A method for tracing the root causes of quality defects by linking standard data with supply chain data.

Abstract

This invention discloses a method for tracing the root causes of quality defects by linking standard data and supply chain data, belonging to the field of quality defect tracing technology. The method includes: constructing a standard parameter model based on product design and processes, and simultaneously constructing a multi-dimensional process data model of the supply chain based on product production data; performing identifier mapping and time series alignment to establish a standard-supply chain dynamic linkage model; detecting product quality defect data and determining product quality deviation indicators and standard indicator deviation values; constructing a directed source map of product defects; quantifying the root cause defects in the directed source map of product defects, generating a defect root cause probability map, and performing root cause visualization tracing. This invention solves the technical problems of inaccurate root cause location of product quality defects and low decision-making efficiency in existing technologies, achieving the technical effect of linking standard data and supply chain data to trace the root causes of quality defects, and improving the accuracy of root cause location and decision-making efficiency.

Description

Technical Field

[0001] This invention relates to the field of quality defect tracing technology, specifically to a method for tracing the root causes of quality defects by linking standard data and supply chain data. Background Technology

[0002] In the product manufacturing process, product quality defects are often related to various factors such as process parameters, material conditions, and equipment operation at each stage of the supply chain. However, in existing technologies, standard parameter data and supply chain process data are often disconnected, making dynamic linkage difficult. This results in challenges in quickly and accurately tracing the root cause after a quality defect occurs. Traditional traceability methods rely heavily on manual inspection or single-dimensional data analysis, which is not only inefficient but also prone to overlooking key influencing factors. They fail to fully reflect the complex relationships that lead to defects, thus affecting the timeliness and accuracy of decision-making and making it difficult to meet the high-precision quality control requirements of modern production.

[0003] Existing technologies suffer from problems such as inaccurate identification of the root causes of product quality defects and low decision-making efficiency. Summary of the Invention

[0004] This application provides a method for tracing the root causes of quality defects by linking standard data and supply chain data, which is used to address the technical problems of inaccurate root cause location of product quality defects and low decision-making efficiency in the prior art.

[0005] In view of the above problems, this application provides a method for tracing the root causes of quality defects by linking standard data and supply chain data, the method comprising:

[0006] A standard parameter model is constructed based on product design and manufacturing processes, and a multi-dimensional supply chain process data model is constructed based on product production data. The standard parameter model and the multi-dimensional supply chain process data model are mapped using identifier codes and aligned with time series to establish a standard-supply chain dynamic linkage model. Product quality defect data is detected, and based on this data, product quality deviation indicators and standard indicator deviation values ​​are determined. Starting with the product quality defect data, the product quality deviation indicators are introduced to perform defect-triggered backtracking on the standard-supply chain dynamic linkage model, constructing a directed product defect tracing graph. Based on the standard indicator deviation values, the root cause defect quantification is performed on the directed product defect tracing graph, generating a defect root cause probability map, and root cause visualization and tracing are performed through the defect root cause probability map.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] A standard parameter model is constructed based on product design and manufacturing processes, and a multi-dimensional process data model of the supply chain is built based on product production data. Identification code mapping and time series alignment are performed to establish a standard-supply chain dynamic linkage model. Product quality defect data is detected, and product quality deviation indicators and standard indicator deviation values ​​are determined. Starting with the product quality defect data, the product quality deviation indicators are introduced to perform defect-triggered backtracking on the standard-supply chain dynamic linkage model, constructing a directed source map of product defects. Root cause defect quantification is performed on the directed source map of product defects to generate a defect root cause probability map, and root cause visualization and tracing are performed through the defect root cause probability map. This achieves the technical effect of linking standard data and supply chain data to trace the root causes of quality defects, improving the accuracy of root cause location and decision-making efficiency. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a flowchart illustrating a method for tracing the root causes of quality defects by linking standard data and supply chain data, as provided in the embodiments of this application.

[0011] Figure 2 This is a schematic diagram of the process for constructing a directed source map of product defects in the quality defect root cause tracing method that links standard data and supply chain data in the embodiments of this application. Detailed Implementation

[0012] This application provides a method for tracing the root causes of quality defects by linking standard data and supply chain data, which is used to address the technical problems of inaccurate root cause location of product quality defects and low decision-making efficiency in the prior art.

[0013] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0014] Examples, such as Figure 1 As shown, this application provides a method for tracing the root causes of quality defects by linking standard data and supply chain data. The method includes:

[0015] Step S100: Construct a standard parameter model based on the product design process, and simultaneously construct a multi-dimensional process data model of the supply chain based on the product production data.

