M1 to M4), the decreasing trend of the overdue rate and the double overdue rate is even more obvious, which indicates that this production scheduling plan can reduce the risk of products exceeding the time limit.

[0262] It is understood that the above embodiments provide multiple implementation schemes, and these implementation schemes can be combined and cross-referenced with each other without conflict, thereby extending to multiple possible implementation schemes. These can all be considered as the implementation schemes disclosed and made public in this application.

[0263] This disclosure also provides an apparatus corresponding to the above-described production scheduling method, which will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0264] See Figure 7 The schematic diagram shown below illustrates the structure of a production scheduling device in an embodiment of this disclosure, as follows: Figure 7 As shown, the production scheduling device 700 may include: The first determining unit 710 is configured to determine the load of all process capacity groups within the current process time chain, and to designate the station corresponding to the process capacity group with the largest load as the bottleneck station; the current process time chain has multiple stations, each of which has a corresponding time length limit, and the bottleneck station is one of the multiple stations. The second determining unit 720 is configured to determine the production cycle of all stations located before the bottleneck station among the plurality of stations based on historical job information; The processing unit 730 is configured to initialize multiple candidate water level coefficients, and based on a pre-established scheduling cost model related to the water level coefficients, as well as the production cycle, the output capacity of the process capacity group, and the time length limit, determine the current water level coefficient when the cost value of the scheduling cost model is minimized, and the allowable unloading quantity under the current water level coefficient. The scheduling unit 740 is configured to determine the amount of work that can be put into operation from the first station based on the allowed unloading volume and the current total amount of work at the bottleneck station and the stations before the bottleneck station in the current time chain.

[0265] about Figure 7 For more information on the working principle, operation method, and beneficial effects of the production scheduling device shown, please refer to the preceding text and... Figures 1 to 6 The specific details will not be repeated here.

[0266] It is understandable that the above division of units is only a logical functional division; in actual implementation, they can be fully or partially integrated into a single physical entity, or they can be physically separated. Furthermore, the above modules can be implemented by the processor calling software.

[0267] This disclosure also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the steps of the production scheduling method described in any of the foregoing embodiments when running the computer program.

[0268] See Figure 8 The diagram illustrates an optional hardware structure of an electronic device provided in an embodiment of this disclosure.

[0269] The electronic device in this embodiment includes at least one processor 81, at least one communication interface 82, at least one memory 83, and at least one communication bus 84.

[0270] In some embodiments, the number of processor 81, communication interface 82, memory 83 and communication bus 84 is at least one, and processor 81, communication interface 82 and memory 83 communicate with each other through communication bus 84.

[0271] Communication interface 82 can be an interface for a communication module used for network communication, such as an interface for a GSM module.

[0272] The processor 81 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the defect detection method of this embodiment.

[0273] The memory 83 may include high-speed RAM memory, and may also include non-volatile memory, such as at least one disk storage.

[0274] The memory 83 stores one or more computer instructions, which are executed by the processor 81 to implement the steps of the aforementioned production scheduling method.

[0275] It should be noted that the above-mentioned electronic device may also include other devices (not shown) that may not be essential to the content of this application; given that these other devices may not be essential for understanding the application content of the embodiments of this invention, this invention will not describe them one by one.

[0276] This disclosure also provides a computer-readable storage medium including instructions that, when the computer-readable storage medium is used by a receiver, cause the receiver to perform the steps of the production scheduling method as described in any of the foregoing embodiments.

[0277] This disclosure also provides a computer program product, including a computer program / instructions, which, when executed by a processor, are used to implement the production scheduling method as described in any of the foregoing embodiments.

[0278] The storage medium may include read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk, etc. The storage medium may also include non-volatile memory or non-transitory memory, etc.

[0279] It should be understood that in the embodiments of this disclosure, the processor can be a central processing unit (CPU), or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0280] It should also be understood that the memory in the embodiments of this disclosure may be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. Non-volatile memory may be ROM, programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DRRAM).

[0281] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer program can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another. For example, the computer program can be transferred from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means.

[0282] While the embodiments disclosed herein are as described above, this disclosure is not limited thereto. Any person skilled in the art can make various alterations and modifications without departing from the spirit and scope of this disclosure; therefore, the scope of protection of this disclosure should be determined by the scope defined in the claims.

