A method and system for intelligent scheduling and optimization of oil cylinder piston rod machining process

Abstract

The present application relates to the technical field of piston rod intelligent scheduling, in particular to an oil cylinder piston rod machining process intelligent scheduling optimization method and system, comprising: obtaining the machining parameter characteristics of multiple to-be-produced batches, and obtaining multiple virtual categories; calculating the comprehensive average deviation rate of the current process flow; calculating the synergistic correlation degree between the current process and the next process corresponding to each virtual category; calculating the machining advantage degree of each virtual category in the current process, and determining the breakpoint process of each virtual category; calculating the dynamic rescheduling trigger factor of each virtual category; when multiple virtual categories contend for the same device, determining the competition priority of each virtual category; determining the arbitration result according to the competition priority of each virtual category, generating a scheduling instruction according to the arbitration result, and issuing the scheduling instruction to the device for execution. The present application can realize closed-loop intelligent scheduling of piston rod multi-process machining, reduce quality defect accumulation, shorten changeover time, and improve the comprehensive efficiency of the device and the product yield.

Description

Technical Field

[0001] This invention relates to the field of intelligent piston rod scheduling technology, specifically to an intelligent scheduling optimization method and system for the machining process of hydraulic cylinder piston rods. Background Technology

[0002] The piston rod of a hydraulic cylinder is the core transmission component, and its machining quality directly affects the sealing performance and service life of the hydraulic system. Piston rods are typically slender shaft parts with a length-to-diameter ratio generally greater than 10. They require multiple consecutive processes, including rough turning, finish turning, grinding, and polishing. Strict constraints exist between these processes regarding machining allowance allocation, cutting heat control, and dimensional accuracy transfer. Furthermore, different batches of piston rods exhibit significant differences in diameter tolerance grades, length specifications, material hardness, and surface roughness requirements. This necessitates changing fixtures, adjusting tool compensation parameters, and resetting machining programs during batch changes, resulting in lengthy turnaround times. How to rationally formulate a scheduling scheme while meeting the constraints of multi-process sequence and accuracy transfer requirements remains a long-standing but unresolved technical challenge in this field.

[0003] Currently, the scheduling methods in the field of hydraulic cylinder piston rod machining mainly include the following: First, using general information tools such as Manufacturing Execution Systems (MES) or Advanced Planning and Scheduling Systems (APS) to collect equipment operation data and combine it with order delivery dates for production scheduling; second, setting up a multi-stage production information chain by acquiring real-time piston rod machining parameters to segment and control machining stages such as roughing, finishing, and grinding. However, the above existing methods all treat each batch as an independent machining unit, using only equipment idle status and order delivery dates as scheduling basis, without considering the dynamic matching relationship between the exit status of the preceding process and the entry requirements of the following process in the continuous multi-process machining of piston rods. In actual production, to catch up with urgent orders, operators often use a "batch transfer" method, directly transferring rods that have not been fully processed in the preceding process to the next process. At this time, the actual state of the rod (such as temperature, dimensional allowance, hardness) may have deviated from the standard entry process requirements of the next process. If there is a lack of quantitative assessment of the degree of deviation, this deviation will accumulate and amplify in subsequent processes, eventually leading to irreversible quality defects such as grinding burns and dimensional deviations.

[0004] Furthermore, existing methods have two shortcomings in formulating scheduling schemes. First, the segmentation strategy is static and fixed, with adjustment nodes pre-set based on initial process data. However, in actual processing, minute fluctuations such as spindle temperature rise and tool wear can cause batch arrival times to shift progressively. Fixed nodes cannot dynamically trigger rescheduling based on the actual tolerance of each batch to time fluctuations, causing the segmentation scheme formulated in the early stages to fail in later execution. Second, when multiple batches conflict in front of a critical piece of equipment (such as a CNC cylindrical grinder), existing rules only compare task priorities or waiting times to determine the processing order, completely neglecting the parameter space differences between competing batches. In actual production, if the diameter, hardness, and other parameters of the batch currently being processed differ significantly from those of the batch to be switched, the changeover adjustment time may differ by several times (from 10 minutes to more than 40 minutes). This neglect results in a formally fair queuing result that is extremely inefficient at the physical execution level. Summary of the Invention

[0005] This invention provides an intelligent scheduling and optimization method and system for the machining process of hydraulic cylinder piston rods to solve existing problems.

[0006] The intelligent scheduling and optimization method and system for the machining process of hydraulic cylinder piston rod of the present invention adopts the following technical solution: One embodiment of the present invention provides an intelligent scheduling and optimization method for the machining process of a hydraulic cylinder piston rod, the method comprising the following steps: The processing parameter features of multiple production batches to be scheduled are obtained, and the multiple production batches to be scheduled are clustered based on the processing parameter features to obtain multiple virtual classifications. Based on the measured values ​​of the key state parameters of the target member in each virtual category when it leaves the current process, and the standard entry requirement value corresponding to the next process, calculate the comprehensive average deviation rate of the current process flow. By using the actual production cycle time of the current process corresponding to each virtual category and the standard average cycle time of the next process, combined with the comprehensive average deviation rate of the current process flow, the degree of synergistic correlation between the current process and the next process corresponding to each virtual category is calculated. Based on the degree of collaboration between the current process and the next process corresponding to each virtual category, the processing advantage of each virtual category in the current process is calculated, and the breakpoint process of each virtual category is determined according to the changing trend of the processing advantage. Based on the deviation between the actual time and the planned time of each batch of production to be scheduled to arrive at the next process in each virtual category, and combined with the position of the breakpoint process of each virtual category in the total process, the dynamic rescheduling trigger factor of each virtual category is calculated. When multiple virtual categories contend for the same device, the competition priority of each virtual category is determined based on the dynamic rescheduling trigger factor of each virtual category and the parameter similarity between each virtual category and the virtual category being processed by the device. The arbitration result is determined based on the competition priority of each virtual category, and a scheduling instruction is generated and issued to the equipment for execution based on the arbitration result.

[0007] Furthermore, the specific steps involved in clustering multiple production batches based on processing parameter features to obtain multiple virtual classifications are as follows: Construct the original processing parameter feature vector for each production batch to be scheduled; wherein the original processing parameter feature vector shall include feature values ​​of at least five dimensions: nominal diameter, tolerance grade, processing length, hardness coefficient, and surface roughness. Normalize the feature values ​​of all production batches to be scheduled in each dimension to obtain the standardized demand feature vector of each production batch to be scheduled. For any two production batches to be scheduled, calculate the Euclidean distance between the standardized demand feature vectors of the two production batches. The reciprocal of the sum of 1 and the Euclidean distance is used as the similarity of processing requirements between the two production batches to be scheduled. When the similarity of processing requirements between two production batches to be scheduled is greater than or equal to a preset similarity threshold, the two production batches to be scheduled are classified into the same virtual category.

[0008] Furthermore, the specific steps for calculating the comprehensive average deviation rate of the current process flow based on the measured values ​​of the key state parameters when the target member in each virtual classification leaves the current process and the standard entry requirement value corresponding to the next process are as follows: Obtain the target rod that has been processed and transferred to the next process in the current process corresponding to the current virtual classification; Obtain the measured values ​​of each key state parameter when each target member leaves the current process, as well as the standard entry requirements for each key state parameter in the next process; Calculate the standard deviation of the measured values ​​of each key state parameter when all target members leave the current process, and use it as the first standard deviation; Calculate the difference between the measured values ​​of each key state parameter when each target member leaves the current process and the standard entry requirements of each key state parameter for the next process, and divide by the first standard deviation to obtain the flow state deviation of each target member; Calculate the mean of the absolute values ​​of the flow state deviations of all target rods, and use it as the flow state deviation rate of the current process corresponding to each key state parameter; The weighted arithmetic mean of the deviation rates of the current process flow parts corresponding to all key status parameters is calculated to obtain the comprehensive average deviation rate of the current process flow.

[0009] Furthermore, the specific steps for calculating the collaborative correlation between the current process and the next process corresponding to each virtual category, by utilizing the actual production cycle time of the current process and the standard average cycle time of the next process, combined with the comprehensive average deviation rate of the current process flow, are as follows: Obtain the actual production cycle time of the current process corresponding to the current virtual category, and the standard average cycle time of the next process to process the current virtual category; Calculate the ratio of the actual production cycle time to the standard average cycle time, and use it as the first ratio. Calculate the difference between 1 and the overall average deviation rate of the current process flow, and use it as the first difference; The product of the first ratio and the first difference is determined as the degree of collaborative association between the current process and the next process corresponding to the current virtual classification.

[0010] Furthermore, the specific steps for calculating the processing advantage of each virtual category in the current process based on the collaborative correlation degree between the current process and the next process corresponding to each virtual category, and determining the breakpoint process for each virtual category according to the changing trend of the processing advantage, are as follows: For the process segment from the current process to the last process corresponding to the current virtual classification, obtain the variance of the collaborative correlation degree between all two adjacent processes within the process segment, and use it as the first variance; Obtain the average of the degree of collaboration between all two adjacent processes within this process segment, and use it as the first average. The square root of the sum of 1 and the first variance is multiplied by the first mean to calculate the processing advantage of the current virtual classification in the current process. If the processing advantage of the current virtual category in the next process is less than that of the current virtual category in the current process, and the decrease exceeds the preset percentage threshold for the first time, the current process will be determined as the breakpoint process of the current virtual category.

[0011] Furthermore, the specific steps for calculating the dynamic rescheduling trigger factor for each virtual category, based on the deviation between the actual and planned times of each batch of production to be scheduled to the next process in each virtual category, and in conjunction with the position of the breakpoint process of each virtual category in the total process, are as follows: For each batch of production to be scheduled within the current virtual category, obtain the actual time when the batch arrives at the next process and the planned time when the batch arrives at the next process; Calculate the absolute difference between the actual time and the planned time for the batch to arrive at the next process, and use this as the first absolute difference for the batch. Calculate the mean of the first absolute differences of all batches, and use it as the production cycle deviation of the current dummy classification. Get the maximum time deviation of the current production line; Calculate the ratio of the current virtual classification's production cycle deviation to the current production line's maximum time deviation, and use this as the second ratio; The ratio of the number of the breakpoint process in the current virtual classification to the total number of processes in the current virtual classification is used as the third ratio. Multiply the sum of 1 and the third ratio by the second ratio to calculate the dynamic rescheduling trigger factor for the current virtual classification.