[0016] Specifically, when constructing a standard parameter model based on product design processes, design requirements related to product quality, such as raw material specifications, processing accuracy thresholds, key process parameter standards, and performance test indicators, are extracted from product design documents. These indicators are then structured to form a model used to measure product qualification benchmarks. Simultaneously, when constructing a multi-dimensional process data model of the supply chain based on product production data, it is necessary to collect actual production data covering the entire supply chain, including material batch information from raw material procurement, equipment operating parameters in the production process, test results of each process, and warehousing and logistics records. By classifying and integrating this data, a multi-dimensional data model that reflects the actual state of each link in the supply chain is formed, laying the foundation for subsequent linkage with the standard parameter model.

[0017] Step S200: Map the standard parameter model to the supply chain multidimensional process data model using identification codes and align the time series to establish a standard-supply chain dynamic linkage model.

[0018] Specifically, the process begins by constructing an identifier mapping rule system, including product-component mapping rules, component-material mapping rules, and equipment-process mapping rules. Based on this system, a hierarchical relationship for product identifier mapping is established, determining the set of product identifier elements and their hierarchical order. A product identifier mapping system is then designed and used to assign and associate identifiers with the standard parameter model and the supply chain multidimensional process data model, resulting in the standard parameter identifier model and the supply chain multidimensional process data identifier model. Next, the supply chain multidimensional process data identifier model is standardized based on the standard parameter identifier model, generating a standard model for supply chain multidimensional process data. Finally, the standard parameter identifier model and the standard model for supply chain multidimensional process data are aligned over time to establish a dynamic linkage model between the standard and the supply chain.

[0019] Step S300: Detect product quality defect data, and determine product quality deviation index and standard index deviation value based on the product quality defect data.

[0020] Specifically, the process begins by acquiring raw data on product surface scratches and dimensional deviations using visual inspection equipment and dimensional measuring instruments. Duplicate data is then removed and outliers corrected according to quality data application standards to obtain standard product quality defect data. Subsequently, feature extraction algorithms are used to extract information such as defect type (e.g., appearance defects, performance defects), defect location (e.g., a specific area of ​​a component), and defect severity (e.g., scratch depth, dimensional deviation) from the standard data, forming a product defect feature set. This feature set is then classified according to a pre-defined product quality index system (e.g., appearance indicators, dimensional accuracy indicators) to determine the corresponding product quality deviation indicators (e.g., appearance roughness deviation). Finally, by calculating the difference or ratio between each data point in the defect feature set and the corresponding standard value in the standard parameter model, the standard index deviation value is obtained (e.g., scratch depth exceeding the standard value by 0.02 mm, dimensional deviation rate of 1.5%).

[0021] Step S400: Starting with the product quality defect data, introduce the product quality deviation index, perform defect-triggered backtracking on the standard-supply chain dynamic linkage model, and construct a product defect directional traceability graph.

[0022] Specifically, taking the detected product quality defect data as the starting point for tracing, and introducing the established product quality deviation indicators as key clues, a defect-triggered backtracking is initiated on the standard-supply chain dynamic linkage model: First, the defect backtracking time range is defined based on the product quality defect data. Within this range, the related processes in the standard-supply chain model backtracking range are searched based on the product quality deviation indicators to determine the product defect-related process set. Then, based on the model, defect paths are associated with these related process sets to obtain a set of related product defect paths. Subsequently, a graph structure element containing graph node elements and graph edge elements is defined. Each path information in the path set is identified by the graph node elements to obtain a product defect path node set. Based on the graph edge elements, defect cascading backtracking is performed on the node set, and finally, a directed source graph of product defects that presents the defect propagation path and node relationships is constructed.

[0023] Step S500: Based on the deviation value of the standard index, perform root cause defect quantification on the directed source map of the product defect, generate a root cause probability map, and perform root cause visualization tracing through the root cause probability map.

[0024] Specifically, when quantifying root cause defects in a directed source map of product defects based on standard indicator deviation values, the process first assigns a defect contribution rate to each process node in the source map according to the historical defect dataset of the product, determining the contribution factor information of the product defect node. Then, the standard-supply chain dynamic linkage model is triggered by the standard indicator deviation value to mark the deviation value of each process node in the source map to obtain the deviation value of the product defect node. The deviation value is then standardized to obtain the node deviation multiple. Subsequently, the contribution factor information of the product defect node is multiplied by the node deviation multiple to obtain the probability of node defect occurrence. Based on this, the probability summation quantification of the association path of each root cause node in the source map is performed to generate a defect root cause probability map. At the same time, a dynamic adjustment mechanism is introduced to adjust and correct the map according to the updated product defect dataset and the updated supply chain data. Finally, the root cause is visualized and traced through this map.

[0025] In one possible implementation, step S200 further includes:

[0026] Step S210: Construct identification code mapping rules, which include product-component mapping rules, component-material mapping rules, and equipment-process mapping rules.

