Production process station arrangement method and system based on big data

Abstract

The invention relates to the technical field of process management, and discloses a production process station arrangement method and system based on big data, and the method comprises the steps: obtaining process information, an employee expectation salary and an employee historical production record, mapping a process name into a process vector through a text embedding model, and generating a process classification name according to a similarity threshold; the method comprises the following steps: calculating employee production efficiency and employee production frequency according to employee historical production records and process classification names, forming an employee skill grade evaluation table, enabling skill evaluation to be stable under double calibers of category and statistics, dividing in-line process information and out-line process information according to in-line marks, and generating an in-line arrangement result; and calculating an off-line station threshold value according to standard working hours and employee expected salaries, comparing off-line station total working hours, executing process migration and employee adjustment in an over-limit manner, directly outputting an off-line arrangement result in a non-over-limit manner, and finally writing an in-line arrangement result and an off-line arrangement result into an MES database, thereby reducing arrangement iteration and improving scheme performability.

Description

Technical Field

[0001] This invention relates to the field of process management technology, and more specifically, to a method and system for arranging production process stations based on big data. Background Technology

[0002] In discrete manufacturing scenarios such as apparel, production organization typically needs to generate executable process and personnel station arrangement results on production line carriers such as hanging lines, facing constraints of multiple processes, multiple stations, and multiple personnel. Existing technologies mostly adopt a combination of rule matching and manual experience verification, supplemented by historical record statistical scoring, similar process merging, constraint screening, and conflict rollback, etc. Its applicable boundaries are usually based on the premise that the process system changes slowly, the skill level of personnel is relatively stable, and the availability of station equipment fluctuates little.

[0003] In actual operation, order styles and process combinations change frequently, and personnel proficiency and output performance fluctuate over time. These unstable factors can cause inconsistencies in skill assessments based on historical statistics and rule-based screening, leading to repeated rollbacks and adjustments during the programming process. This results in a decrease in the execution stability of programming results both inside and outside the line. Therefore, the technical problem that needs to be solved is how to improve the consistency of skill assessment and reduce the number of programming iterations to achieve stable and executable solutions both inside and outside the line.

[0004] In view of this, the present invention proposes a production process station arrangement method and system based on big data to solve the above problems. Summary of the Invention

[0005] To overcome the above-mentioned defects of the prior art, the present invention provides a production process station arrangement method and system based on big data.

[0006] To achieve the above objectives, the present invention provides the following technical solution: Firstly, a method for arranging production process stations based on big data is provided, including: Obtain process information, employee expected salary, and employee historical production records. Process information includes process name, line marking, standard working hours, and equipment type. The text embedding model is used to map the process name to the process vector, and the process vector is classified according to the preset similarity threshold to generate the process classification name; Employee production efficiency and frequency are calculated based on employees' historical production records and process classification names. Based on employee production efficiency and frequency, an employee skill level assessment table is generated. The process information is divided according to the markings within the line to obtain the process information within the line and the process information outside the line. The process information within the line is arranged according to the employee skill level assessment table, and the threshold for the station outside the line is calculated according to the employee's expected salary and standard working hours. The total working hours of off-line stations are calculated based on off-line process information and standard working hours. When the total working hours of off-line stations exceed the off-line station threshold, process migration and employee adjustment are performed, and off-line scheduling results are generated. When the total working hours of off-line stations do not exceed the off-line station threshold, off-line scheduling results are generated directly based on off-line process information. The on-line scheduling results and off-line scheduling results are written into the MES database.

[0007] In some embodiments, the process vectors are categorized according to a preset similarity threshold, and the generated process category names include: A text embedding model is used to map each process name to obtain a 384-dimensional process vector, and the process names and 384-dimensional process vectors are stored in a process vector table; the process names in the process vector table are deduplicated to obtain a set of deduplicated process names. Calculate the cosine similarity between each pair of process vectors in the deduplicated process name set to obtain a similarity mapping table; Process names that meet the preset similarity threshold in the similarity mapping table are grouped into the same category, and a process category name is determined for each category.

[0008] In some embodiments, the similarity threshold is 95%, and the cosine similarity is calculated from the angle between the 384-dimensional process vectors corresponding to the process names.

[0009] In some embodiments, grouping process names that meet a preset similarity threshold in the similarity mapping table into the same category includes: Before classifying based on the similarity mapping table, the process name set is first merged according to the literal consistency of the process name to form the first classification group set; For the representative process names of the first categorized group set, query the similarity mapping table to generate a group similarity matrix table. For the group similarity matrix table, perform cross-group merging on the group pairs that meet the preset similarity threshold to form a second categorized group set, and determine the process categorization name for the second categorized group set.

[0010] In some embodiments, the in-line arrangement results for generating in-line process information based on the employee skill level assessment form include: Obtain team information, divide the in-line process information into special process sets and non-special process sets based on in-line tags, associate the special process sets with the special station IDs in the team information, assign employee information to the special process sets according to the employee skill level assessment table, and generate special process arrangement results; The process retrieves the work ticket numbers corresponding to each process in the set of non-special processes, sorts the set of non-special processes in reverse order according to the work ticket numbers, and divides it into a set of combined processes and a set of non-combined processes. Based on the employee skill level assessment table, the process arrangement results of combined processes and non-combined processes are generated respectively, and together they form the in-line arrangement result.

