A big data-based manufacturing execution scheduling optimization system and method

Abstract

The present application relates to the technical field of production scheduling optimization, in particular to a manufacturing execution scheduling optimization system and method based on big data, which comprises an order parameter standardization module, a process version boundary solidification module, a multi-parameter linkage verification module, an execution condition offset monitoring module and a local rebinding confirmation module.In the present application, the heterogeneous matching obstacles are eliminated by unified numbering, and the joint verification efficiency is improved.The process version configuration is converted into interval boundary and associated rules to avoid parameter fluctuation misjudgment.A multi-parameter joint permission matrix is established to break through the single-dimensional verification limitation and coordinate multiple constraints such as equipment, tooling, materials and beat.The offset difference calculation and covariance analysis control covariant are introduced to exclude mixed interference such as temperature and wear, calculate the trend coefficient, upgrade to trend perception and predict the drift direction.After the offset is weighted and corrected, the matching and binding are re-performed to realize adaptive updating and reduce manual intervention.The manufacturing execution precision is enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of production scheduling optimization, and in particular to a big data-based manufacturing execution scheduling optimization system and method. Background Art

[0002] The technical field of production scheduling optimization includes production planning preparation, work order scheduling, equipment resource allocation, material supply coordination, production process monitoring, operation execution feedback, and production data analysis. Its core is to carry out unified management and scheduling around production tasks, equipment status, material inventory, personnel allocation, process routes, order demands, production takt, equipment operation records, quality inspection records and logistics circulation information. By collecting equipment operation data at the manufacturing site, work order execution data, material inbound and outbound data, equipment maintenance records, process parameter data and quality inspection data, systematic management is performed on production plan formulation, operation sequence arrangement, resource allocation adjustment, production progress tracking, abnormal production handling and production execution control, so as to provide data support for production organization and scheduling in the manufacturing process.

[0003] Wherein, the big data-based manufacturing execution scheduling optimization system refers to a system that utilizes large-scale production data generated at the manufacturing site to manage scheduling activities in the manufacturing execution process. The technical matters it addresses cover production order information collection, process route information sorting, equipment operation status collection, equipment maintenance record sorting, material inventory information collection, material distribution information recording, personnel post information management, process execution information recording, production progress information update, quality inspection data collection, abnormal event recording and historical production record analysis. Manufacturing execution scheduling is completed by establishing the correspondence between production order data and process route data, determining equipment availability according to equipment operation status and equipment maintenance records, determining material supply status according to material inventory information and material distribution information, arranging operation tasks according to personnel post information and process execution information, adjusting process execution sequence according to production progress information, checking process execution status according to quality inspection data, updating production task arrangement according to abnormal event records, and configuring subsequent production scheduling according to historical production records.

[0004] In the prior art, data formats are not uniform, and cross-system association requires manual comparison, which is time-consuming and error-prone. Process versions are stored as fixed values and not converted to interval boundaries, so fluctuations within tolerances cannot be identified, resulting in implicit deviations. Parameter verification is limited to a single dimension, lacking joint relationship verification between equipment and tooling, materials and processes, etc., leading to frequent conflicts exposed during the execution of scheduling plans. Offset monitoring adopts fixed thresholds, and alarms lag behind the drift process. When production conditions change, only local fields are updated without linkage adjustment of associated relationships, and subsequent scheduling still uses old bindings. Historical analysis is limited to statistical reports, and trend data is not used for dynamic calibration. These deficiencies weaken the system's ability to cope with fluctuations, coordinate constraints and make predictions. Summary of the Invention

[0005] The purpose of this invention is to address the shortcomings of existing technologies by proposing a manufacturing execution scheduling optimization system and method based on big data.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A manufacturing execution scheduling optimization system based on big data includes: The order parameter standardization module collects product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters, and operation execution parameters and categorizes them into corresponding sets. It generates a data number in a unified format for each set and combines the data number with each parameter set into a structured data packet, which is then transmitted as a standardized parameter set. The process version boundary solidification module reads the process version configuration parameters, extracts the upper and lower limits of dimensions and the start and end values ​​of batches and retains them as boundary pairs, establishes an independent mapping relationship between each equipment identifier and the cycle value, and generates an execution boundary dataset for merging and transmission. The multi-parameter linkage verification module receives the standardized parameter set and the execution boundary data set, performs item-by-item threshold comparison and calculates the matching pass rate, establishes a joint licensing relationship based on the comparison result of the matching pass rate and the preset threshold, generates a linkage licensing matrix and transmits it. The execution condition offset monitoring module receives the linkage permission matrix, obtains the current parameter value and performs offset difference calculation with the initial parameter. It adopts a multiple regression model based on the principle of covariance analysis, incorporates the time index as a continuous independent variable and the covariate as a control variable into the equation, fits the adjusted offset trend coefficient, and combines the offset difference sequence with the offset trend prediction value into an offset trend transmission array and transmits it.

[0007] As a further embodiment of the present invention, the order parameter standardization module includes product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters, and operation execution parameters; the process version boundary solidification module includes process version configuration parameters, size upper and lower limits, batch start and end values, and execution boundary data; the multi-parameter linkage verification module specifically refers to the standardized parameter set, the execution boundary data set, and the linkage permission matrix; the execution condition offset monitoring module includes offset amount, offset trend coefficient, and offset trend transmission array.

[0008] As a further aspect of the present invention, the order parameter standardization module includes: The parameter collection submodule collects the size and weight values ​​from product attribute parameters and assigns them to the product attribute set; it collects the temperature, speed, and pressure values ​​from equipment status parameters and assigns them to the equipment status set; it collects the wear and calibration offset values ​​from tooling status parameters and assigns them to the tooling status set; it collects the time value from process execution parameters and assigns it to the process execution set; it collects the time consumption value from process execution parameters and assigns it to the process execution set; it normalizes the original values ​​in each set and converts them into dimensionless standardized values ​​in the [0,1] interval; it sums all the standardized values ​​in each set to obtain five sub-totals and adds them together to obtain the total parameter set. Number generation submodule: According to the preset collection traversal order, assign an incremental sequence number from 001 to 005 to each collection, combine the last two digits of the current year with the month and date to form a six-digit number, combine it with the sequence number to form a nine-digit number, and concatenate the five nine-digit numbers in order to generate the total value of the data number; Standard Set Submodule: Establishes key-value pair mappings between the five nine-digit numbers and their corresponding normalized parameter value lists according to the set name, and combines all mappings into a JSON structure or relational data table. Each record contains a unique record number, a set type label, and all normalized parameter values, which are used as a standardized parameter set for transmission. The unique record number is generated by concatenating the last two digits of the year, the month and date, and the set sequence number in sequence, with a fixed total of 9 digits.