[0027] Step S220: According to the identification code mapping rules, the standard parameter model and the supply chain multidimensional process data model are assigned and associated with identification codes to obtain the standard parameter identification model and the supply chain multidimensional process data identification model.

[0028] Step S230: Based on the standard parameter identification model, perform data standardization processing on the supply chain multidimensional process data identification model to generate a supply chain multidimensional process data standard model.

[0029] Step S240: Align the standard parameter identification model with the standard model of supply chain multidimensional process data in time series to establish the standard-supply chain dynamic linkage model.

[0030] Specifically, the constructed identification code mapping rules aim to establish the association logic between various elements throughout the product lifecycle. Among them, the product-component mapping rule is used to clarify the unique identification correspondence between the finished product and its constituent components, ensuring that each product can be accurately associated with all the components that make it up; the component-material mapping rule realizes the identification binding of components with the raw materials, parts and other materials required to produce the component, so that the component can be traced back to the corresponding material source; the equipment-process mapping rule identifies and associates the equipment used in the production process with the processing process performed by the equipment, thereby corresponding the equipment operation data with the specific process links. These three types of rules together constitute a complete identification association system covering products, components, materials, equipment and processes.

[0031] Based on the established identifier code mapping rules, including product-component mapping rules, component-material mapping rules, and equipment-process mapping rules, a hierarchical relationship for product identifier code mapping is first established. This clarifies the identifier code element set and hierarchical order of each element, such as products, components, materials, equipment, and processes, and then designs the corresponding product identifier code mapping system. Based on this mapping system, unique identifier codes are assigned to design indicators and standard parameters involved in the standard parameter model. At the same time, corresponding identifier codes are assigned to data elements such as material batches, equipment operating parameters, and process inspection results in the supply chain multidimensional process data model. The identifier codes are then used to bind related elements in the two models, realizing the association mapping between standard parameters and supply chain process data. Finally, a standard parameter identifier model and a supply chain multidimensional process data identifier model with unified identifiers are obtained.

[0032] Using the standard parameter identification model as a reference, the various types of data in the supply chain multidimensional process data identification model are standardized. This means that, in accordance with the parameter definitions, data format specifications, and statistical standards set in the standard parameter identification model, the raw material procurement information, production process parameters, equipment operation records, warehousing and logistics data included in the supply chain multidimensional process data identification model are uniformly standardized. This eliminates problems such as format differences, inconsistent units, and inconsistent indicator names caused by different data sources. This ensures that the processed supply chain multidimensional process data matches the standard parameter identification model in terms of structure and standards, ultimately generating a standard supply chain multidimensional process data model. This provides a consistent data foundation for subsequent time series alignment with the standard parameter identification model.

[0033] By extracting the standard execution timestamps (such as the standard start and end times of a certain process) from each process node in the standard parameter identification model and the actual operation time records (such as material input time, equipment start time, and process completion time) of the corresponding links in the standard model of supply chain multidimensional process data, a time series matching algorithm is used to accurately align the two models in the time dimension, ensuring that the standard parameters and the actual process data of the supply chain correspond one-to-one in chronological order. For example, the standard processing temperature parameter of a certain component is associated with its actual temperature monitoring data during production at the minute-level time granularity, so that the standard requirements of each time node can match the corresponding actual execution data, and finally a standard-supply chain dynamic linkage model that can reflect the dynamic correlation between standard parameters and supply chain process data in real time is established.

[0034] In one possible implementation, step S220 further includes:

[0035] Step S221: Establish a hierarchical relationship for product identification code mapping according to the identification code mapping rules.

[0036] Step S222: Based on the product identification code mapping hierarchy, determine the product identification code element set and the product identification code hierarchy order.

[0037] Step S223: Design a product identification code mapping system based on the product identification code element set and the product identification code hierarchical order.

[0038] Step S224: Based on the product identification code mapping system, perform identification code allocation and association mapping on the standard parameter model and the supply chain multidimensional process data model to obtain the standard parameter identification model and the supply chain multidimensional process data identification model.

[0039] Specifically, based on the established identifier code mapping rules, including product-component mapping rules, component-material mapping rules, and equipment-process mapping rules, the hierarchical association logic between various coding elements is sorted out to establish a hierarchical relationship for product identifier code mapping. This hierarchical relationship takes the product as the highest level, and associates it with each component that makes up the product in sequence. Each component further corresponds to the materials required for its production, and is also associated with the equipment used to process the component. The equipment is then bound to the specific process and related process parameters, forming a hierarchical structure of "product → component → material → equipment → process parameters". This clarifies the subordinate relationship and association path of each coding element in the traceability chain, providing a basic framework for the subsequent determination of the identifier code element set and the design of the mapping system.

