A cross-regional metering production base collaborative scheduling method, device and computer program product

Abstract

The application discloses a cross-region measurement production base collaborative scheduling method and device and a computer program product, wherein the method comprises the following steps: acquiring multi-base equipment heterogeneous data, extracting a first bearing capacity feature representing the production resource bearing capacity of each base through a machine learning model; decomposing the original order according to the process, acquiring the emergency data when the resource demand exceeds the available amount, determining the dynamic production capacity attenuation of each type of emergency event by using a classification algorithm, and updating the second bearing capacity feature; acquiring upstream quality perception sequence data, analyzing the upstream deviation prediction value by using a time series prediction model, and determining the downstream rework time according to the comparison between the deviation and the threshold; combining the second bearing capacity feature and the rework time, and performing load balancing and re-distribution on the initial order decomposition data to generate collaborative scheduling instructions of each base. Through dynamic correction of production capacity evaluation and pre-intervention of quality risk, the scheduling accuracy and resource utilization efficiency of multi-base collaborative manufacturing are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of metrology production line scheduling technology, specifically to a method, device, and computer program product for collaborative scheduling of cross-regional metrology production bases. Background Technology

[0002] The ongoing digital transformation of the manufacturing industry is driving the evolution of cross-regional metrology and production bases from decentralized, independent operations to a multi-base collaborative operation model. Multi-base collaboration has become a key path to enhance the overall competitiveness of enterprises. With increasing market demand volatility and product customization, the inherent limitations of a single production base in terms of equipment capacity, process coverage, and production capacity flexibility are becoming increasingly apparent. This makes it difficult to independently cope with complex and ever-changing order demands. Enterprises urgently need to optimize the allocation of manufacturing resources and improve overall production flexibility through multi-base collaboration. However, traditional collaborative manufacturing methods mainly rely on manual experience and fixed rules for cross-base task allocation and schedule coordination. This approach is essentially a static open-loop scheduling paradigm, which is ill-suited to the dynamic changes in key elements such as production capacity, equipment status, and material supply among different bases. This results in significant problems of low collaboration efficiency and resource waste.

[0003] The core challenge currently facing cross-base collaborative scheduling stems primarily from the significant heterogeneity among production bases in terms of equipment capacity, process routes, and quality standards. This heterogeneity makes it difficult to accurately quantify and match the actual carrying capacity of each base when orders are broken down by process across multiple bases, often resulting in initial allocation plans deviating from the true available capacity of each base. Further complicating matters, unforeseen events such as equipment failures, material shortages, and personnel changes during production execution can alter the available capacity of each base in real time, and the transmission and response to these dynamic changes across bases exhibit significant lags. For example, if a critical processing equipment failure at the East China base causes a sudden decrease in the base's effective capacity, urgent orders already allocated to that base will be unable to be completed on schedule. Meanwhile, other bases, failing to obtain this capacity change information in a timely manner, will continue to execute non-urgent orders according to their original plans, leading to overall delivery delays and decreased customer satisfaction. This imbalance between the real-time adaptation of dynamic capacity changes caused by unforeseen events and order demand further induces a loss of control over cross-base production progress.

[0004] The essence of the loss of control over cross-base production progress coordination lies in the lack of a real-time perception and linkage mechanism for the completion status and quality information of upstream and downstream processes when different bases undertake different processes of the same order. When the semi-finished products produced by the upstream base have quality deviations or delivery delays, the downstream base cannot obtain deviation information in advance and proactively adjust its own production plan. It can only passively respond after the downstream process starts, taking measures such as stopping work and waiting or centralized rework, which significantly extends the cross-base production cycle of the entire order. In a real-world scenario, a certain model of electricity meter product requires the precision machining of core metering components to be completed by the North China base, and then the overall assembly and performance testing to be carried out by the South China base. If the dimensional accuracy indicators of the components at the North China base exceed the standard deviation, the South China base can only discover the problem after receiving the unqualified materials and putting them online for testing. At this time, reorganizing processing and cross-regional logistics have already caused the cumulative delivery time to exceed the delivery window agreed upon by the customer, resulting in a significant increase in the risk of order default. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method, device and computer program product for collaborative scheduling of cross-regional metrological production bases, so as to improve the scheduling accuracy, resource utilization efficiency and on-time order delivery capability of multi-base collaborative manufacturing.