[0011] In some embodiments, generating the combined process arrangement result includes: The screening and allocation steps are repeated until all combined processes have been allocated. The screening and allocation steps include: using consecutive work ticket numbers as the connection condition, screening candidate work sets that can be continuously completed by the same employee from the remaining unallocated combined work sets, sorting the candidate work sets according to the number of work processes, and selecting the candidate work set with the most work processes from the sorting results as the current target combined work set; Obtain the number of equipment types involved in the current target combination process set. If the number of equipment types is less than or equal to 3, assign employee information to the current target combination process set according to the employee skill level assessment table, and write the process number, production sequence number, employee number and station number of the current target combination process set into the in-line arrangement result. If the number of equipment types is greater than 3, mark the current target combination process set as a non-combination process or remove it from the candidate process set.

[0012] In some embodiments, the total working hours of off-line stations are calculated based on off-line process information and standard working hours, including: Filter the set of off-line processes with automatic template machine as the equipment type from the off-line process information, and adjust the standard working hours according to the average available working hours per person in the employee's historical production records; Based on the revised standard working hours, initial stations are allocated to the set of off-line processes, and the station working hours for each initial station are calculated. The station working hours are then used as the total working hours for off-line stations.

[0013] In some embodiments, when the total working hours at off-line stations exceed an off-line station threshold, the method for performing process migration and employee adjustments, and generating off-line scheduling results includes: The last inserted process is determined from the set of off-line processes allocated to the initial station where the total working hours of the off-line stations exceed the off-line station threshold. The last inserted process is then moved to the previous station that is idle or not fully loaded, while keeping the process number and production sequence number unchanged after the migration. Based on the employee skill level assessment table, the employees with the lowest skill level ranking are determined, and these employees are assigned to the last inserted process of the target station. The employee number and process information of the target station are updated, and the final off-line arrangement result is generated. The target station refers to the previous station that is in an idle state or not fully loaded.

[0014] Secondly, a production process station layout system based on big data is provided, which is used to implement the aforementioned production process station layout method based on big data, including: Data acquisition module: used to acquire process information, employee expected salary and employee historical production records. Process information includes process name, line marking, standard working hours and equipment type. Vector Classification Module: This module uses a text embedding model to map process names to process vectors, classifies the process vectors according to a preset similarity threshold, and generates process classification names. Skills assessment module: Used to calculate employee production efficiency and employee production frequency based on employee historical production records and process classification names, and generate employee skill level assessment table based on employee production efficiency and employee production frequency; The division and arrangement module is used to divide the process information according to the markings in the line, obtain the process information in the line and the process information outside the line, generate the in-line arrangement result of the process information in the line according to the employee skill level assessment table, and calculate the threshold of the station outside the line according to the employee's expected salary and standard working hours. Offline orchestration module: Used to calculate the total working hours of offline stations based on offline process information and standard working hours. When the total working hours of offline stations exceed the offline station threshold, process migration and employee adjustment are performed, and offline orchestration results are generated. When the total working hours of offline stations do not exceed the offline station threshold, offline orchestration results are generated directly based on offline process information, and the online orchestration results and offline orchestration results are written into the MES database.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention acquires process information, employee expected salaries, and employee historical production records. It uses a text embedding model to map process names to process vectors and categorizes them according to a preset similarity threshold to generate process classification names. Based on employee historical production records and process classification names, it calculates employee production efficiency and frequency, generating an employee skill level assessment table to reduce the impact of single fluctuations on skill assessments. It divides in-line and out-of-line process information based on in-line markers. First, it generates in-line scheduling results according to the employee skill level assessment table. Then, it calculates out-of-line station thresholds based on employee expected salaries and standard working hours, comparing them with the total out-of-line station working hours. If the threshold is exceeded, process migration and employee adjustment are performed; otherwise, out-of-line scheduling results are directly generated. The process classification name plus the employee skill level assessment table stabilizes skill assessments, while the out-of-line station threshold plus process migration and employee adjustment converges out-of-line load, reducing scheduling iterations. The in-line and out-of-line scheduling results are written to the MES database for stable execution. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a production process station arrangement method based on big data in this invention. Figure 2 This is a schematic diagram of a production process station arrangement system based on big data according to the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. In the following detailed description, many specific details are set forth to provide a thorough understanding of the exemplary embodiments described. However, it will be apparent to those skilled in the art that the described embodiments may be practiced without some or all of these specific details. In other exemplary embodiments, well-known structures have not been described in detail to avoid unnecessarily obscuring the concepts of this disclosure. It should be understood that the specific embodiments described herein are merely illustrative of the present invention and are not intended to limit the present invention. Furthermore, the various aspects described in the embodiments may be combined arbitrarily without conflict.