[0040] Based on the established product identification code mapping hierarchy, core elements with unique identifiers are extracted from each level to determine the product identification code element set. This set covers key information such as product model code, unique component number, material batch code, equipment-specific ID, and process parameter identifier. Simultaneously, according to the hierarchical logic of each element in the hierarchy, the product identification code hierarchy order is clarified, i.e., arranged in the order of "product → component → material → equipment → process parameter," ensuring that each element has a clear position within the hierarchy. This provides a clear element basis and sequence specification for the subsequent design of the product identification code mapping system.

[0041] Based on the established set of product identification code elements (including product model code, component unique number, material batch code, equipment unique ID, process parameter identifier, etc.) and the hierarchical order of product identification codes, a product identification code mapping system is designed. This system uses unified coding rules to structurally link elements at each level, for example, employing a combination of "product code - component code - material code - equipment code - process parameter code." This ensures that each element's identification code not only reflects its own uniqueness but also allows for tracing back to its higher-level related elements through the coding structure. This achieves seamless data linkage across all stages of the product lifecycle at the coding level, providing an executable coding framework for the allocation and mapping of identification codes between standard parameter models and multi-dimensional process data models in the supply chain.

[0042] Based on the designed product identification code mapping system, firstly, unique identification codes are assigned to elements in the standard parameter model, such as product design indicators, component standard parameters, material specification standards, and equipment process parameter standards, according to the coding rules in the system, forming a standard parameter identification model with identification codes. Simultaneously, corresponding identification codes are assigned to data elements in the supply chain multidimensional process data model, such as product production records, component processing data, material procurement information, equipment operating parameters, and process execution results, according to the same mapping system, forming a supply chain multidimensional process data identification model. During this process, through the association relationships of the identification codes (such as binding product identification codes with corresponding component identification codes, matching component identification codes with related material identification codes, etc.), the two models are associated and mapped at the data level, ensuring that standard parameters and supply chain process data can accurately correspond through identification codes, ultimately resulting in a mutually associative standard parameter identification model and a supply chain multidimensional process data identification model.

[0043] In one possible implementation, step S300 further includes:

[0044] Step S310: Clean the product quality defect data according to the quality data application standard to obtain standard product quality defect data.

[0045] Step S320: Extract key features from the standard product quality defect data to determine the product defect feature set, which includes defect type, defect location, and defect severity.

[0046] Step S330: Classify the product defect feature set according to the product quality index system to determine the product quality deviation index.

[0047] Step S340: Calculate the deviation value of the product defect feature set based on the product quality deviation index, and determine the standard index deviation value.

[0048] Specifically, in accordance with the quality data application standards, data cleaning is performed on the collected product quality defect data. This process includes removing duplicate defect information, correcting obvious errors or anomalies in the test data, supplementing and improving key missing information in the data (such as the time and location of defect detection), and converting defect data in different formats into a standardized data format. After these processes, standardized, accurate, and complete product quality defect data is finally obtained.

[0049] For the standard product quality defect data obtained after cleaning, feature extraction methods are used to extract key information that reflects the essence of the defects, thereby determining the product defect feature set. This feature set specifically includes: defect type (identifying whether the defect belongs to the category of appearance damage, performance non-compliance, or dimensional non-compliance); defect location (precisely locating the specific component or area of ​​the product where the defect occurs); and defect severity (quantifying the severity of the defect, such as the depth of scratches, the amount of dimensional deviation, and the extent of performance parameter non-compliance). These key features comprehensively depict the specific situation of the product quality defects.

[0050] Based on a pre-constructed product quality indicator system (covering multiple dimensions of indicators such as appearance quality, dimensional accuracy, performance parameters, and material properties), the various features in the product defect feature set are matched and classified: the defect type (such as appearance scratches, dimensional deviations), defect location (such as the surface of a component, a certain assembly step), and defect severity (such as scratch depth, dimensional deviation) are matched one by one with the specific indicators in the indicator system (such as appearance integrity indicators, key dimensional tolerance indicators, performance compliance indicators, etc.). Through the correlation and matching of features and indicators, the scattered defect features are classified into the corresponding quality indicator dimensions, and finally the product quality deviation indicators reflecting the relationship between defects and standard deviations are determined, such as "appearance roughness deviation", "core component dimensional tolerance deviation", and "performance parameter compliance rate deviation".

[0051] Based on established product quality deviation indicators, and combined with defect severity information from the product defect feature set, the deviation value between the defect and the standard is calculated. Referring to the standard requirements of the corresponding quality indicators in the standard parameter model, the actual defect severity reflected in the defect feature set is quantitatively compared with the standard requirements. By calculating the difference between the two, a value that accurately reflects the degree of defect deviation from the standard is obtained. Finally, the standard indicator deviation value is determined, providing a quantitative basis for subsequent root cause analysis.