[0009] As a further aspect of the present invention, the process version boundary curing module includes: Interval calculation submodule: Reads the upper limit value of the size, the lower limit value of the size, the batch start value, and the batch end value, and retains the upper limit value and the lower limit value as a pair as a size boundary pair, and retains the batch start value and the end value as a pair as a batch boundary pair; The correspondence submodule reads the device identifier value and its corresponding standard beat value, pairs each device identifier with a beat value independently, and does not perform cross-device regression fitting; it then indexes each pair by the device identifier value to generate a device-beat mapping table. Boundary Set Submodule: Appends the size boundary pairs and batch boundary pairs to the end of each record in the device-cycle mapping table to form an execution boundary dataset containing seven fields: 'Device ID, Cycle Time, Upper Size Limit, Lower Size Limit, Batch Start, Batch End, and Compatible Tooling List'.

[0010] As a further aspect of the present invention, the multi-parameter linkage verification module includes: Matching and Validation Submodule: The size value, batch value, equipment identifier value, and tooling identifier value of each record in the standardized parameter set are compared item by item with the corresponding size upper and lower limits, batch start and end, equipment existence, and compatible tooling list in the execution boundary dataset. Each item is marked as 1 if it passes the comparison, and 0 otherwise. The four marked values ​​are summed to obtain the matching success count. The joint relationship submodule calculates the matching pass rate by dividing the matching pass count by 4; if the pass rate is greater than or equal to a preset threshold, the record is determined to be eligible for joint permission; otherwise, it is determined not to be eligible. The license matrix submodule arranges the joint license qualification determination results of all records in the original parameter order to generate a one-dimensional Boolean license vector; the license vector is combined with the corresponding record number and parameter value to form a linked license matrix, which is a two-dimensional table structure containing the record number, all parameter values ​​and license Boolean values.

[0011] As a further aspect of the present invention, the execution condition offset monitoring module includes: Offset Difference Submodule: Receives the linkage permission matrix, extracts the current parameter value of each record, performs a subtraction operation with the initial parameter value of the corresponding record to obtain a single offset, then arranges all the single offsets in ascending order of record number, and binds these record numbers to obtain the offset difference sequence. Trend coefficient submodule: Based on the offset difference sequence, read the time index sequence and at least one covariate sequence, where the covariates include cumulative output, equipment temperature, or tooling usage frequency; fit the offset difference using a multiple linear regression model, with the time index as a continuous independent variable and the covariates as control variables included in the model; estimate the regression parameters using the least squares method; and extract the regression coefficients corresponding to the time indexes, which are the adjusted offset trend coefficients, representing the average rate of change of the offset over time after excluding the influence of covariates. State sequence submodule: The offset difference sequence is arranged in the order of record number as the first column, and the local estimated offset at the corresponding time point of each record is arranged as the second column. The two columns are combined into an offset trend transmission array.

[0012] As a further aspect of the present invention, the system further includes: The local rebinding confirmation module receives the offset trend transmission array, extracts the offset difference sequence and the local estimated offset sequence from it, performs weighted correction according to preset weights, and compares it with the allowable range. For records whose corrected offset does not exceed the allowable range, the original binding is maintained. For records whose corrected offset exceeds the allowable range, boundary matching is re-executed to generate a local binding identifier. After summarizing the binding status sequence to determine the validity, the final execution license number is output.

[0013] As a further aspect of the present invention, the local rebinding confirmation module specifically comprises an offset difference sequence, a local estimated offset sequence, a local binding identifier, and a final execution license number.

[0014] As a further aspect of the present invention, the local rebinding confirmation module includes: Data extraction submodule: receives the offset trend transmission array, extracts the offset difference sequence from the first column of the offset trend transmission array, and extracts the local estimated offset sequence from the second column; Weighted correction submodule: Based on the local estimated offset sequence and the offset difference sequence, the offset of each record is weighted and corrected according to a preset weight to obtain the corrected offset value; Binding Determination Submodule: Each corrected offset value is compared with the upper and lower limits of the allowed offset range one by one. Records within the range maintain the original binding relationship; records outside the range are unbound, and the execution boundary data set is called. Based on the size value and batch value of the current record, the device cycle value corresponding to the matching size boundary and batch boundary is retrieved, and a new local binding identifier is generated. If no matching device is found, the rebinding of the record is determined to be unsuccessful, the local binding identifier of the record is set to an invalid null value, and the number of failed records is accumulated. Summarize the binding status of all records: Extract all records with non-empty local binding identifiers, combine their record numbers with the binding device tick value, and generate a list of executable license numbers; If the ratio of the number of failed records to the total number of records is less than the preset safety redundancy threshold (e.g., 25%), and the number of failed records is not the total number of records, then the overall scheduling is deemed effective, the list of executable license numbers is output, and the failed records are marked as "pending manual review". If the proportion of failure records exceeds the safety threshold, the system will not output an execution permission, but will instead trigger a global alarm signal and output the detailed numbers of all failure records, thus suspending the scheduling process.

[0015] A manufacturing execution scheduling optimization method based on big data includes the following steps: S1: By collecting product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters and operation execution parameters and classifying them into corresponding sets, a data number with a unified format is generated for each set, and the data number and each parameter set are transmitted as a standardized parameter set; S2: By reading the process version configuration parameters, the difference between the upper and lower limits of the size and the start and end values ​​of the batch is calculated to form an interval, the correspondence between the equipment and the cycle time is established, and the execution boundary data set is generated and transmitted. S3: Based on the standardized parameter set and the execution boundary data set, compare the size, batch size, equipment existence and tooling compatibility item by item, calculate the matching pass rate, establish a joint license relationship based on a preset threshold, and generate a linkage license matrix for transmission. S4: Based on the linkage permission matrix, obtain the current parameter value and perform offset difference calculation with the initial parameter. Use a multiple regression model based on the principle of covariance analysis, take the time index as a continuous independent variable and the covariate as a control variable into the equation, fit the adjusted offset trend coefficient, and combine the offset difference sequence and the offset trend prediction value into an offset trend transmission array for transmission. S5: Based on the local estimated offset sequence and the offset difference sequence, the offset of each record is weighted and corrected according to a preset weight; S6: Compare each corrected offset value with the allowed offset range. If the offset is within the range, maintain the original binding. If the offset is outside the range, re-match and generate a new local binding identifier. Extract all records with non-empty local binding identifiers to generate an executable license number list. Calculate the ratio of the number of failed records to the total number of records. If the ratio is less than the preset safety redundancy threshold, the overall scheduling is determined to be valid and the executable license number list is output. At the same time, the failed records are marked as pending manual review. If the proportion of failed records exceeds the safety threshold, a global alarm signal is triggered and the scheduling process is terminated.