[0018] The user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0019] Example 1 Please see Figure 1 As shown, this embodiment discloses a production process station layout method based on big data, including: S10: Obtain process information, employee expected salary, and employee historical production records. Process information includes process name, line marking, standard working hours, and equipment type. Understandably, process information specifically refers to a set of data used to clearly describe the execution status of each process in the production workshop. This includes process name, line marking, standard working hours, and equipment type. The process name clearly identifies the specific content of each production task, such as "sewing the front piece" or "overlocking." The line marking clearly distinguishes whether the process belongs to an in-line or out-of-line process, facilitating clear division of labor during subsequent station arrangement. Standard working hours refer to the standard time required to execute the process under specific process conditions, clearly reflecting the basic basis for production efficiency assessment. The equipment type clearly indicates the specific type of production equipment used to execute the process, such as a flat sewing machine, overlocking machine, or automatic template machine, providing a basis for subsequent determination of equipment configuration and employee skill allocation.

[0020] Furthermore, in this embodiment, the employee expected salary is explicitly obtained as the salary standard that a specific employee hopes to receive for participating in a specific production process. This information is explicitly used to reasonably allocate employee positions during process scheduling to ensure consistency between the employee's actual income and expected salary, reduce the frequency of subsequent scheduling and personnel adjustments, and improve the stability of the process scheduling plan.

[0021] Furthermore, the employee historical production records explicitly obtained in this embodiment refer to a set of historical data that clearly records the types of processes an employee participated in during past workshop production, their actual production efficiency, and the actual output quantity. For example, it can be clearly recorded that an employee's historical production efficiency in the process of "sewing cuffs" was 85%, and that the historical output stability was relatively high. This data is specifically used for the accurate assessment of employee skill levels in the future, in order to support more accurate and reasonable employee process allocation and production station arrangement.

[0022] S20: Use a text embedding model to map process names to process vectors, classify the process vectors according to a preset similarity threshold, and generate process classification names; In this embodiment, the text embedding model is a semantic representation model based on deep learning technology, which can transform the process names represented in natural language into a high-dimensional vector form that can be processed by machines. The text embedding model used in this embodiment maps each specific process name into a specific 384-dimensional process vector. The resulting process vector can clearly represent the semantic similarity and intrinsic relationship between processes.

[0023] It should be noted that in actual production processes, process names often exhibit clear diversity and inconsistency. For example, processes with different names but similar meanings may exist. This inconsistency increases the complexity of production process arrangement in existing technologies. In this embodiment, a clear text embedding model is used to uniformly represent processes with different names as clear process vectors. Then, the similarity between these process vectors is calculated to determine the specific degree of association between processes.

[0024] Specifically, this embodiment employs the cosine similarity method to calculate the similarity between every two process vectors. Furthermore, it categorizes the process vectors based on a predefined similarity threshold (e.g., 95%). In particular, when the cosine similarity of two process vectors exceeds the threshold, the two process names are explicitly categorized into the same explicit process category set, and a specific process category name is determined for each category set, thereby explicitly achieving structured classification management of processes.

[0025] For example, when the process names are "sewing the front piece" and "front piece stitching", although the descriptions are different, the actual processes are clearly similar. After mapping through the text embedding model, the calculated similarity value reaches 97%, which exceeds the preset similarity threshold of 95%. The two process names are clearly classified into the "front piece sewing" category, and the unified category name is used for subsequent production process station arrangement. This clearly reduces the uncertainty and repeated adjustments in subsequent arrangement, and significantly improves the stability and execution efficiency of process arrangement.

[0026] Based on a preset similarity threshold, the process vectors are categorized, and the generated process category names include: A text embedding model is used to map each process name to obtain a 384-dimensional process vector, and the process names and 384-dimensional process vectors are stored in a process vector table; the process names in the process vector table are deduplicated to obtain a set of deduplicated process names. Calculate the cosine similarity between each pair of process vectors in the deduplicated process name set to obtain a similarity mapping table; Process names that meet the preset similarity threshold in the similarity mapping table are grouped into the same category, and a process category name is determined for each category.

[0027] The similarity threshold is 95%, and the cosine similarity is calculated by the angle between the 384-dimensional process vectors corresponding to the process names.

[0028] In this embodiment, the explicit process classification name specifically refers to the unified name of multiple process names belonging to the same type of process steps after explicit semantic analysis. This explicit process classification method can significantly improve the clarity, stability and execution consistency of production process arrangement. In order to explicitly generate process classification names, this embodiment first uses a text embedding model to perform semantic mapping on the explicit process names. Each explicit process name is specifically mapped to an explicit 384-dimensional process vector, clearly forming a clear correspondence between process names and corresponding process vectors, and storing them in the process vector table to provide a data foundation for subsequent process vector analysis.

[0029] After clearly forming the process vector table, the process names stored in the process vector table are further clearly deduplicated. Specifically, redundant process names with exactly the same literal meaning are explicitly deleted, thereby obtaining a set of deduplicated process names. This set of deduplicated process names is used for subsequent process vector similarity analysis.

[0030] Specifically, for every two processes in the deduplicated process name set, a pairwise cosine similarity calculation is performed on the 384-dimensional process vectors. The cosine similarity is explicitly represented as the angle between the two process vectors in the 384-dimensional space, thus obtaining the specific similarity value between each pair of process vectors. This generates a similarity mapping table, which clearly records the explicit semantic similarity between each pair of process names. In this embodiment, a 95% explicit similarity threshold is explicitly used as the explicit process classification standard. Process names in the similarity mapping table that explicitly meet or exceed the preset threshold are explicitly classified into the same explicit classification group. Furthermore, a unified and explicit process classification name is explicitly assigned to each classification group to clearly indicate the unified technological meaning of the processes under that group.