[0052] In one possible implementation, such as Figure 2 As shown, step S400 further includes:

[0053] Step S410: Starting from the product quality defect data, determine the product defect backtracking time range based on the product quality defect data.

[0054] Step S420: Based on the product quality deviation index, perform defect path association on the standard-supply chain dynamic linkage model according to the product defect backtracking time range to obtain a set of associated product defect paths.

[0055] Step S430: Based on the set of associated product defect paths, perform defect-triggered backtracking to construct a directed source graph for product defects.

[0056] Specifically, starting with product quality defect data, based on information such as the defect discovery time, the product production batch involved, and the execution period of the corresponding production process contained in the data, combined with the time nodes of the entire product production process (such as the raw material input time, the start and end time of each process, and the finished product inspection time), the time interval from the earliest process time that may have caused the defect to the time when the defect was discovered is defined. This determines the product defect backtracking time range, ensuring that the range can cover all supply chain links and production process periods related to the occurrence of the defect.

[0057] Based on the defined product defect backtracking timeframe, and using product quality deviation indicators as the core correlation element, the standard-supply chain dynamic linkage model matches and retrieves standard parameter data (such as standard values ​​for each process and quality inspection standards) with multi-dimensional supply chain process data (such as material procurement records, equipment operating parameters, and process execution results) within that timeframe. By precisely comparing the quality deviation indicators with data paths such as production process nodes, material flow paths, and equipment operation procedures for the corresponding time period in the model, the correlation paths related to defects are screened out. These paths include the raw material supply links, each production process, equipment operation process, and warehousing and logistics links. This path information is then integrated and summarized to ultimately obtain a set of related product defect paths.

[0058] Based on the set of associated product defect paths, and taking product quality defects as the trigger point, the system traces back along each path from the defect discovery stage to the upstream stage that may have caused the defect. In this process, graph node elements (such as nodes of each production process, material nodes, equipment nodes, standard parameter nodes, etc.) and graph edge elements (relationships between nodes, such as the supply and demand relationship between processes and materials, and the operation relationship between equipment and processes) are defined. According to the sequence and association logic in the path, the information of each path in the set of associated product defect paths is identified as a node, forming a set of product defect path nodes. Then, the node set is cascaded and traced back through graph edge elements, and each node is connected according to the traceability relationship. Finally, a directed traceability graph of product defects is constructed that can intuitively show the defect propagation path and the relationship between each stage.

[0059] In one possible implementation, step S420 further includes:

[0060] Step S421: Divide the backtracking range of the standard-supply chain dynamic linkage model according to the product defect backtracking time range to obtain the backtracking range of the standard-supply chain model.

[0061] Step S422: Based on the product quality deviation index, perform a related process search on the backtracking range of the standard-supply chain model to determine the set of related processes for product defects.

[0062] Step S423: Based on the standard-supply chain dynamic linkage model, perform defect path association on the product defect-related process set to obtain the related product defect path set.

[0063] Specifically, based on the established product defect backtracking time range, the standard parameter data and multi-dimensional supply chain process data contained in the standard-supply chain dynamic linkage model are divided into time-dimension ranges. Model data within this backtracking time range are selected, including product standard parameters, component processing standards, material supply information, equipment operation data, and process execution records for the corresponding period. These time-related model data related to defect backtracking are integrated to form the standard-supply chain model backtracking range, ensuring that this range accurately covers the time interval of standard parameters and supply chain process data that may be related to product quality defects.

[0064] Based on the established product quality deviation indicators, within the backtracking range of the standard-supply chain model, each deviation indicator is matched and retrieved with the process standard parameters (such as the quality standard values ​​corresponding to the process) and the process execution records (such as the actual processing parameters and test results) in the supply chain process data within that time range. Using the pre-established identification code mapping relationship in the model, the process associated with the standard parameter corresponding to the deviation indicator is located. At the same time, the execution data of the process within the backtracking time range is filtered out by combining time series alignment information. Then, the search for related processes is expanded through the correlation between processes (such as the connection between upstream and downstream processes, equipment sharing relationship, etc.). All processes directly or indirectly related to the deviation indicators are summarized and integrated to finally determine the set of processes associated with product defects.

[0065] Based on the relationships between various elements in the standard-supply chain dynamic linkage model (including the correspondence between products, components, materials, equipment, and processes formed by identification code mapping, and the temporal relationships established by time series alignment), with each process in the product defect-related process set as the core node, the model traces upwards to the related raw material procurement links (such as the supply path and supplier information of the materials used in the process), material warehousing and circulation paths, and the equipment operation links involved in the process (such as equipment numbers and operating parameter records); it extends downwards to the subsequent processing processes and testing links through which the output of the process flows. Through the dynamic linkage mechanism established in the model, these scattered related paths are structured and connected, and finally integrated to form a complete set of related product defect paths covering all relevant links from raw material supply to finished product processing.

