A manufacturing quality data management method and system based on an MES and a PDM system

Abstract

The application provides a manufacturing quality data management method and system based on MES and PDM systems, which comprises the following steps: performing unified semantic ontology modeling on two types of quality characteristic elements to generate standardized quality characteristic entities; taking the hierarchical affiliation of a product BOM and the logical sequence of manufacturing processes as a topological skeleton, mapping each standardized quality characteristic entity to a corresponding BOM node and process node, and constructing a cross-system quality characteristic pedigree network; based on the correlation strength of each entity in the quality characteristic pedigree network and the process transmission logic, pre-constructing a set of deviation transmission links from design quality characteristics to finished product quality performance; when receiving finished product quality abnormal data, matching the abnormal characteristics to the corresponding nodes of the quality characteristic pedigree network, and based on the set of deviation transmission links, performing reverse root cause reasoning to locate the design characteristic parameters and original quality requirements associated with the abnormality. The application can greatly improve the management efficiency of manufacturing quality data.

Description

Technical Field

[0001] This invention relates to the field of manufacturing technology, and in particular to a manufacturing quality data management method and system based on MES and PDM systems. Background Technology

[0002] With the deep application of Model-Based Definition (MBD) technology in high-end manufacturing, 3D models containing Complete Product Manufacturing Information (PMI) have become the core data carrier throughout the product lifecycle. The PDM system, as the central hub for product data management, carries the MBD model annotated with PMI, completing the definition and version control of design-side quality requirements. The MES system, on the other hand, is responsible for manufacturing quality execution processes such as process inspection, defect recording, and measured data collection on the production floor, serving as the core carrier for manufacturing-side quality data management. Currently, in the process of digital transformation in the manufacturing industry, enterprises generally rely on the collaboration of PDM and MES systems to conduct manufacturing quality data management, thereby promoting the implementation of design quality requirements on the production floor.

[0003] In existing technologies, the application of MBD 3D quality information is mostly focused on fragmented optimization of single links: some technical solutions have achieved structured extraction and classification management of MBD model PMI information within the PDM system, some solutions focus on the automatic generation of measurement programs for single inspection equipment such as coordinate measuring machines, and others have achieved the visualization and viewing of lightweight 3D models on the MES terminal. However, the above technologies are all isolated single-point improvements and have not yet formed a closed-loop quality data chain across the PDM and MES systems: PMI quality requirements such as dimensional tolerances, geometric tolerances, and inspection specifications in the MBD model cannot be automatically decomposed and transformed into process-level inspection tasks that can be executed by the MES system, and still rely on manual transcription and configuration, which is not only inefficient but also prone to distortion of quality standard transmission; at the same time, the measured quality data collected on the production site cannot be back-linked to the corresponding features of the 3D model on the PDM terminal, making it difficult to carry out feature-level quality deviation comparison and quantitative analysis.

[0004] This data gap between quality definition at the design end and quality execution at the manufacturing end further leads to derivative problems such as delayed collaboration on engineering changes and broken quality traceability chains across the entire supply chain: after design changes are released, the inspection standards at the MES end are difficult to update in a timely manner, which can easily lead to batch quality risks; when quality anomalies occur in finished products, it is impossible to quickly reverse-locate to the corresponding design features and original quality requirements, resulting in low efficiency of root cause analysis and difficulty in supporting the quality closed-loop improvement of the design-manufacturing integration. This has become the core technical bottleneck restricting the improvement of manufacturing quality data management efficiency. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a manufacturing quality data management method and system based on MES and PDM systems, so as to solve the problem that when a quality abnormality occurs in a finished product, it is impossible to quickly reverse locate the corresponding design features and original quality requirements, which leads to a reduction in the efficiency of manufacturing quality data management.

[0006] The first aspect of the present invention proposes: A manufacturing quality data management method based on MES and PDM systems, wherein the method includes: Quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system are extracted respectively. Unified semantic ontology modeling is performed on the two types of quality feature elements to generate standardized quality feature entities. Using the hierarchical relationship of the product BOM and the sequential logic of the manufacturing process as the topological skeleton, each standardized quality feature entity is mapped to the corresponding BOM node and process node to construct a cross-system quality feature spectrum network. Based on the correlation strength and process transmission logic of each entity in the quality feature spectrum network, a set of deviation transmission links from design quality features to finished product quality performance is pre-constructed. When receiving data on abnormal finished product quality, the abnormal features are matched to the corresponding nodes in the quality feature spectrum network. Based on the deviation propagation link set, reverse root cause reasoning is performed to locate the design feature parameters and original quality requirements associated with the abnormality.

[0007] The beneficial effects of this invention are as follows: This technical solution eliminates semantic barriers between quality data across systems by conducting unified semantic ontology modeling of quality feature elements in the design dimension of the PDM system and the manufacturing dimension of the MES system, achieving standardized and unified expression of quality features; a quality feature spectrum network constructed with the product BOM hierarchical relationship and the sequential logic of manufacturing processes as the topological framework connects the links between quality elements in the design and manufacturing stages; and a pre-constructed deviation transmission link set provides a clear reasoning basis for quality traceability. When receiving abnormal finished product quality data, the system can quickly match abnormal features with spectrum network nodes and accurately locate the corresponding design feature parameters and original quality requirements along the transmission link, effectively solving the pain point of difficulty in tracing the source of finished product quality abnormalities back to the design end, significantly shortening the time for quality root cause location, significantly improving the efficiency of manufacturing quality data management and the response speed of abnormality handling, and enhancing the integrated design and manufacturing quality control capabilities.