[0012] Furthermore, when multiple virtual categories contend for the same device, the competition priority of each virtual category is determined based on the dynamic rescheduling trigger factor of each virtual category and the parameter similarity between each virtual category and the virtual category being processed by the device. The specific steps include the following: For the two virtual categories that are competing for the equipment, other than the virtual category that is currently processing the equipment, one of them will be designated as the first virtual category and the other as the second virtual category. Calculate the parameter similarity between the first virtual classification and the virtual classification of equipment in the current processing stage, and use it as the first similarity. Calculate the parameter similarity between the second virtual classification and the virtual classification of equipment in the current processing stage, and use it as the second similarity. Calculate the ratio of the dynamic rescheduling trigger factor of the first virtual classification to the dynamic rescheduling trigger factor of the second virtual classification, and use it as the fourth ratio; The fifth ratio is calculated by taking the sum of 1 and the first similarity as the numerator and the sum of 1 and the second similarity as the denominator. The product of the fourth ratio and the fifth ratio is determined as the competitive priority of the first virtual classification relative to the second virtual classification.

[0013] Furthermore, the specific steps for determining the arbitration result based on the competitive priority of each virtual category are as follows: When the competition priority of the first virtual category relative to the second virtual category is greater than 1, the first virtual category is determined to have the right to use the equipment. When the competition priority of the first virtual category relative to the second virtual category is less than or equal to 1, the second virtual category is determined to have the right to use the equipment.

[0014] Furthermore, the specific steps involved in generating scheduling instructions based on the arbitration result and issuing them to the equipment for execution are as follows: The arbitration result is converted into a scheduling instruction; the scheduling instruction includes the processing program number, tool compensation parameters and tooling fixture number corresponding to the virtual classification that has obtained the right to use the equipment. The scheduling command is sent to the programmable controller or CNC system of the corresponding equipment, forcing the equipment to load the process file of the specified virtual category; The system collects completion signals and actual production rhythm in real time, and transmits them back to the scheduling server via the edge gateway. The data is then compared with the planned values ​​to update the dynamic rescheduling trigger factors. The scheduling server scans all dynamic rescheduling trigger factors for the virtual production category at fixed intervals. When any dynamic rescheduling trigger factor is greater than 1, the rescheduling process is triggered.

[0015] This invention proposes an intelligent scheduling and optimization system for the machining process of a hydraulic cylinder piston rod, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is executed by the processor, it implements the steps of the intelligent scheduling and optimization method for the machining process of a hydraulic cylinder piston rod as described above.

[0016] The beneficial effects of the technical solution of this invention are as follows: This invention proposes an intelligent scheduling optimization method and system for the machining process of hydraulic cylinder piston rods. Through quantitative evaluation of the inter-process synergy, it can accurately identify segmentation points based on the deviation between the measured values ​​of key state parameters of the target rod when it leaves the current process and the standard entry requirements of the next process, combined with the matching degree of the cycle time of the preceding and following processes. This allows for early prediction of the risk of discontinuity before the state deviation accumulates to the quality defect threshold, effectively reducing rework or scrap due to inconsistent states. By constructing a dynamic rescheduling trigger factor that integrates the degree of time offset and the position of the breakpoint process, virtual classifications in the later stages of processing automatically obtain lower trigger tolerance due to higher accumulated risk, achieving differentiation. Re-scheduling triggers overcome the shortcomings of traditional static segmentation strategies in handling actual fluctuations. By introducing parameter similarity between each category and the current processing state of the equipment as an arbitration dimension when multiple virtual categories contend for equipment, the system prioritizes virtual categories with shorter changeover adjustment times when urgency levels are similar, significantly reducing equipment non-output time. Furthermore, scheduling instructions are directly sent to the equipment PLC or CNC system in the form of structured messages, and closed-loop control is formed in conjunction with real-time data feedback from equipment sensors, achieving second-level response from scheduling decisions to equipment execution. Ultimately, significant improvements are made in overall equipment efficiency, product yield, and delivery cycle, providing a systematic intelligent scheduling solution for multi-process continuous processing scenarios of piston rods. 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 the steps of an intelligent scheduling and optimization method for the machining process of a hydraulic cylinder piston rod according to the present invention. Figure 2 This is a block diagram of an intelligent scheduling and optimization system for the machining process of a hydraulic cylinder piston rod according to the present invention. Detailed Implementation

[0019] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an intelligent scheduling optimization method and system for hydraulic cylinder piston rod processing proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0020] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0021] The following description, in conjunction with the accompanying drawings, details the specific scheme of the intelligent scheduling and optimization method and system for the machining process of hydraulic cylinder piston rod provided by this invention.

[0022] Please see Figure 1 The diagram illustrates a flowchart of an intelligent scheduling optimization method for the machining process of a hydraulic cylinder piston rod according to an embodiment of the present invention. The method includes the following steps: Step S001: Obtain the processing parameter features of multiple production batches to be scheduled, and perform clustering processing on the multiple production batches to be scheduled based on the processing parameter features to obtain multiple virtual classifications.

[0023] It should be noted that piston rods from different orders or batches often differ in nominal diameter, tolerance grade, machining length, material hardness, and surface roughness requirements. If processing is done sequentially according to arrival order, frequent batch switching will lead to a surge in equipment downtime. Each switch requires changing fixtures, adjusting tool compensation parameters, and resetting the machining program, severely compressing actual processing time. To reduce changeover frequency and improve equipment utilization from the outset, this embodiment first clusters all batches to be scheduled before initiating specific process scheduling calculations. Batches with similar overall machining parameters are grouped into the same virtual category. Batch processing within the same virtual category requires minimal changeover adjustments during continuous processing. Subsequent dynamic scheduling and equipment contention arbitration are then performed between different virtual categories.

[0024] A batch awaiting production refers to one or more independent batches that are waiting to enter the piston rod processing production line within the same production planning cycle. Each batch contains several piston rods with the same drawing number, process route, and quality requirements.

[0025] The processing parameter characteristics refer to five key technical indicators used to characterize the processing process attributes and quality objectives of each batch, namely: nominal diameter (unit mm), tolerance grade (IT6 to IT11), processing length (unit mm), hardness coefficient (dimensionless value based on HRC normalization) and surface roughness (reciprocal of Ra value).

[0026] The above-mentioned processing parameter characteristics are obtained in the following ways: directly read from the database of the Manufacturing Execution System (MES) or Enterprise Resource Planning System (ERP); entered by process engineers through the human-machine interface before production scheduling; or automatically parsed and extracted from the piston rod process card in the Product Lifecycle Management System (PLM) and stored in the original data table of the scheduling server.

[0027] In step S001, multiple production batches to be scheduled are clustered based on processing parameter features to obtain multiple virtual classifications, specifically including steps S0011-S0015: Step S0011: Construct the original processing parameter feature vector for each production batch to be scheduled; wherein, the original processing parameter feature vector shall include feature values ​​of at least five dimensions: nominal diameter, tolerance grade, processing length, hardness coefficient, and surface roughness.

[0028] It should be noted that the original processing parameter feature vector refers to a numerical vector used to characterize the processing attributes and quality targets of each production batch to be scheduled, and it consists of feature values ​​in multiple dimensions. It must include feature values ​​in at least the following five dimensions: Nominal diameter: The design diameter of the piston rod, in millimeters. It is used to determine the spindle speed range and cutting parameters of the machining equipment. Piston rods of different diameters have significant differences in clamping methods and tool selection. When changing production, it is necessary to change the fixture of the corresponding specification.

[0029] Tolerance grade: The ISO 286 international tolerance grade standard (such as IT6 to IT11) is adopted to reflect the allowable variation range of the piston rod diameter. The higher the tolerance grade (the smaller the value), the more stringent the allocation of finishing allowance, and the machining parameters of subsequent grinding processes need to be adjusted accordingly.

[0030] Machining length: The effective length of the piston rod to be machined, in millimeters. It is used to estimate the machining time of a single piece and to determine the clamping method and support position. When switching batches with large length differences, the position of the center frame needs to be adjusted, which increases the changeover time.

[0031] Hardness coefficient: A dimensionless value based on the Rockwell hardness value of the piston rod material after normalization. The value range is [0,1]. It is used to reflect the machinability of the material. The higher the hardness, the faster the tool wears. When changing production, the cutting parameters and tool compensation values ​​need to be adjusted accordingly.

[0032] Surface roughness: The reciprocal of the final surface roughness Ra value of the piston rod. The larger the value, the higher the roughness requirement, and the more difficult and time-consuming the grinding and polishing processes become.

[0033] In addition to the five dimensions mentioned above, the original processing parameter feature vector may also include, depending on actual production needs: Initial straightness: The straightness deviation of the piston rod blank or the part transferred from the previous process before entering the current process, in millimeters per meter (mm / m). Rods with large initial straightness deviations need to be corrected by adding a straightening process or increasing the single cutting depth during the roughing stage, which will result in the actual production cycle of this batch being longer than the standard cycle.

[0034] For example, if the production batch B001 is a piston rod with a diameter of 50 mm, a tolerance grade of IT7, a machining length of 400 mm, a material hardness of 0.65 after normalization, and a surface roughness requirement of 0.50 (the reciprocal of the Ra value), then the original machining parameter feature vector of batch B001 is (50, 7, 400, 0.65, 0.50).

[0035] Step S0012: Normalize the feature values ​​of all production batches to be scheduled in each dimension to obtain the standardized demand feature vector of each production batch to be scheduled.

[0036] It should be noted that normalization refers to linearly scaling the original feature values ​​of each batch in the five dimensions according to each dimension, so that the values ​​of all batches in the same dimension fall within the range of [0,1].

[0037] Specifically, for each of the five dimensions—nominal diameter, tolerance grade, machining length, hardness coefficient, and surface roughness—the maximum and minimum values ​​for all batches to be produced in that dimension are first calculated. Then, the minimum value for that dimension is subtracted from the original characteristic value of each batch, and then divided by the difference between the maximum and minimum values ​​for that dimension. The result is the standardized characteristic value of that batch in that dimension. After the above processing, the numerical range of each dimension is uniformly compressed to between 0 and 1, which preserves the relative size relationship and degree of difference between batches in that dimension, while eliminating the interference caused by the different dimensions and numerical magnitudes between different dimensions in subsequent calculations.

[0038] For example, batch B001 has a nominal diameter of 50 mm, batch B002 has a nominal diameter of 52 mm, and batch B003 has a nominal diameter of 75 mm. The minimum nominal diameter among the three batches is 50 mm, and the maximum is 75 mm. The nominal diameter of batch B001 is normalized to 0, that of batch B002 is approximately 0.08, and that of batch B003 is normalized to 1. Similarly, the other four dimensions are also linearly scaled according to their respective minimum and maximum values. Finally, each batch obtains a standardized demand feature vector with values ​​in all five dimensions between 0 and 1, which serves as the basis for subsequent similarity calculations and cluster analysis.