[0006] To address the aforementioned technical problems, this invention provides a method for collaborative scheduling of cross-regional metrology production bases, comprising: Step S1: Obtain heterogeneous equipment data from multiple bases, extract features from the heterogeneous equipment data, and obtain a first carrying capacity feature characterizing the production resource carrying capacity of each base. Step S2: Obtain initial order decomposition data formed by decomposing the original production order according to the process dimension. In response to the resource demand corresponding to the initial order decomposition data being greater than the resource availability corresponding to the first carrying capacity feature, obtain data on emergencies during the production execution process and determine the dynamic capacity reduction corresponding to various emergencies. Step S3: Update the first carrying capacity feature based on the dynamic capacity reduction to obtain the second carrying capacity feature; Step S4: Obtain the quality perception sequence data of the upstream process, perform time series feature analysis on the quality perception sequence data using a time series prediction model to obtain the upstream deviation prediction value, and determine the downstream rework time based on the comparison result of the upstream deviation prediction value and the preset quality deviation threshold. Step S5: Based on the second carrying capacity characteristic and the downstream rework time, the initial order decomposition data is redistributed to obtain the target order decomposition data; Step S6: Generate collaborative scheduling instructions for the multiple bases based on the target order decomposition data.

[0007] Preferably, in step S1, feature extraction is performed on the heterogeneous data of the equipment to obtain a first carrying capacity feature characterizing the production resource carrying capacity of each base, specifically including: The first carrying capacity feature is obtained by performing feature extraction processing on the heterogeneous data of the device through a machine learning model. The heterogeneous equipment data includes information on differences in equipment types and differences in equipment operating status between different bases.

[0008] Preferably, step S2 specifically includes: Extract the production resource carrying capacity values ​​of each base from the first carrying capacity feature to determine the resource availability of each base; Obtain the original production order, decompose the original production order according to the process dimension to form a sub-task data set, and extract the resource requirements corresponding to the sub-task data set; The resource demand is compared with the resource availability, and a first capacity warning is generated when the resource demand is greater than the resource availability. The emergency event data is obtained based on the first production capacity early warning information.

[0009] Preferably, obtaining the emergency event data based on the first production capacity early warning information specifically includes: Analyze the warning time node corresponding to the first capacity warning information; Retrieve real-time monitoring records during the production process based on the aforementioned warning time points; Extract abnormal equipment operation records and abnormal material flow records from the real-time monitoring records; Based on the equipment operation abnormality record, generate equipment failure event information; based on the material flow abnormality record, generate material supply interruption event information. The equipment failure event information and the material supply interruption event information are combined and processed to obtain the emergency event data.

[0010] Preferably, in step S2, determining the dynamic capacity reduction corresponding to various emergencies specifically includes: The emergency event data is input into a pre-trained classification algorithm model, and the event feature vector in the emergency event data is extracted through the classification algorithm model. The emergency event data is categorized based on the event feature vectors to obtain emergency event categories. For each type of emergency, retrieve the corresponding historical capacity reduction records, and calculate the dynamic capacity reduction amount corresponding to that type of emergency based on the historical capacity reduction records.

[0011] Preferably, step S3 specifically includes: By adopting a resource capacity reduction method, the cumulative value of the product of dynamic capacity reduction and corresponding impact coefficient corresponding to various emergencies is subtracted from the value of the first carrying capacity characteristic to obtain the second carrying capacity characteristic; These various emergencies include equipment failure events and material supply interruption events.

[0012] Preferably, in step S4, the time-series prediction model is used to perform time-series feature analysis on the quality-sensing sequence data to obtain the upstream deviation prediction value, specifically including: The quality-sensing sequence data are arranged in chronological order to construct a time-series data set; The time series data set is input into the time series prediction model, and the time series prediction model is used to extract the time series fluctuation characteristics in the time series data set. Calculate the predicted values ​​of the quality inspection indicators of the upstream process at future time nodes based on the aforementioned time-series fluctuation characteristics; The upstream deviation prediction value is calculated based on the difference between the predicted value of the quality inspection index and the standard quality index.

[0013] Preferably, step S6 specifically includes: Extract the order quantity and decomposition table from the target order decomposition data, and classify the order quantity and decomposition table using a classification algorithm to obtain the first task item set; Based on the first task item set, obtain the base number and capacity value of each production base, filter the base numbers whose capacity values ​​are greater than a preset threshold, and determine the target base number set; For the set of target base numbers, the load rate corresponding to each target base number is obtained, and the load rate is clustered using a clustering algorithm to obtain the first allocation order; Extract the priority and material list from the first allocation order; when the priority meets a preset condition, parse the material list to obtain the second allocation order. Obtain the execution period corresponding to the second allocation order, calculate the degree of coordination between each production base based on the execution period, and generate a coordinated scheduling instruction for the multiple bases based on the degree of coordination.