[0008] Furthermore, the step of performing unified semantic ontology modeling on the two types of quality feature elements to generate standardized quality feature entities includes: The quality characteristic elements in the design dimension are categorized into geometric tolerances, material properties, and assembly constraints; the quality characteristic elements in the manufacturing dimension are categorized into process inspections, equipment parameters, and environmental conditions. Define unified attribute fields and association rules for each type of quality feature element to form a quality feature ontology system; The two types of quality feature elements extracted are mapped to the quality feature ontology system respectively, and standardized quality feature entities are generated accordingly.

[0009] Furthermore, the step of using the hierarchical relationship of the product BOM and the sequential logic of manufacturing processes as the topological skeleton, mapping each standardized quality feature entity to the corresponding BOM node and process node, and constructing a cross-system quality feature hierarchy network includes: Set each component node in the product BOM and each process node in the manufacturing process as a topological skeleton node, and establish dependent edges and temporal edges between skeleton nodes according to the BOM hierarchy and process sequence. Standardized quality feature entities are treated as attribute nodes and attached to their corresponding skeleton nodes. Calculate the feature transfer correlation degree between different attribute nodes, establish the transfer edges between attribute nodes based on the correlation degree threshold, and form a corresponding quality feature spectrum network.

[0010] Furthermore, the step of pre-constructing a set of deviation transmission links from design quality features to finished product quality performance based on the association strength of each entity in the quality feature spectrum network and the process transmission logic includes: Along the forward flow direction of the process and the top-to-bottom assembly direction of the BOM, traverse all design dimension attribute nodes in the quality feature spectrum network to generate all reachable paths from design features to finished product features. Retrieve historical quality data and calculate the deviation transmission coefficient on each reachable path; The reachable paths with deviation propagation coefficients are organized into a deviation propagation link set.

[0011] Furthermore, the step of performing reverse root cause reasoning based on the deviation propagation link set to locate the design feature parameters and original quality requirements associated with the anomaly includes: Map the abnormal indicators in the finished product quality anomaly data to the finished product attribute nodes in the quality feature spectrum network to determine the anomaly initiation node; Backtrack all upstream nodes along the deviation propagation link set, and calculate the deviation contribution of each upstream design dimension node by combining the deviation propagation coefficient of each link; Sort the design dimension nodes by deviation contribution from high to low, and extract the design feature parameters and original quality requirements corresponding to the top-ranked design dimension nodes.

[0012] Furthermore, after the steps of locating and associating the design feature parameters and original quality requirements with the anomalies, the method further includes: The actual deviation data of this anomaly is compared with the design feature parameters obtained from the location to verify the accuracy of the transmission coefficient of the corresponding deviation transmission link; When the verification deviation exceeds the preset range, the deviation propagation coefficient of the corresponding link is updated, and the deviation propagation link set is corrected synchronously.

[0013] Furthermore, the steps of extracting quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system respectively include: Named entity recognition is performed on unstructured design drawings and technical requirement documents in the PDM system to extract quality feature elements of the design dimension; Field parsing is performed on structured inspection reports and equipment operation logs in the MES system to extract quality feature elements from the manufacturing dimension; The two types of quality feature elements extracted were preprocessed by deduplication and normalization.

[0014] The second aspect of the present invention proposes: A manufacturing quality data management system based on MES and PDM systems, wherein the system includes: The modeling module is used to extract quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system respectively, perform unified semantic ontology modeling on the two types of quality feature elements, and generate standardized quality feature entities. The building module is used to map each standardized quality feature entity to the corresponding BOM node and process node using the hierarchical relationship of the product BOM and the sequential logic of the manufacturing process as the topological skeleton, thereby constructing a cross-system quality feature spectrum network. The link module is used to pre-build a set of deviation transmission links from design quality features to finished product quality performance based on the association strength of each entity in the quality feature spectrum network and the process transmission logic. The inference module is used to match abnormal features to the corresponding nodes in the quality feature spectrum network when receiving abnormal finished product quality data. Based on the deviation propagation link set, it performs reverse root cause inference to locate the design feature parameters and original quality requirements associated with the abnormality.

[0015] Furthermore, the step of performing unified semantic ontology modeling on the two types of quality feature elements to generate standardized quality feature entities includes: The quality characteristic elements in the design dimension are categorized into geometric tolerances, material properties, and assembly constraints; the quality characteristic elements in the manufacturing dimension are categorized into process inspections, equipment parameters, and environmental conditions. Define unified attribute fields and association rules for each type of quality feature element to form a quality feature ontology system; The two types of quality feature elements extracted are mapped to the quality feature ontology system respectively, and standardized quality feature entities are generated accordingly.

[0016] Furthermore, the step of using the hierarchical relationship of the product BOM and the sequential logic of manufacturing processes as the topological skeleton, mapping each standardized quality feature entity to the corresponding BOM node and process node, and constructing a cross-system quality feature hierarchy network includes: Set each component node in the product BOM and each process node in the manufacturing process as a topological skeleton node, and establish dependent edges and temporal edges between skeleton nodes according to the BOM hierarchy and process sequence. Standardized quality feature entities are treated as attribute nodes and attached to their corresponding skeleton nodes. Calculate the feature transfer correlation degree between different attribute nodes, establish the transfer edges between attribute nodes based on the correlation degree threshold, and form a corresponding quality feature spectrum network.