[0016] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, by generating unified numbers for each parameter set, heterogeneous data matching barriers are eliminated, improving joint verification efficiency. Process version configurations are converted into interval boundaries and association rules to avoid misjudgments caused by parameter fluctuations. A multi-parameter joint permission relationship is established to generate a linkage matrix, overcoming the limitations of single-dimensional verification and ensuring synchronous coordination of multiple constraints such as equipment, tooling, materials, and cycle time, reducing scheduling conflicts. By introducing offset difference calculation and covariance analysis to control covariates, adjusted trend coefficients are calculated, upgrading fixed threshold alarms to trend perception, eliminating interference from confounding factors such as temperature and wear, and predicting drift direction in advance. Weighted correction of offset states followed by re-matching and binding enables adaptive updates, reducing manual intervention and response delays. Overall, this enhances manufacturing execution accuracy. Attached Figure Description

[0017] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart of the present invention; Figure 3 This is a flowchart of the method steps of the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] Please see Figure 1 A manufacturing execution scheduling optimization system based on big data includes: The order parameter standardization module collects product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters, and operation execution parameters and categorizes them into corresponding sets. It generates a data number in a unified format for each set and combines the data number with each parameter set into a structured data package, which is then transmitted as a standardized parameter set. Product attribute parameters: Parameters describing product characteristics, such as size, color, and material. Equipment status parameters: Parameters reflecting the current operating status of the equipment, such as temperature, pressure, and speed. Tooling status parameters: Indicating the use and status of tooling, such as wear level and position calibration. Process execution parameters: Includes various operational parameters during the process, such as temperature setpoints and pressure setpoints. Operation execution parameters: Parameters describing the current status of each operation, such as operation completion time and operation pass rate.

[0021] The process version boundary solidification module reads the process version configuration parameters, extracts the upper and lower limits of dimensions and the start and end values ​​of batches and retains them as boundary pairs, establishes an independent mapping relationship between each equipment identifier and the cycle value, and generates an execution boundary dataset for merging and transmission. Process Version Configuration Parameters: These parameters configure the version information of the process flow, ensuring that specific processes are executed according to the predetermined version during production. Size Limits: These set the size range of the product or component, ensuring it meets design requirements during production or inspection. Batch Start and End Values: These define the start and end quantities of a production batch, used to manage and monitor the progress of production. Equipment and Cycle Time: These describe the operating rhythm of production equipment, ensuring stable and efficient operation during production. Execution Boundary Data Set: This set contains calculated data defining the compliant production range and conditions, used to verify and control the effectiveness of the production process.

[0022] The multi-parameter linkage verification module receives a standardized parameter set and an execution boundary data set, performs item-by-item threshold comparison and calculates the matching pass rate, establishes a joint licensing relationship based on the comparison result of the matching pass rate and the preset threshold, generates a linkage licensing matrix and transmits it. **Interval Position:** Represents the positional relationship within a numerical range, used to determine the specific location of a parameter within a specified range. **Numbering Consistency:** Ensures accurate numbering correspondence across different parameter sets, avoiding confusion caused by incorrect numbering. **Priority Position:** Indicates the priority or importance level of a parameter or state, used to ensure parameter selection for the optimal path. **Sequence Matching:** Verifies whether the execution order of parameters conforms to the preset program, preventing production problems caused by process chaos. **Linkage Permission Matrix:** A matrix generated based on the linkage relationships of multiple parameters, aiding in management and decision-making regarding whether to proceed with the production process.

[0023] The execution condition offset monitoring module receives the linkage permission matrix, obtains the current parameter value and performs offset difference calculation with the initial parameter. It adopts a multiple regression model based on the principle of covariance analysis, incorporates the time index as a continuous independent variable and the covariate as a control variable into the equation, fits the adjusted offset trend coefficient, and combines the offset difference sequence with the offset trend prediction value into an offset trend transmission array and transmits it. Current parameter values: These represent parameter data collected in real time during actual production and are crucial for monitoring production status. Initial parameters: Parameter values ​​set before production begins, serving as a baseline for subsequent monitoring and analysis. Offset: The difference between the current parameter values ​​and the initial parameters, indicating parameter changes during production. Offset trend coefficient: A coefficient used to represent the trend of parameter changes, helping to assess the stability of the production process.

[0024] The local rebinding confirmation module receives the offset trend transmission array, extracts the offset difference sequence and the local estimated offset sequence from it, performs weighted correction according to preset weights, and compares it with the allowable range. For records whose corrected offset does not exceed the allowable range, the original binding is maintained. For records whose corrected offset exceeds the allowable range, boundary matching is re-executed to generate a local binding identifier. After summarizing the binding status sequence to determine the validity, the final execution license number is output.

[0025] Boundary matching: Compares the current state with preset conditions to ensure that the production process meets the approval standards. Local binding identifier: Represents the confirmation and marking of a specific parameter state after reassessment, used for subsequent control and decision-making.

[0026] The order parameter standardization module includes product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters, and operation execution parameters; the process version boundary solidification module includes process version configuration parameters, size upper and lower limits, batch start and end values, and execution boundary data; the multi-parameter linkage verification module specifically refers to the standardized parameter set, execution boundary data set, and linkage permission matrix; the execution condition offset monitoring module includes offset, offset trend coefficient, and offset trend transmission array; the local rebinding confirmation module specifically includes offset difference sequence, local estimated offset sequence, local binding identifier, and final execution permission number.