Claims

1. A production scheduling method, characterized in that, include: Determine the load of all process capacity groups within the current process time chain, and designate the station corresponding to the process capacity group with the highest load as the bottleneck station; the current process time chain has multiple stations, each of which has a corresponding time length limit, and the bottleneck station is one of the multiple stations. Based on historical operation information, determine the production cycle of all stations located before the bottleneck station among the multiple stations; Multiple candidate water level coefficients are initialized, and based on a pre-established scheduling cost model related to the water level coefficients, as well as the production cycle, the output capacity of the process capacity group, and the time length limit, the current water level coefficient with the lowest cost value of the scheduling cost model and the allowable unloading quantity under the current water level coefficient are determined. Based on the allowed unloading volume and the current total workload of the bottleneck station and the stations before the bottleneck station in the current time chain, the available workload from the first station is determined.

2. The production scheduling method according to claim 1, characterized in that, The step of determining the load of all process capacity groups within the current time chain and designating the site corresponding to the process capacity group with the highest load as the bottleneck site includes: Determine the work-in-process level at each station corresponding to each process capacity group, and the real-time estimated output of available equipment within the process capacity group; Based on the work-in-process water level and the real-time estimated output, the load of each process capacity group is determined, and the station corresponding to the process capacity group with the maximum load is designated as the bottleneck station.

3. The production scheduling method according to claim 2, characterized in that, Meet one or more of the following conditions: Determining the real-time projected output of available equipment within each of the process capacity groups includes: acquiring the real-time status of each available equipment, as well as the utilization rate and hourly output of each available equipment; and determining the real-time projected output based on the real-time status of the available equipment, as well as the utilization rate and hourly output of each available equipment. When determining the real-time estimated output, the real-time status of the available equipment is converted into equipment availability data according to a preset status determination rule. The equipment availability data is used to characterize whether the equipment is available. The equipment availability data includes at least a first value indicating that the equipment is available and a second value indicating that the equipment is unavailable. When determining the load for each of the process capacity groups, a normalization process is also performed, and the maximum load is determined based on the normalized load. When there are multiple sites corresponding to the maximum load, the site with the smallest position relative to the first site is determined as the bottleneck site based on the site order in the current time chain. The expression for determining the load of each of the aforementioned process capacity groups is as follows: Bottleneck site in, Indicates the water level of the product, Device availability data indicating the real-time status of the available devices. This indicates the hourly output of the available equipment. This indicates the utilization rate of the available equipment. Indicates the output capacity of the process capacity group; This indicates the real-time estimated output.

4. The production scheduling method according to claim 1, characterized in that, The historical operation information includes: product batches processed by all stations within the current process time chain within a preset time period, and the running time status information of each station under each product batch; The step of determining the production cycle of all stations preceding the bottleneck station among the multiple stations based on historical operation information includes: Based on the running time status information of each station under each product batch, determine the average production cycle of each station in the current process time chain; The average production cycle of all stations preceding the bottleneck station is summed up to obtain the production cycle.

5. The production scheduling method according to claim 4, characterized in that, The runtime status information includes: runtime, queuing time, and suspension time; The step of determining the average production cycle of each station within the current process time chain based on the running time status information of each station under each product batch includes: Based on the running time of each site under all product batches, determine the first interval and the corresponding average running time; select the first running time that meets the first interval and the second running time that does not meet the first interval from all running times, and use the average running time to replace all the second running times; determine the average running time based on the first running time, all the average running times, and the number of all product batches. Based on the queuing time of each station under all product batches, determine the second interval corresponding to the queuing time and the corresponding average queuing time; select the first queuing time that meets the second interval and the second queuing time that does not meet the second interval from all queuing times, and use the average queuing time to replace all the second queuing times; determine the average queuing time based on the first queuing time, all the average queuing times, and the number of all product batches. Based on the suspension time of each site under all product batches, determine the third interval corresponding to the suspension time and the corresponding suspension average value; select the first suspension time that meets the third interval and the second suspension time that does not meet the third interval from all suspension times, and use the suspension average value to replace all the second suspension times; determine the average suspension time based on the first suspension time, all the suspension average values ​​and the number of all product batches; The sum of the average running time, the average queuing time, and the average hang-up time is taken as the average production cycle of the corresponding station within the current process time chain.

6. The production scheduling method according to claim 5, characterized in that, Meet one or more of the following conditions: The step of determining the first interval based on the running time of each site under all product batches includes: sorting all the running times to obtain a first sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the first sequence respectively; and determining the first interval based on the first quartile and the third quartile corresponding to the first sequence. The step of determining the second interval based on the queuing time of each station under all product batches includes: sorting all queuing times to obtain a second sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the second sequence respectively; and determining the second interval based on the first quartile and the third quartile corresponding to the second sequence. The step of determining the third interval based on the suspension time of each site under all product batches includes: sorting all suspension times to obtain a third sequence arranged in ascending order; calculating the first quartile and the third quartile corresponding to the third sequence respectively; and determining the third interval based on the first quartile and the third quartile corresponding to the third sequence.