[0017] Furthermore, the step of pre-constructing a set of deviation transmission links from design quality features to finished product quality performance based on the association strength of each entity in the quality feature spectrum network and the process transmission logic includes: Along the forward flow direction of the process and the top-to-bottom assembly direction of the BOM, traverse all design dimension attribute nodes in the quality feature spectrum network to generate all reachable paths from design features to finished product features. Retrieve historical quality data and calculate the deviation transmission coefficient on each reachable path; The reachable paths with deviation propagation coefficients are organized into a deviation propagation link set.

[0018] Furthermore, the step of performing reverse root cause reasoning based on the deviation propagation link set to locate the design feature parameters and original quality requirements associated with the anomaly includes: Map the abnormal indicators in the finished product quality anomaly data to the finished product attribute nodes in the quality feature spectrum network to determine the anomaly initiation node; Backtrack all upstream nodes along the deviation propagation link set, and calculate the deviation contribution of each upstream design dimension node by combining the deviation propagation coefficient of each link; Sort the design dimension nodes by deviation contribution from high to low, and extract the design feature parameters and original quality requirements corresponding to the top-ranked design dimension nodes.

[0019] Furthermore, after the steps of locating and associating the design feature parameters and original quality requirements with the anomalies, the method further includes: The actual deviation data of this anomaly is compared with the design feature parameters obtained from the location to verify the accuracy of the transmission coefficient of the corresponding deviation transmission link; When the verification deviation exceeds the preset range, the deviation propagation coefficient of the corresponding link is updated, and the deviation propagation link set is corrected synchronously.

[0020] Furthermore, the steps of extracting quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system respectively include: Named entity recognition is performed on unstructured design drawings and technical requirement documents in the PDM system to extract quality feature elements of the design dimension; Field parsing is performed on structured inspection reports and equipment operation logs in the MES system to extract quality feature elements from the manufacturing dimension; The two types of quality feature elements extracted were preprocessed by deduplication and normalization.

[0021] The third aspect of the present invention proposes: A computer device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the manufacturing quality data management method based on the MES and PDM system as described above.

[0022] The fourth aspect of the present invention proposes: A readable storage medium having a computer program stored thereon, wherein the program, when executed by a processor, implements as described above.

[0023] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0024] Figure 1 A flowchart illustrating the manufacturing quality data management method based on MES and PDM systems provided in the first embodiment of the present invention; Figure 2 The structural block diagram of the manufacturing quality data management system based on MES and PDM systems provided in the third embodiment of the present invention.

[0025] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0026] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Several embodiments of the invention are illustrated in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.

[0027] It should be noted that when a component is said to be "fixed to" another component, it can be directly on the other component or there may be an intervening component. When a component is said to be "connected to" another component, it can be directly connected to the other component or there may be an intervening component. The terms "vertical," "horizontal," "left," "right," and similar expressions used in this document are for illustrative purposes only.

[0028] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0029] Please see Figure 1 The image shows a manufacturing quality data management method based on MES and PDM systems provided in the first embodiment of the present invention. The manufacturing quality data management method based on MES and PDM systems provided in this embodiment can significantly shorten the time for root cause location of quality issues, significantly improve the efficiency of manufacturing quality data management and the response speed of abnormal handling, and enhance the integrated quality control capabilities of design and manufacturing.

[0030] Specifically, this embodiment provides: A manufacturing quality data management method based on MES and PDM systems, wherein the method includes: Step S10: Extract the quality feature elements of the design dimension in the PDM system and the quality feature elements of the manufacturing dimension in the MES system respectively, perform unified semantic ontology modeling on the two types of quality feature elements, and generate standardized quality feature entities. It's important to note that in the traditional business architecture of manufacturing enterprises, PDM systems carry quality requirements during the product design phase, often in the form of unstructured drawings and technical documents. MES systems, on the other hand, carry quality data from the manufacturing process, typically in the form of scattered process inspection records and equipment parameters. These two types of data belong to different business domains, with significant differences in terminology, description methods, and data formats. For example, the design-level description of "cylindricity tolerance 0.02mm" and the manufacturing-level record of "outer circle runout detection value" cannot be automatically matched due to inconsistent terminology. This necessitates manual verification, which is extremely inefficient and prone to errors. By performing unified semantic ontology modeling on both types of quality characteristic elements, quality characteristics from different systems and with different expressions are mapped to standardized entities with unified specifications. This breaks down the data barriers between design and manufacturing at the semantic level, providing a unified and comparable foundation for cross-system quality characteristics, and offering a unified data benchmark for subsequent cross-system quality correlation analysis.

[0031] Step S20: Using the hierarchical relationship of the product BOM and the sequential logic of the manufacturing process as the topological skeleton, map each standardized quality feature entity to the corresponding BOM node and process node to construct a cross-system quality feature spectrum network. It's important to note that product quality characteristics do not exist independently. Instead, they are progressively transferred along the BOM hierarchy, from parts to components to finished products, and simultaneously formed through sequential manufacturing processes. Quality deviations are propagated along both structural and process dimensions. By using the BOM hierarchy and process logic as a topological framework, standardized quality characteristic entities are attached to corresponding nodes. This allows all quality characteristics to be linked to the product's physical structure and manufacturing process, forming a hierarchical network that combines spatial hierarchical relationships with temporal process sequences. This clearly presents the stage to which each quality characteristic belongs and its upstream and downstream connections, fundamentally changing the fragmented and unconnected state of quality characteristics and providing a structured network foundation for deviation propagation analysis.