[0014] The present invention also provides a cross-regional metering production base collaborative scheduling device, comprising: One or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to execute the cross-regional metering production base collaborative scheduling method.

[0015] The present invention also provides a computer program product, including computer instructions, which instruct computer equipment to perform operations corresponding to the cross-regional metering production base collaborative scheduling method.

[0016] The beneficial effects of this invention are as follows: This invention extracts features from the differences in equipment type and operating status across different bases using a machine learning model, constructing a first carrying capacity feature that reflects the actual production resource carrying capacity. This effectively solves the problem of distorted capacity assessment caused by neglecting equipment heterogeneity in traditional methods, providing an accurate decision-making benchmark for cross-base order allocation. When the resource demand after order decomposition exceeds the available capacity, a capacity warning is automatically triggered. A classification algorithm is used to classify and quantify capacity decay in sudden events such as equipment failure and material supply interruptions, dynamically updating the second carrying capacity feature in a decreasing resource capacity manner. This allows the system to perceive the instantaneous changes in available capacity at each base in real time, avoiding allocation failures and delivery delays caused by information lag. This invention also analyzes the quality perception sequence data of upstream processes using a time-series prediction network, predicting the magnitude of quality deviations before they actually occur, and calculating the additional rework hours that downstream bases need to reserve. This achieves early exposure and proactive intervention of quality risks, changing the passive situation of downstream units passively waiting and temporarily stopping work for rework in the traditional model. This invention combines updated resource availability and rework hours to load balance and redistribute orders, enabling task scheduling to simultaneously consider capacity constraints and quality compensation. The generated collaborative scheduling instructions achieve precise process-level coordination between different bases. Overall, this invention achieves a dual upgrade in cross-regional, multi-base capacity assessment from static to dynamic, and in quality management from passive response to proactive prediction, significantly improving the scheduling accuracy, resource utilization efficiency, and on-time order delivery capabilities of multi-base collaborative manufacturing. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating a cross-regional metering production base collaborative scheduling method according to Embodiment 1 of the present invention.

[0019] Figure 2 This is a schematic diagram of the process for obtaining the first bearing capacity feature in an embodiment of the present invention.

[0020] Figure 3 This is a schematic diagram of the process for dynamically updating the carrying capacity in an embodiment of the present invention. Detailed Implementation

[0021] The following description of the embodiments is taken with reference to the accompanying drawings, which illustrate specific embodiments in which the invention can be implemented.

[0022] Please refer to Figure 1 As shown, Embodiment 1 of the present invention provides a method for collaborative scheduling of cross-regional metering production bases, including: Step S1: Obtain heterogeneous equipment data from multiple bases, extract features from the heterogeneous equipment data, and obtain a first carrying capacity feature characterizing the production resource carrying capacity of each base. Step S2: Obtain initial order decomposition data formed by decomposing the original production order according to the process dimension. In response to the resource demand corresponding to the initial order decomposition data being greater than the resource availability corresponding to the first carrying capacity feature, obtain data on emergencies during the production execution process and determine the dynamic capacity reduction corresponding to various emergencies. Step S3: Update the first carrying capacity feature based on the dynamic capacity reduction to obtain the second carrying capacity feature; Step S4: Obtain the quality perception sequence data of the upstream process, perform time series feature analysis on the quality perception sequence data using a time series prediction model to obtain the upstream deviation prediction value, and determine the downstream rework time based on the comparison result of the upstream deviation prediction value and the preset quality deviation threshold. Step S5: Based on the second carrying capacity characteristic and the downstream rework time, the initial order decomposition data is redistributed to obtain the target order decomposition data; Step S6: Generate collaborative scheduling instructions for the multiple bases based on the target order decomposition data.

[0023] The cross-regional metering production base collaborative scheduling method provided in this invention is applied to a production scheduling system in a cross-regional multi-base collaborative manufacturing scenario. This scenario includes multiple metering equipment production bases distributed across different geographical regions, such as the North China base, the East China base, and the South China base. Each base exhibits significant differences in equipment type, process routes, and quality standards, and unexpected events such as equipment failures and material supply interruptions frequently occur during production, resulting in dynamic changes in the actual available capacity of each base. Furthermore, the metering products (such as smart meters and water meters) produced collaboratively across bases require close coordination between upstream and downstream processes; the processing quality of upstream processes directly affects the rework volume and overall delivery cycle of downstream processes.