[0039] The above normalization process does not change the order of superiority and inferiority of each batch in the same dimension, but only changes the absolute size and dimension of the values. Its purpose is to make the feature values ​​of the five dimensions have the same weight contribution in the subsequent Euclidean distance calculation, so as to avoid a certain dimension from dominating the similarity judgment result due to the original value being too large (such as the processing length, which is hundreds of millimeters), causing the actual influence of other dimensions (such as the hardness coefficient, which is between 0 and 1) to be submerged.

[0040] Step S0013: For any two production batches to be scheduled, calculate the Euclidean distance between the standardized demand feature vectors of the two production batches.

[0041] It should be noted that Euclidean distance refers to the geometric distance between the standardized demand feature vectors of any two production batches, used to quantitatively measure the overall difference between the two batches in terms of processing parameters. Since the feature values ​​of the five dimensions have all been normalized to the [0,1] interval through step S0012, each dimension has the same numerical magnitude. Therefore, the calculated Euclidean distance can truly reflect the comprehensive differences between batches in terms of processing parameters, and there will be no situation where a certain dimension dominates the distance calculation result due to its original numerical magnitude being too large.

[0042] For example, the standardized demand feature vector of batch B001 is (0, 0.2, 0, 0.333, 0.25), which means that the standardized values ​​of this batch in the five dimensions are 0, 0.2, 0, 0.333 and 0.25, respectively.

[0043] The standardized demand feature vector for batch B002 is (0.08, 0, 0.1, 0.267, 0.75), which means that the standardized values ​​of this batch in the five dimensions are 0.08, 0, 0.1, 0.267 and 0.75, respectively.

[0044] When calculating the Euclidean distance between batches B001 and B002, the squares of the differences in each of the five dimensions are first calculated: the first dimension (nominal diameter) is 0.0064; the second dimension (tolerance grade) is 0.04; the third dimension (processing length) is 0.01; the fourth dimension (hardness coefficient) is approximately 0.0044; and the fifth dimension (surface roughness) is 0.25. The sum of these five squared values ​​is approximately 0.3108. Taking the square root of this sum yields an Euclidean distance of approximately 0.557 between batches B001 and B002.

[0045] If the calculated Euclidean distance between batch B001 and batch B003 is approximately 1.439, and the Euclidean distance between batch B002 and batch B003 is approximately 1.245, then comparing these three Euclidean distances reveals that the Euclidean distance between batch B001 and batch B002 is the smallest, indicating that their overall processing parameters are the closest, requiring the fewest equipment adjustments during production changeovers. Conversely, the Euclidean distance between batch B001 and batch B003 is the largest, indicating that their overall processing parameters differ the most, requiring the most equipment adjustments during production changeovers.

[0046] Step S0014: Use the reciprocal of the sum of 1 and the Euclidean distance as the similarity of processing requirements between the two production batches to be scheduled.

[0047] Step S0015: When the similarity of processing requirements between the two production batches to be scheduled is greater than or equal to the preset similarity threshold, the two production batches to be scheduled are classified into the same virtual category.

[0048] It should be noted that the preset similarity threshold is the minimum similarity standard for determining whether two production batches meet the clustering criteria. When the processing requirements similarity of two batches is greater than or equal to this threshold, it indicates that their overall processing parameters are sufficiently similar, and the equipment requires less adjustment of tools, fixtures, and processing programs during production changeover, making continuous processing economical. Therefore, these two batches are classified into the same virtual category. When the similarity is less than this threshold, it indicates that their processing parameters differ significantly. If forced to aggregate, the production changeover adjustment would be time-consuming, resulting in significant equipment efficiency losses, making them unsuitable for being classified into the same virtual category.

[0049] The formation of virtual categories employs a strong classification strategy. A batch is only allowed to join a virtual category if its processing requirements similarity to all existing batches within that category is greater than or equal to a preset threshold. For example, if the similarity between batch A and batch B is greater than or equal to the threshold, batch A and batch B first form an initial virtual category C1. When evaluating whether batch C can join virtual category C1, it must simultaneously satisfy the requirement that its similarity to both batch A and batch B is greater than or equal to the threshold. If batch C's similarity to batch A meets the threshold requirement but its similarity to batch B does not, then batch C cannot join virtual category C1 and must either create a new virtual category or join another virtual category that meets the conditions. Through this method, the system iteratively traverses all batches to be scheduled, ultimately outputting multiple mutually exclusive virtual category sets. Batches within each virtual category satisfy the condition that their pairwise similarity is greater than or equal to the threshold, ensuring that the processing parameters of each batch within a category are highly consistent, and that the changeover adjustment time between any two batches is within an acceptable range.

[0050] Each batch can only belong to one virtual category in the final classification; a batch cannot belong to multiple virtual categories simultaneously. The output result is a set of virtual categories {C1, C2, ..., CK}, where each virtual category consists of one or more batches with highly similar processing requirements. The specific value of the preset similarity threshold is configured by the process engineer based on actual changeover statistics. In this embodiment, the threshold is set to 0.75 by default and is adjustable. A higher threshold setting results in stricter clustering conditions, higher consistency of processing parameters among batches within a virtual category, and shorter changeover adjustment time, but may increase the number of virtual categories. A lower threshold setting results in more lenient clustering conditions, which may reduce the number of virtual categories, but increases the parameter differences among batches within a category, and correspondingly prolongs the changeover adjustment time. In practical applications, this threshold can be dynamically adjusted based on equipment performance, product precision requirements, and production capacity targets.

[0051] Step S002: Based on the measured values ​​of the key state parameters of the target member in each virtual category when it leaves the current process, and the standard entry requirement value corresponding to the next process, calculate the comprehensive average deviation rate of the current process flow.

[0052] It should be noted that in the actual piston rod machining process, although the machining parameters of batches within the same virtual category are generally similar, a critical issue remains when transferring between processes. When batches are completed in the preceding process and transferred to the following process, due to factors such as equipment condition, tool wear, and heat accumulation during actual machining, the key state parameters (such as temperature, dimensional allowance, and hardness) of the rod leaving the current process often deviate from the standard entry requirements of the subsequent process. If this deviation is not quantitatively assessed and transfer is carried out blindly, the rod may suffer from grinding burns, dimensional deviations, and other quality defects during subsequent machining due to inconsistent condition. More seriously, this deviation will accumulate and amplify in subsequent processes, ultimately causing batch rework or scrap.

[0053] Step S002 specifically includes steps S0021-S0026: Step S0021: Obtain the target rod that has been processed and transferred to the next process in the current process corresponding to the current virtual classification.

[0054] It should be noted that: the target rod refers to the piston rod that has been processed and transferred to the entrance of the next process in the current process corresponding to the current virtual classification. The acquisition of the target rod is achieved by a sensor installed at the discharge end of the equipment in the current process, in conjunction with a programmable controller (PLC): whenever a rod completes processing in the current process and is discharged from the discharge end, the PLC records the completion time of the rod and, combined with its virtual classification information, marks the rod as a target rod, and simultaneously sends a completion event message to the scheduling server. Based on the received message, the scheduling server, using the virtual classification number and process number as indexes, associates and stores the rod's identification information, batch number, completion timestamp, and subsequently collected key status parameter measured values ​​in the flow record table for subsequent steps to read and calculate.

[0055] Step S0022: Obtain the measured values ​​of each key state parameter when each target member leaves the current process, as well as the standard entry requirements for each key state parameter in the next process.

[0056] It should be noted that: critical state parameters refer to the physical or geometric properties of the piston rod that affect the processing quality of the next process. Depending on the process requirements of different processes, critical state parameters include, but are not limited to, temperature, dimensional allowance, and hardness.

[0057] Among the aforementioned key state parameters, temperature refers to the real-time temperature of the target rod when it leaves the current process, which is collected by an infrared temperature sensor or contact thermocouple installed at the discharge end of the current process equipment at the instant the target rod is discharged; dimensional allowance refers to the remaining cutting thickness in the diameter direction of the target rod after processing in the current process, which is automatically detected by a laser measuring instrument or pneumatic measuring instrument installed at the discharge end when the target rod leaves the current process; hardness refers to the surface hardness value of the target rod when it leaves the current process, which is detected by an online hardness testing device during the flow of the target rod.

[0058] Standard entry requirements refer to the process standard values ​​set for the incoming material condition to ensure normal processing quality in the next process. These values ​​are pre-set by process engineers according to process specifications and stored in the manufacturing execution system. For example, if the next process is a fine grinding process, the standard entry requirement for incoming material temperature is 30 degrees Celsius, and the standard entry requirement for dimensional allowance is 0.15 mm; if the next process is a heat treatment process, the standard entry requirement for incoming material hardness is 45 HRC. These standard entry requirements are indexed by the process routes of the virtual classification and associated with the standard process parameters of each virtual classification in each process, so that the scheduling server can read them when it needs to calculate the deviation rate.

[0059] The correspondence between measured values ​​and standard entry requirements is as follows: there is a one-to-one correspondence between the measured values ​​and standard entry requirements of the same key state parameter. For example, the measured temperature value is paired with the standard temperature entry requirement, and the measured dimensional allowance value is paired with the standard dimensional allowance entry requirement. This ensures that each deviation in the subsequent deviation rate calculation is compared based on the same physical parameter, avoiding confusion of physical meaning caused by cross-parameter calculation.

[0060] Step S0023: Calculate the standard deviation of the measured values ​​of each key state parameter when all target members leave the current process, and use it as the first standard deviation.

[0061] It should be noted that standard deviation is a statistic that measures the dispersion between measured values ​​of various target components on the same critical state parameter, and is used to reflect the quality of consistency in the current process. For a certain critical state parameter, if the measured values ​​of multiple target components are close to each other, the standard deviation is small, indicating that the current process is stable and the differences between the components are small; if the measured values ​​of multiple target components are widely dispersed, the standard deviation is large, indicating that the current process fluctuates significantly and the consistency between the components is poor.

[0062] Taking temperature parameters as an example, if the current process has processed three target rods with measured temperatures of 32 degrees Celsius, 35 degrees Celsius, and 29 degrees Celsius, the system calculates the standard deviation of these three temperature values ​​to obtain the dispersion of the set of measured temperature values. Other key state parameters such as dimensional allowances and hardness are calculated using the same method to determine their respective standard deviations.

[0063] Step S0024: Calculate the difference between the measured values ​​of each key state parameter when each target member leaves the current process and the standard entry requirements of each key state parameter for the next process, and divide by the first standard deviation to obtain the flow state deviation of each target member.

[0064] It should be noted that the flow state deviation is used to quantify the degree to which each target rod deviates from the standard entry requirement of the next process in a certain key state parameter. It is measured in units of standard deviation, so that deviations of different parameters and different dimensions can be compared on the same scale.