[0032] Step S30: Based on the correlation strength of each entity in the quality feature spectrum network and the process transmission logic, pre-construct a set of deviation transmission links from design quality features to finished product quality performance; It's important to note that the transmission of quality deviations exhibits a clear directionality and path dependence. Parameter deviations at the design stage gradually propagate and amplify along the manufacturing process, ultimately manifesting as quality anomalies in the finished product. In traditional traceability models, the occurrence of anomalies necessitates temporary adjustments to the process route, verification of design requirements, and investigation of the impact of each step, resulting in long traceability cycles and reliance on personnel experience. By forward-looking analysis of all reachable paths from design features to finished product features, and quantifying the deviation transmission coefficient of each path using historical data, a complete set of deviation transmission links is pre-constructed. This is equivalent to digitally analyzing and quantifying all quality causal relationships in advance, transforming post-event manual investigation into pre-event digital pre-configuration. Once an anomaly occurs, there's no need for temporary causal analysis; traceability can be directly conducted based on the pre-built links, significantly improving the efficiency and accuracy of root cause identification.

[0033] Step S40: When receiving finished product quality anomaly data, match the anomaly features to the corresponding nodes in the quality feature spectrum network, perform reverse root cause reasoning based on the deviation propagation link set, and locate the design feature parameters and original quality requirements associated with the anomaly.

[0034] It's important to note that traditional quality anomaly tracing often remains at the manufacturing process level, only pinpointing which process deviation occurred. It struggles to quickly link this to the original design parameters and quality requirements, leading to a lack of precise data support for design optimization and the recurrence of similar quality issues. By matching finished product anomalies to corresponding nodes in a spectral network and tracing back along a pre-established deviation propagation path, the contribution of each upstream design feature to the anomaly can be quantified. This allows for precise identification of the core design parameters and corresponding original quality requirements that triggered the anomaly, providing clear targeting for design optimization and process improvement. This truly establishes a complete traceability channel from "finished product anomaly - manufacturing process - design source," driving quality control from "post-event rectification" to "source optimization."

[0035] Second Embodiment Furthermore, the step of performing unified semantic ontology modeling on the two types of quality feature elements to generate standardized quality feature entities includes: The quality characteristic elements in the design dimension are categorized into geometric tolerances, material properties, and assembly constraints; the quality characteristic elements in the manufacturing dimension are categorized into process inspections, equipment parameters, and environmental conditions. Define unified attribute fields and association rules for each type of quality feature element to form a quality feature ontology system; The two types of quality feature elements extracted are mapped to the quality feature ontology system respectively, and standardized quality feature entities are generated accordingly.

[0036] It's important to note that the quality characteristics of the design and manufacturing dimensions have completely different attribute systems and business meanings. Without unified classification and modeling, feature attributes will be mixed, association rules will be chaotic, and subsequent association analysis will be unsuitable. Based on the business attributes and data characteristics of the quality characteristics, the design dimension is divided into three core feature categories: geometric tolerances (including dimensional tolerances, form and position tolerances, etc.), material properties (including strength, hardness, material composition, etc.), and assembly constraints (including fit clearances, assembly torque, etc.). The manufacturing dimension is divided into three core feature categories: process inspection (including machining dimensions and form and position inspection results for each process), equipment parameters (including machine tool speed, feed rate, tool parameters, etc.), and environmental conditions (including workshop temperature, humidity, cleanliness, etc.). Through classification modeling, each feature category can be matched with corresponding attribute fields and association rules, avoiding attribute redundancy or missing attributes caused by a one-size-fits-all modeling approach, and significantly improving the adaptability and accuracy of the semantic ontology.

[0037] Attribute fields describe the core information of each quality feature. Common fields such as unique feature identifier, feature name, nominal value, upper and lower tolerances, unit of measurement, associated object, and data source are uniformly set. Specific fields are also set for different categories; for example, a datum element field is added for geometric tolerances, and an equipment number field is added for equipment parameters. Association rules define the relationships between different features, such as dependency, mapping, influence, and causality. For example, "there is a mapping verification relationship between geometric tolerance features at the design end and corresponding process inspection features at the manufacturing end," and "there is a causal influence relationship between material performance features and heat treatment process parameters." By constructing this quality feature ontology system that is independent of specific systems, a "universal language" for cross-system quality features is established. Whether it's design requirements in PDM or inspection data in MES, it can be described using this unified language, fundamentally eliminating semantic barriers between systems.

[0038] The mapping process employs multi-dimensional matching rules, combining semantic similarity of feature names, part numbers, corresponding process numbers, and parameter units for comprehensive matching, effectively avoiding mapping errors caused by homonyms or synonyms. The original quality feature elements extracted from PDM and MES are mapped to their corresponding categories in the ontology system according to their categories, populated with unified attribute fields, and standardized quality feature entities with unique identifiers are generated. After mapping, quality features that originally belonged to two separate systems and had different descriptions have become standardized entities with unified formats, attributes, and identifiers. This not only fully preserves the core information of the original features but also provides a foundation for cross-system correlation analysis, offering standardized node elements for the subsequent construction of the quality feature hierarchy network.

[0039] Furthermore, the step of using the hierarchical relationship of the product BOM and the sequential logic of manufacturing processes as the topological skeleton, mapping each standardized quality feature entity to the corresponding BOM node and process node, and constructing a cross-system quality feature hierarchy network includes: Set each component node in the product BOM and each process node in the manufacturing process as a topological skeleton node, and establish dependent edges and temporal edges between skeleton nodes according to the BOM hierarchy and process sequence. Standardized quality feature entities are treated as attribute nodes and attached to their corresponding skeleton nodes. Calculate the feature transfer correlation degree between different attribute nodes, establish the transfer edges between attribute nodes based on the correlation degree threshold, and form a corresponding quality feature spectrum network.