[0066] In one possible implementation, step S430 further includes:

[0067] Step S431: Define graph structure elements, which include graph node elements and graph edge elements.

[0068] Step S432: Identify each path information in the associated product defect path set according to the graph node elements to obtain the product defect path node set.

[0069] Step S433: Based on the graph edge elements, perform defect cascading backtracking on the product defect path node set to construct the product defect directed source graph.

[0070] Specifically, graph structure elements are defined, including graph node elements and graph edge elements. Graph node elements are further divided into four types: defect nodes are represented by red circles, labeled with the defect ID and defect type; component nodes are yellow hexagons, labeled with the component ID and component name; process nodes are blue rectangles, labeled with the process ID and operation name; and resource nodes are green rhombuses, labeled with the equipment or material ID and related parameters. The rules for graph edge elements are as follows: solid arrows indicate direct relationships, such as "defect → component"; dashed arrows indicate indirect relationships; and the thickness of the arrow is proportional to the deviation matching degree; for example, when the deviation matching degree is 92%, the arrow corresponds to a thick line.

[0071] Based on the defined graph node elements (including red circular defect nodes, yellow hexagonal component nodes, blue rectangular process nodes, and green diamond-shaped resource nodes, and their respective labeling rules), each type of path information (such as involved defects, related components, processes, equipment, and materials) contained in the associated product defect path set is matched and labeled one by one. Defect-related information in the path is mapped to defect nodes and labeled with defect ID and type; involved component information is mapped to component nodes and labeled with component ID and name; production process information is mapped to process nodes and labeled with process ID and operation name; equipment and material information is mapped to resource nodes and labeled with equipment / material ID and parameters. Through this node labeling process, a product defect path node set containing all relevant nodes is formed.

[0072] Based on defined graph elements (solid arrows representing direct relationships and dashed arrows representing indirect relationships, with arrow thickness proportional to deviation matching degree), a cascading defect backtracking is performed on each node in the product defect path node set. Starting from the defect node, according to the relationship logic between nodes, nodes with direct relationships are connected by solid arrows (such as defect nodes and corresponding component nodes), and nodes with indirect relationships are connected by dashed arrows (such as process nodes and resource nodes of related equipment). The arrow thickness is adjusted according to the deviation matching degree (e.g., thicker arrows are used to connect nodes with high matching degrees). During the reverse tracing process, the relationship between each node is established sequentially, ultimately constructing a directed source graph of product defects that clearly shows the propagation path and relationship logic of defects from the source stage to the final manifestation stage.

[0073] In one possible implementation, step S500 further includes:

[0074] Step S510: Allocate the defect contribution rate of each process node in the product defect directed tracing graph according to the product historical defect dataset, and determine the product defect node contribution factor information.

[0075] Step S520: Based on the deviation value of the standard indicator, trigger the standard-supply chain dynamic linkage model to identify the deviation value of each process node in the product defect directional traceability diagram, and obtain the deviation value of the product defect node.

[0076] Step S530: Based on the product defect node contribution factor information and the product defect node deviation value, perform defect quantification on each root cause node in the product defect directed source map to obtain a defect root cause probability map.

[0077] Specifically, the random forest algorithm is used to train the historical defect dataset of the product. The parameter features of each process node in the historical data (such as running time, temperature fluctuation, parameter deviation, etc.) are used as input variables, and whether the defect occurs is used as the output label. The algorithm constructs a correlation model between node parameters and defect occurrence. Then, the model is used to calculate the feature importance score of each process node in the current defect scenario in the product defect directed tracing graph. The score is normalized and used as the defect contribution rate of each node, thereby determining the product defect node contribution factor information.

[0078] The standard-supply chain dynamic linkage model is activated by using the established standard indicator deviation value as a trigger condition. This model locates the standard parameters (such as the standard operating range of the process, the quality acceptance threshold, etc.) and actual supply chain process data (such as the actual operating parameters and test results of the node) corresponding to each process node in the product defect directional traceability diagram through the internal identification code mapping relationship and time series alignment mechanism. By comparing and calculating the actual data of each process node with the standard parameters, the deviation value between the actual value and the standard value of each node is obtained, and these deviation values ​​are marked on the corresponding process nodes in the traceability diagram, thereby obtaining the product defect node deviation value, which intuitively reflects the degree of deviation of each node from the standard requirements in actual production.

[0079] The deviation values ​​of product defect nodes are standardized by comparing and calculating the deviation value of each node with a preset standard range to determine the node deviation multiple, thereby uniformly quantifying the degree of deviation of each node. Then, the contribution factor information of product defect nodes (the contribution ratio of each node to the defect) is multiplied by the node deviation multiple to obtain the probability of defect occurrence for each node. This probability comprehensively reflects the contribution weight of the node to the defect and the actual degree of deviation. Finally, based on these node defect occurrence probabilities, the probability summation quantification is performed on the association paths of each root cause node in the product defect directed tracing graph. That is, the probability values ​​of related nodes are accumulated along the path, ultimately forming a defect root cause probability map that can intuitively show the likelihood of each root cause node and its association path causing defects.

