A method for quality traceability of automotive aluminum trim products

Abstract

This invention discloses a method for quality traceability of automotive aluminum trim panels, belonging to the field of product traceability technology. Specifically, it includes: first, generating a unique root identifier code containing basic information for the raw materials of the aluminum trim panels. The root identifier code includes three basic pieces of information: supplier number, material batch number, and warehousing time. Terminals are deployed in key processes such as warehousing, stamping, and oxidation to collect process data bound to the identifier code in real time. Then, a data aggregation module integrates the data to generate a complete quality file for each product and establishes a causal mapping model of adjacent process parameters to form an anomaly traceability link. Based on this, parameter monitoring boundaries are set to achieve out-of-specification warnings and abnormal batch locking and blocking. Finally, a two-way traceability query channel is constructed, supporting forward trajectory tracking based on the identifier code and reverse raw material and process positioning based on problem characteristics, achieving full-process quality control.

Description

Technical Field

[0001] This invention relates to the field of product traceability technology, and specifically to a method for tracing the quality of automotive aluminum trim products. Background Technology

[0002] As a key component enhancing the aesthetics and texture of vehicle interiors, automotive aluminum trim panels typically involve a long, high-precision manufacturing process encompassing raw material pretreatment, precision stamping, surface anodizing, quality inspection, and final assembly and warehousing. Currently, with the automotive industry's increasing demands for refined component manufacturing, mainstream manufacturers have widely adopted automated production equipment and basic information management systems (such as MES or ERP systems) to collect and store key process parameters and product testing data. This digital approach aims to record the production history and ensure that the basic quality of components meets the OEM's (Automotive Equipment Manufacturer) standards.

[0003] However, while existing manufacturing systems possess basic data recording capabilities, they face challenges when confronted with deeper quality control requirements. The current automotive aluminum trim production process suffers from fragmented data across multiple processes, lacking logical connections. This leads to a broken quality traceability chain, hindering the precise identification of anomalies through cross-process parameter analysis. Furthermore, the absence of real-time monitoring and interception mechanisms makes it difficult to meet the demands for accurate two-way traceability and closed-loop quality control throughout the entire product lifecycle. Specifically, data from each process is often stored in isolated equipment or sub-systems, creating de facto "data silos." For instance, when a final product exhibits surface "white spot" defects, the lack of a causal mapping model between anodizing current parameters and preceding stamping stress data makes it difficult for technicians to determine the true cause of the defect, often relying solely on post-hoc speculation based on experience. More seriously, the lack of a real-time interception mechanism based on parameter boundaries means that if a subtle, imperceptible deviation occurs in a preceding stage, the affected batch cannot be immediately identified. This allows substandard semi-finished products to continue flowing to high-cost subsequent processing stages, ultimately resulting in mass scrapping and significant waste of production resources. Summary of the Invention

[0004] The purpose of this invention is to provide a method for quality traceability of automotive aluminum trim panels, thereby solving the problems in the background art:

[0005] The objective of this invention can be achieved through the following technical solutions:

[0006] A method for tracing the quality of automotive aluminum trim products includes the following steps:

[0007] S1: Generate a unique root identification code for each batch of aluminum decorative panel raw materials. This root identification code contains three basic information: supplier number, material batch number, and warehousing time.

[0008] S2: Deploy data acquisition terminals at five process nodes: raw material warehousing, stamping, anodizing, quality inspection, and assembly outbound. After the terminal is bound to the root identification code, it records the process parameters and inspection values ​​of each node in real time.

[0009] S3: Construct a quality data collection module to automatically match and integrate the parameter data collected from each process node through the root identifier code, and generate a complete quality data archive for a single aluminum decorative panel product in chronological order;

[0010] S4: Establish a parameter causal mapping model between adjacent processes, establish association rules between the output parameters of the preceding process and the quality performance of the following process, and form a cross-process parameter transmission link for anomaly tracing.

[0011] S5: Set parameter monitoring boundary values ​​for each process node. When the collected parameters exceed the boundary range, the system triggers an early warning signal and locks the current batch of products to prevent abnormal products from flowing to the downstream process.

[0012] S6: Construct a two-way traceability query channel. The forward channel traces the complete processing trajectory and final destination of the product from the root identification code, while the reverse channel locates the problematic product to the raw material batch and specific process node.

[0013] As a further aspect of the present invention: in step S1, the specific method for generating a unique root identifier code for the batch of aluminum decorative panel raw materials is as follows:

[0014] It receives the original text data of supplier number and material batch number uploaded by the raw material storage entry terminal, and at the same time, it captures the timestamp of the current moment as the entry time parameter through the time synchronization interface of the server.

[0015] The warehousing time parameter is converted into a long integer pure numeric character sequence. At the same time, a zero-digit padding operation is performed on the supplier number and material batch number so that the three sets of data reach the preset fixed character width.

[0016] Following a fixed sequence of supplier number, material batch number, and warehousing time, three sets of standardized data are sequentially concatenated using specific meaningless delimiters to generate a unique root identifier.