[0040] It's important to note that the product BOM is divided into three levels of nodes based on physical composition: part level, component level, and finished product level, reflecting the spatial structural hierarchy of the product. Manufacturing processes are categorized according to the process route, including machining, heat treatment, surface treatment, assembly, and final inspection, reflecting the temporal processing flow of the product. The BOM nodes and process nodes together form the topological skeleton. Hierarchical inclusion relationships within the BOM are represented by subordinate edges (e.g., a finished product node contains multiple component nodes, and a component node contains multiple part nodes); sequential relationships are represented by temporal edges (e.g., roughing processes precede finishing processes). This two-layered skeleton structure establishes the physical and process framework of the entire quality feature network, ensuring that subsequent quality features correspond to specific parts and process steps, allowing the network structure to perfectly align with the actual manufacturing logic of the product.

[0041] Each standardized quality feature entity has its corresponding associated object and generation process. For example, the dimensional tolerance feature of a part is simultaneously attached to the corresponding BOM part node and the corresponding machining process node, while the material hardness feature is simultaneously attached to the corresponding BOM part node and the heat treatment process node. This "dual-node attachment" method allows each quality feature to be precisely located to "which part + which process," clearly defining the generation process and associated object of each quality feature. This completely avoids the problem of quality features being disconnected from business processes and also provides a precise location benchmark for subsequent analysis of the transfer of quality features between different nodes.

[0042] Different quality characteristics exhibit varying degrees of influence from each other. For example, the dimensional tolerance of the same part directly affects the assembly precision, exhibiting a very high correlation. Conversely, the material hardness of a part has a weaker impact on surface roughness, resulting in a lower correlation. The correlation calculation employs a weighted approach combining statistical data and expert experience. First, the Pearson correlation coefficient and mutual information values ​​between characteristics are calculated based on historical quality data. Then, the weighted correction is applied using the experience and knowledge of process experts, ultimately yielding a quantified correlation value. Setting a reasonable correlation threshold, and establishing transitive edges only for nodes with acceptable correlation, accurately characterizes the influence relationships between quality characteristics, filtering out weakly correlated and invalid connections. This ensures the network's correlation while avoiding invalid connections from interfering with subsequent analysis, ultimately forming a well-structured, clearly correlated, and quantifiable quality characteristic spectrum network.

[0043] Furthermore, the step of pre-constructing a set of deviation transmission links from design quality features to finished product quality performance based on the association strength of each entity in the quality feature spectrum network and the process transmission logic includes: Along the forward flow direction of the process and the top-to-bottom assembly direction of the BOM, traverse all design dimension attribute nodes in the quality feature spectrum network to generate all reachable paths from design features to finished product features. Retrieve historical quality data and calculate the deviation transmission coefficient on each reachable path; The reachable paths with deviation propagation coefficients are organized into a deviation propagation link set.

[0044] It's important to note that the traversal process employs a depth-first search algorithm. Starting from the root node of all design dimensions, it rigorously traverses layer by layer along the forward flow of processes and the upward assembly of the BOM, until all finished product quality characteristic nodes are reached. This fully covers the entire transmission path of "design parameter deviation → process deviation → component assembly deviation → finished product quality deviation." This complete traversal ensures that all possible deviation transmission paths are covered without omission, preventing blind spots in subsequent root cause tracing due to missing paths. This guarantees the integrity of the transmission link set and provides a complete set of paths for subsequent deviation quantification.

[0045] The deviation transmission coefficient represents the degree of influence of upstream feature deviation on downstream feature. Physically, it represents the number of units of deviation in the downstream feature that result from a one-unit deviation in the upstream feature. The calculation process is based on multiple batches of historical quality data accumulated by the enterprise. Statistical methods such as multiple linear regression and maximum likelihood estimation are used to fit the correspondence between upstream and downstream feature deviations on each path. Simultaneously, the differences in transmission coefficients under normal and abnormal operating conditions are distinguished to ensure the accuracy of coefficient quantification. After quantification, each path is no longer merely a logical causal relationship but possesses calculable and comparable quantitative attributes, providing a scientific numerical basis for calculating the contribution during subsequent reverse tracing.

[0046] Each organized link contains structured fields such as a unique link ID, starting design node, intermediate process node, ending finished product node, segmented transmission coefficient, and total transmission coefficient, forming a standardized link database that supports rapid retrieval by node and link. This link set serves as a centralized carrier of quality deviation transmission patterns. After pre-compiling and quantifying all links, when finished product quality anomalies occur, there is no need to temporarily analyze causal relationships and calculate the degree of impact; the link set can be directly invoked for reverse reasoning, significantly improving the response speed and analysis accuracy of anomaly tracing.

[0047] Furthermore, the step of performing reverse root cause reasoning based on the deviation propagation link set to locate the design feature parameters and original quality requirements associated with the anomaly includes: Map the abnormal indicators in the finished product quality anomaly data to the finished product attribute nodes in the quality feature spectrum network to determine the anomaly initiation node; Backtrack all upstream nodes along the deviation propagation link set, and calculate the deviation contribution of each upstream design dimension node by combining the deviation propagation coefficient of each link; Sort the design dimension nodes by deviation contribution from high to low, and extract the design feature parameters and original quality requirements corresponding to the top-ranked design dimension nodes.

