Search

CN-122022614-A - Keyboard leather sheath production data model construction method and system

CN122022614ACN 122022614 ACN122022614 ACN 122022614ACN-122022614-A

Abstract

The invention discloses a keyboard leather sheath production data model construction method which comprises the following steps of collecting leather sheath production data and preprocessing to generate a standardized production data set, constructing a production entity node set and distributing unique identifiers, extracting procedure event records to construct an over-edge and establish an apex-over-edge incidence matrix to generate a basic production hypergraph, calculating over-edge weights to generate a weighted production hypergraph, constructing a cross-granularity mapping matrix and a process route sequence constraint matrix and embedding the incidence matrix to generate a time sequence constraint weighted production hypergraph, constructing a super-graph Laplace operator, executing consistency diffusion update to generate an updated production entity node feature set, and executing quantity conservation correction and index consistency check to the updated production entity node feature set to generate a trusted index data table and a cross-granularity traceability relation data table. The invention realizes the improvement of the consistency and the traceability reliability of the production data structure.

Inventors

  • SHEN SHOUMING
  • ZHANG LIU

Assignees

  • 凯晖科技股份有限公司

Dates

Publication Date
20260512
Application Date
20260415

Claims (10)

  1. 1. The method for constructing the keyboard leather sheath production data model is characterized by comprising the following steps of: Collecting leather sheath production data, and preprocessing to generate a standardized production data set; constructing leather sheath production nodes based on the standardized production data set, generating a production entity node set, and distributing a unique identifier for each production entity node; The method comprises the steps of extracting procedure event records from a standardized production data set, constructing superedges, establishing vertex-superedge incidence matrixes between a production entity node set and the superedge set, generating a basic production supergraph structure, calculating the weight of each superedge according to data source priority, time integrity scores and quantity conservation consistency scores, constructing a superedge weight matrix, a node degree matrix and a superedge degree matrix, generating a weighted production supergraph structure, constructing a cross-granularity mapping matrix from material batch nodes to semi-finished product batch nodes to finished product serial number nodes and a process route sequence constraint matrix between procedure nodes, embedding the cross-granularity mapping matrix and the process route sequence constraint matrix into the vertex-superedge incidence matrix, generating a time constraint weighted production supergraph, constructing a supergraph Laplacian based on the time constraint weighted production supergraph, introducing constraint terms of the cross-granularity mapping matrix and the process route sequence constraint matrix into the supergraph Laplacian, performing consistency diffusion updating on yield characteristics, quantity characteristics and working hour characteristics of the production entity node characteristics after updating, generating a production entity node characteristic set after updating, performing quantity conservation coefficient and consistency calibration index data and traceability data, and finishing a data structure table, and completing the construction of a data table.
  2. 2. The method for constructing a keyboard leather sheath production data model according to claim 1, wherein the acquisition of the leather sheath production data comprises acquisition order data, material batch data, process route data, work order report data, equipment operation data and quality inspection records, the preprocessing comprises the steps of performing field mapping unification, unique identification verification, time stamp standardization and quantity consistency preliminary verification processing on the acquisition of the leather sheath production data to generate a standardized production data set with consistent field names, unique numbers and time alignment and containing quantity mark fields, and the construction of the leather sheath production nodes comprises the construction of order nodes, material batch nodes, semi-finished batch nodes, finished product serial number nodes, work order nodes, process nodes, equipment nodes and personnel nodes.
  3. 3. The method for constructing the keyboard leather sheath production data model is characterized in that the method specifically comprises the steps of analyzing each procedure event record in a standardized production data set, extracting associated order nodes, material batch nodes, semi-finished product batch nodes, finished product serial number nodes, work order nodes, procedure nodes, equipment nodes and personnel nodes in the procedure event records, positioning corresponding production entity nodes according to corresponding unique identifiers in a production entity node set, distributing unique superside numbers to each procedure event record, enabling the production entity nodes associated with the procedure event records to form a corresponding superside set, constructing vertex-superside association matrixes of the row corresponding production entity node set and the column corresponding superside set, assigning 1 at corresponding positions of the matrixes when the production entity nodes participate in the corresponding superside, assigning 0 when the production entity nodes do not participate in the corresponding superside, and generating the basic production superside map structure based on the vertex-superside association matrixes, the production entity node set and the superside set.