[0024] The specific implementation methods for each step are described in detail below.

[0025] Step S1 is used to obtain carrying capacity characteristics, that is, to obtain heterogeneous equipment data from multiple bases, extract features from the heterogeneous equipment data, and obtain the first carrying capacity characteristics that characterize the production resource carrying capacity of each base.

[0026] Please combine Figure 2 As shown, in this embodiment, the heterogeneous equipment data includes information on differences in equipment types and differences in equipment operating status between different bases. For example, equipment type differences might include a base equipped with a high-precision five-axis machining center, while another base only has ordinary CNC machine tools. Equipment operating status differences might include a base where critical equipment is operating at full capacity, while similar equipment at another base is idle or operating at half capacity.

[0027] In practical implementation, machine learning models can be used to extract features from the heterogeneous equipment data. Taking into account factors such as differences in equipment type, operating status, and resource availability at each base, a first carrying capacity feature value is calculated, which quantitatively characterizes the current production resource carrying capacity of each base. This carrying capacity reflects the actual ability of each base to accept new orders, considering equipment heterogeneity.

[0028] As a preferred embodiment of this invention, the first load-bearing capacity characteristic can be calculated using the following formula:

[0029] in, Indicates the first load-bearing capacity characteristic. Represents a machine learning model function. This indicates heterogeneous data from the device. Indicates the total number of bases. Indicates the first Information on differences in equipment types at each base Indicates the first Information on differences in equipment operating status at each base Indicates the first Number of devices at each base Indicates the first The amount of resources available at each base. and This represents the weighting coefficient. The formula calculates the production resource carrying capacity of each base by extracting features from heterogeneous equipment data from multiple sites and comprehensively considering differences in equipment type and operating status.

[0030] Step S2 is used for capacity early warning and emergency response.

[0031] The process involves obtaining initial order decomposition data from original production orders, broken down by process dimension. Specifically, firstly, the production resource carrying capacity values ​​of each base are extracted from the first carrying capacity feature, and the resource availability of each base is determined based on these values. Then, the original production orders are obtained and decomposed according to process dimension. The decomposed data is combined to form a sub-task data set, which is designated as the initial order decomposition data to be processed. Finally, the resource requirements corresponding to the initial order decomposition data are extracted.

[0032] In response to the resource demand corresponding to the initial order decomposition data being greater than the resource availability corresponding to the first carrying capacity feature, data on sudden events during the production execution process are acquired.

[0033] During implementation, the scheduling system compares the resources required for each subtask in the initial order breakdown data with the current resource availability at each base. When it is found that the resource demand of one or more bases exceeds their availability, a first capacity warning is generated, and real-time contingency data is further acquired to dynamically revise the assessment of the capacity of each base.

[0034] The process of obtaining emergency event data based on the first production capacity warning information is as follows: receiving the first production capacity warning information and parsing the warning time node corresponding to the first production capacity warning information; retrieving real-time monitoring records during the production execution process based on the warning time node; extracting equipment operation abnormality records and material flow abnormality records from the real-time monitoring records; generating equipment failure event information based on the equipment operation abnormality records and generating material supply interruption event information based on the material flow abnormality records; and combining the equipment failure event information and the material supply interruption event information to obtain the emergency event data.

[0035] After acquiring emergency event data, the dynamic capacity reduction corresponding to each type of emergency event is determined. Specifically, the emergency event data is input into a pre-trained classification algorithm model, which extracts event feature vectors from the emergency event data. The emergency event data is then categorized based on the event feature vectors to obtain emergency event categories. For each emergency event category, corresponding historical capacity reduction records are retrieved. The dynamic capacity reduction corresponding to that emergency event category is calculated based on the historical capacity reduction records.

[0036] In practical applications, classification algorithms can be modeled using support vector machines, decision trees, or gradient boosting trees. For example, if a critical stamping machine at a certain production base suddenly fails and shuts down, the classification algorithm identifies it as a "critical equipment failure" based on historical failure records. It then retrieves the historical average capacity reduction rate for this category and, combined with the proportion of this equipment in the base's total capacity, calculates the dynamic capacity reduction caused by the sudden event.

[0037] Step S3 is used for carrying capacity update, that is, updating the first carrying capacity feature based on the dynamic capacity decay to obtain the second carrying capacity feature.