[0048] It should be noted that finished product quality anomalies are typically presented as non-conforming items or out-of-tolerance indicators. The mapping process employs multi-dimensional matching rules, combining information such as the anomaly indicator name, test item number, and the corresponding finished product model to accurately match it with the finished product attribute nodes in the genealogical network. If multiple anomalies exist simultaneously, they are located to multiple starting nodes, supporting parallel tracing of multiple anomalies. Only by accurately locating the anomaly starting node can subsequent reverse tracing proceed along the correct path, avoiding tracing errors caused by misaligned starting points. This is the primary prerequisite for ensuring the accuracy of root cause localization.

[0049] Starting from the initial node of the anomaly, all upstream nodes are traced backward along the pre-established deviation propagation path, essentially searching for all possible causes along the path of deviation transmission. The calculation logic for the deviation contribution is as follows: Contribution of a single link = Abnormal deviation value of the finished product × Propagation coefficient of the corresponding link; if multiple propagation links for the same upstream node all point to the anomaly node, the contributions of all links are summed to obtain the total deviation contribution of that node. The higher the contribution, the greater the probability that the design feature corresponding to that node is the root cause of the anomaly. Through quantitative contribution calculation, the empirical bias of manual tracing can be completely avoided, providing clear numerical basis for root cause judgment and significantly improving the accuracy of tracing.

[0050] The cumulative contribution percentage method is typically used to extract the top 80% of nodes by cumulative contribution percentage as the core root cause. The output includes not only specific design parameter values ​​but also the corresponding original design quality requirements, the relevant PDM document number, and the drawing location. Design and process engineers can directly locate the original design data and quickly determine whether the problem stems from unreasonable design tolerance settings, mismatched material requirements, or deviations in the manufacturing process from design requirements. This allows for targeted design optimization or process adjustments, completely establishing a traceability channel from finished product anomalies to the design source and effectively shortening the rectification cycle for quality issues.

[0051] Furthermore, after the steps of locating and associating the design feature parameters and original quality requirements with the anomalies, the method further includes: The actual deviation data of this anomaly is compared with the design feature parameters obtained from the location to verify the accuracy of the transmission coefficient of the corresponding deviation transmission link; When the verification deviation exceeds the preset range, the deviation propagation coefficient of the corresponding link is updated, and the deviation propagation link set is corrected synchronously.

[0052] It should be noted that the pre-built deviation transmission coefficient is calculated based on historical data. However, factors such as equipment aging, material batch changes, and process improvements during production can all cause changes in the actual deviation transmission relationship. If a static transmission coefficient is always used, the accuracy of subsequent root cause localization will gradually decrease. The verification process uses a deviation comparison method: the actual upstream deviation value of the current anomaly is multiplied by the original transmission coefficient to obtain the theoretical downstream deviation value, which is then compared with the actual downstream deviation value to calculate the relative error. If the error exceeds a preset reasonable range (usually 15%), the transmission coefficient of that link is determined to be inaccurate and needs to be updated. By verifying the link with each anomaly event, links that do not conform to actual production can be identified in a timely manner, providing a basis for subsequent coefficient updates.

[0053] The coefficient update employs incremental learning, adding the actual deviation data from the current data set to the historical dataset while assigning higher weights to recent data. This refits the propagation coefficients, ensuring they adapt to the current production status while preventing significant fluctuations caused by single outliers, thus achieving smooth iteration of the propagation set. Through this closed-loop mechanism of "use-verify-update," the deviation propagation set can continuously iterate and optimize, always staying aligned with the current production situation. This continuously improves the accuracy of reverse root cause reasoning, enabling the entire quality data management system to possess self-learning and dynamic adaptation capabilities, allowing it to continuously optimize alongside the production process.

[0054] Furthermore, the steps of extracting quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system respectively include: Named entity recognition is performed on unstructured design drawings and technical requirement documents in the PDM system to extract quality feature elements of the design dimension; Field parsing is performed on structured inspection reports and equipment operation logs in the MES system to extract quality feature elements from the manufacturing dimension; The two types of quality feature elements extracted were preprocessed by deduplication and normalization.

[0055] It's important to note that quality requirements in PDM systems are mostly scattered across annotations on design drawings and natural language clauses in technical requirement documents, lacking unified structured storage. Traditional methods require manual retrieval of quality requirements from each drawing and document, resulting in extremely low efficiency and a high risk of omissions. For design drawings, optical character recognition (OCR) technology is first used to extract annotation text, followed by named entity recognition (NAME) technology to identify tolerance types, values, and datum information. For technical requirement documents, natural language processing (NLP) technology is used to extract quality-related entities and attributes; for example, material performance features with a nominal value of HRC58 are extracted from "surface hardness not less than HRC58". This automated extraction technology transforms unstructured design information into structured quality feature elements, significantly improving the efficiency and accuracy of quality data extraction from the design side.

[0056] Quality data in the MES system is scattered across multiple modules such as quality management, equipment management, and environmental monitoring. Different reports and different devices have different data field definitions, making direct cross-system analysis impossible. By using a pre-defined feature classification mapping table, fields from different data sources are parsed and matched. This maps detection values ​​in inspection reports to process inspection features, operating parameters in equipment logs to equipment parameter features, and environmental monitoring data to environmental condition features. This integrates the quality data scattered across various MES modules into a unified feature set, achieving standardized extraction of multi-source quality data from the manufacturing end.