  4. 4. The method for constructing a keyboard leather sheath production data model according to claim 3, wherein the generation of the weighted production hypergraph structure specifically comprises: Reading a data source field of a corresponding procedure event record for each superside in the superside set, mapping the data source field into a data source priority value according to a preset source level table, and writing the data source priority value into a superside attribute table; Reading a start time field and a finish time field in a process event record, calculating a time span value, calculating a difference value between the time span value and a standard beat value in process route data, mapping the difference value range into a time integrity score, and writing the time integrity score into a superb attribute table; Reading input quantity fields, output quantity fields and bad quantity fields corresponding to material batch nodes, semi-finished product batch nodes, finished product serial number nodes and work order nodes which are associated with the superside, calculating a quantity difference value between the input quantity and the output quantity plus the bad quantity, mapping the quantity difference value interval into a quantity conservation consistency score, and writing into a superside attribute table; The method comprises the steps of carrying out weighted summation calculation on data source priority values, time integrity scores and quantity conservation consistency scores corresponding to the same superside according to preset weight coefficients to obtain superside weights of the superside, constructing all the superside weights into superside weight matrixes in a diagonal form according to superside number sequences, carrying out summation calculation on the superside weights corresponding to rows of each production entity node based on vertex-superside incidence matrixes and superside weight matrixes to obtain weighted degree values of each production entity node, constructing node degree matrixes according to the number sequences of the production entity nodes, carrying out statistics on the participation quantity of each production entity node corresponding to each column in the vertex-superside incidence matrixes to obtain node number values of each superside, constructing the superside degree matrixes according to the superside number sequences, and generating a weighted production supergraph structure by taking the vertex-superside incidence matrixes, the superside weight matrixes, the node degree matrixes and the superside weight matrixes as parameters.
  5. 5. The method for constructing a keyboard leather production data model according to claim 4, wherein the construction process of the cross-granularity mapping matrix and the process route sequence constraint matrix comprises the following steps: Reading a material feeding record, a semi-finished product circulation record and a finished product serial number generation record based on a standardized production data set, extracting two-stage corresponding relations among material batch nodes, semi-finished product batch nodes and finished product serial number nodes, taking the material batch nodes as a first dimension index, taking the finished product serial number nodes as a second dimension index, constructing a cross-granularity mapping matrix, and assigning 1 to a matrix corresponding position when the material batch nodes generate corresponding finished product serial number nodes through the semi-finished product batch nodes, and assigning 0 to the matrix corresponding position when no corresponding relation exists; Carrying out row normalization processing on the cross-granularity mapping matrix to ensure that the mapping relation of the finished product serial number nodes corresponding to the nodes of the same material batch meets the constraint of consistent quantity proportion, and keeping the mapping proportion between the input quantity and the number of the finished product serial numbers consistent; Based on the process route data reading procedure number sequence, constructing an initial process route sequence matrix by taking the previous procedure node as a row index and the subsequent procedure node as a column index according to the process route sequence, and assigning 1 to the corresponding position and 0 to the rest positions when the subsequent procedure directly receives the previous procedure; Performing path closure computation on the initial process route sequence matrix to generate a process route sequence constraint matrix containing an indirect sequence relationship; And embedding the cross-granularity mapping matrix and the process route sequence constraint matrix as structure constraint parameters into a weighted production hypergraph structure.
  6. 6. The method for constructing a keyboard leather production data model according to claim 5, wherein the generating of the time sequence constraint weighted production hypergraph specifically comprises: According to the mapping relation from the material batch node in the cross-granularity mapping matrix to the finished product serial number node, carrying out association enhancement processing on the corresponding material batch node row and the finished product serial number node row in the vertex-cross-granularity mapping matrix, when the value of the corresponding position of the cross-granularity mapping matrix is larger than 0, multiplying the element value of the corresponding cross-border column in the vertex-cross-granularity mapping matrix by the mapping value to generate the cross-granularity enhancement association matrix; And combining the vertex-superside constraint matrix with the superside weight matrix, and combining the node degree matrix with the superside degree matrix together to generate a time sequence constraint weighted production supergraph.