[0038] This embodiment uses a resource capacity reduction method to update the numerical value of the first carrying capacity feature. Please refer to... Figure 3 As shown, the specific update process is as follows: Obtain the initial value of the first carrying capacity feature; determine the total number of emergency event categories and extract the impact coefficient of each type of emergency event; obtain the dynamic capacity reduction corresponding to each type of emergency event; calculate the product of the impact coefficient of each type of emergency event and the corresponding dynamic capacity reduction; extract the weight of equipment failure events and the capacity reduction caused by equipment failure, and calculate their product; extract the weight of material supply interruption events and the capacity reduction caused by material supply interruption, and calculate their product; sum the product results of all emergency event categories to obtain the total capacity reduction value; subtract the total capacity reduction value from the initial value of the first carrying capacity feature; set the result of the subtraction as the second carrying capacity feature reflecting the impact of the emergency event.

[0039] As a preferred embodiment of this invention, the second load-bearing capacity characteristic value can be calculated using the following formula:

[0040] in, This indicates the secondary carrying capacity characteristic reflecting the impact of a sudden event. This indicates the total number of emergency categories. Indicates the first Impact coefficient of similar emergencies Indicates the first Dynamic capacity reduction corresponding to sudden events The weights of equipment failure events are indicated. This indicates the amount of production capacity reduction caused by equipment failure. Indicates the weight of material supply disruption events. This represents the reduction in production capacity caused by material supply disruptions. The formula dynamically updates the carrying capacity using a resource capacity reduction method.

[0041] Through the aforementioned resource capacity reduction update method, the second carrying capacity characteristic accurately reflects the actual available capacity of each base after the impact of a sudden event. For example, if a base originally had an available capacity of 1000 units / day (corresponding to the first carrying capacity characteristic), after a reduction of 150 units / day due to equipment failure and a reduction of 50 units / day due to material interruption, its second carrying capacity characteristic corresponding to the second carrying capacity characteristic will be updated to 800 units / day. This dynamic update mechanism ensures that subsequent order allocation decisions are always based on the latest actual capacity.

[0042] Step S4 is used for process quality perception and rework assessment.

[0043] The quality perception sequence data of the upstream process is obtained, and the time-series prediction model is used to perform time-series feature analysis on the quality perception sequence data to obtain the upstream deviation prediction value.

[0044] The quality perception sequence data includes quality inspection index data at each time node of the upstream process. For example, in the processing of the core components of the electricity meter, the time sequence formed by arranging the dimensional accuracy values ​​and electrical performance parameters of each batch of sampling inspections in chronological order.

[0045] The specific process for performing time-series feature analysis on quality-sensing sequence data is as follows: The quality-sensing sequence data is arranged according to the time node order to construct a time-series data set; the time-series data set is input into the time-series prediction model; the time-series prediction model extracts the time-series fluctuation characteristics from the time-series data set; the predicted values ​​of the quality inspection indicators of upstream processes at future time nodes are calculated based on the time-series fluctuation characteristics; and the predicted upstream deviation value is calculated based on the difference between the predicted quality inspection indicator values ​​and the standard quality indicators.

[0046] The time-series prediction model can employ time-series prediction networks such as Long Short-Term Memory (LSTM) or Gated Recurrent Unit (GRU). The model learns the temporal evolution of quality fluctuations based on historical quality inspection data from various time points, predicting the magnitude of potential quality deviations at future time points, i.e., the upstream deviation prediction value.

[0047] After obtaining the upstream deviation prediction value, the downstream rework time is determined based on the comparison between the upstream deviation prediction value and the preset quality deviation threshold. The specific determination process is as follows: Calculate the difference between the upstream deviation prediction value and the preset quality deviation threshold; obtain a pre-set rework time conversion coefficient; determine the total number of time nodes and extract quality inspection index data for each time node; average the quality inspection index data for each time node to obtain the average value of the quality inspection index; calculate the product of the rework time conversion coefficient and the difference between the upstream deviation prediction value and the quality deviation threshold; correct the product result by combining the total number of time nodes and the average value of the quality inspection index; and set the corrected result as the downstream rework time.

[0048] In a preferred embodiment of this invention, downstream rework time can be calculated using the following formula:

[0049] in, Indicates downstream rework time. This represents the rework time conversion factor. This represents the predicted upstream deviation. This represents a pre-established quality deviation threshold. This represents the total number of time points. Indicates the first Quality inspection index data at each time point This represents the average value of the quality inspection indicators. The formula determines the additional rework hours required based on the difference between the upstream deviation prediction value and the quality deviation threshold, combined with the average level of the quality perception sequence data.