[0027] Please see Figure 2 The order parameter standardization module includes: First, the system reads the first batch of records from the daily order sequence table in the metal processing database. The actual measured physical dimension of this record is extracted and set to 150.5. Simultaneously, using column name indexes, the following values ​​are extracted sequentially for the same record: bearing weight (set to 45.2), spindle operating temperature (set to 210.0), tool operating speed (set to 1200.0), hydraulic system pressure (set to 5.5), standard single-piece machining time (set to 30.0), and total process time (set to 45.0). Then, the system memory calls the pre-entered upper and lower boundary arrays of limit parameters, extracting the maximum dimension (300.0 and minimum dimension (0.0), the maximum weight (100.0 and minimum weight (0.0), and the temperature... The maximum reference value is 300.0 and the minimum reference value is 0.0; the maximum reference value of rotational speed is 3000.0 and the minimum reference value is 0.0; the maximum reference value of pressure is 10.0 and the minimum reference value is 0.0; the maximum reference value of time is 60.0 and the minimum reference value is 0.0; and the maximum reference value of time consumption is 100.0 and the minimum reference value is 0.0. Then, linear mapping calculations are performed on each of the above physical quantities. Specifically, a subtractor is called to subtract the corresponding minimum value from the currently extracted actual measured value, and then a divider is called to divide the subtraction difference by the difference between the corresponding maximum and minimum values. After basic arithmetic operations, the normalized result of the size is 0.502 ((150.5-0) / (300.0-0)), and the normalized result of the weight is... The normalized values ​​are 0.452 for temperature, 0.700 for speed, 0.400 for pressure, 0.550 for time, and 0.500 for duration. This converts all the original values ​​with drastically different physical units into pure floating-point normalized values ​​that are forcibly confined to the closed interval [0,1]. Next, the normalized values ​​are grouped and summed according to the five sets: the normalized value of size (0.502) and the normalized value of weight (0.452) belonging to the first set (product attributes) are added together to obtain the first subtotal value of 0.954; the normalized values ​​of temperature (0.700), speed (0.400), and pressure (0.450) belonging to the second set (equipment status) are summed together. The values ​​0.550 are added together to obtain the second sub-total value of 1.650. The normalized wear value of 0.350 assigned to the third set (tooling status) and the normalized calibration offset value of 0.200 are added together (in this example, the wear value and calibration offset are normalized to 0.350 and 0.200 respectively) to obtain the third sub-total value of 0.550. The normalized time value of 0.500 assigned to the fourth set (process execution) is used as the fourth sub-total value of 0.500. The normalized time value of 0.450 assigned to the fifth set (process execution) is used as the fifth sub-total value of 0.450. Finally, the results of these five sub-total values ​​are sent to the arithmetic logic operation module of the central processing unit to perform sequential addition and accumulation, i.e., 0.954 + 1.650 + 0.550 + 0.500 + 0.450, after carry-over and accumulation, yields a total parameter set value of 4.104.

[0028] The system calls a loop allocation instruction, which, according to the preset traversal order of five parameter sets (product attributes → equipment status → tooling status → process execution → operation execution), sequentially allocates string constants 001 to 005. Then, it calls the underlying operating system's clock function to retrieve the timestamp, reads the current server-calibrated year value, and sets it to... The built-in string extraction function extracts the last two characters of the year value to obtain the last two characters of the year string. Then, the clock pointer is called to read the current month and date values ​​and set them. moon After filtering out special format characters, the characters are concatenated into four consecutive numeric characters. The last two digits of the extracted year string With month and date strings Concatenate and merge to generate a fixed-length numeric string of six digits. After this, the system controller enters a second loop traversal process to process the six-character fixed prefix string. These are respectively compared with the incrementing sequence numbers of the three character lengths already allocated to the memory nodes of each set in the previous order. to Perform a merge operation, for example, for the first set... and By concatenating the beginning and end of the string, the first nine-digit number containing nine consecutive characters can be accurately calculated. The second and ninth digits are generated in the same manner. The third and ninth digits The fourth and ninth digits and the fifth and ninth digits Next, the system starts the merging queue of the character array buffer, arranges the five independently generated fixed-length nine-digit numbers in a linear order strictly according to the numerical value from smallest to largest, calls the system-level string concatenation machine to concatenate them end to end and perform continuous assembly, and finally synthesizes a super long string with a total length of forty-five characters in memory, and successfully outputs the total value of the data number.

[0029] For the assembly node of the standard set submodule, the system controller accesses the mapping list previously residing in memory via pointers and extracts the five independent nine-digit numbers generated in the aforementioned calculation. to Simultaneously, the system retrieves a list of five normalized parameter values ​​generated in the previous steps according to the index offset, such as the floating-point array [0.502, 0.452] of the first set (normalized values ​​for size and weight). The system calls the hash mapping library function, using each nine-digit number as a key identifier, and its corresponding floating-point array and the preset text set name as value data, forcibly establishing a one-to-one textual key-value pair association structure. After executing this mapping creation operation five times, five independent memory hash table nodes are generated. Then, the serialization and transcoding module of the standard data exchange format is called to merge and reconstruct these five hash table nodes, uniformly writing them into an empty file with a tree-like data structure. The syntax parser then converts and encapsulates the entire structure into a JSON structure object that supports cross-platform parsing. This structure is forcibly standardized to contain five independent data object records. During the serialization writing process, the checker performs a validity check to verify that the dictionary of each written record strictly contains a record number as a unique identifier primary key, similar to... Such fields, text-type collection type labels, and a list of normalized parameter values, where the unique record number is again extracted by an independent validation thread from the last two digits of the year in the system calendar. Extracting the month and date And read the set index determined by the current object's ranking, such as To perform verification-based reconstruction, the system reads the character length attribute of the concatenated string and compares it with a preset value. After confirming that the two are equal, the total number of digits is determined to be consistent with the fixed value. Bit constraints are applied, and the validated complete JSON structure is ultimately published to the public data bus via a message queue publishing command, serving as a standardized parameter set to be passed to subsequent business modules.

[0030] Please see Figure 2 The process version boundary curing module includes: The processor accesses the process configuration relational database according to process scheduling instructions, uses a select query to match the product processing specification identifier currently being executed, and first reads the upper limit value of the dimension pointed to by the cursor from the first column of the database query result set and sets it to... Read the lower limit value of the size pointed to by the cursor from the second column of the result set and set it to... The system then initiates the process of instantiating the memory array, extracting the upper limit value. and lower limit value Following the rule of arranging from largest to smallest, the data is directly filled into a contiguous array variable containing two double-precision floating-point storage spaces. The reference name of this array object is then registered and bound as a size boundary pair in the variable table, thus preserving it as an indivisible whole at the physical memory level. The processor then changes the address offset of the query cursor, extracts the batch start integer value for the current equipment capacity configuration from the production planning and scheduling data table, and sets it to... And extract the corresponding batch termination integer value and set it as Using identical memory array construction instructions, the collected initial values ​​are... With termination value Following the logic of filling in smaller to larger values, the data is then inserted into another contiguous array containing two integer data storage spaces. The reference name of this new array is registered as a batch boundary pair in the variable table, thus preserving it as a... The module ultimately transmits the memory address pointers of the two independently constructed array entity objects through the system's internal registers, completing the instantiation calculation and construction of the boundary range array object.