[0065] For example, the current process is rough turning, and the next process is fine grinding. The key state parameter is temperature. If the measured temperature of the target rod G001 is 32 degrees Celsius, and the standard inlet temperature requirement for the fine grinding process is 30 degrees Celsius, the difference between the two is 2 degrees Celsius. The first standard deviation of the temperature parameter calculated in step S0023 is 3 degrees Celsius. Then, the flow state deviation of the target rod G001 in terms of temperature parameter is 2 divided by 3, which is about 0.67 standard deviations. This means that the temperature of the rod deviates from the standard inlet requirement by about 0.67 times the standard deviation range.

[0066] It is important to note that for the same parameter, if the measured value is higher than the standard entry requirement, the difference is positive, and the flow state deviation is positive; if the measured value is lower than the standard entry requirement, the difference is negative, and the flow state deviation is negative. The positive and negative signs only indicate the direction of deviation (too high or too low). For other key state parameters such as dimensional allowance and hardness, the corresponding flow state deviations are calculated separately in the same way. For example, the flow state deviation of target rod G001 in terms of dimensional allowance parameter is the difference between the measured dimensional allowance value of the rod and the standard entry requirement value of the dimensional allowance for the fine grinding process, divided by the standard deviation of the dimensional allowance parameter to obtain the flow state deviation corresponding to the dimensional allowance. Each key state parameter has its own independent flow state deviation value, which is not confused with each other.

[0067] The larger the deviation value, the further the target member deviates from the standard entry requirement in terms of the critical state parameter, and the greater the risk of quality defects when entering the next process.

[0068] Step S0025: Calculate the mean of the absolute values ​​of the flow state deviations of all target rods, and use it as the current process flow state deviation rate corresponding to each key state parameter.

[0069] It should be noted that the current process flow-through component status deviation rate is used to characterize the average deviation of a certain key status parameter across all flow-through target components. Its value directly reflects the consistency of the current process's processing quality on that parameter. The larger the value, the further the measured value of that parameter deviates from the standard entry requirement, and the more unstable the processing status of the current process.

[0070] Specifically, for each critical state parameter, the absolute value of the flow state deviation of all target rods on that parameter is taken one by one, and then these absolute values ​​are added together and divided by the total number of target rods. The average value obtained is the current process flow state deviation rate corresponding to that critical state parameter.

[0071] Taking temperature parameters as an example, the current process flows through three target rods G001, G002, and G003. The differences between their measured temperatures and the standard inlet requirements are 2 degrees Celsius, 5 degrees Celsius, and -1 degree Celsius, respectively. In step S0024, these differences are divided by the standard deviation of the temperature parameter, 3 degrees Celsius, respectively, yielding flow status deviations of approximately 0.67, 1.67, and -0.33. Taking the absolute values, these are 0.67, 1.67, and 0.33, respectively. The average of these three is approximately 0.89, meaning the current process flow status deviation rate for the temperature parameters is 0.89, indicating that the average deviation of this group of target rods from the standard inlet requirements in terms of temperature parameters is approximately 0.89 standard deviations. Similarly, the dimensional allowance parameter is calculated in the same way to obtain the dimensional allowance-related dimensional allowance deviation rate.

[0072] The aforementioned state deviation rate is calculated independently for each key state parameter. Each parameter, such as temperature, dimensional allowance, and hardness, has its own corresponding state deviation rate value, which are not confused with each other.

[0073] Step S0026: Calculate the weighted arithmetic mean of the current process flow status deviation rates corresponding to all key status parameters to obtain the comprehensive average deviation rate of the current process flow.

[0074] It should be noted that the comprehensive average deviation rate is used to aggregate the deviation rates of multiple key state parameters such as temperature, dimensional allowance, and hardness of the components in the current process into a single comprehensive index by weighted arithmetic average. This comprehensively measures the overall deviation of all target components in the current process from the next process in multiple quality dimensions.

[0075] Specifically, first, the deviation rate of the current process flow item corresponding to each key status parameter is obtained. Then, the deviation rates are weighted and summed (i.e., weighted arithmetic mean) according to preset weighting coefficients. The result is the comprehensive average deviation rate of the current process flow. The weighting coefficient for each key status parameter is preset by the process engineer based on the parameter's influence on the processing quality of the next process; parameters with higher influence have greater weights. The sum of the weighting coefficients for all key status parameters equals 1.

[0076] For example, the deviation rate of the component state for temperature parameter is 0.89, for dimensional allowance parameter it is 0.61, and for hardness parameter it is 0.43. If the three parameters are weighted equally, with each parameter having a weight of one-third, the overall average deviation rate is the sum of the three divided by three, resulting in approximately 0.64. If the process engineer judges based on experience that temperature parameter has the greatest impact on the processing quality of the finishing grinding process, followed by dimensional allowance, and then hardness has a relatively small impact, the weights can be set as 0.50 for temperature, 0.30 for dimensional allowance, and 0.20 for hardness. The overall average deviation rate is then 0.89 multiplied by 0.50 plus 0.61 multiplied by 0.30 plus 0.43 multiplied by 0.20, resulting in approximately 0.71. The larger the above-mentioned overall average deviation rate, the further the overall state of the component in the current process deviates from the standard entry requirements of the next process, the worse the inter-process coordination, and the more likely it is to cause subsequent quality defects.

[0077] Step S003: Using the actual production cycle time of the current process corresponding to each virtual category and the standard average cycle time of the next process, combined with the comprehensive average deviation rate of the current process flow, calculate the degree of synergy between the current process and the next process corresponding to each virtual category.

[0078] It should be noted that the overall average deviation rate reflects the degree of deviation in the quality status of the components in the current process, i.e., whether the components meet the entry requirements of the next process. However, the coordination between processes depends not only on the matching of quality status but also on the matching of production capacity. If the production speed of the current process is inconsistent with the processing speed of the next process, even if the quality status of each component is qualified, it will inevitably lead to buffer overflow or equipment waiting for materials, which will also interrupt the continuity of production and cause unexpected downtime.

[0079] Step S003 specifically includes steps S0031-S0034: Step S0031: Obtain the actual production cycle time of the current process corresponding to the current virtual category, and the standard average cycle time of the next process for processing the current virtual category.

[0080] It should be noted that the actual production cycle time refers to the average time required to process each piston rod under the current production conditions in the current process, expressed in minutes per piece. This time is collected and calculated in real-time by the programmable logic controller (PLC) on the equipment. Specifically, the PLC records the start and end times of processing each rod on the current process equipment, subtracting the two times to obtain the single-piece processing time for that rod. When the cumulative processing reaches a preset sample size, the system uses the average single-piece processing time of the most recently completed preset number of rods as the actual production cycle time at the current moment, eliminating the interference of outliers on cycle time evaluation. This preset sample size is set by default in the system, with a default value of 50 pieces, and can be modified by process engineers on the configuration interface based on actual production cycle time fluctuations. The calculated actual production cycle time is uploaded to the scheduling server in real-time via an edge gateway.

[0081] The standard average cycle time refers to the standard time required for the next process to process each piston rod under normal operating conditions. The unit is minutes per piece. It is pre-determined by the process engineer based on the rated processing capacity of the equipment used in the next process, the process parameters of the cutting tools or grinding wheels, and the standard process procedure for the piston rod in that process. It is stored in the process standard process parameter library of the manufacturing execution system.

[0082] For example, for virtual category C1, if the current process is rough turning, the programmable controller records the time taken per piece for the 50 most recently processed rods in this process. The average of these 50 values ​​is taken as the actual production cycle time for the rough turning process, which is measured to be 2.0 minutes / piece. If the next process is fine grinding, the standard average cycle time for a single machine in the fine grinding process stored in the manufacturing execution system is 1.5 minutes / piece. The actual production cycle time of the rough turning process is 2.0 minutes / piece, while the standard average cycle time of the fine grinding process is 1.5 minutes / piece. The former is greater than the latter, indicating that the output speed of the current process is slower than the average processing capacity of the next process, and there is a risk of feed shortage.

[0083] Step S0032: Calculate the ratio of the actual production cycle time to the standard average cycle time, and use it as the first ratio.

[0084] It should be noted that this ratio reflects the relative size of the actual output capacity of the current process to the processing capacity of the next process. When the first ratio is greater than 1, it means that the average time for the current process to process one piston rod is greater than the standard time for the next process to process one piston rod. In other words, the output speed of the current process is slower than the processing speed of the next process, resulting in a supply shortage and a risk of the next process having to stop due to material shortage. When the first ratio is equal to 1, it means that the actual production cycle of the current process is completely consistent with the standard average cycle of the next process, the capacity of the preceding and following processes is perfectly matched, and the production line connection is the smoothest. When the first ratio is less than 1, it means that the average time for the current process to process one piston rod is less than the standard time for the next process to process one piston rod. In other words, the output speed of the current process is faster than the processing speed of the next process, resulting in a supply exceeding demand. This poses a risk of material blockage due to buffer accumulation. If the buffer capacity is limited, the overflowing work-in-process will force the preceding process to slow down or stop.

[0085] Step S0033: Calculate the difference between 1 and the overall average deviation rate of the current process flow, and use it as the first difference.

[0086] It should be noted that the first difference is used to quantify the impact of the quality status of the current process flow on the inter-process coordination. Specifically, the first difference is obtained by subtracting the overall average deviation rate of the current process flow from the value 1.

[0087] The overall average deviation rate reflects the degree to which the current process's components deviate from the standard entry requirements of the next process in multiple key state parameters. A larger value indicates a worse quality state of the components and a greater negative impact on the processing quality of the next process. Therefore, subtracting the deviation rate from 1 results in a larger difference, indicating a better quality state of the components and a greater contribution to the synergy; a smaller difference indicates a worse quality state and a stronger weakening effect on the synergy. When the overall average deviation rate is 0, it means that the measured values ​​of all target components are completely consistent with the standard entry requirements. In this case, the first difference is 1, and there is no reduction in synergy. When the overall average deviation rate is greater than 0 and less than 1, the first difference is between 0 and 1, and the synergy decreases proportionally. When the overall average deviation rate is greater than or equal to 1, the first difference is less than or equal to 0, indicating that the deviation has become unacceptable, and forced transfer will lead to batch quality accidents.

[0088] Step S0034: The product of the first ratio and the first difference is determined as the degree of collaboration between the current process and the next process corresponding to the current virtual classification.

[0089] It should be noted that the degree of synergy is a comprehensive indicator used to quantify the smoothness of the connection between the current process and the next process. Its value reflects the overall level of synergy between the preceding and following processes in terms of capacity matching and quality matching.