[0080] In one possible implementation, step S530 further includes:

[0081] Step S531: Standardize the deviation values ​​of the product defect nodes and determine the node deviation multiple.

[0082] Step S532: The product of the product defect node contribution factor information and the node deviation multiple is used as the node defect occurrence probability.

[0083] Step S533: Based on the probability of the occurrence of the node defect, perform probability summation and quantification on the association path of each root cause node in the directed tracing graph of the product defect to obtain the defect root cause probability map.

[0084] Specifically, the deviation values ​​of product defect nodes are standardized. This involves converting the actual deviation value of each node into a deviation factor relative to the standard value by combining the standard value range in the standard parameter model corresponding to that node. For example, if the standard parameter range for a certain process node is 10±2, and the actual detected deviation value is 3, then by calculating the ratio of 3 to the upper limit of the standard allowable deviation of 2, the deviation factor for that node is 1.5. This standardizes and quantifies the degree of deviation at different nodes, providing comparable basic data for subsequent defect probability calculations.

[0085] The product defect node contribution factor information obtained through historical data allocation (e.g., a contribution factor of 0.4 for a certain process node, representing that the node bears an average of 40% of the influence weight in similar defects in history) is multiplied by the node deviation multiple determined after standardization (e.g., the node's actual parameters deviate from the standard by a multiple of 2.5), i.e., 0.4 × 2.5 = 1.0. The result is the probability of the defect occurring at that node. This probability value comprehensively reflects the likelihood of a defect being caused by the combined effect of the node's inherent influence weight on the defect and the current actual deviation.

[0086] Based on the calculated probability of defects occurring at each node, for each root cause node and its associated path in the directed tracing graph of product defects, the probability of defects occurring at all nodes involved is cumulatively calculated along the path direction. For example, if a root cause node is sequentially associated with process A, process B, and component C through a path, and its node defect occurrence probabilities are 0.6, 0.3, and 0.5 respectively, then the total probability of this path is 0.6 + 0.3 + 0.5 = 1.4. By performing this kind of probability summation and quantification on the associated paths of all root cause nodes, a defect root cause probability map is finally formed that can intuitively show the likelihood of defects caused by each root cause node and its path.

[0087] In one possible implementation, step S530 further includes:

[0088] Step S534: Introduce a dynamic adjustment mechanism to obtain product update defect dataset and supply chain update data based on the dynamic adjustment mechanism.

[0089] Step S535: Adjust and correct the defect root cause probability map based on the product update defect dataset and supply chain update data.

[0090] Specifically, a dynamic adjustment mechanism is introduced. This mechanism pre-sets the trigger conditions for data updates (such as a fixed time period, new defect data reaching a preset quantity threshold, or changes in key links of the supply chain). When the trigger conditions are met, the mechanism automatically retrieves newly added product defect records and defect feature parameters from the production quality inspection system to form a product update defect dataset. At the same time, it synchronously obtains the latest material procurement information, process execution data, equipment maintenance records, supplier change information, and other supply chain update data from the supply chain management system to ensure that the data obtained can reflect the latest status of production and the supply chain in a timely manner.

[0091] Based on the acquired product update defect dataset and supply chain update data, the defect root cause probability map is adjusted and corrected: the defect contribution rate of each process node is recalculated using the product update defect data, and the node deviation value is updated in combination with the supply chain update data (such as newly added material parameters, equipment operation data, etc.), thereby adjusting the node defect occurrence probability; at the same time, the association path of the root cause node is updated according to the changes in the supply chain links, and the path probability is re-quantified, so that the defect root cause probability map can reflect the latest defect association situation and impact weight, improving the accuracy and timeliness of root cause tracing.