[0031] For example, regarding the similarity mapping relationship of process vectors in the process name set, explicit classification examples are shown in Table 1: Table 1 The cosine similarity data mentioned above all exceed the 95% threshold. Therefore, "sewing the front piece" and "front piece stitching" are clearly classified under the process category name "front piece sewing," "overlocking" and "binding" are clearly classified under the process category name "overlocking and binding," and "sewing the cuff" and "cuff stitching" are clearly classified under the process category name "cuff sewing." Through the above clear classification process, this embodiment significantly improves the clarity and reliability of the process station arrangement, effectively avoids process confusion, errors, and repeated backtracking in production arrangement, and further realizes the stability and executability of the production arrangement results.

[0032] The process names that meet the preset similarity threshold in the similarity mapping table are grouped into the same category, including: Before classifying based on the similarity mapping table, the process name set is first merged according to the literal consistency of the process name to form the first classification group set; For the representative process names of the first categorized group set, query the similarity mapping table to generate a group similarity matrix table. For the group similarity matrix table, perform cross-group merging on the group pairs that meet the preset similarity threshold to form a second categorized group set, and determine the process categorization name for the second categorized group set.

[0033] In this embodiment, when clearly classifying process names into a unified "process classification group", since there are many differences or similarities in the expression of process names in actual production, and process names often show little difference in literal meaning and high semantic overlap in text description, it is necessary to clearly classify the process name set in stages and levels to achieve more accurate and reliable process classification results.

[0034] This embodiment proposes a specific phased classification method. First, before classifying process names using a similarity mapping table, a first round of merging is explicitly performed on the process name set based on the explicit literal consistency of the process names. Specifically, if two or more process names are explicitly identical in literal meaning or have only minor differences (such as spaces, punctuation, etc.), they are explicitly merged into the same explicit classification group to form the first classification group set. This explicit first-round merging method can effectively reduce the amount of redundant processing in subsequent similarity calculations and ensure the clear consistency of the basic process name data.

[0035] After obtaining the first classification group set, this embodiment further explicitly queries the representative process name of each explicit first classification group set from the similarity mapping table, and explicitly generates an explicit group similarity matrix table. This group similarity matrix table specifically represents the explicit cosine similarity value between the representative process name of each classification group and the representative names of other groups, thereby clearly reflecting the specific degree of similarity of different classification groups at the semantic level.

[0036] Furthermore, this embodiment uses a predetermined similarity threshold (e.g., 95%) as a benchmark, and explicitly performs cross-group merging processing on group pairs in the group similarity matrix table that explicitly meet or exceed the predetermined threshold. Specifically, if the similarity between the representative process names of two different groups explicitly reaches or exceeds the predetermined threshold of 95%, then the two groups are explicitly merged into a unified second classification group, and a predetermined second classification group set is formed after merging.

[0037] Finally, a unified process classification name is specifically determined for the aforementioned clearly formed second classification group set. For example, for a certain second classification group set after clear merging, the process names in it clearly include "cuff sewing", "sewing cuffs", "cuff stitching", etc. The unified process classification name for this classification group set is clearly determined as "cuff sewing", thereby clearly achieving the accuracy and consistency of process name classification. Compared with traditional methods, the phased classification method proposed in this embodiment significantly improves the accuracy of process classification, effectively avoids production arrangement errors and instability caused by differences in process name expression, and clearly improves the stability and executability of production process station arrangement.

[0038] S30: Calculate employee production efficiency and employee production frequency based on employee historical production records and process classification names, and generate an employee skill level assessment table based on employee production efficiency and employee production frequency. In this embodiment, the employee skill level assessment table is a data structure that clearly reflects the production efficiency and proficiency level of employees in a specific process category. By clearly assessing the employee skill level, it can effectively provide accurate personnel skill data support for the arrangement of production process positions, reduce the number of arrangement iterations and adjustments caused by fluctuations in personnel skills and inaccurate assessments, and thus ensure the stability of the process arrangement plan.

[0039] Furthermore, this embodiment first explicitly calculates employee skill data based on employee historical production records and process category names. The employee historical production records explicitly include the employee's production efficiency, quality, and production frequency when performing different processes within a specific time period. The process category names explicitly indicate the specific process categories formed through the aforementioned steps. Therefore, it is first necessary to explicitly extract each employee's production performance data under each specific process category from their historical production records based on the process category names, including specific information such as production efficiency, quantity completed, and number of operations.

[0040] For example, as shown in Table 2: Table 2 Furthermore, the specific skill assessment indicators calculated based on the aforementioned employee productivity and production frequency are clearly defined, and employee skill levels are clearly classified according to explicit scoring rules. For example, skill levels are explicitly set as "Excellent," "Good," and "Pass." Specific scoring rules are illustrated below: Skill level "Excellent": Production efficiency ≥90%, and production frequency ≥200 times / month; Skill level "Good": Production efficiency ≥80% and <90%, production frequency ≥150 times / month and <200 times / month; Skill level "qualified": Production efficiency ≥70% and <80%, production frequency ≥100 times / month and <150 times / month.