7. The production scheduling method according to claim 1 or 4, characterized in that, When the maximum load of multiple process time chains is the same, the contribution ratio of each process time chain to the production cycle is determined based on the work-in-process level of the bottleneck station in each process time chain; the production cycle from the first station to the bottleneck station in the current process time chain is determined based on the contribution ratio of each process time chain and the production cycle from the first station in each process time chain to the bottleneck station.

8. The production scheduling method according to claim 1, characterized in that, The process of initializing multiple candidate water level coefficients and determining the current water level coefficient when the cost value of the scheduling cost model is minimized, and the allowable shipment quantity under the current water level coefficient, based on a pre-established scheduling cost model related to the water level coefficient, the production cycle, and the output capacity of the process capacity group, includes: Based on each candidate water level coefficient, as well as the production cycle and the output capacity of the process capacity group, determine the allowable shipment quantity under each candidate water level coefficient; Based on each candidate water level coefficient and its corresponding production cycle, allowable unloading volume, and output capacity of the process capacity group, determine the operational prediction parameters of the bottleneck site under each candidate water level coefficient; The operation prediction parameters are input into the scheduling cost model to determine the cost value of the scheduling cost model under each candidate water level coefficient. The candidate water level coefficient corresponding to the minimum cost value is taken as the current water level coefficient, and the allowable shipment quantity corresponding to the candidate water level coefficient is taken as the allowable shipment quantity under the current water level coefficient.

9. The production scheduling method according to claim 8, characterized in that, The step of determining the allowable shipment quantity under each candidate water level coefficient based on each candidate water level coefficient, the production cycle, the output capacity of the process capacity group, and the time length limit includes: The total production time is determined based on the production cycle and the time limit corresponding to the bottleneck site. The estimated total production volume is determined based on the total production time and the output capacity of the process capacity group; Based on the estimated total production volume and the candidate water level coefficient, the corresponding allowable shipment volume is determined.

10. The production scheduling method according to claim 8, characterized in that, The step of determining the operational prediction parameters of the bottleneck station under each candidate water level coefficient, based on each candidate water level coefficient and its corresponding production cycle, allowable output, and output capacity of the process capacity group, includes: A multilayer perceptron neural network model is used to perform nonlinear regression on the candidate water level coefficients and their corresponding production cycles, allowable unloading quantities, and output capacity of process capacity groups to obtain the operating prediction parameters under each candidate water level coefficient. The multilayer perceptron neural network model is trained using each actual water level coefficient within multiple historical periods, the actual allowable loading volume, actual production cycle, output capacity of the actual process capacity group, and actual operating parameters corresponding to the actual water level coefficient.

11. The production scheduling method according to claim 8, characterized in that, Within a preset search space, a particle swarm optimization algorithm is used to initialize multiple candidate water level coefficients, including: randomly generating multiple initial particles within the preset search space, and setting an initial position and initial velocity for each initial particle; wherein each initial particle represents a candidate water level coefficient, and the initial velocity represents the change of the candidate water level coefficient in adjacent iterations; The step of inputting the operational prediction parameters into the scheduling cost model to determine the cost value of the scheduling cost model under each candidate water level coefficient includes: Substitute the running prediction parameters corresponding to each initial particle into the scheduling cost model to obtain the cost value of each initial particle, and record the individual optimal cost value and corresponding position of each initial particle; Compare the individual optimal cost values ​​of all the initial particles, select the particle with the smallest individual optimal cost value as the global optimal particle, and record its corresponding global optimal cost value and global optimal position. Perform at least one update iteration operation to update the velocity and position of each initial particle, and apply boundary constraints to the updated particle positions so that each updated particle is still within the search space, and recalculate the global optimal cost value and the global optimal position. The step of using the candidate water level coefficient corresponding to the minimum cost value as the current water level coefficient includes: Calculate the change in cost difference between the two global optimal cost values ​​in any two adjacent update iterations, and in response to the change in cost difference being less than a preset threshold, use the current global optimal position as the current water level coefficient.

12. The production scheduling method according to claim 11, characterized in that, Meet one or more of the following conditions: The preset search space is selected from [0.5, 2]; The preset threshold is 0.

01.

13. The production scheduling method according to claim 8, characterized in that, The scheduling cost model includes: a first cost function and a second cost function, wherein the first cost function has a first weight coefficient, the second cost function has a second weight coefficient, and the sum of the first weight coefficient and the second weight coefficient is a constant. Wherein, the first cost function reflects the loss cost when the candidate water level coefficient is in an over-configured state, the second cost function reflects the idle cost when the candidate water level coefficient is in an under-configured state, and the time length limit of the bottleneck station serves as the time constraint information of the first cost function.