[0090] The two independent evaluation dimensions of capacity matching and quality matching are integrated into a single comprehensive index through multiplication. Deterioration in the performance of either dimension will lead to a proportional decrease in the degree of synergy.

[0091] When the collaboration correlation degree is greater than 1, it indicates that the current process and the next process have matching capacity and qualified quality, and the collaboration between processes is good. The current flow rhythm can continue without scheduling intervention. When the collaboration correlation degree is approximately equal to 1, it indicates that the processes are in a critical state. The system suggests paying attention but will not trigger intervention for the time being. When the collaboration correlation degree is less than 1, it indicates insufficient capacity or excessive quality deviation, and there is a risk of disconnection between processes. The system automatically marks this virtual category as having insufficient collaboration problems at the current process level.

[0092] For example, if the first ratio is 1.33 (indicating a risk of food shortage) and the first difference is 0.29 (indicating a serious quality deviation), then the degree of synergy is 1.33 multiplied by 0.29 equals 0.386. This value is much less than 1, indicating that there are serious problems in both the capacity and quality dimensions, and the synergy between processes is extremely poor. It is advisable to prioritize setting up a breakpoint process at this location for intervention.

[0093] Step S004: Based on the degree of collaboration between the current process and the next process corresponding to each virtual category, calculate the processing advantage of each virtual category in the current process, and determine the breakpoint process of each virtual category according to the changing trend of the processing advantage.

[0094] It should be noted that while synergy reflects the smoothness of connection between any two adjacent processes, relying solely on the synergy of adjacent processes, even if the synergy of each pair of adjacent processes is within an acceptable range individually, the cumulative decline in the synergy of multiple processes may still lead to a concentrated outbreak of quality risks in the final process. Therefore, it is necessary to take a more macroscopic perspective of "the overall processing potential from the current process to the final process," comprehensively evaluate the synergy of the current process and all subsequent processes, and quantify the feasibility of each process as a segmentation adjustment point.

[0095] Step S004 specifically includes steps S0041-S0044: Step S0041: For the process segment from the current process to the last process corresponding to the current virtual classification, obtain the variance of the collaborative correlation degree between all two adjacent processes within the process segment, and use it as the first variance.

[0096] It should be noted that the first variance is used to quantify the degree of fluctuation in the degree of collaboration between adjacent processes within a process segment from the current process to the last process, that is, the stability of the level of collaboration within that process segment.

[0097] Specifically, the algorithm extracts all collaboration correlation values ​​corresponding to the process segment from the current process to the last process (i.e., the collaboration correlation between the current process and the next process, the collaboration correlation between the next process and the next process after that, up to the collaboration correlation between the penultimate process and the last process). Then, it calculates the mean of all these collaboration correlation values, and then calculates the sum of squares of deviations of each collaboration correlation value from the mean. Finally, it divides the sum of squares of deviations by the total number of collaboration correlation values ​​within the process segment. The result is the first variance. The smaller this value, the more consistent the collaboration level among the processes within the process segment, indicating a stable workflow and no significant fluctuations in the smoothness of the connections between processes. The larger this value, the more uneven the collaboration level among the processes within the process segment, with at least one or more abnormal process pairs exhibiting significantly low or high collaboration.

[0098] Step S0042: Obtain the average value of the collaborative correlation degree between all two adjacent processes within the process segment, and use it as the first average value.

[0099] It should be noted that the first mean is used to quantify the overall level of collaboration between adjacent processes within a process segment from the current process to the last process, i.e., the central tendency of collaboration within that process segment. The larger the value, the higher the average smoothness of connection between adjacent processes within that process segment, and the better the overall collaboration level; the smaller the value, the lower the average smoothness of connection between adjacent processes within that process segment, and the worse the overall collaboration level.

[0100] Step S0043: Multiply the square root of the reciprocal of the sum of 1 and the first variance by the first mean to calculate the processing advantage of the current virtual classification in the current process.

[0101] It should be noted that the processing advantage score quantifies the feasibility of continuing processing from the current virtual category to the next process step. Its value comprehensively reflects the average level and stability of collaboration within the process segment from the current step to the last step. A higher processing advantage score indicates better overall conditions for continuing the flow from the current step, reducing the need for stopping for status adjustments; a lower processing advantage score indicates worse overall conditions for continuing the flow from the current step, making it more appropriate to consider setting a breakpoint process at that point.

[0102] The first mean reflects the overall level of synergy in the process segment; a higher mean indicates a higher degree of advantage. The first variance reflects the degree of fluctuation in the synergy of the process segment; a larger variance indicates a more unstable level of synergy and a more severe reduction in the degree of advantage. Multiplying the two, the processing advantage is maximized when the mean is high and the variance is small; when the mean is low or the variance is large, the processing advantage decreases accordingly.

[0103] Step S0044: If the processing advantage of the current virtual category in the next process is less than the processing advantage of the current virtual category in the current process, and the decrease exceeds the preset percentage threshold for the first time, the current process is determined as the breakpoint process of the current virtual category.

[0104] It should be noted that a breakpoint process refers to the process position in which the current virtual classification needs to pause and adjust its status during continuous processing. The determination is based on the changing trend of the processing advantage. When the processing advantage shows a significant decrease for the first time from a certain process, it indicates that the coordination level of the subsequent process segments has deteriorated significantly. Continuing to process will lead to an increase in quality risk. Therefore, a breakpoint should be set at the process position.

[0105] Specifically, starting from the first process in the current virtual classification, the system scans process by process. For each pair of adjacent processes (current process and next process), the processing advantage of each process is compared, and the rate of decrease is calculated. The rate of decrease is calculated as follows: the processing advantage of the current process minus the processing advantage of the next process, then divided by the processing advantage of the current process, yielding a percentage decrease. If the processing advantage of the next process is less than that of the current process, and the percentage decrease exceeds a preset percentage threshold for the first time, the current process is identified as a breakpoint process in the virtual classification, and the scanning process stops.

[0106] The preset percentage threshold is a percentage value pre-set by process engineers based on product accuracy requirements and actual production line operating experience, with a default value of 15%. This threshold should not be set too low (otherwise, the system will become overly sensitive, frequently triggering breakpoint judgments and increasing unnecessary scheduling interventions), nor should it be set too high (otherwise, the system will become sluggish and miss real disconnection risk points). The default value of 15% is an empirical value determined after comprehensively considering the general amplitude of equipment fluctuations and the cumulative characteristics of quality risks in piston rod processing. In actual applications, it can be dynamically adjusted based on the statistical results of production line operating data.

[0107] If no decrease exceeding the threshold is found after traversing all adjacent processes, the system automatically sets the last process as the default breakpoint process for that virtual category, ensuring that each virtual category has a clear adjustment node.

[0108] Step S005: Based on the deviation between the actual time and the planned time of each batch of production to be scheduled to arrive at the next process in each virtual category, and combined with the position of the breakpoint process of each virtual category in the total process, calculate the dynamic rescheduling trigger factor for each virtual category.

[0109] It should be noted that the breakpoint process indicates the location at the planning level where the current virtual classification should undergo state adjustment, i.e., "where to stop and adjust." However, in actual production, minute fluctuations such as spindle temperature rise, tool wear, and coolant temperature changes can cause the arrival time of each batch to deviate from the subsequent processes step by step. This time deviation accumulates and amplifies as the machining process progresses, often resulting in a discrepancy between the planned arrival time and the actual arrival time of the breakpoint process. If adjustments are performed at the breakpoint process solely based on the static plan, without considering the actual time deviation, two situations may occur: first, the batch arrives at the breakpoint process before the adjustment is ready, causing equipment to wait; second, the batch arrives with a severe delay, exceeding the waiting time allowed by the process window, leading to further deterioration of the component's state. Therefore, a dynamic rescheduling trigger factor needs to be constructed, combining the actual time deviation with the breakpoint process location, so that the system can automatically trigger a rescheduling request when the deviation exceeds the virtual classification's tolerance, achieving a leap from "when to adjust according to the plan" to "dynamically deciding when to adjust based on the actual deviation."

[0110] Step S005 specifically includes steps S0051-S0057: Step S0051: For each batch to be scheduled in the current virtual category, obtain the actual time when the batch arrives at the next process and the planned time when the batch arrives at the next process.

[0111] It should be noted that: the actual time refers to the moment when the last piece of each batch of equipment to be scheduled actually arrives at the entrance of the next process. This is triggered by the position sensor located at the feed end of the equipment in the next process when it detects the arrival of the last piece of the batch. The time is recorded by the programmable controller and then uploaded to the scheduling server via industrial Ethernet, accurate to the second.

[0112] Planned time refers to the theoretical arrival time calculated by adding the planned start time of each batch in the first process to the standard average cycle time of each process. Specifically, the scheduling server starts with the planned start time of the first process of the batch and sequentially adds the standard average cycle time of the batch in each process until the entry point of the next process being examined. The resulting time is the planned arrival time of the batch in that process. This planned time is stored in the scheduling plan table of the manufacturing execution system and is pre-calculated and fixed by the scheduling server during the production scheduling phase.

[0113] For example, suppose that the planned start time of batch B001 in virtual category C1 for the first process (roughing process) is 8:00 AM, the standard average cycle time for the roughing process is 10 minutes, and the next process is the semi-finishing process. If we are currently considering the planned start time of the semi-finishing process (the next process), then the planned time is 8:00 AM plus 10 minutes, that is, 8:10 AM.

[0114] Step S0052: Calculate the absolute difference between the actual time and the planned time for the batch to arrive at the next process, and use it as the first absolute difference for the batch.

[0115] It should be noted that the first absolute difference is the absolute value of the deviation between the actual time and the planned time when the same batch arrives at the same process, which is used to quantify the time offset of the batch in the current circulation stage.

[0116] Step S0053: Calculate the mean of the first absolute differences of all batches as the production cycle deviation of the current virtual classification.

[0117] It should be noted that the production cycle deviation is used to quantify the degree to which the current virtual classification deviates from the plan in the time dimension.

[0118] Specifically, the first absolute difference of each batch to be scheduled within the virtual category is summed, and then divided by the total number of batches within the virtual category. The result is the production cycle deviation of the virtual category. This value is in minutes. The larger the value, the more serious the deviation of the virtual category from the plan, indicating that the deviation between the actual operating status of the equipment and the planned baseline is already quite significant, and it is urgent to determine whether to trigger rescheduling. The smaller the value, the more the virtual category is basically progressing according to the planned rhythm, and the flow between processes is smooth.

[0119] Step S0054: Obtain the maximum time deviation of the current production line.

[0120] It should be noted that the maximum time deviation refers to the maximum tolerable time deviation of each virtual category relative to the planned timetable, which is allowed by the current production line, and the unit is minutes.