[0041] Finally, this embodiment explicitly uses the aforementioned clearly defined employee skill level assessment indicators and scoring rules to specifically organize the clearly defined skill level results for each employee into a complete employee skill level assessment table, as shown in Table 3 below: Table 3 The generated employee skill level assessment table clearly and accurately reflects the actual skill level of employees under each specific process category, effectively supporting the precise matching of personnel and positions in the subsequent production process scheduling, thereby significantly reducing the frequency of production scheduling iterations and adjustments, and significantly improving the stability and execution efficiency of production process position scheduling.

[0042] S40: Divide the process information according to the markings inside the line to obtain the process information inside the line and the process information outside the line. Generate the arrangement result of the process information inside the line according to the employee skill level assessment table, and calculate the threshold of the station outside the line according to the employee's expected salary and standard working hours. The in-line arrangement results of generating in-line process information based on the employee skill level assessment form include: Obtain team information, divide the in-line process information into special process sets and non-special process sets based on in-line tags, associate the special process sets with the special station IDs in the team information, assign employee information to the special process sets according to the employee skill level assessment table, and generate special process arrangement results; The process retrieves the work ticket numbers corresponding to each process in the set of non-special processes, sorts the set of non-special processes in reverse order according to the work ticket numbers, and divides it into a set of combined processes and a set of non-combined processes. Based on the employee skill level assessment table, the process arrangement results of combined processes and non-combined processes are generated respectively, and together they form the in-line arrangement result.

[0043] Understandably, the in-line marking is used to clearly distinguish whether a production process belongs to a hanging line or other processes that must be continuously executed within the production line. This marking directly affects the priority of the process in the station arrangement and the personnel allocation method. Therefore, it is necessary to first clearly divide the process information based on the in-line marking to obtain in-line process information and out-of-line process information. The two types of processes adopt different processing strategies in the subsequent arrangement logic.

[0044] Furthermore, in this embodiment, for the in-line process information, the in-line scheduling result is generated based on the employee skill level assessment table. The core of this process is to ensure the continuity and stability of the in-line production rhythm as much as possible, and to avoid production line fluctuations caused by frequent personnel switching or skill mismatch. To this end, the team information is first obtained. The team information clearly records the available positions, special position identifiers and position numbers within the team. On this basis, the in-line process information is further divided into special process sets and non-special process sets according to the in-line markings. Among them, special processes usually refer to processes with clear restrictions on equipment, skills or positions.

[0045] When processing a set of special processes, this embodiment associates the set of special processes with the special station IDs in the team information one by one, and combines them with the employee skill level assessment table to assign employee information with matching skill levels to each special process, thereby generating a special process arrangement result. This result can ensure that key processes are completed by employees with the corresponding skill levels at designated stations, reducing uncertainty in the production process.

[0046] For non-special process sets, this embodiment further associates and obtains the work ticket number corresponding to each process. The work ticket number is used to reflect the sequential relationship of the processes in the production process. Then, the non-special process sets are arranged in reverse order of the work ticket numbers, and on this basis, they are divided into combined process sets and non-combined process sets. The combined process set represents a combination of processes that are continuous in production sequence and have the possibility of being completed continuously by the same employee.

[0047] Subsequently, this embodiment performs employee allocation processing on the combined process set and the non-combined process set according to the employee skill level assessment table, generating combined process arrangement results and non-combined process arrangement results. The above two types of arrangement results are then integrated with the aforementioned special process arrangement results to finally form the in-line arrangement result. Through the above method, this embodiment takes into account process continuity, personnel skill matching, and station position constraints when generating the in-line arrangement scheme. Compared with the existing technology that simply relies on rule matching or manual adjustment, it can effectively reduce the repeated rollback of in-line arrangement and improve the stability and executability of the in-line production scheme.

[0048] The generated combined process arrangement results include: The screening and allocation steps are repeated until all combined processes have been allocated. The screening and allocation steps include: using consecutive work ticket numbers as the connection condition, screening candidate work sets that can be continuously completed by the same employee from the remaining unallocated combined work sets, sorting the candidate work sets according to the number of work processes, and selecting the candidate work set with the most work processes from the sorting results as the current target combined work set; Obtain the number of equipment types involved in the current target combination process set. If the number of equipment types is less than or equal to 3, assign employee information to the current target combination process set according to the employee skill level assessment table, and write the process number, production sequence number, employee number and station number of the current target combination process set into the in-line arrangement result. If the number of equipment types is greater than 3, mark the current target combination process set as a non-combination process or remove it from the candidate process set.

[0049] In this embodiment, the combined process arrangement aims to achieve a precise match between employee skills and process continuity, reducing the efficiency loss caused by frequent switching between employees and equipment during the production process. Therefore, it is necessary to repeatedly execute specific screening and allocation steps until all processes in the combined process set are clearly allocated.