[0050] For example, assuming a preset quality deviation threshold of 0.05mm, if the time-series prediction model predicts that the dimensional deviation of a certain batch in the upstream process will reach 0.08mm, exceeding the threshold of 0.03mm, then the rework time conversion factor will be applied. By averaging the quality-sensing sequence data, the additional rework hours required by downstream facilities can be calculated. This allows downstream facilities to schedule rework periods in their production plans in advance, avoiding involuntary shutdowns and waiting after receiving substandard semi-finished products.

[0051] Step S5 is used for order reassignment.

[0052] Based on the second carrying capacity characteristic and the downstream rework hours, the initial order decomposition data is reallocated to obtain the target order decomposition data. The specific reallocation process is as follows: The second carrying capacity feature is analyzed to extract the current resource availability status of each base; the downstream rework hours are added to the initial order decomposition data to update the subtask data set; the updated subtask data set and the current resource availability status of each base are input into the load balancing algorithm; the task allocation weight of each base is calculated through the load balancing algorithm; the updated subtask data set is redistributed according to the task allocation weight to obtain the target order decomposition data.

[0053] In practical implementation, load balancing algorithms can employ optimization algorithms such as weighted round-robin, minimum connection count, or genetic algorithms. The algorithm uses the resource availability status corresponding to the second carrying capacity characteristic of each base as a capacity constraint, downstream rework hours as an additional increment to the total task volume, and the optimal load balance of each base as the objective function. It calculates the task allocation weights for each base and redistributes the updated sub-task data set to obtain target order decomposition data with a relatively balanced task volume across bases.

[0054] Step S6 is used for generating coordinated scheduling instructions.

[0055] Based on the target order decomposition data, coordinated scheduling instructions are generated for the multiple bases. The specific generation process is as follows: Extract the order quantity and decomposition table from the target order decomposition data; classify the order quantity and decomposition table using a random forest algorithm to obtain a first task item set; based on the first task item set, obtain the base number and capacity value of each production base; determine whether the capacity value is greater than a preset threshold; if the capacity value is greater than the preset threshold, extract the corresponding base number to determine a target base number set; for the target base number set, obtain the load rate corresponding to each target base number; perform cluster analysis on the load rate using a hierarchical clustering algorithm to obtain a first allocation order; using the first allocation order, extract the priority and material order from the first allocation order; determine whether the priority meets the pre-established conditions; if the priority meets the pre-established conditions, parse the material order to obtain a second allocation order; obtain the second allocation order and extract the execution period corresponding to the second allocation order; calculate the coordination degree between each production base based on the execution period to determine the target coordination degree; generate a first scheduling order for multiple bases based on the target coordination degree; perform format conversion processing on the first scheduling order to obtain a second scheduling order, and set the second scheduling order as a coordinated scheduling instruction for the multiple bases.

[0056] For example, in a specific scheduling operation, upon receiving an order for 5,000 pieces of metering equipment parts, the scheduling system immediately activated its distributed computing engine to analyze the data. By invoking a genetic algorithm model to cross-reference the current capacity load and material inventory of each manufacturing base and assess their fitness, the optimal allocation was determined: 3,000 production tasks were assigned to the first manufacturing base with sufficient capacity, and the remaining 2,000 tasks were allocated to the second manufacturing base, which was closer in terms of logistics. Subsequently, based on this allocation result, collaborative scheduling instructions containing specific start times and equipment parameters were automatically generated. These instructions were then transmitted in real-time to the programmable logic controllers (PLCs) at the corresponding bases via the enterprise's internal industrial IoT bus, directly driving the production line operation.

[0057] Through the above steps, this embodiment of the invention achieves dynamic perception of production capacity and real-time optimized matching of orders across multiple regional bases, while establishing a process-level progress and quality information collaboration mechanism. When a sudden event occurs at a base leading to a reduction in production capacity, the system can promptly update the base's carrying capacity assessment and reallocate orders accordingly. When upstream processes show a deviation trend in quality, the system can predict and quantify downstream rework hours in advance, allowing downstream bases to adjust their production plans ahead of time. The entire method effectively solves the problem of uncontrolled collaboration caused by dynamic capacity changes and lagging quality information across regional metering production bases, significantly shortening order delivery cycles and improving the overall resource utilization rate and customer satisfaction of multiple bases.

[0058] Furthermore, this embodiment of the invention provides an alternative solution for the method of order reallocation and collaborative scheduling instruction generation.