[0092] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0093] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0094] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for tracing the root causes of quality defects by linking standard data and supply chain data, characterized in that: The method includes: A standard parameter model is constructed based on the product design process, and a multi-dimensional process data model of the supply chain is constructed based on the product production data. The standard parameter model is mapped with the supply chain multidimensional process data model by identification code mapping and time series alignment to establish a standard-supply chain dynamic linkage model. The product quality defect data is obtained through detection, and the product quality deviation index and standard index deviation value are determined based on the product quality defect data. Starting with the product quality defect data, the product quality deviation index is introduced to perform defect-triggered backtracking on the standard-supply chain dynamic linkage model and construct a product defect directional traceability graph. Based on the standard index deviation value, the root cause defect quantification is performed on the product defect directional tracing map to generate a defect root cause probability map, and the root cause is visualized and traced through the defect root cause probability map. The establishment of the standard-supply chain dynamic linkage model includes: Construct identification code mapping rules, which include product-component mapping rules, component-material mapping rules, and equipment-process mapping rules; According to the identification code mapping rule, the standard parameter model and the supply chain multidimensional process data model are assigned and associated with identification codes to obtain the standard parameter identification model and the supply chain multidimensional process data identification model. Based on the standard parameter identification model, the supply chain multidimensional process data identification model is subjected to data standardization processing to generate a supply chain multidimensional process data standard model. The standard parameter identification model is aligned with the supply chain multidimensional process data standard model over time to establish the standard-supply chain dynamic linkage model. The obtained standard parameter identification model and supply chain multidimensional process data identification model include: Based on the aforementioned identification code mapping rules, a hierarchical relationship for product identification code mapping is established; Based on the product identification code mapping hierarchy, determine the product identification code element set and the product identification code hierarchy order; Based on the product identification code element set and product identification code hierarchical order, design a product identification code mapping system; Based on the product identification code mapping system, the standard parameter model and the supply chain multidimensional process data model are assigned and associated with identification codes to obtain the standard parameter identification model and the supply chain multidimensional process data identification model. The construction of the directed source graph for product defects includes: Starting with the product quality defect data, the product defect backtracking time range is determined based on the product quality defect data; Based on the product quality deviation index, the standard-supply chain dynamic linkage model is used to associate defect paths according to the product defect backtracking time range to obtain a set of associated product defect paths. Based on the set of defect paths of the associated products, defect-triggered backtracking is performed to construct a directed source graph of product defects.

2. The method for tracing the root causes of quality defects by linking standard data and supply chain data as described in claim 1, characterized in that, The determination of product quality deviation indicators and standard indicator deviation values ​​includes: The product quality defect data is cleaned according to the quality data application standards to obtain standard product quality defect data. Key features are extracted from the standard product quality defect data to determine the product defect feature set, which includes defect type, defect location, and defect severity. The product defect feature set is classified according to the product quality index system to determine the product quality deviation index. Based on the product quality deviation index, the deviation value of the product defect feature set is calculated to determine the standard index deviation value.

3. The method for tracing the root causes of quality defects by linking standard data and supply chain data as described in claim 1, characterized in that, The method of obtaining the set of related product defect paths includes: The standard-supply chain dynamic linkage model is divided into backtracking ranges according to the product defect backtracking time range to obtain the standard-supply chain model backtracking range. Based on the product quality deviation index, the relevant process search is performed on the backtracking range of the standard-supply chain model to determine the set of processes associated with product defects; Based on the aforementioned standard-supply chain dynamic linkage model, defect paths are associated with the product defect-related process set to obtain a set of associated product defect paths.

4. The method for tracing the root causes of quality defects by linking standard data and supply chain data as described in claim 1, characterized in that, The construction of the directed source graph for product defects includes: Define graph structure elements, which include graph node elements and graph edge elements; The path information in the associated product defect path set is identified by the graph node elements to obtain the product defect path node set. Based on the graph edge elements, the product defect path node set is cascaded backtracked to construct the product defect directed source graph.

5. The method for tracing the root causes of quality defects by linking standard data and supply chain data as described in claim 1, characterized in that, The generation of the defect root cause probability map includes: Based on the product historical defect dataset, the defect contribution rate of each process node in the product defect directed tracing graph is allocated to determine the product defect node contribution factor information. Based on the deviation value of the standard indicator, the standard-supply chain dynamic linkage model is triggered to identify the deviation value of each process node in the product defect directional traceability diagram, thereby obtaining the deviation value of the product defect node. Based on the product defect node contribution factor information and the product defect node deviation value, the root cause nodes in the product defect directed source graph are quantified to obtain a defect root cause probability map.

6. The method for tracing the root causes of quality defects by linking standard data and supply chain data as described in claim 5, characterized in that, The process of obtaining the defect root cause probability map includes: The deviation values ​​of the product defect nodes are standardized to determine the node deviation multiple; The product of the product defect node contribution factor information and the node deviation multiple is used as the probability of node defect occurrence. Based on the probability of the occurrence of the node defects, the associated paths of each root cause node in the directed tracing graph of product defects are probabilistically summed and quantified to obtain the defect root cause probability map.

7. The method for tracing the root causes of quality defects by linking standard data and supply chain data as described in claim 6, characterized in that, The process of obtaining the defect root cause probability map includes: A dynamic adjustment mechanism is introduced, and product update defect dataset and supply chain update data are obtained based on the dynamic adjustment mechanism. The defect root cause probability map is adjusted and corrected based on the product update defect dataset and supply chain update data.

Images