[0050] Specifically, this embodiment uses the continuity of work order numbers as an explicit connection condition. This ensures that the selected processes in the combined process set are clearly continuous in the production process, and selects a set of candidate processes that can be continuously completed by the same employee from the remaining unassigned combined process sets. Furthermore, to prioritize the combined process sets that meet the high requirements for process continuity in the production process, this embodiment explicitly sorts the candidate process sets according to the number of processes in the set. From the sorting results, the candidate process set with the most processes is selected as the current target combined process set to maximize the continuity of employees working continuously at the same station.

[0051] For example, consider the following process combination: cuff sewing, shoulder sewing, and front panel sewing are performed consecutively. The specific equipment types for cuff sewing, shoulder sewing, and front panel sewing are an automatic template machine, a buttonhole machine, and an overlock machine, respectively. The number of equipment types involved is clearly counted as three, meeting the defined equipment type limit. Therefore, based on the employee skill level assessment table, the corresponding employee information is clearly assigned to the current target process combination. Specifically, for example, selecting employee E001 with an "Excellent" skill level in the skill level assessment table, the process number, production sequence number, employee number (E001), and specific station number for cuff sewing, shoulder sewing, and front panel sewing are recorded in the in-line arrangement results, ensuring clear skill matching and process continuity.

[0052] Furthermore, if the number of equipment types involved in the current target combination process set exceeds a specified limit (e.g., more than 3 types), it means that the combination process set may cause employees to frequently switch equipment and positions, which clearly violates the production efficiency requirements. In this embodiment, the set is clearly marked as a non-combination process set, or directly removed from the candidate process set, in order to avoid the waste of equipment and personnel resources or the instability of process execution during the process allocation process.

[0053] S50: Calculate the total working hours of off-line stations based on off-line process information and standard working hours. When the total working hours of off-line stations exceed the off-line station threshold, perform process migration and employee adjustment, and generate off-line scheduling results. When the total working hours of off-line stations do not exceed the off-line station threshold, directly generate off-line scheduling results based on off-line process information, and write the on-line scheduling results and off-line scheduling results into the MES database.

[0054] Calculate the total working hours for off-line stations based on off-line process information and standard working hours, including: Filter the set of off-line processes with automatic template machine as the equipment type from the off-line process information, and adjust the standard working hours according to the average available working hours per person in the employee's historical production records; Based on the revised standard working hours, initial stations are allocated to the set of off-line processes, and the station working hours for each initial station are calculated. The station working hours are then used as the total working hours for off-line stations.

[0055] In this embodiment, the total working hours of off-line stations are used to clearly assess the load of off-line production process stations and determine whether subsequent process relocation and employee adjustment are necessary. This ensures the rationality of the scheduling scheme and the stability of production execution. The off-line process set with the equipment type clearly defined as automatic template machine is selected from the off-line process information. This is because the standard working hours of automatic template machines usually have high consistency and stability, which can be used to accurately assess the load of each station when calculating the total working hours of off-line stations, avoiding the impact of working hour fluctuations of other equipment types on the overall calculation accuracy.

[0056] Furthermore, since the production efficiency of each employee varies significantly in the actual production process, this embodiment explicitly introduces the average available working hours per person from the employee's historical production records to correct the standard working hours corresponding to the process, so as to obtain a more specific actual executable standard working hours. This process is specifically manifested in the explicit analysis of the actual production time data of employees in the historical production records to explicitly calculate the corrected standard working hours.

[0057] For example, several processes are explicitly extracted from the set of off-line processes, such as the "cuff sewing" process corresponding to the automatic template machine. Its standard working time is explicitly set at 5 minutes, and the average available working time per employee in the historical production records is explicitly set at 6 minutes. The working time after adjusting the standard working time for the above processes is explicitly set at 6 minutes. Based on the explicitly adjusted standard working time, the processes in the off-line process set are explicitly assigned to initial stations, and the station working time corresponding to each initial station is explicitly calculated. The sum of the station working times is the explicitly defined total working time for the off-line stations.

[0058] Through the above-described correction and calculation process, this embodiment can clearly and accurately determine the actual load status of off-line production stations, effectively avoiding situations where the load is excessively concentrated or the process allocation is uneven during actual production execution. It clearly improves the rationality, stability, and executability of production scheduling results, and reduces repeated adjustments and efficiency losses caused by inaccurate station time estimation during production.

[0059] When the total working hours at off-line stations exceed the off-line station threshold, the methods for performing process migration and employee adjustments, and generating off-line scheduling results include: The last inserted process is determined from the set of off-line processes allocated to the initial station where the total working hours of the off-line stations exceed the off-line station threshold. The last inserted process is then moved to the previous station that is idle or not fully loaded, while keeping the process number and production sequence number unchanged after the migration. Based on the employee skill level assessment table, the employees with the lowest skill level ranking are determined, and these employees are assigned to the last inserted process of the target station. The employee number and process information of the target station are updated, and the final off-line arrangement result is generated. The target station refers to the previous station that is in an idle state or not fully loaded.

[0060] Understandably, when the total working hours at off-line stations exceed a specific off-line station threshold, it indicates that the current off-line stations are experiencing excessive production load and uneven working hour distribution. Therefore, it is necessary to appropriately relocate and adjust the processes and employees to effectively balance the load between stations, ensure that the production scheduling plan can be executed smoothly, and avoid production line efficiency losses due to station overload.