[0059] In this embodiment, the load balancing algorithm employs particle swarm optimization (PSO). Unlike the genetic algorithm used, PSO simulates bird flock foraging behavior, rapidly converging to the globally optimal task allocation scheme through group information sharing in a multi-dimensional search space. In constructing the fitness function, this embodiment comprehensively considers three dimensions: the current load rate of each base, the historical on-time delivery rate, and logistics costs. These are assigned different weights and then summed to obtain an allocation result that better reflects actual business needs.

[0060] Corresponding to the cross-regional metrology production base collaborative scheduling method in Embodiment 1 of the present invention, Embodiment 2 of the present invention also provides a cross-regional metrology production base collaborative scheduling device, comprising: One or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, and the one or more applications are configured to execute the cross-regional metering production base collaborative scheduling method.

[0061] Corresponding to the cross-regional metrology production base collaborative scheduling method in Embodiment 1 of the present invention, Embodiment 3 of the present invention also provides a computer program product, including computer instructions, which instruct computer equipment to perform the operation corresponding to the cross-regional metrology production base collaborative scheduling method.

[0062] Preferably, the processor can be a central processing unit (CPU), or 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 the processor can be any conventional processor. The processor is the control center of the device, connecting various parts of the device through various interfaces and lines.

[0063] The memory mainly includes a program storage area and a data storage area. The program storage area can store the operating system, applications required for at least one function, etc., while the data storage area can store related data, etc. Furthermore, the memory can be a high-speed random access memory, or a non-volatile memory, such as a plug-in hard drive, a SmartMedia Card (SMC), a Secure Digital (SD) card, and a Flash Card, or other volatile solid-state storage devices.

[0064] It should be noted that the above-mentioned devices may include, but are not limited to, processors and memory, as will be understood by those skilled in the art.

[0065] For the working principle and process of the above embodiments, please refer to the description of Embodiment 1 of the present invention, which will not be repeated here.

[0066] Compared with existing technologies, this invention has the following significant advantages: This invention uses a machine learning model to extract features from the differences in equipment types and operating states at various bases, constructing a first carrying capacity feature that reflects the actual production resource carrying capacity. This effectively solves the problem of distorted capacity assessment caused by neglecting equipment heterogeneity in traditional methods, providing an accurate decision-making benchmark for cross-base order allocation. When the resource demand after order decomposition exceeds the available capacity, a capacity warning is automatically triggered. A classification algorithm is used to classify and quantify capacity decay in sudden events such as equipment failure and material supply interruptions, dynamically updating the second carrying capacity feature in a decreasing resource capacity manner. This allows the system to perceive the instantaneous changes in available capacity at each base in real time, avoiding allocation failures and delivery delays caused by information lag. This invention also analyzes the quality perception sequence data of upstream processes through a time-series prediction network, predicting the magnitude of quality deviations before they actually occur, and calculating the additional rework hours that downstream bases need to reserve. This achieves early exposure and proactive intervention of quality risks, changing the passive situation of downstream units passively waiting and temporarily stopping work for rework in the traditional model. This invention combines updated resource availability and rework hours to load balance and redistribute orders, enabling task scheduling to simultaneously consider capacity constraints and quality compensation. The generated collaborative scheduling instructions achieve precise process-level coordination between different bases. Overall, this invention achieves a dual upgrade in cross-regional, multi-base capacity assessment from static to dynamic, and in quality management from passive response to proactive prediction, significantly improving the scheduling accuracy, resource utilization efficiency, and on-time order delivery capabilities of multi-base collaborative manufacturing.

[0067] The above description is merely a preferred embodiment of the present invention and should not be construed as limiting the scope of the invention. Therefore, any equivalent variations made in accordance with the claims of the present invention are still within the scope of the present invention.

Claims

1. A method for collaborative scheduling of cross-regional metering production bases, characterized in that, include: Step S1: Obtain heterogeneous equipment data from multiple bases, extract features from the heterogeneous equipment data, and obtain a first carrying capacity feature characterizing the production resource carrying capacity of each base. Step S2: Obtain initial order decomposition data formed by decomposing the original production order according to the process dimension. In response to the resource demand corresponding to the initial order decomposition data being greater than the resource availability corresponding to the first carrying capacity feature, obtain data on emergencies during the production execution process and determine the dynamic capacity reduction corresponding to various emergencies. Step S3: Update the first carrying capacity feature based on the dynamic capacity reduction to obtain the second carrying capacity feature; Step S4: Obtain the quality perception sequence data of the upstream process, perform time series feature analysis on the quality perception sequence data using a time series prediction model to obtain the upstream deviation prediction value, and determine the downstream rework time based on the comparison result of the upstream deviation prediction value and the preset quality deviation threshold. Step S5: Based on the second carrying capacity characteristic and the downstream rework time, the initial order decomposition data is redistributed to obtain the target order decomposition data; Step S6: Generate collaborative scheduling instructions for the multiple bases based on the target order decomposition data.