  7. 7. The method for constructing a keyboard leather sheath production data model according to claim 6, wherein the construction process of the hypergraph laplacian comprises: Reading a vertex-superside constraint matrix, a superside weight matrix, a node degree matrix and a superside degree matrix, and performing diagonal element normalization processing on the node degree matrix and the superside degree matrix to generate a node normalization matrix and a superside normalization matrix; Sequentially performing matrix multiplication operation on the node normalization matrix, the vertex-superside constraint matrix, the superside weight matrix, the superside normalization matrix and the transpose matrix of the vertex-superside constraint matrix to generate a structure propagation matrix, and subtracting the structure propagation matrix from the unit matrix to generate a basic supergraph Laplacian; Reading a cross-granularity mapping matrix, performing dimension alignment and symmetrical expansion processing on the cross-granularity mapping matrix according to the number of the production entity nodes, constructing a cross-granularity propagation modulation matrix, performing characteristic scaling processing on the cross-granularity propagation modulation matrix, and forming propagation intensity adjusting factors by the mapping relation between nodes with different granularities; Reading a process route sequence constraint matrix, generating a sequence propagation inhibition matrix based on the process route sequence constraint matrix, assigning zero to a propagation path corresponding to a node pair violating a process route sequence relation, and maintaining an original propagation weight value for the node pair conforming to the sequence relation to generate a sequence propagation modulation matrix; And performing matrix element-by-element modulation operation on the basic super-graph Laplacian and the cross-granularity propagation modulation matrix and the sequential propagation modulation matrix respectively, enhancing the propagation channel of the basic super-graph Laplacian according to the cross-granularity mapping relation, inhibiting an illegal propagation path according to the process route sequential relation, and generating the time sequence constraint weighting production super-graph Laplacian.
  8. 8. The method for constructing a keyboard leather production data model according to claim 7, wherein the step of performing consistent diffusion update specifically comprises: Extracting yield fields, good quantity fields and man-hour fields for each production entity node in the production entity node set based on the standardized production data set, and constructing an initial node feature matrix according to the unique identification alignment of the production entity node; reading a time sequence constraint weighted production supergraph Laplacian, and carrying out matrix multiplication update on the initial node feature matrix and the time sequence constraint weighted production supergraph Laplacian to obtain a primary diffusion node feature matrix; Performing order consistency check on the feasible node feature matrix based on the process route order constraint matrix, and when the working hour characteristic corresponding to the subsequent working hour node is smaller than the working hour characteristic corresponding to the preceding working hour node, updating the working hour characteristic corresponding to the subsequent working hour node to be the sum of the working hour characteristic corresponding to the preceding working hour node and the preset minimum working hour increment, so as to generate an order correction node feature matrix; Performing cross-granularity consistency check on the sequence correction node feature matrix based on the cross-granularity mapping matrix, calculating the sum of good quality number features of a finished product serial number node set mapped by the same material batch node, scaling the good quality number features of the finished product serial number node set to be consistent with the yield features corresponding to the material batch node when the sum of the good quality number features is larger than the yield features corresponding to the material batch node, generating a cross-granularity correction node feature matrix, performing iterative convergence control on the cross-granularity correction node feature matrix, stopping iteration when the difference of two adjacent iterative node feature matrices is smaller than a preset difference threshold, and outputting an updated production entity node feature set.
  9. 9. The method for constructing the keyboard leather sheath production data model according to claim 8, wherein the step of performing quantity conservation correction and index consistency verification comprises the steps of counting the sum of input quantity characteristics, output quantity characteristics and bad quantity characteristics of material batch nodes, semi-finished product batch nodes and finished product serial number nodes according to a cross-granularity mapping relation, when the sum of the output quantity characteristics and the bad quantity characteristics is not equal to the input quantity characteristics, carrying out proportional adjustment on the output quantity characteristics of lower nodes according to a mapping proportion, enabling the sum of the output quantity characteristics and the bad quantity characteristics after adjustment to be equal to the input quantity characteristics, comparing the good quantity characteristics and the output characteristics of the same production entity node, and when the good quantity characteristics are larger than the output characteristics, adjusting the good quantity characteristics to be the output characteristics, and generating a trusted index data table.
  10. 10. A keyboard leather production data model construction system, which is applied to the keyboard leather production data model construction method as claimed in any one of claims 1 to 9, and is characterized by comprising the following modules: the data acquisition and preprocessing module is used for acquiring leather sheath production data and generating a standardized production data set; The production entity node construction module is used for constructing a production entity node set based on the standardized production data set and distributing unique identifiers; the hypergraph construction module is used for extracting the process event records, constructing hyperedges and generating a vertex-hyperedge correlation matrix; the weighted hypergraph generation module is used for calculating the hyperedge weight and constructing a hyperedge weight matrix, a node degree matrix and a hyperedge degree matrix; The time sequence constraint embedding module is used for constructing a cross-granularity mapping matrix and a process route sequence constraint matrix and generating a time sequence constraint weighted production hypergraph; The operator construction and diffusion updating module is used for constructing a super-graph Laplace operator and executing consistent diffusion updating; And the correction and output module is used for executing quantity conservation correction and index consistency verification and generating a trusted index data table and a cross-granularity traceability relation data table.