[0031] The system uses the Simple Network Management Protocol (SMLP) at the LAN level to send broadcast query packets to production terminal devices in the current online-ready state within the workshop network. After receiving feedback messages, it sequentially extracts the active device identifier strings, setting the first device identifier value as Device A, the second device identifier value as Device B, and the third device identifier value as Device C. Subsequently, by accessing the central process time database and tooling allocation table, it uses the device identifier as the query primary key to extract the current operating specifications and compatible tooling information of each device. The standard cycle time value corresponding to Device A is set to 42.5, and the compatible tooling list is set to [T-01, T-02, T-03]; the standard cycle time value corresponding to Device B is 46.0, and the compatible tooling list is [T-01, T-04]; the standard cycle time value corresponding to Device C is 44.5, and the compatible tooling list is [T-02, T-03, T-05]. Using the system's built-in key-value pair constructor, the string-type device identifier, the double-precision floating-point tick second value, and the string array-type compatible tooling list are strongly type-merged and bound, forming an independent triplet structure containing device A, 42.5, and [T-01, T-02, T-03]. During this operation, the system's operation controller forcibly raises the interrupt priority of process isolation, locks the currently processed device index query range, and directly blocks memory read requests across different device historical production data tables at the physical layer. In this way, at the program logic level, the cross-device regression fitting multivariate equation calculation operation containing different device feature column independent variables is resolutely not executed. Only the current single-point static performance value of the device is extracted. Then, the three binary paired data are re-sorted using the basic bubble sort algorithm, strictly following the internal code increment rule of the device identifier string in the character encoding table, and finally a standardized two-dimensional relationship table, namely the device and tick mapping table, is generated in the memory stack area.

[0032] Table 1 Equipment and Cycle Mapping Data Table , As shown in Table 1, the system completed the index binding operation of the initial corresponding relationship nodes based on this two-dimensional structure.

[0033] The system controller retrieves the memory handle of the device-to-clock mapping table, which has been established and maintained in memory as shown in Table 1. It then initiates a pointer loop to traverse the control flow, accessing each row of the two-dimensional table object. When the pointer reaches the first row, i.e., the data record containing device A, the system calls the horizontal append function to extract the size boundary pairs previously temporarily stored in the cache constant area. and the temporary batch boundary pair array, i.e. The two array objects are directly decomposed into four independent scalar numerical elements. These elements are then appended to the end of the current device A record row in strict accordance with the preset column order of upper and lower size limits, batch start, and batch end. The same decomposition and appending operation is performed on device B and device C record rows. After all three loop append instructions for the three devices have passed verification and been executed without errors, the system calls the table object's reconstruction and formatting module to redefine the data dictionary attributes of the entire two-dimensional memory table column header. This generates a data table object where each data row is forced to contain a unique device ID (e.g., device A), a floating-point standard tick (e.g., 42.5), an upper size limit (e.g., 151.5), a lower size limit (e.g., 148.5), a batch scheduling start point (e.g., 200), a batch scheduling end point (e.g., 800), and a compatible tooling list (string array) (e.g., [T-01, T-02, T-03]), totaling seven independent fields. Finally, the table's persistent output instruction is executed, delivering and archiving the entire reconstructed and encapsulated table data structure to form a standardized execution boundary dataset.

[0034] Please see Figure 2 The multi-parameter linkage verification module includes: The processor first uses an internal row index pointer to locate the first record within the normalized parameter set structure, extracts the measured values ​​of the dimensional parameters contained in that record, and sets them as follows: The actual production batch size is set as follows: The currently bound standby device identifier is set to Device A, and the pre-assigned tooling fixture identifier is set to... Then, using device A as the primary key, the corresponding row is retrieved from the execution boundary dataset to extract the upper limit of the restricted size of device A. and lower limit Batch order With batch stop It also reads the complete set of tooling capabilities that device A currently has compatibility with from the underlying hardware parameter mapping table and sets it as an array. The collected values ​​are then sent to the arithmetic logic unit for item-by-item comparison and verification. The first item comparison is for the size, and a floating-point number judgment is performed. Are they both greater than or equal to? and less than or equal to The comparator returns a true value, and the value is compared against the given constraint. The first item's flag is then marked as an integer in the temporary stack. The second comparison is performed on a batch basis, checking for integers. Are they both greater than or equal to? and less than or equal to The comparator returns a true value. After comparing the values ​​within the specified range, the second flag is also set to an integer. The third comparison checks for device existence, executing a string matching search command to determine if the device identifier value "Device A" of the current record exists in the Device ID column of the execution boundary dataset. If the retrieval system reports a successful match, the third flag is then marked as an integer 1. The fourth comparison checks for tooling compatibility, executing a string matching search command to determine if the identifier T-01 exists in the compatible tooling list [T-01, T-02, T-03] corresponding to Device A. If the retrieval system reports a successful match, the fourth flag is then marked as an integer 1. When traversing any record, if any of the above single-item comparison results throw a logical false, the flag is directly forcibly assigned the integer 0. Finally, the system calls the basic accumulator to directly add the four integer flag values ​​just generated for the current record, i.e., 1+1+1+1, calculating and outputting the integer result 4, thus obtaining the matching pass count for this record.

[0035] The processor receives the matching result of the current test record from the previous stage, which is set to 4 via a count integer. This value is then sent to the floating-point division unit as the dividend. Simultaneously, the system reads the total number of mandatory verification items hard-coded in the underlying configuration file, setting it to 4 as the divisor. The arithmetic operation of dividing the two is performed, i.e., double-precision floating-point number 4.0 divided by 4.0, yielding a quotient of 1.0, thus obtaining the matching pass rate. The system background then starts a file configuration extractor, reading the preset threshold parameter for permission control from the historical rule constraint text. This threshold is specifically set based on the statistical distribution of data from 20,000 past normal production records, and its lower limit is set to [value missing]. Extract the pre-set baseline threshold. It is then fed into the logic comparator module and compared with the calculated actual matching success rate. Perform absolute value comparisons using the greater than or equal to comparison operator. Is it greater than or equal to the threshold? If the conditional judgment is true in the current comparison example, the controller writes the bit value into the currently recorded permission status mask. This allows the program to explicitly determine within itself whether a record qualifies for joint licensing; conversely, if the calculated ratio is lower than a set threshold, the program will not grant a joint license. If the lower limit value is reached, the comparator will return a logical false value, and the controller will write the bit value into the mask. Therefore, it was determined that the record does not currently qualify for joint production approval.

[0036] The system controller initiates a global collection task, iterating through and extracting the final joint license qualification determination results of all records that have undergone joint decision-making operations within the batch. It then converts the acquired series of Boolean logic true / false indicators into machine-readable integer Boolean values, such as a result array consisting of five nodes. It forces the array element index sequence to be arranged in one dimension strictly according to the natural reading order of the underlying original parameters of each record, and retains its output as a one-dimensional Boolean permission vector object. Then, it starts the matrix construction underlying application interface, allocates two-dimensional data table storage space in the cache, and stores the previously generated data in the cache. The permission vector of the state is transposed as an independent data vector. Simultaneously, using foreign key relationships, a sequence of record numbers based on a nine-bit encoding rule, corresponding one-to-one with these decision results, is synchronously extracted from a copy of the original database. The system extracts a list of all standardized parameter measurement values ​​that were previously detected and recorded. It then performs a horizontal combination and splicing instruction on the one-dimensional permission vector generated earlier, along with the record number column and the parameter value matrix column. After splicing, it synthesizes a huge two-dimensional data structure in a memory snapshot. The output data table matrix structure contains a unique nine-digit record primary key number from the first column to the last column, a numerical set covering all collected physical quantity parameter values, and a sequence of permission Boolean feature values ​​representing the final approval conclusion. The final overall output is a linkage permission matrix.