2. The method according to claim 1, characterized in that, In step S1, feature extraction is performed on the heterogeneous data of the equipment to obtain the first carrying capacity feature characterizing the production resource carrying capacity of each base, specifically including: The first carrying capacity feature is obtained by performing feature extraction processing on the heterogeneous data of the device through a machine learning model. The heterogeneous equipment data includes information on differences in equipment types and differences in equipment operating status between different bases.

3. The method according to claim 1, characterized in that, Step S2 specifically includes: Extract the production resource carrying capacity values ​​of each base from the first carrying capacity feature to determine the resource availability of each base; Obtain the original production order, decompose the original production order according to the process dimension to form a sub-task data set, and extract the resource requirements corresponding to the sub-task data set; The resource demand is compared with the resource availability, and a first capacity warning is generated when the resource demand is greater than the resource availability. The emergency event data is obtained based on the first production capacity early warning information.

4. The method according to claim 3, characterized in that, The step of obtaining the emergency event data based on the first production capacity early warning information specifically includes: Analyze the warning time node corresponding to the first production capacity warning information; Retrieve real-time monitoring records during the production process based on the aforementioned warning time points; Extract abnormal equipment operation records and abnormal material flow records from the real-time monitoring records; Based on the equipment operation abnormality record, generate equipment failure event information; based on the material flow abnormality record, generate material supply interruption event information. The equipment failure event information and the material supply interruption event information are combined and processed to obtain the emergency event data.

5. The method according to claim 1, characterized in that, In step S2, determining the dynamic capacity reduction corresponding to various emergencies specifically includes: The emergency event data is input into a pre-trained classification algorithm model, and the event feature vector in the emergency event data is extracted through the classification algorithm model. The emergency event data is categorized based on the event feature vectors to obtain emergency event categories. For each type of emergency, retrieve the corresponding historical capacity reduction records, and calculate the dynamic capacity reduction amount corresponding to that type of emergency based on the historical capacity reduction records.

6. The method according to claim 1, characterized in that, Step S3 specifically includes: By adopting a resource capacity reduction method, the cumulative value of the product of dynamic capacity reduction and corresponding impact coefficient corresponding to various emergencies is subtracted from the value of the first carrying capacity characteristic to obtain the second carrying capacity characteristic; These various emergencies include equipment failure events and material supply interruption events.

7. The method according to claim 1, characterized in that, In step S4, a time-series prediction model is used to perform time-series feature analysis on the quality-sensing sequence data to obtain the upstream deviation prediction value, specifically including: The quality-sensing sequence data are arranged in chronological order to construct a time-series data set; The time series data set is input into the time series prediction model, and the time series prediction model is used to extract the time series fluctuation characteristics in the time series data set. Calculate the predicted values ​​of the quality inspection indicators of the upstream process at future time nodes based on the aforementioned time-series fluctuation characteristics; The upstream deviation prediction value is calculated based on the difference between the predicted value of the quality inspection index and the standard quality index.

8. The method according to claim 1, characterized in that, Step S6 specifically includes: Extract the order quantity and decomposition table from the target order decomposition data, and classify the order quantity and decomposition table using a classification algorithm to obtain the first task item set; Based on the first task item set, obtain the base number and capacity value of each production base, filter the base numbers whose capacity values ​​are greater than a preset threshold, and determine the target base number set; For the set of target base numbers, the load rate corresponding to each target base number is obtained, and the load rate is clustered using a clustering algorithm to obtain the first allocation order; Extract the priority and material list from the first allocation order; when the priority meets a preset condition, parse the material list to obtain the second allocation order. Obtain the execution period corresponding to the second allocation order, calculate the degree of coordination between each production base based on the execution period, and generate a coordinated scheduling instruction for the multiple bases based on the degree of coordination.

9. A cross-regional metering production base collaborative scheduling device, characterized in that, include: One or more processors; Memory; One or more applications, wherein the one or more applications are stored in the memory and configured to be executed by the one or more processors, the one or more applications being configured to perform the cross-regional metering production base collaborative scheduling method as described in any one of claims 1 to 8.

10. A computer program product, characterized in that, It includes computer instructions that instruct computer equipment to perform operations corresponding to the cross-regional metering production base collaborative scheduling method as described in any one of claims 1 to 8.

Images