[0121] Specifically, the maximum time deviation is determined by both the production line buffer capacity and the inter-process buffering capacity. A larger buffer capacity allows for a greater number of work-in-process items and a wider allowable time fluctuation range; conversely, a smaller buffer capacity results in a lower tolerance for time deviations. In this embodiment, the maximum time deviation is calculated based on the physical capacity of the production line buffer and the average processing time per unit in the next process. Specifically, it is calculated as the maximum number of work-in-process items the buffer can hold multiplied by the processing time per unit in the next process, then divided by 2. This value is preset by the process engineer based on production line design parameters and actual operating experience and stored in the production line configuration parameter table of the manufacturing execution system. If the production line process layout or buffer configuration changes, this value can be adjusted accordingly by the process engineer in the system.

[0122] For example, if the current production line has a buffer zone before the fine grinding process, which can accommodate up to 20 piston rods simultaneously, and the single-piece processing time for the fine grinding process is 1.5 minutes, then the maximum time deviation is 20 multiplied by 1.5 divided by 2, which equals 15 minutes. This value indicates that the production line allows the arrival time of work-in-process before the fine grinding process to fluctuate by a maximum of 15 minutes relative to the plan. When the actual arrival time deviates from the planned time but the absolute value is less than 15 minutes, the buffer zone can absorb the fluctuation through its own work-in-process inventory and buffering capacity, and the system does not need to trigger rescheduling. When the deviation exceeds 15 minutes, the buffer zone will face the risk of overflow (arriving too early) or material shortage (arriving too late), at which point rescheduling needs to be triggered for intervention.

[0123] The maximum time deviation is a fixed configuration parameter at the production line level. Different processes or different virtual categories on the same production line share the same maximum time deviation value, which does not change with changes in virtual categories or processes.

[0124] Step S0055: Calculate the ratio of the current virtual classification's production cycle deviation to the current production line's maximum time deviation, and use this as the second ratio.

[0125] It should be noted that the second ratio is used to quantify the relative magnitude of the actual time deviation of the current virtual classification with respect to the maximum allowable deviation of the production line. The two quantities are divided in minutes to obtain a dimensionless value.

[0126] The smaller the ratio, the less severe the current offset, and the more capable the production line is of handling it; the larger the value, the more severe the current offset, and the less capable the production line is of absorbing the fluctuation.

[0127] When the second ratio is less than 1, it indicates that the current virtual classification time deviation has not exceeded the maximum allowable deviation range of the production line and is still within the absorbable range of the production line's buffer and buffer capacity; the system does not need to intervene immediately. When the second ratio is equal to 1, it indicates that the current virtual classification time deviation has just reached the maximum allowable deviation limit of the production line and is in a critical state of the tolerable range; monitoring should be strengthened. When the second ratio is greater than 1, it indicates that the current virtual classification time deviation has exceeded the maximum allowable deviation range of the production line, and the buffer is about to face the risk of overflow or material shortage; a rescheduling process should be triggered for intervention.

[0128] Step S0056: Use the ratio of the number of the breakpoint process in the current virtual category to the total number of processes in the current virtual category as the third ratio.

[0129] It should be noted that the third ratio is used to characterize the relative position of the current virtual classification breakpoint process in the entire processing flow, reflecting the proportion of processes that have been completed in this virtual classification.

[0130] The breakpoint process numbers are sequentially assigned starting from the first process, numbered 1, with subsequent processes numbered sequentially. The total number of processes is the total number of processes required for the virtual category to complete all processing. The third ratio ranges from 0 to 1 as a positive number. A larger value indicates that the breakpoint process is closer to the end of the processing flow, meaning that the current virtual category is already in the later stage of processing when the rescheduling decision is triggered; a smaller value indicates that the breakpoint process is closer to the beginning of the processing flow, meaning that the current virtual category is still in the early stage of processing.

[0131] Step S0057: Multiply the sum of 1 and the third ratio by the second ratio to calculate the dynamic rescheduling trigger factor for the current virtual classification.

[0132] It should be noted that the dynamic rescheduling trigger factor is a dimensionless quantitative indicator used to comprehensively evaluate the urgency of triggering rescheduling in the time dimension of the current virtual classification. Its value is determined by two factors: the degree of time deviation and the location of the breakpoint process.

[0133] The time deviation degree (second ratio) and the breakpoint process location (third ratio) are integrated into a single comprehensive index through multiplication. The time deviation degree reflects the severity of the current time offset, while the breakpoint location coefficient reflects the sensitivity of the current processing stage to the time offset. The closer the breakpoint is to the end of processing, the more serious the impact can be, even if the time offset is small. When the trigger factor is greater than 1, it indicates that the time offset of the current virtual classification has exceeded the tolerable range, and the system automatically triggers a rescheduling request. When the trigger factor is less than or equal to 1, it indicates that the time offset of the current virtual classification is still within the tolerable range, and the system continues to maintain the current scheduling plan unchanged.

[0134] Step S006: When multiple virtual categories contend for the same device, determine the competition priority of each virtual category based on the dynamic rescheduling trigger factor of each virtual category and the parameter similarity between each virtual category and the virtual category being processed by the device.

[0135] It should be noted that the dynamic rescheduling trigger factor provides each virtual category with an independent basis for determining whether it needs rescheduling. When the trigger factor is greater than 1, the virtual category sends a rescheduling request to the scheduling server. However, in actual production, multiple virtual categories may trigger rescheduling requests within similar time periods (for example, multiple virtual categories may trigger over-limits successively due to their respective time offsets, all waiting for the release of the same critical equipment). In this case, an arbitration mechanism is needed to determine which virtual category has priority in obtaining the right to use the equipment.

[0136] While sorting virtual categories solely by trigger factor magnitude can reflect their time urgency, it overlooks the changeover adjustment time when switching between different virtual categories. If the next virtual category to be processed differs significantly in processing parameters while the equipment is currently processing a particular category, it's necessary to change fixtures, adjust tool compensation parameters, and modify the processing program, significantly increasing changeover time. If two virtual categories have similar urgency, selecting the category closer to the current equipment status parameters as the next processing target can effectively shorten changeover time and improve overall equipment utilization.

[0137] Step S006 specifically includes steps S0061-S0066: Step S0061: For the two virtual categories that are competing for the equipment, other than the virtual category that the equipment is currently processing, designate one of them as the first virtual category and the other as the second virtual category.

[0138] It should be noted that: the "equipment currently processing virtual classification" refers to the virtual classification that is currently occupying the critical equipment, that is, the virtual classification in which the equipment is currently performing processing tasks. The feature vector of this virtual classification, as a representation of the current state of the equipment, has been stored in the equipment status table of the scheduling server. It is updated by the programmable controller on the equipment side after the last type change is completed and reported to the scheduling server. It is also refreshed synchronously when the virtual classification completes all processing tasks and the equipment switches to the next virtual classification.

[0139] Two virtual categories vying for the equipment refer to any two virtual categories that, at the current moment or within a preset time window (default 30 seconds to 5 minutes), have both issued rescheduling requests (i.e., their respective dynamic rescheduling trigger factors are greater than 1) and both have the same critical equipment as the target processing equipment. When three or more virtual categories are vying for the same equipment, the system arbitrates according to the principle of "pairwise comparison and round-by-round elimination": two virtual categories that have not yet completed the comparison are randomly selected as the first virtual category and the second virtual category, respectively. The competition priority between the two is calculated through the process from step S0066 to determine the winner; then the winner is paired with the next virtual category that has not yet been compared, and the above process is repeated until all virtual categories vying for the equipment have participated in the comparison, and the final winning virtual category obtains the priority right to use the equipment.

[0140] Specifically, two virtual categories are selected from the set of candidate virtual categories competing for the same device. One of them is designated as the first virtual category, and the other as the second virtual category. The naming of the first and second virtual categories is only for distinguishing different objects and does not imply superiority or priority.

[0141] For example, suppose the device is currently processing virtual category Cd. At this time, the dynamic rescheduling trigger factors of virtual categories Ca, Cb, and Cc are all greater than 1, and all request to use the device. The system first selects Ca and Cb for pairing and comparison, and sets Ca as the first virtual category and Cb as the second virtual category.

[0142] Step S0062: Calculate the parameter similarity between the first virtual classification and the virtual classification of the equipment in the process of processing, and use it as the first similarity.

[0143] It should be noted that the first similarity score is used to quantify the overall similarity between the first virtual classification and the virtual classification currently being processed by the equipment in terms of processing parameters, and its value ranges from 0 to 1. The closer the similarity score is to 1, the closer the processing parameters of the first virtual classification and the virtual classification currently being processed by the equipment are, and the less adjustment is required when switching from the current processing state of the equipment to the first virtual classification; the closer the similarity score is to 0, the greater the difference in processing parameters between the two, and the longer the adjustment time is.

[0144] Specifically, firstly, the average demand feature vector of the virtual category of equipment processing is obtained. This vector is obtained by taking the arithmetic mean of the standardized demand feature vectors of all batches within this virtual category along its dimensions. Then, the average demand feature vector of the first virtual category is obtained, and the Euclidean distance between the average demand feature vector of the first virtual category and the average demand feature vector of the virtual category of equipment processing is calculated in the same way as in step S0013. Then, the reciprocal of the sum of 1 and the Euclidean distance is used as the parametric similarity between the two. This calculation method is exactly the same as the batch similarity calculation method in step S0014, the only difference being that the calculation object is changed from "two batches" to "one virtual category and the virtual category of equipment processing".

[0145] Step S0063: Calculate the parameter similarity between the second virtual classification and the virtual classification of the equipment in the process of processing, and use it as the second similarity.

[0146] Specifically, the second similarity is obtained in exactly the same way as in step S0062.

[0147] Step S0064: Calculate the ratio of the dynamic rescheduling trigger factor of the first virtual classification to the dynamic rescheduling trigger factor of the second virtual classification, and use it as the fourth ratio.

[0148] It should be noted that the fourth ratio quantifies the relative time urgency of the first virtual classification compared to the second virtual classification. A fourth ratio greater than 1 indicates that the dynamic rescheduling trigger factor for the first virtual classification is greater than that for the second, meaning the first virtual classification has a higher time urgency at the current moment, and its request should be responded to first from a time perspective. A fourth ratio less than 1 indicates that the dynamic rescheduling trigger factor for the second virtual classification is greater than that for the first, meaning the second virtual classification has a higher time urgency. A fourth ratio equal to 1 indicates that both virtual classifications have the same time urgency.

[0149] Step S0065: Calculate the fifth ratio by taking the sum of 1 and the first similarity as the numerator and the sum of 1 and the second similarity as the denominator.