14. The production scheduling method according to claim 13, characterized in that, The operational prediction parameters include: the actual average queuing time of the bottleneck site, the output of the process capacity group corresponding to the bottleneck site, and the idle time. In response to the actual average queuing time being less than or equal to the time length limit of the bottleneck site, 0 is used as the result of the first cost function; In response to the actual average queuing time exceeding the time limit of the bottleneck station, a first cost function is determined based on the actual average queuing time. Each candidate water level coefficient has its own actual average queuing time. The process includes: weighting the deviation of the actual average queuing time from the time limit according to a preset third weighting coefficient, where the deviation is measured in squared form; dynamically adjusting the weighted result based on the ratio between the number of work-in-process exceeding the time limit and the corresponding total quantity within multiple historical periods, wherein the dynamic adjustment includes nonlinearly amplifying the ratio according to a preset proportional adjustment coefficient and an exponential adjustment parameter; and combining the weighted result with the nonlinearly amplified adjustment result to obtain the first cost function.

15. The production scheduling method according to claim 14, characterized in that, Based on the output capacity and idle time of the process capacity group corresponding to each of the candidate water level coefficients, a second cost function is determined, including: Based on a preset fourth weighting coefficient, the deviation between the output capacity and the theoretical output capacity of the process capacity group is processed, and the deviation is nonlinearly amplified to obtain the deviation result; wherein, the output capacity of the process capacity group is determined based on the output of the process capacity group. Based on the idle time of the machines in the process capacity group and its proportion in the preset production cycle, the deviation result is adjusted to obtain the second cost function.

16. The production scheduling method according to claim 15, characterized in that, The expression for the first cost function is: The second cost function The expression is: The expression for the scheduling cost model is: in, Represents any of the candidate water level coefficients; This represents the actual average queuing time at the bottleneck station corresponding to the candidate water level coefficient. This indicates the time limit for the bottleneck site; Indicates the third weighting coefficient; Indicates the amplification factor for historical events; This indicates the number of work-in-process items that exceed the time length limit over multiple historical periods; This indicates the total quantity of work-in-process over multiple historical periods; Indicates the order of the amplification factor for historical events; This represents the fourth weighting coefficient; This indicates the output capacity of the process capacity group corresponding to the candidate water level coefficient; This indicates the theoretical output capacity of the process capacity group; This indicates the idle time of the machines in the process capacity group within a preset production cycle; This indicates the preset production cycle; Indicates the penalty coefficient; This represents the first weighting coefficient. This represents the second weighting coefficient.

17. The production scheduling method according to claim 1, characterized in that, The step of determining the available workload from the first station based on the allowed unloading volume and the current total workload of the bottleneck station and the stations preceding the bottleneck station in the current time chain includes: In response to the allowable delivery quantity being greater than the current total operation quantity, the difference between the allowable delivery quantity and the current total operation quantity is taken as the available operation quantity; In response to the fact that the allowed unloading quantity is less than the current total operation quantity, the available operation quantity is determined to be 0.

18. A production scheduling device, characterized in that, include: The first determining unit is configured to determine the load of all process capacity groups within the current process time chain, and to designate the station corresponding to the process capacity group with the largest load as the bottleneck station; the current process time chain has multiple stations, each of which has a corresponding time length limit, and the bottleneck station is one of the multiple stations. The second determining unit is configured to determine the production cycle of all stations located before the bottleneck station among the plurality of stations based on historical operation information; The processing unit is configured to initialize multiple candidate water level coefficients, and based on a pre-established scheduling cost model related to the water level coefficients, as well as the production cycle, the output capacity of the process capacity group, and the time length limit, determine the current water level coefficient when the cost value of the scheduling cost model is minimized, and the allowable unloading quantity under the current water level coefficient. The scheduling unit is configured to determine the amount of work that can be put into operation from the first station based on the allowed unloading volume and the current total workload of the bottleneck station and the stations before the bottleneck station in the current time chain.

19. An electronic device, characterized in that, The method includes a memory and a processor, wherein the memory stores a computer program that can run on the processor, characterized in that the processor executes the steps of the production scheduling method according to any one of claims 1 to 17 when running the computer program.

20. A computer program product comprising a computer program / instructions, characterized in that, When the computer program / instructions are executed by the processor, they are used to implement the production scheduling method as described in any one of claims 1 to 17; And / or, A storage medium, characterized in that, The storage medium stores one or more computer instructions, which are used to implement the production scheduling method as described in any one of claims 1 to 17.

Images