[0037] Please see Figure 2 The execution condition offset monitoring module includes: For the error extraction process of the offset difference submodule, the module process captures and parses the complete two-dimensional linkage permission matrix table structure passed to the cache, activates cursor positioning and reads the business data object row by row, and extracts the latest measured parameter value associated with the first identification record. For example, the extracted current state size parameter value is set to... Immediately use the primary key encoding inherent in the record to initiate a query request to the cloud-based historical snapshot server, precisely retrieving the initial theoretical rated parameter value of the filing record under the corresponding primary key. The collected and extracted current parameter values Compared with the initial recorded parameter values Loading the algebraic subtraction unit to perform high-precision difference calculations, i.e., executing... minus The single absolute offset is obtained by the difference result output by the arithmetic unit. This process iterates through other valid records within the batch, performing the same historical value retrieval and subtraction operations one after another to obtain a series of independent values ​​generated by other records, such as... , , , After collecting all the differences, the system calls the built-in array sorting function package to sort the high-precision nine-bit record numbers corresponding to all the individual offsets obtained previously. to Perform a strict sorting process in ascending order of numerical values. During the sorting process, a strong reference locking instruction is used to ensure that all offset floating-point values ​​are always tightly bound to the key-value pairs of these sorted nine-bit record numbers. After the sorting and reorganization is complete, directly output the offset difference sequence with numerical continuity, which is organized by sequence number and accompanied by hash index labels. .

[0038] During the multivariate modeling calculation process, the system calculates based on the constructed... The offset difference sequence reads a time-aligned independent time index sequence set as an integer sequence. Simultaneously, multi-threaded concurrent access to the equipment manufacturing execution interface is enabled, and the covariate sequence of the actual cumulative output of the machine in the corresponding running cycle is set as follows. The device temperature covariate sequence uploaded by the spindle sensor is set as follows: The covariate sequence for the number of times the tooling is used in database maintenance is set as follows. The system calls the data analysis library to set the offset difference column as the dependent variable vector. It then uses a low-level multiple linear regression mathematical model to perform regression fitting calculation matrix operations on the aforementioned data. During the calculation construction, the extracted time integer sequence is forcibly entered as the first feature column of the design matrix as the continuous independent variable of the dominant model. The remaining acquired sequences of production, temperature, and tooling frequency are uniformly added as peripheral control variables to the multiple regression design matrix. The least squares error partial derivative is iteratively solved using the inverse method. Through matrix inner product and inverse matrix multiplication operations, various regression parameters of the current model are accurately estimated. Subsequently, the local partial regression coefficients corresponding to the continuous independent variables (i.e., time indices) are filtered out by feature name and set as constants. The slope value extracted directly is the offset trend coefficient after adjusting for external interference factors. The value of represents, at the arithmetic level, the pure rate change of the target offset over time, assuming that the effects of production accumulation, abnormal temperature changes, and blade wear are completely excluded from the coupling effects of the aforementioned covariates.

[0039] The system calls the array transformation tool to convert the sequence of five offset difference elements established and bound in the previous steps into... Strictly follow the original key-value record numbering. to The ascending order is converted into a column vector, and the arrangement is set as the first data column of a brand new empty memory matrix object. Next, the system callback has already established a multiple linear regression fitting equation containing control variables and fixed-time slope coefficients. The current discrete-time index value and the measured control variable value corresponding to each record are substituted into the equation model to start the forward theoretical calculation process. Based on algebraic substitution operations, the local theoretical estimated expected offset value at the current time node is calculated for each of the five records, and the theoretical value set is set as follows. The newly calculated expected set is also transformed using the array transformation tool and arranged in the row order of the corresponding index. It is written into the second data column that forms the new matrix object. The controller activates the table object combination wrapper and performs seamless memory combination of the first column storing the actual difference vector and the second column storing the theoretically derived offset through the row primary key matching relationship. It directly splices and outputs a final offset trend transmission array structure object containing the correspondence between the two columns of physical attributes in the system.

[0040] Please see Figure 2 The partial rebinding confirmation module includes: The main microprocessor receives a two-dimensional offset trend transmission array containing five rows and two columns of features from the predecessor task via the memory bus monitoring pipe. It then activates its internal row and column stripper to perform individual read operations according to the column offset index. First, it points to the first column address of the two-dimensional array memory block and copies out the five sequentially arranged specific difference values ​​stored within it all at once. Finally, it extracts the offset difference sequence generated by the predecessor from the first column of the offset trend transmission array, thus extracting the array. The data is then sent to the local register set. The offset index is then changed to point to the second column of the array. The sequence representing the local estimated offset stored in the node is read, and the local estimated offset sequence, i.e., the array [0.65, 0.45, 0.25, 0.55, 0.35], is accurately extracted from the second column and stored in a temporary list.

[0041] The processor reads the preset algorithm correction weights for environmental vibration interference from the equipment experience configuration table. Based on the long-term equipment aging statistical model, it sets the actual measurement difference retention weight to 0.8 and the prediction increase weight based on local estimation to 0.2. It then calls the arithmetic unit to extract the element value (0.65) corresponding to the first record in the local estimation offset sequence stored in the current temporary list, and the element value (0.7) corresponding to the first record in the offset difference sequence pointed to by the traversal. According to the set two-parameter composite logic, it performs a weighted correction operation on the offsets involved in these elements, specifically executing the first step of algebraic multiplication, i.e., performing... Multiplying 0.7 by the measured weight 0.8 and multiplying 0.65 by the predicted weight 0.2, we get temporary products of 0.56 and 0.13. Then, we perform the second step of algebraic addition, which adds the two temporary products together using an adder. After this four-operation mixture, we finally get the exact value of the offset of the record after weight smoothing, which is 0.69. The system loop control pointer then increments sequentially. According to the same rule, we perform weighted calculations on all the remaining elements in the difference sequence array. For each record, we generate the final corrected offset value that meets the model adjustment expectations and store it in a one-dimensional state table.