Description

Keyboard leather sheath production data model construction method and system Technical Field The invention relates to the technical field of production data modeling, in particular to a method and a system for constructing a keyboard leather sheath production data model. Background The production of the keyboard leather sheath belongs to a discrete manufacturing process, a production link relates to various data such as order management, material batch management, work order execution, process route control, equipment operation monitoring, quality inspection recording and the like, an existing manufacturing execution system stores and manages order data, batch data, process data and equipment data through a relational database, a main external key mode is adopted to establish an association relation among entities, indexes such as yield, good product quantity, working hours and the like are generated through a report statistics mode, and part of the system introduces a graph structure or a data warehouse model to integrate data for realizing production tracing and index analysis. In the technical scheme, the association between production entities is established in a pairwise relationship mode, the multi-element coupling relationship of a plurality of entities is difficult to express in the same process event, the cross-granularity tracing path is dependent on multi-table connection to be realized, the structure is complex, the stability is insufficient, the conventional system adopts a rule checking or post-processing correction mode for the data source difference, the time sequence constraint and the quantity conservation relationship, the cross-granularity mapping relationship and the process route sequence relationship are not embedded into the structure calculation process, the index consistency is dependent on manual rule setting, the propagation result is difficult to meet the quantity conservation and sequence constraint in the structure layer, the production data lacks a uniform consistency diffusion mechanism in the updating process, and when the conflict exists among the yield characteristics, the good product quantity characteristics and the man-hour characteristics, the data reliability and the tracing accuracy are influenced by a plurality of times of manual adjustment. Therefore, how to provide a method and a system for constructing a keyboard leather sheath production data model is a problem that needs to be solved by those skilled in the art. Disclosure of Invention The invention aims to provide a method and a system for constructing a keyboard leather sheath production data model, and the method and the system realize consistent diffusion updating and quantity conservation correction of the characteristics of production entity nodes by constructing a time sequence constraint weighted production hypergraph and embedding a cross-granularity mapping matrix and a process route sequence constraint matrix, and have the advantages of clear traceability relationship, strong index calculation consistency and high data reliability. The method for constructing the keyboard leather sheath production data model according to the embodiment of the invention comprises the following steps: Collecting leather sheath production data, and preprocessing to generate a standardized production data set; constructing leather sheath production nodes based on the standardized production data set, generating a production entity node set, and distributing a unique identifier for each production entity node; The method comprises the steps of extracting procedure event records from a standardized production data set, constructing superedges, establishing vertex-superedge incidence matrixes between a production entity node set and the superedge set, generating a basic production supergraph structure, calculating the weight of each superedge according to data source priority, time integrity scores and quantity conservation consistency scores, constructing a superedge weight matrix, a node degree matrix and a superedge degree matrix, generating a weighted production supergraph structure, constructing a cross-granularity mapping matrix from material batch nodes to semi-finished product batch nodes to finished product serial number nodes and a process route sequence constraint matrix between procedure nodes, embedding the cross-granularity mapping matrix and the process route sequence constraint matrix into the vertex-superedge incidence matrix, generating a time constraint weighted production supergraph, constructing a supergraph Laplacian based on the time constraint weighted production supergraph, introducing constraint terms of the cross-granularity mapping matrix and the process route sequence constraint matrix into the supergraph Laplacian, performing consistency diffusion updating on yield characteristics, quantity characteristics and working hour characteristics of the production entity node characteristics after upd