[0057] The raw feature elements extracted from the two systems may contain duplicate records, inconsistent descriptions, and inconsistent units. Directly using these for semantic modeling can lead to entity duplication and mapping errors. Deduplication addresses this by retaining the most recent and authoritative record for the same feature, based on its source and collection time. Normalization includes standardizing units of measurement (e.g., unifying inches and millimeters to millimeters), standardizing terminology (e.g., standardizing "bouncing" and "oscillation" to "circular bouncing"), and standardizing data format and precision to ensure consistent format across all feature data. This preprocessing significantly improves data quality, providing clean and standardized data input for subsequent semantic ontology modeling and effectively ensuring the accuracy of the modeling results.

[0058] Please see Figure 2 The third embodiment of the present invention provides: A manufacturing quality data management system based on MES and PDM systems, wherein the system includes: The modeling module is used to extract quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system respectively, perform unified semantic ontology modeling on the two types of quality feature elements, and generate standardized quality feature entities. The building module is used to map each standardized quality feature entity to the corresponding BOM node and process node using the hierarchical relationship of the product BOM and the sequential logic of the manufacturing process as the topological skeleton, thereby constructing a cross-system quality feature spectrum network. The link module is used to pre-build a set of deviation transmission links from design quality features to finished product quality performance based on the association strength of each entity in the quality feature spectrum network and the process transmission logic. The inference module is used to match abnormal features to the corresponding nodes in the quality feature spectrum network when receiving finished product quality abnormality data. Based on the deviation transmission link set, it performs reverse root cause inference to locate the design feature parameters and original quality requirements associated with the abnormality.

[0059] Furthermore, the step of performing unified semantic ontology modeling on the two types of quality feature elements to generate standardized quality feature entities includes: The quality characteristic elements in the design dimension are categorized into geometric tolerances, material properties, and assembly constraints; the quality characteristic elements in the manufacturing dimension are categorized into process inspections, equipment parameters, and environmental conditions. Define unified attribute fields and association rules for each type of quality feature element to form a quality feature ontology system; The two types of quality feature elements extracted are mapped to the quality feature ontology system respectively, and standardized quality feature entities are generated accordingly.

[0060] Furthermore, the step of using the hierarchical relationship of the product BOM and the sequential logic of manufacturing processes as the topological skeleton, mapping each standardized quality feature entity to the corresponding BOM node and process node, and constructing a cross-system quality feature hierarchy network includes: Set each component node in the product BOM and each process node in the manufacturing process as a topological skeleton node, and establish dependent edges and temporal edges between skeleton nodes according to the BOM hierarchy and process sequence. Standardized quality feature entities are treated as attribute nodes and attached to their corresponding skeleton nodes. Calculate the feature transfer correlation degree between different attribute nodes, establish the transfer edges between attribute nodes based on the correlation degree threshold, and form a corresponding quality feature spectrum network.

[0061] Furthermore, the step of pre-constructing a set of deviation transmission links from design quality features to finished product quality performance based on the association strength of each entity in the quality feature spectrum network and the process transmission logic includes: Along the forward flow direction of the process and the top-to-bottom assembly direction of the BOM, traverse all design dimension attribute nodes in the quality feature spectrum network to generate all reachable paths from design features to finished product features. Retrieve historical quality data and calculate the deviation transmission coefficient on each reachable path; The reachable paths with deviation propagation coefficients are organized into a deviation propagation link set.

[0062] Furthermore, the step of performing reverse root cause reasoning based on the deviation propagation link set to locate the design feature parameters and original quality requirements associated with the anomaly includes: Map the abnormal indicators in the finished product quality anomaly data to the finished product attribute nodes in the quality feature spectrum network to determine the anomaly initiation node; Backtrack all upstream nodes along the deviation propagation link set, and calculate the deviation contribution of each upstream design dimension node by combining the deviation propagation coefficient of each link; Sort the design dimension nodes by deviation contribution from high to low, and extract the design feature parameters and original quality requirements corresponding to the top-ranked design dimension nodes.

[0063] Furthermore, after the steps of locating and associating the design feature parameters and original quality requirements with the anomalies, the method further includes: The actual deviation data of this anomaly is compared with the design feature parameters obtained from the location to verify the accuracy of the transmission coefficient of the corresponding deviation transmission link; When the verification deviation exceeds the preset range, the deviation propagation coefficient of the corresponding link is updated, and the deviation propagation link set is corrected synchronously.

[0064] Furthermore, the steps of extracting quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system respectively include: Named entity recognition is performed on unstructured design drawings and technical requirement documents in the PDM system to extract quality feature elements of the design dimension; Field parsing is performed on structured inspection reports and equipment operation logs in the MES system to extract quality feature elements from the manufacturing dimension; The two types of quality feature elements extracted were preprocessed by deduplication and normalization.

[0065] The fourth embodiment of the present invention provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the manufacturing quality data management method based on the MES and PDM system as described above.

[0066] The fifth embodiment of the present invention provides a readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, it implements the manufacturing quality data management method based on the MES and PDM system as described above.

[0067] In summary, the manufacturing quality data management method and system based on MES and PDM provided in the above embodiments of the present invention can significantly shorten the time for root cause analysis of quality issues, significantly improve the efficiency of manufacturing quality data management and the response speed for handling anomalies, and enhance the integrated quality control capabilities of design and manufacturing.

[0068] It should be noted that the above modules can be functional modules or program modules, and can be implemented through software or hardware. For modules implemented through hardware, the above modules can reside in the same processor; or the above modules can be located in different processors in any combination.