[0042] The controller reads the set legal range scale, i.e., the absolute upper limit of the allowed offset range is set to 0.60 and the absolute lower limit is set to -0.60. It extracts each calculated corrected offset value, such as 0.69 as mentioned in the first point above, and executes a comparison command. The comparator determines that the condition of the value being greater than or equal to -0.60 and less than or equal to 0.60 is not met (0.69 > 0.60). For records exceeding the legal range of the safety threshold, the database management system is triggered to execute an unbinding command on the original relation table to remove the established association binding relationship between the record and the initial device, and a new query is initiated to load the execution boundary data set for re-matching. If a record's correction value exceeds the limit, for example, the calculated value is... And compare beyond the aforementioned If the upper limit is reached, the system kernel will forcibly unbind the device from its original binding relationship, re-initiate a call to query and load the entire cached execution boundary data set, and extract the current actual size value of the out-of-bounds suspended record, such as... And the current batch value, such as The system utilizes verification operations to perform a search and matching comparison of the legal dimensions and batch constraint boundaries of all idle candidate devices. Once it is confirmed that the allowable size range and batch tolerance interval of a new device, such as device B, successfully encompass the aforementioned two feature parameters, the standard device cycle time value associated with this new device is extracted. After suspending the primary key of the record, a new connection pointer for device B is written to generate a new locally bound valid identifier. After the entire set is traversed, the controller summarizes the number of legal records in its internal register that have a normal status bit that has not been unbound or successfully rebound. The first record (with a corrected offset of 0.69 that exceeds the range) failed to rebind, while the other four are within the range. Therefore, the number of legal records is 4. The calculation shows that this number 4 accounts for 4% of the original The total number of received and processed sequences, i.e., 5, has a floating-point ratio of 0.8. The system proposes that the operating red line threshold set in the long-term downtime accident analysis and statistics table is 0.75. The logic judge compares and determines that the calculated ratio of 0.8 exceeds the preset validity threshold of 0.75. The system then determines that this processing batch is valid as a whole and retrieves the instruction bus to output the formatted final execution license number instruction sequence to the underlying control system. If the opposite occurs, i.e., the calculated ratio drops below the set safety threshold of 0.75, an abnormality trigger is immediately thrown. The error port of the control board outputs a stop operation alarm level signal and permanently marks all unbound records as dangerous states in the abnormality log database.

[0043] A manufacturing execution scheduling optimization method based on big data includes the following steps: S1: By collecting product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters and operation execution parameters and classifying them into corresponding sets, a data number with a unified format is generated for each set, and the data number and each parameter set are transmitted as a standardized parameter set; S2: By reading the process version configuration parameters, the difference between the upper and lower limits of the size and the start and end values ​​of the batch is calculated to form an interval, the correspondence between the equipment and the cycle time is established, and the execution boundary data set is generated and transmitted. S3: Based on the standardized parameter set and execution boundary data set, it compares the size, batch size, equipment existence, and tooling compatibility item by item, calculates the matching pass rate, establishes a joint license relationship based on the preset threshold, and generates a linkage license matrix for transmission. S4: Based on the linkage permission matrix, obtain the current parameter value and perform offset difference calculation with the initial parameter. Use a multiple regression model based on the principle of covariance analysis, take the time index as a continuous independent variable and the covariate as a control variable into the equation, fit to obtain the adjusted offset trend coefficient, and combine the offset difference sequence and the offset trend prediction value into an offset trend transmission array for transmission. S5: Based on the locally estimated offset sequence and the offset difference sequence, the offset of each record is weighted and corrected according to a preset weight; S6: Compare each corrected offset value with the allowed offset range. If the value is within the range, maintain the original binding. If the value is outside the range, re-match and generate a new local binding identifier. Extract all records with non-empty local binding identifiers to generate an executable license number list. Calculate the ratio of the number of failed records to the total number of records. If the ratio is less than the preset safety redundancy threshold, the overall scheduling is deemed valid and the executable license number list is output. At the same time, the failed records are marked as pending manual review. If the proportion of failed records exceeds the safety threshold, a global alarm signal is triggered and the scheduling process is terminated.

[0044] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A manufacturing execution scheduling optimization system based on big data, characterized in that: The system includes: The order parameter standardization module collects product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters, and operation execution parameters and categorizes them into corresponding sets. It generates a data number in a unified format for each set and combines the data number with each parameter set into a structured data packet, which is then transmitted as a standardized parameter set. The process version boundary solidification module reads the process version configuration parameters, extracts the upper and lower limits of dimensions and the start and end values ​​of batches and retains them as boundary pairs, establishes an independent mapping relationship between each equipment identifier and the cycle value, and generates an execution boundary dataset for merging and transmission. The multi-parameter linkage verification module receives the standardized parameter set and the execution boundary data set, performs item-by-item threshold comparison and calculates the matching pass rate, establishes a joint licensing relationship based on the comparison result of the matching pass rate and the preset threshold, generates a linkage licensing matrix and transmits it. The execution condition offset monitoring module receives the linkage permission matrix, obtains the current parameter value and performs offset difference calculation with the initial parameter. It adopts a multiple regression model based on the principle of covariance analysis, incorporates the time index as a continuous independent variable and the covariate as a control variable into the equation, fits the adjusted offset trend coefficient, and combines the offset difference sequence with the offset trend prediction value into an offset trend transmission array and transmits it.

2. The manufacturing execution scheduling optimization system based on big data according to claim 1, characterized in that: The order parameter standardization module includes product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters, and operation execution parameters; the process version boundary solidification module includes process version configuration parameters, size upper and lower limits, batch start and end values, and execution boundary data; the multi-parameter linkage verification module specifically refers to the standardized parameter set, the execution boundary data set, and the linkage permission matrix; the execution condition offset monitoring module includes offset amount, offset trend coefficient, and offset trend transmission array.

3. The manufacturing execution scheduling optimization system based on big data according to claim 1, characterized in that: The order parameter standardization module includes: The parameter collection submodule collects the size and weight values ​​from product attribute parameters and assigns them to the product attribute set; it collects the temperature, speed, and pressure values ​​from equipment status parameters and assigns them to the equipment status set; it collects the wear and calibration offset values ​​from tooling status parameters and assigns them to the tooling status set; it collects the time value from process execution parameters and assigns it to the process execution set; it collects the time consumption value from process execution parameters and assigns it to the process execution set; it normalizes the original values ​​in each set and converts them into dimensionless standardized values ​​in the [0,1] interval; it sums all the standardized values ​​in each set to obtain five sub-totals and adds them together to obtain the total parameter set. Number generation submodule: According to the preset collection traversal order, assign an incremental sequence number from 001 to 005 to each collection, combine the last two digits of the current year with the month and date to form a six-digit number, combine it with the sequence number to form a nine-digit number, and concatenate the five nine-digit numbers in order to generate the total value of the data number; Standard Set Submodule: Establishes key-value pair mappings between the five nine-digit numbers and their corresponding normalized parameter value lists according to the set name, and combines all mappings into a JSON structure or relational data table. Each record contains a unique record number, a set type label, and all normalized parameter values, which are used as a standardized parameter set for transmission. The unique record number is generated by concatenating the last two digits of the year, the month and date, and the set sequence number in sequence, with a fixed total of 9 digits.