[0061] First, we analyze the set of off-line processes that were explicitly assigned to the initial station where the total working hours of the off-line stations exceeded the preset threshold. We then clearly identify the last inserted process from this set of processes. The reason for choosing the last inserted process for migration is that this process has high flexibility in the station allocation order and will not affect the continuity and stability of the processes before the station. Migrating this process can effectively reduce the load of the current station without disrupting the overall production sequence.

[0062] Subsequently, in this embodiment, the last inserted process is explicitly moved to the previous target station that is either idle or not fully loaded, in order to ensure load balancing. Specifically, the process after the migration still retains its original process number and production sequence number. This method clearly avoids production chaos and misoperation problems caused by changes in process number or sequence.

[0063] For example, if the total working hours of an initial station exceed a specific threshold, and its last inserted process is clearly "cuff sewing", the process is moved to the previous target station that is clearly idle. The process number (e.g., 012) and production sequence number (e.g., 120) of this process remain unchanged, which effectively reduces the load pressure on the initial station and clearly maintains the stability of the overall production process.

[0064] Furthermore, this embodiment makes clear adjustments to employees based on the employee skill level assessment table. Specifically, it redistributes employees to target positions and selects employees with lower skill levels to perform the last inserted process that has been moved. This clear redistribution of employees ensures a precise match between employee skills and process complexity, effectively improving production efficiency after the position adjustment.

[0065] For example, if the employee skill level assessment table clearly shows that the employee ranked last is E003, this employee is clearly assigned to the target station to perform the above-mentioned migration process, and the employee ID and process information corresponding to the target station are clearly updated, thus completing the clear employee adjustment.

[0066] Through the aforementioned clear process migration and employee adjustment operations, this embodiment generates an optimized off-line scheduling result. Unlike the static scheduling scheme in the prior art, this embodiment proposes a dynamic and clear adjustment mechanism, which can effectively balance the production load, significantly reduce the number of repeated adjustments caused by station working hours exceeding the threshold, and significantly improve the stability and production efficiency of the production scheduling result.

[0067] Example 2 Please see Figure 2 As shown, based on the same inventive concept, this embodiment discloses a production process station arrangement system based on big data. For details not covered in this embodiment, please refer to the relevant parts of Embodiment 1. The system includes: Data acquisition module: used to acquire process information, employee expected salary and employee historical production records. Process information includes process name, line marking, standard working hours and equipment type. Vector Classification Module: This module uses a text embedding model to map process names to process vectors, classifies the process vectors according to a preset similarity threshold, and generates process classification names. Skills assessment module: Used to calculate employee production efficiency and employee production frequency based on employee historical production records and process classification names, and generate employee skill level assessment table based on employee production efficiency and employee production frequency; The division and arrangement module is used to divide the process information according to the markings in the line, obtain the process information in the line and the process information outside the line, generate the in-line arrangement result of the process information in the line according to the employee skill level assessment table, and calculate the threshold of the station outside the line according to the employee's expected salary and standard working hours. Offline orchestration module: Used to calculate the total working hours of offline stations based on offline process information and standard working hours. When the total working hours of offline stations exceed the offline station threshold, process migration and employee adjustment are performed, and offline orchestration results are generated. When the total working hours of offline stations do not exceed the offline station threshold, offline orchestration results are generated directly based on offline process information, and the online orchestration results and offline orchestration results are written into the MES database.

[0068] The detailed description above, in conjunction with the accompanying drawings, describes examples but does not represent all examples that can be implemented or fall within the scope of the claims. The terms “example” and “exemplary” are used in this specification to mean “serving as an example, instance or illustration” and do not mean “superior to or better than other examples”.

[0069] Throughout this specification, the phrase "an embodiment" or "an embodiment" means that a particular feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment of the invention. Therefore, the use of these phrases may refer to more than one embodiment. Furthermore, the described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0070] It should also be noted that these embodiments may be described as processes depicted as flowcharts, structural diagrams, or block diagrams. Although a flowchart may describe the operations as sequential processes, many of these operations can be performed in parallel or concurrently, and the order of these operations may be rearranged.

Claims

1. A production process station layout method based on big data, characterized in that, include: Obtain process information, employee expected salary, and employee historical production records. Process information includes process name, line marking, standard working hours, and equipment type. The text embedding model is used to map the process name to the process vector, and the process vector is classified according to the preset similarity threshold to generate the process classification name; Employee production efficiency and frequency are calculated based on employees' historical production records and process classification names. Based on employee production efficiency and frequency, an employee skill level assessment table is generated. The process information is divided according to the markings within the line to obtain the process information within the line and the process information outside the line. The process information within the line is arranged according to the employee skill level assessment table, and the threshold for the station outside the line is calculated according to the employee's expected salary and standard working hours. The total working hours of off-line stations are calculated based on off-line process information and standard working hours. When the total working hours of off-line stations exceed the off-line station threshold, process migration and employee adjustment are performed, and off-line scheduling results are generated. When the total working hours of off-line stations do not exceed the off-line station threshold, off-line scheduling results are generated directly based on off-line process information. The on-line scheduling results and off-line scheduling results are written into the MES database.