[0069] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0070] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0071] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0072] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0073] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A manufacturing quality data management method based on MES and PDM systems, characterized in that, The method includes: Quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system are extracted respectively. Unified semantic ontology modeling is performed on the two types of quality feature elements to generate standardized quality feature entities. Using the hierarchical relationship of the product BOM and the sequential logic of the manufacturing process as the topological skeleton, each standardized quality feature entity is mapped to the corresponding BOM node and process node to construct a cross-system quality feature spectrum network. Based on the correlation strength and process transmission logic of each entity in the quality feature spectrum network, a set of deviation transmission links from design quality features to finished product quality performance is pre-constructed. When receiving data on abnormal finished product quality, the abnormal features are matched to the corresponding nodes in the quality feature spectrum network. Based on the deviation propagation link set, reverse root cause reasoning is performed to locate the design feature parameters and original quality requirements associated with the abnormality.

2. The manufacturing quality data management method based on MES and PDM systems according to claim 1, characterized in that, The steps of performing unified semantic ontology modeling on the two types of quality feature elements to generate standardized quality feature entities include: The quality characteristic elements in the design dimension are categorized into geometric tolerances, material properties, and assembly constraints; the quality characteristic elements in the manufacturing dimension are categorized into process inspections, equipment parameters, and environmental conditions. Define unified attribute fields and association rules for each type of quality feature element to form a quality feature ontology system; The two types of quality feature elements extracted are mapped to the quality feature ontology system respectively, and standardized quality feature entities are generated accordingly.

3. The manufacturing quality data management method based on MES and PDM systems according to claim 2, characterized in that, The steps of constructing a cross-system quality feature hierarchy network by using the hierarchical relationship of the product BOM and the sequential logic of manufacturing processes as the topological skeleton, mapping each standardized quality feature entity to the corresponding BOM node and process node, include: Set each component node in the product BOM and each process node in the manufacturing process as a topological skeleton node, and establish dependent edges and temporal edges between skeleton nodes according to the BOM hierarchy and process sequence. Standardized quality feature entities are treated as attribute nodes and attached to their corresponding skeleton nodes. Calculate the feature transfer correlation degree between different attribute nodes, establish the transfer edges between attribute nodes based on the correlation degree threshold, and form a corresponding quality feature spectrum network.

4. The manufacturing quality data management method based on MES and PDM systems according to claim 3, characterized in that, The steps for pre-constructing a set of deviation transmission links from design quality features to finished product quality performance based on the association strength and process transmission logic of each entity in the quality feature spectrum network include: Along the forward flow direction of the process and the top-to-bottom assembly direction of the BOM, traverse all design dimension attribute nodes in the quality feature spectrum network to generate all reachable paths from design features to finished product features. Retrieve historical quality data and calculate the deviation transmission coefficient on each reachable path; The reachable paths with deviation propagation coefficients are organized into a deviation propagation link set.

5. The manufacturing quality data management method based on MES and PDM systems according to claim 4, characterized in that, The steps for performing reverse root cause reasoning based on the deviation propagation link set to locate the design feature parameters and original quality requirements associated with anomalies include: Map the abnormal indicators in the finished product quality anomaly data to the finished product attribute nodes in the quality feature spectrum network to determine the anomaly initiation node; Backtrack all upstream nodes along the deviation propagation link set, and calculate the deviation contribution of each upstream design dimension node by combining the deviation propagation coefficient of each link; Sort the design dimension nodes by deviation contribution from high to low, and extract the design feature parameters and original quality requirements corresponding to the top-ranked design dimension nodes.

6. The manufacturing quality data management method based on MES and PDM systems according to claim 5, characterized in that, Following the steps of defining the design feature parameters and original quality requirements for locating and associating anomalies, the method further includes: The actual deviation data of this anomaly is compared with the design feature parameters obtained from the location to verify the accuracy of the transmission coefficient of the corresponding deviation transmission link; When the verification deviation exceeds the preset range, the deviation propagation coefficient of the corresponding link is updated, and the deviation propagation link set is corrected synchronously.

7. The manufacturing quality data management method based on MES and PDM systems according to claim 1, characterized in that, The steps of extracting quality feature elements of the design dimension from the PDM system and quality feature elements of the manufacturing dimension from the MES system respectively include: Named entity recognition is performed on unstructured design drawings and technical requirement documents in the PDM system to extract quality feature elements of the design dimension; Field parsing is performed on structured inspection reports and equipment operation logs in the MES system to extract quality feature elements from the manufacturing dimension; The two types of quality feature elements extracted were preprocessed by deduplication and normalization.

8. A manufacturing quality data management system based on MES and PDM systems, characterized in that, The system includes: The modeling module is used to extract quality feature elements from the design dimension of the PDM system and the manufacturing dimension of the MES system respectively, perform unified semantic ontology modeling on the two types of quality feature elements, and generate standardized quality feature entities. The building module is used to map each standardized quality feature entity to the corresponding BOM node and process node using the hierarchical relationship of the product BOM and the sequential logic of the manufacturing process as the topological skeleton, thereby constructing a cross-system quality feature spectrum network. The link module is used to pre-build a set of deviation transmission links from design quality features to finished product quality performance based on the association strength of each entity in the quality feature spectrum network and the process transmission logic. The inference module is used to match abnormal features to the corresponding nodes in the quality feature spectrum network when receiving abnormal finished product quality data. Based on the deviation propagation link set, it performs reverse root cause inference to locate the design feature parameters and original quality requirements associated with the abnormality.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the manufacturing quality data management method based on the MES and PDM system as described in any one of claims 1 to 7.

10. A readable storage medium having a computer program stored thereon, characterized in that, When executed by the processor, the program implements the manufacturing quality data management method based on the MES and PDM system as described in any one of claims 1 to 7.

Images