4. The manufacturing execution scheduling optimization system based on big data according to claim 1, characterized in that: The process version boundary curing module includes: Interval calculation submodule: Reads the upper limit value of the size, the lower limit value of the size, the batch start value, and the batch end value, and retains the upper limit value and the lower limit value as a pair as a size boundary pair, and retains the batch start value and the end value as a pair as a batch boundary pair; The correspondence submodule reads the device identifier value and its corresponding standard beat value, pairs each device identifier with a beat value independently, and does not perform cross-device regression fitting; it then indexes each pair by the device identifier value to generate a device-beat mapping table. Boundary Set Submodule: Appends the size boundary pairs and batch boundary pairs to the end of each record in the device-cycle mapping table to form an execution boundary dataset containing seven fields: 'Device ID, Cycle Time, Upper Size Limit, Lower Size Limit, Batch Start, Batch End, and Compatible Tooling List'.

5. The manufacturing execution scheduling optimization system based on big data according to claim 1, characterized in that: The multi-parameter linkage verification module includes: Matching and Validation Submodule: The size value, batch value, equipment identifier value, and tooling identifier value of each record in the standardized parameter set are compared item by item with the corresponding size upper and lower limits, batch start and end, equipment existence, and compatible tooling list in the execution boundary dataset. Each item is marked as 1 if it passes the comparison, and 0 otherwise. The four marked values ​​are summed to obtain the matching success count. The joint relationship submodule calculates the matching pass rate by dividing the matching pass count by 4; if the pass rate is greater than or equal to a preset threshold, the record is determined to be eligible for joint permission; otherwise, it is determined not to be eligible. The license matrix submodule arranges the joint license qualification determination results of all records in the original parameter order to generate a one-dimensional Boolean license vector; the license vector is combined with the corresponding record number and parameter value to form a linked license matrix, which is a two-dimensional table structure containing the record number, all parameter values ​​and license Boolean values.

6. The manufacturing execution scheduling optimization system based on big data according to claim 1, characterized in that: The execution condition offset monitoring module includes: Offset Difference Submodule: Receives the linkage permission matrix, extracts the current parameter value of each record, performs a subtraction operation with the initial parameter value of the corresponding record to obtain a single offset, then arranges all the single offsets in ascending order of record number, and binds these record numbers to obtain the offset difference sequence. Trend coefficient submodule: Based on the offset difference sequence, read the time index sequence and at least one covariate sequence, where the covariates include cumulative output, equipment temperature, or tooling usage frequency; fit the offset difference using a multiple linear regression model, with the time index as a continuous independent variable and the covariates as control variables included in the model; estimate the regression parameters using the least squares method; and extract the regression coefficients corresponding to the time indexes, which are the adjusted offset trend coefficients, representing the average rate of change of the offset over time after excluding the influence of covariates. State sequence submodule: The offset difference sequence is arranged in the order of record number as the first column, and the local estimated offset at the corresponding time point of each record is arranged as the second column. The two columns are combined into an offset trend transmission array.

7. The manufacturing execution scheduling optimization system based on big data according to claim 1, characterized in that: The system also includes: The local rebinding confirmation module receives the offset trend transmission array, extracts the offset difference sequence and the local estimated offset sequence from it, performs weighted correction according to preset weights, and compares it with the allowable range. For records whose corrected offset does not exceed the allowable range, the original binding is maintained. For records whose corrected offset exceeds the allowable range, boundary matching is re-executed to generate a local binding identifier. After summarizing the binding status sequence to determine the validity, the final execution license number is output.

8. The manufacturing execution scheduling optimization system based on big data according to claim 7, characterized in that: The local rebinding confirmation module specifically includes an offset difference sequence, a local estimated offset sequence, a local binding identifier, and a final execution license number.

9. The manufacturing execution scheduling optimization system based on big data according to claim 7, characterized in that: The local rebinding confirmation module includes: Data extraction submodule: receives the offset trend transmission array, extracts the offset difference sequence from the first column of the offset trend transmission array, and extracts the local estimated offset sequence from the second column; Weighted correction submodule: Based on the local estimated offset sequence and the offset difference sequence, the offset of each record is weighted and corrected according to a preset weight to obtain the corrected offset value; Binding Determination Submodule: Each corrected offset value is compared one by one with the upper and lower limits of the allowed offset range. Records within the range maintain their original binding relationship; records outside the range are unbound, and the execution boundary data set is called. Based on the size value and batch value of the current record, the device cycle value corresponding to the matching size boundary and batch boundary is retrieved, and a new local binding identifier is generated. If no matching device is found, the rebinding of the record is determined to be unsuccessful, the local binding identifier of the record is set to an invalid null value, and the number of failed records is accumulated.

10. A manufacturing execution scheduling optimization method based on big data, characterized in that, The execution of a manufacturing execution scheduling optimization system based on big data according to any one of claims 1-9 includes the following steps: S1: By collecting product attribute parameters, equipment status parameters, tooling status parameters, process execution parameters and operation execution parameters and classifying them into corresponding sets, a data number with a unified format is generated for each set, and the data number and each parameter set are transmitted as a standardized parameter set; S2: By reading the process version configuration parameters, the difference between the upper and lower limits of the size and the start and end values ​​of the batch is calculated to form an interval, the correspondence between the equipment and the cycle time is established, and the execution boundary data set is generated and transmitted. S3: Based on the standardized parameter set and the execution boundary data set, compare the size, batch size, equipment existence and tooling compatibility item by item, calculate the matching pass rate, establish a joint license relationship based on a preset threshold, and generate a linkage license matrix for transmission. S4: Based on the linkage permission matrix, obtain the current parameter value and perform offset difference calculation with the initial parameter. Use a multiple regression model based on the principle of covariance analysis, take the time index as a continuous independent variable and the covariate as a control variable into the equation, fit the adjusted offset trend coefficient, and combine the offset difference sequence and the offset trend prediction value into an offset trend transmission array for transmission. S5: Based on the local estimated offset sequence and the offset difference sequence, the offset of each record is weighted and corrected according to a preset weight; S6: Compare each corrected offset value with the allowed offset range. If the offset is within the range, maintain the original binding. If the offset is outside the range, re-match and generate a new local binding identifier. Extract all records with non-empty local binding identifiers to generate an executable license number list. Calculate the ratio of the number of failed records to the total number of records. If the ratio is less than the preset safety redundancy threshold, the overall scheduling is determined to be valid and the executable license number list is output. At the same time, the failed records are marked as pending manual review. If the proportion of failed records exceeds the safety threshold, a global alarm signal is triggered and the scheduling process is terminated.

Images