2. The production process station arrangement method based on big data according to claim 1, characterized in that, Based on a preset similarity threshold, the process vectors are categorized, and the generated process category names include: A text embedding model is used to map each process name to obtain a 384-dimensional process vector, and the process names and 384-dimensional process vectors are stored in a process vector table; the process names in the process vector table are deduplicated to obtain a set of deduplicated process names. Calculate the cosine similarity between each pair of process vectors in the deduplicated process name set to obtain a similarity mapping table; Process names that meet the preset similarity threshold in the similarity mapping table are grouped into the same category, and a process category name is determined for each category.

3. The production process station arrangement method based on big data according to claim 2, characterized in that, The similarity threshold is 95%, and the cosine similarity is calculated by the angle between the 384-dimensional process vectors corresponding to the process names.

4. The production process station arrangement method based on big data according to claim 2, characterized in that, The process names that meet the preset similarity threshold in the similarity mapping table are grouped into the same category, including: Before classifying based on the similarity mapping table, the process name set is first merged according to the literal consistency of the process name to form the first classification group set; For the representative process names of the first categorized group set, query the similarity mapping table to generate a group similarity matrix table. For the group similarity matrix table, perform cross-group merging on the group pairs that meet the preset similarity threshold to form a second categorized group set, and determine the process categorization name for the second categorized group set.

5. The production process station arrangement method based on big data according to claim 1, characterized in that, The in-line arrangement results of generating in-line process information based on the employee skill level assessment form include: Obtain team information, divide the in-line process information into special process sets and non-special process sets based on in-line tags, associate the special process sets with the special station IDs in the team information, assign employee information to the special process sets according to the employee skill level assessment table, and generate special process arrangement results; The process retrieves the work ticket numbers corresponding to each process in the set of non-special processes, sorts the set of non-special processes in reverse order according to the work ticket numbers, and divides it into a set of combined processes and a set of non-combined processes. Based on the employee skill level assessment table, the process arrangement results of combined processes and non-combined processes are generated respectively, and together they form the in-line arrangement result.

6. The production process station arrangement method based on big data according to claim 5, characterized in that, The generated combined process arrangement results include: The screening and allocation steps are repeated until all combined processes have been allocated. The screening and allocation steps include: using consecutive work ticket numbers as the connection condition, screening candidate work sets that can be continuously completed by the same employee from the remaining unallocated combined work sets, sorting the candidate work sets according to the number of work processes, and selecting the candidate work set with the most work processes from the sorting results as the current target combined work set; Obtain the number of equipment types involved in the current target combination process set. If the number of equipment types is less than or equal to 3, assign employee information to the current target combination process set according to the employee skill level assessment table, and write the process number, production sequence number, employee number and station number of the current target combination process set into the in-line arrangement result. If the number of equipment types is greater than 3, mark the current target combination process set as a non-combination process or remove it from the candidate process set.

7. The production process station arrangement method based on big data according to claim 6, characterized in that, Calculate the total working hours for off-line stations based on off-line process information and standard working hours, including: Filter the set of off-line processes with automatic template machine as the equipment type from the off-line process information, and adjust the standard working hours according to the average available working hours per person in the employee's historical production records; Based on the revised standard working hours, initial stations are allocated to the set of off-line processes, and the station working hours for each initial station are calculated. The station working hours are then used as the total working hours for off-line stations.

8. The production process station arrangement method based on big data according to claim 7, characterized in that, When the total working hours at off-line stations exceed the off-line station threshold, the methods for performing process migration and employee adjustments, and generating off-line scheduling results include: The last inserted process is determined from the set of off-line processes allocated to the initial station where the total working hours of the off-line stations exceed the off-line station threshold. The last inserted process is then moved to the previous station that is idle or not fully loaded, while keeping the process number and production sequence number unchanged after the migration. Based on the employee skill level assessment table, the employees with the lowest skill level ranking are determined, and these employees are assigned to the last inserted process of the target station. The employee number and process information of the target station are updated, and the final off-line arrangement result is generated. The target station refers to the previous station that is in an idle state or not fully loaded.

9. A production process station layout system based on big data, used to implement the production process station layout method based on big data as described in any one of claims 1-8, characterized in that, include: Data acquisition module: used to acquire process information, employee expected salary and employee historical production records. Process information includes process name, line marking, standard working hours and equipment type. Vector Classification Module: This module uses a text embedding model to map process names to process vectors, classifies the process vectors according to a preset similarity threshold, and generates process classification names. Skills assessment module: Used to calculate employee production efficiency and employee production frequency based on employee historical production records and process classification names, and generate employee skill level assessment table based on employee production efficiency and employee production frequency; The division and arrangement module is used to divide the process information according to the markings in the line, obtain the process information in the line and the process information outside the line, generate the in-line arrangement result of the process information in the line according to the employee skill level assessment table, and calculate the threshold of the station outside the line according to the employee's expected salary and standard working hours. Offline orchestration module: Used to calculate the total working hours of offline stations based on offline process information and standard working hours. When the total working hours of offline stations exceed the offline station threshold, process migration and employee adjustment are performed, and offline orchestration results are generated. When the total working hours of offline stations do not exceed the offline station threshold, offline orchestration results are generated directly based on offline process information, and the online orchestration results and offline orchestration results are written into the MES database.

Images