[0150] It should be noted that the fifth ratio is used to quantify the relative advantage of the first virtual category over the second virtual category in terms of changeover adjustment costs. When the ratio is greater than 1, it means that the processing parameters of the first virtual category are closer to the current state of the equipment, and the changeover adjustment time to the first virtual category is shorter. When the ratio is less than 1, it means that the processing parameters of the second virtual category are closer to the current state of the equipment, and the changeover adjustment time to the second virtual category is shorter. When the ratio is equal to 1, it means that the two virtual categories are equally close to the processing parameters of the current state of the equipment, and the changeover adjustment time is comparable.

[0151] The "add 1" operation in the numerator and denominator ensures that when the similarity value is small, the ratio will not become abnormal as the denominator approaches zero. At the same time, it physically reflects the non-linear effect of similarity on the production adjustment time. When the similarity is close to 1, the value after adding 1 is approximately equal to 2, which makes the influence weight of the affinity term reasonably amplified. When the similarity is close to 0, the value after adding 1 is approximately equal to 1, avoiding numerical instability caused by an excessively small denominator.

[0152] Step S0066: The product of the fourth ratio and the fifth ratio is determined as the competitive priority of the first virtual classification relative to the second virtual classification.

[0153] It should be noted that the competition priority is used to comprehensively measure the priority of the first virtual category in obtaining the right to use the equipment relative to the second virtual category. Its value is determined by two dimensions: the urgency of time and the cost of production adjustment.

[0154] When the competition priority is greater than 1, it means that the first virtual category is superior to the second virtual category in the comprehensive evaluation of time urgency and production adjustment costs, and should be given priority to obtain the right to use the equipment; when the value is less than 1, it means that the second virtual category is superior in the overall evaluation and should be given priority to obtain the right to use the equipment; when the value is equal to 1, it means that the two virtual categories are comparable in the comprehensive evaluation, and the winner is determined according to the preset rules (such as first trigger first response or random selection).

[0155] The core of competitive priority lies in the physical meaning of the multiplicative relationship: the ratio of time urgency (fourth ratio) and the ratio of production affinity (fifth ratio) are multiplicative rather than additive. This means that a significant difference in either dimension will cause the competitive priority to decrease exponentially. If the production adjustment cost of the first dummy category is much higher than that of the second dummy category (the fifth ratio is significantly less than 1), even if its time urgency is slightly higher (the fourth ratio is slightly greater than 1), the multiplied competitive priority may still be less than 1, indicating that the high production cost is sufficient to offset the advantage of time urgency. Conversely, if the time urgency of the first dummy category is much higher than that of the second dummy category (the fourth ratio is significantly greater than 1), even if its production adjustment cost is slightly higher (the fifth ratio is slightly less than 1), the multiplied competitive priority may still be greater than 1, indicating that the high time urgency is sufficient to offset the disadvantage of production cost. This multiplicative design allows the system to flexibly trade off between "time urgency" and "production cost".

[0156] Step S007: Determine the arbitration result based on the competition priority of each virtual category, generate scheduling instructions based on the arbitration result, and send them to the equipment for execution.

[0157] Specifically, step S007, which determines the arbitration result based on the competition priority of each virtual category, includes: When the competition priority of the first virtual category relative to the second virtual category is greater than 1, the first virtual category is determined to have the right to use the equipment.

[0158] When the competition priority of the first virtual category relative to the second virtual category is less than or equal to 1, the second virtual category is determined to have the right to use the equipment.

[0159] Specifically, step S007, which generates a scheduling instruction based on the arbitration result and sends it to the device for execution, includes steps S0071-S0073: Step S0071: Convert the arbitration result into a scheduling instruction; wherein the scheduling instruction includes the machining program number, tool compensation parameters and tooling fixture number corresponding to the virtual classification that has obtained the right to use the equipment.

[0160] It should be noted that: the arbitration result refers to the virtual classification number for obtaining the right to use the equipment, as determined in step S0066. The scheduling instruction refers to the structured control message issued by the scheduling server to the programmable controller or CNC system on the device side.

[0161] Specifically, the scheduling server, based on the winning virtual classification number in the arbitration result, uses that number as an index to query the process document mapping table pre-installed in the Manufacturing Execution System (MES) to obtain the corresponding machining program number (used by the CNC system to select the correct machining program), tool compensation parameters (including tool radius compensation value, tool length compensation value, and tool tip radius compensation value, etc., used to ensure machining accuracy), and tooling fixture number (used by the automated loading and unloading robot to pick up the correct fixture from the quick-change tooling library). These three parameters are encapsulated in a structured message format, containing the device identifier, instruction type ("changeover"), virtual classification number, and the specific values ​​of the three parameters. This message is transmitted via industrial Ethernet to the corresponding device's edge gateway for parsing and execution. The process document mapping table is pre-configured by process engineers before production scheduling based on the machining requirements of each virtual classification, stored in the MES, and synchronized to the scheduling server's local cache with each production scheduling update to ensure rapid response when changeover instructions are issued. Through this transformation, the arbitration result is converted from a logical judgment of "which virtual classification wins" into a set of physical control parameters that the equipment can directly recognize and execute.

[0162] Step S0072: Send the scheduling instruction to the programmable controller or CNC system of the corresponding equipment to force the equipment to load the process file of the specified virtual category.

[0163] It should be noted that: "Deployment" refers to the process by which the scheduling server transmits the scheduling instruction message generated in step S0071 to the corresponding device edge gateway via industrial Ethernet (using Profinet or OPC UA protocol), and the edge gateway parses and forwards it to the device programmable controller or CNC system.

[0164] Forced loading refers to the automatic invocation of the corresponding machining program by the programmable logic controller (PLC) or CNC system after receiving a scheduling instruction, based on the machining program number in the instruction message. It also involves automatically updating the values ​​in the tool compensation register according to the tool compensation parameters, and driving an automated loading / unloading robot to retrieve the corresponding tooling from the quick-change tooling library based on the tooling fixture number via the equipment-side changeover control logic. The entire loading process requires no manual intervention, achieving an automated closed loop from scheduling decision-making to equipment execution.

[0165] Step S0073: Real-time acquisition of completion signals and actual production cycle time, which are transmitted back to the scheduling server via the edge gateway, compared with the planned values, and the dynamic rescheduling trigger factors are updated; wherein, the scheduling server scans all dynamic rescheduling trigger factors of the in-production virtual category at fixed intervals, and when any dynamic rescheduling trigger factor is greater than 1, the rescheduling process is triggered.

[0166] It should be noted that the completion signal refers to the digital trigger signal generated by the programmable controller of the equipment when each rod is processed and discharged from the equipment. It is generated by the position sensor or proximity switch installed at the discharge end of the equipment when it detects the discharge of the rod. After being timestamped by the programmable controller, it is uploaded to the scheduling server through the edge gateway.

[0167] Specifically, the programmable controller records the start time when processing each link begins and the end time when the completion signal is triggered. The difference between the two times is the actual processing time per link. The scheduling server, using the equipment as a dimension and the actual processing time of the most recently completed preset number of links (default 50) as samples, calculates a moving average as the current actual production cycle time for the equipment, and transmits this value back to the scheduling server via the edge gateway. The scheduling server compares the transmitted actual production cycle time with the planned value stored in the manufacturing execution system cycle by cycle. If the deviation between the actual cycle time and the planned value exceeds a preset threshold (default ±10%), a cycle time anomaly warning is triggered. Simultaneously, the actual cycle time value is substituted into the deviation calculation in step S0053 to dynamically update the production cycle time deviation and dynamic rescheduling trigger factor for this virtual category.

[0168] The scheduling server scans all dynamic rescheduling trigger factors for in-production virtual categories at a fixed period (default 30 seconds). The scanning method is as follows: the scheduling server traverses the trigger factor storage table for all currently in-production virtual categories, reads the latest trigger factor value for each virtual category, and compares it with a threshold of 1. If the trigger factors for all virtual categories are less than or equal to 1, the system maintains the current scheduling plan without any intervention. When any dynamic rescheduling trigger factor is greater than 1, the system immediately triggers the rescheduling process, re-executing the decision chain from steps S002 to S006: recalculating the overall average deviation rate, collaborative correlation, processing advantage, breakpoint process location, and trigger factor for that virtual category, and re-arbitrating the multi-virtual category equipment contention based on the updated data, generating a new scheduling instruction and issuing it to the equipment for execution. This scanning, comparison, and triggering mechanism operates continuously at a fixed period, ensuring the system can respond promptly to dynamic changes in actual production, achieving continuous closed-loop monitoring from data acquisition and status assessment to scheduling decisions.

[0169] In summary, in this embodiment of the invention, by quantitatively evaluating the inter-process collaborative correlation, the segmentation point can be accurately identified based on the deviation between the measured value of the key state parameter when the target component leaves the current process and the standard entry requirement value of the next process, combined with the matching degree of the cycle time of the preceding and following processes. This allows for early prediction of the risk of disconnection before the state deviation accumulates to the quality defect threshold, effectively reducing rework or scrap due to state discrepancies. Furthermore, by constructing a dynamic rescheduling trigger factor that integrates the degree of time offset and the location of the breakpoint process, virtual classifications in the later stages of processing automatically receive a lower trigger tolerance due to higher accumulated risk, achieving differentiated rescheduling triggers and overcoming the limitations of traditional static segmentation strategies. The system cannot handle actual fluctuations. By introducing parameter similarity between each category and the current processing state of the equipment as an arbitration dimension when multiple virtual categories contend for equipment, the system prioritizes the virtual category with shorter changeover adjustment time when the urgency is similar, significantly reducing the non-output time of the equipment. Furthermore, the scheduling instructions are directly sent to the equipment PLC or CNC system in the form of structured messages, and closed-loop control is formed with real-time data feedback from the equipment sensors. This achieves a second-level response from scheduling decision to equipment execution, ultimately resulting in significant improvements in overall equipment efficiency, product yield, and delivery cycle. This provides a systematic intelligent scheduling solution for multi-process continuous processing scenarios of piston rods.

[0170] This invention also proposes an intelligent scheduling and optimization system for the machining process of hydraulic cylinder piston rods. Please refer to [link / reference]. Figure 2 The diagram illustrates a block diagram of an intelligent scheduling and optimization system for the machining process of a hydraulic cylinder piston rod according to an embodiment of the present invention. The system includes: The virtual classification acquisition module 100 is used to acquire the processing parameter features of multiple production batches to be scheduled, and to perform clustering processing on the multiple production batches to be scheduled based on the processing parameter features to acquire multiple virtual classifications. The virtual classification analysis module 200 is used to calculate the comprehensive average deviation rate of the current process flow based on the measured values ​​of the key state parameters of the target member in each virtual classification when it leaves the current process and the standard entry requirement value corresponding to the next process. By using the actual production cycle time of the current process corresponding to each virtual category and the standard average cycle time of the next process, combined with the comprehensive average deviation rate of the current process flow, the degree of synergistic correlation between the current process and the next process corresponding to each virtual category is calculated. Based on the degree of collaboration between the current process and the next process corresponding to each virtual category, the processing advantage of each virtual category in the current process is calculated, and the breakpoint process of each virtual category is determined according to the changing trend of the processing advantage. Based on the deviation between the actual time and the planned time of each batch of production to be scheduled to arrive at the next process in each virtual category, and combined with the position of the breakpoint process of each virtual category in the total process, the dynamic rescheduling trigger factor of each virtual category is calculated. When multiple virtual categories contend for the same device, the competition priority of each virtual category is determined based on the dynamic rescheduling trigger factor of each virtual category and the parameter similarity between each virtual category and the virtual category being processed by the device. The scheduling module 300 is used to determine the arbitration result based on the competition priority of each virtual category, generate scheduling instructions based on the arbitration result, and send them to the equipment for execution.

[0171] It should be noted that the system provided in the above embodiments is only an example of the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the computer device can be divided into different functional modules to complete all or part of the functions described above. In addition, the intelligent scheduling and optimization system for hydraulic cylinder piston rod processing provided in the above embodiments and the intelligent scheduling and optimization method for hydraulic cylinder piston rod processing are based on the same concept. The specific implementation process is detailed in the method embodiment and will not be repeated here.

[0172] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0173] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0174] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent scheduling and optimization of the machining process of a hydraulic cylinder piston rod, characterized in that, The method includes the following steps: The processing parameter features of multiple production batches to be scheduled are obtained, and the multiple production batches to be scheduled are clustered based on the processing parameter features to obtain multiple virtual categories; wherein, the virtual category is a set of production batches to be scheduled with similar processing requirements. Based on the measured values ​​of the key state parameters of the target component in each virtual classification when it leaves the current process, and the standard entry requirement value corresponding to the next process, calculate the comprehensive average deviation rate of the current process flow; whereby the comprehensive average deviation rate is a comprehensive index used to measure the degree to which the overall key state parameters of the components flowing in the current process deviate from the standard entry requirement of the next process. By using the actual production cycle time of the current process corresponding to each virtual category and the standard average cycle time of the next process, combined with the comprehensive average deviation rate of the current process flow, the degree of synergy between the current process and the next process corresponding to each virtual category is calculated; whereby, the degree of synergy is a comprehensive indicator used to quantify the smoothness of the connection between the current process and the next process. Based on the degree of synergy between the current process and the next process corresponding to each virtual category, the processing advantage of each virtual category in the current process is calculated, and the breakpoint process of each virtual category is determined according to the changing trend of the processing advantage. Among them, the processing advantage is an indicator used to measure the feasibility of continuing processing from the current process; the breakpoint process is the process position where the processing advantage decreases significantly. Based on the deviation between the actual time and the planned time of each batch of production to be scheduled to arrive at the next process in each virtual category, and combined with the position of the breakpoint process of each virtual category in the total process, the dynamic rescheduling trigger factor of each virtual category is calculated; among which, the dynamic rescheduling trigger factor is a quantitative indicator used to comprehensively evaluate the urgency of triggering rescheduling in the current virtual category. When multiple virtual categories contend for the same device, the competition priority of each virtual category is determined based on the dynamic rescheduling trigger factor of each virtual category and the parameter similarity between each virtual category and the virtual category being processed by the device. The arbitration result is determined based on the competition priority of each virtual category, and a scheduling instruction is generated and issued to the equipment for execution based on the arbitration result.

2. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 1, characterized in that, The specific steps involved in clustering multiple production batches based on processing parameter features to obtain multiple virtual classifications are as follows: Construct the original processing parameter feature vector for each production batch to be scheduled; wherein the original processing parameter feature vector shall include feature values ​​of at least five dimensions: nominal diameter, tolerance grade, processing length, hardness coefficient, and surface roughness. Normalize the feature values ​​of all production batches to be scheduled in each dimension to obtain the standardized demand feature vector of each production batch to be scheduled. For any two production batches to be scheduled, calculate the Euclidean distance between the standardized demand feature vectors of the two production batches. The reciprocal of the sum of 1 and the Euclidean distance is used as the similarity of processing requirements between the two production batches to be scheduled. When the similarity of processing requirements between two production batches to be scheduled is greater than or equal to a preset similarity threshold, the two production batches to be scheduled are classified into the same virtual category.

3. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 2, characterized in that, The specific steps for calculating the comprehensive average deviation rate of the current process flow based on the measured values ​​of the key state parameters when the target member in each virtual classification leaves the current process and the standard entry requirement value corresponding to the next process are as follows: Obtain the target rod that has been processed and transferred to the next process in the current process corresponding to the current virtual classification; Obtain the measured values ​​of each key state parameter when each target member leaves the current process, as well as the standard entry requirements for each key state parameter in the next process; Calculate the standard deviation of the measured values ​​of each key state parameter when all target members leave the current process, and use it as the first standard deviation; Calculate the difference between the measured values ​​of each key state parameter when each target member leaves the current process and the standard entry requirements of each key state parameter for the next process, and divide by the first standard deviation to obtain the flow state deviation of each target member; Calculate the mean of the absolute values ​​of the flow state deviations of all target rods, and use it as the flow state deviation rate of the current process corresponding to each key state parameter; The weighted arithmetic mean of the deviation rates of the current process flow parts corresponding to all key status parameters is calculated to obtain the comprehensive average deviation rate of the current process flow.

4. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 3, characterized in that, The specific steps for calculating the collaborative correlation between the current process and the next process corresponding to each virtual category, by using the actual production cycle time of the current process and the standard average cycle time of the next process, combined with the comprehensive average deviation rate of the current process flow, are as follows: Obtain the actual production cycle time of the current process corresponding to the current virtual category, and the standard average cycle time of the next process to process the current virtual category; Calculate the ratio of the actual production cycle time to the standard average cycle time, and use it as the first ratio. Calculate the difference between 1 and the overall average deviation rate of the current process flow, and use it as the first difference; The product of the first ratio and the first difference is determined as the degree of collaborative association between the current process and the next process corresponding to the current virtual classification.

5. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 4, characterized in that, The specific steps for calculating the processing advantage of each virtual category in the current process based on the collaborative correlation degree between the current process and the next process corresponding to each virtual category, and determining the breakpoint process for each virtual category based on the changing trend of the processing advantage degree, are as follows: For the process segment from the current process to the last process corresponding to the current virtual classification, obtain the variance of the collaborative correlation degree between all two adjacent processes within the process segment, and use it as the first variance; Obtain the average of the degree of collaboration between all two adjacent processes within this process segment, and use it as the first average. The square root of the sum of 1 and the first variance is multiplied by the first mean to calculate the processing advantage of the current virtual classification in the current process. If the processing advantage of the current virtual category in the next process is less than that of the current virtual category in the current process, and the decrease exceeds the preset percentage threshold for the first time, the current process will be determined as the breakpoint process of the current virtual category.

6. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 5, characterized in that, The process of calculating the dynamic rescheduling trigger factor for each virtual category, based on the deviation between the actual and planned times of each batch of production to be scheduled to the next process in each virtual category, and considering the position of the breakpoint process in the overall process, includes the following specific steps: For each batch of production to be scheduled within the current virtual category, obtain the actual time when the batch arrives at the next process and the planned time when the batch arrives at the next process; Calculate the absolute difference between the actual time and the planned time for the batch to arrive at the next process, and use this as the first absolute difference for the batch. Calculate the mean of the first absolute differences of all batches, and use it as the production cycle deviation of the current dummy classification. Get the maximum time deviation of the current production line; Calculate the ratio of the current virtual classification's production cycle deviation to the current production line's maximum time deviation, and use this as the second ratio; The ratio of the number of the breakpoint process in the current virtual classification to the total number of processes in the current virtual classification is used as the third ratio. Multiply the sum of 1 and the third ratio by the second ratio to calculate the dynamic rescheduling trigger factor for the current virtual classification.

7. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 6, characterized in that, When multiple virtual categories contend for the same device, the competition priority of each virtual category is determined based on the dynamic rescheduling trigger factor of each virtual category and the parameter similarity between each virtual category and the virtual category being processed by the device. The specific steps include the following: For the two virtual categories that are competing for the equipment, other than the virtual category that is currently processing the equipment, one of them will be designated as the first virtual category and the other as the second virtual category. Calculate the parameter similarity between the first virtual classification and the virtual classification of equipment in the current processing stage, and use it as the first similarity. Calculate the parameter similarity between the second virtual classification and the virtual classification of equipment in the current processing stage, and use it as the second similarity. Calculate the ratio of the dynamic rescheduling trigger factor of the first virtual classification to the dynamic rescheduling trigger factor of the second virtual classification, and use it as the fourth ratio; The fifth ratio is calculated by taking the sum of 1 and the first similarity as the numerator and the sum of 1 and the second similarity as the denominator. The product of the fourth ratio and the fifth ratio is determined as the competitive priority of the first virtual classification relative to the second virtual classification.

8. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 7, characterized in that, The specific steps for determining the arbitration result based on the competitive priority of each virtual category are as follows: When the competition priority of the first virtual category relative to the second virtual category is greater than 1, the first virtual category is determined to have the right to use the equipment. When the competition priority of the first virtual category relative to the second virtual category is less than or equal to 1, the second virtual category is determined to have the right to use the equipment.

9. The intelligent scheduling and optimization method for the machining process of the hydraulic cylinder piston rod according to claim 8, characterized in that, The specific steps involved in generating scheduling instructions based on the arbitration result and issuing them to the equipment for execution are as follows: The arbitration result is converted into a scheduling instruction; the scheduling instruction includes the processing program number, tool compensation parameters and tooling fixture number corresponding to the virtual classification that has obtained the right to use the equipment. The scheduling command is sent to the programmable controller or CNC system of the corresponding equipment, forcing the equipment to load the process file of the specified virtual category; The system collects completion signals and actual production rhythm in real time, and transmits them back to the scheduling server via the edge gateway. The data is then compared with the planned values ​​to update the dynamic rescheduling trigger factors. The scheduling server scans all dynamic rescheduling trigger factors for the virtual production category at fixed intervals. When any dynamic rescheduling trigger factor is greater than 1, the rescheduling process is triggered.

10. A smart scheduling and optimization system for the machining process of a hydraulic cylinder piston rod, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is executed by the processor, it implements the steps of the intelligent scheduling and optimization method for the machining process of a hydraulic cylinder piston rod as described in any one of claims 1-9.

Images