[0017] As a further aspect of the present invention: In step S2, the process of deploying data acquisition terminals at the five process nodes of raw material warehousing, stamping, anodizing, quality inspection, and assembly outbound, and recording the process parameters and inspection values ​​of each node in real time after the terminals are bound to the root identifier code, is as follows:

[0018] Data acquisition terminals are fixedly installed at the raw material warehousing, stamping, anodizing, quality inspection, and assembly and delivery processes. The industrial fieldbus is used to establish physical communication connections between each data acquisition terminal and the controller of the corresponding process equipment.

[0019] The data acquisition terminal drives the external optical scanning module to read the root identification code pasted on the aluminum decorative panel transfer fixture; the data acquisition terminal locks the read root identification code as the data index primary key within the current process time window to complete the logical binding;

[0020] The data acquisition terminal reads the process parameters and inspection values ​​output by the equipment controller in real time through a physical communication connection; it timestamps the read data packets and writes them into a local database field with the root identifier as the primary key of the data index.

[0021] As a further aspect of the present invention: In step S3, the process of constructing the quality data collection module, which automatically matches and integrates the parameter data collected from each process node through the root identifier code, and generates a complete quality data archive for a single aluminum decorative panel product in chronological order, is as follows:

[0022] A data access adapter unit is used to connect to the local database of the data acquisition terminal of each process, extract data packets containing root identifiers, and convert them into a preset standard data format to ensure data structure consistency.

[0023] A root identifier hierarchical indexing mechanism is constructed, which decomposes the root identifier into sub-indexes such as supplier number, material batch number, and warehousing time. Data is filtered according to the sub-index hierarchy to achieve accurate matching and association between the root identifier and the data of each process.

[0024] A timestamp calibration and sorting module was built to extract the timestamps of each matching data, and time calibration was completed based on the server time synchronization interface. The associated data were then integrated according to the calibrated time order to form a preliminary quality data sequence.

[0025] Set up an archive integration and output unit to complete and associate the data of each process after sorting by time according to the process logic from raw material warehousing to assembly and delivery, and generate a complete quality data archive of a single aluminum decorative panel with the root identifier code as the unique identifier.

[0026] As a further aspect of the present invention: In step S4, the process of establishing a parameter causal mapping model between adjacent processes, establishing association rules between the output parameters of the preceding process and the quality performance of the subsequent process, and forming a cross-process parameter transmission link for anomaly tracing is as follows:

[0027] Extract parameter data corresponding to adjacent processes from the quality data collection module, associate the preceding output parameters and subsequent quality performance data of the same product according to the root identifier code, clarify the corresponding fields of the two types of data and complete the data alignment;

[0028] Design a gradient correlation screening mechanism for process parameters. Based on the process logic of adjacent processes, eliminate preceding parameters that have no process correlation with the quality performance of subsequent processes, and retain core influencing parameters to form a candidate correlation set.

[0029] A causal inference algorithm is used to analyze the candidate association set, construct a parameter causal mapping model, quantify the causal association strength between the preceding output parameters and the subsequent quality performance, and generate deterministic association rules.

[0030] By linking the association rules of each adjacent process according to the processing sequence, and using the root identifier code as the data thread, a cross-process parameter transmission link is constructed to clarify the transmission path and association relationship of each parameter for anomaly tracing.

[0031] As a further aspect of the present invention: the specific method for analyzing the candidate association set using a causal inference algorithm, constructing a parameter causal mapping model, quantifying the causal correlation strength between preceding output parameters and subsequent quality performance, and generating deterministic association rules is as follows:

[0032] Clean the preceding output parameters and subsequent quality performance data in the candidate association set, remove missing values ​​and outliers, and unify the data format and units to ensure that the data meets the input requirements of the causal inference algorithm.

[0033] The propensity score algorithm is used to analyze the preprocessed candidate association set. By calculating the propensity score values ​​between parameters, the direct causal relationship between the preceding output parameters and the subsequent quality performance is identified.

[0034] Based on the identified direct causal relationships, a parametric causal mapping model is constructed, and the tendency score is transformed into a causal association strength index. Based on this index, deterministic association rules that satisfy the process logic are extracted.

[0035] As a further aspect of the present invention: In step S5, the process of setting parameter monitoring boundary values ​​for each process node, and triggering an early warning signal and locking the current batch of products when the collected parameters exceed the boundary range, thereby preventing abnormal products from flowing to downstream processes, is as follows:

[0036] Extract parameter data of historical qualified products from each process and corresponding quality performance data of subsequent processes, classify them by process and associate them with the complete data chain corresponding to the same identification code;

[0037] Design a process-related boundary value calibration mechanism, which combines the process logic of each process and, based on the influence weight of the previous output parameters on the current process, calibrates the historical parameter range to form the monitoring boundary values ​​of each process parameter;

[0038] The monitoring boundary values ​​of parameters for each process are entered into the system monitoring module. The system compares the parameters uploaded by the acquisition terminal with the boundary values ​​in real time. If the values ​​exceed the boundary values, an early warning is triggered. The batch is locked by the root identifier code and a prohibition instruction is sent to the downstream terminal.

[0039] As a further aspect of the present invention: In step S6, the process of constructing a two-way traceability query channel, where the forward channel tracks the complete processing trajectory and final destination of the product from the root identifier code, and the reverse channel locates the problematic product to the raw material batch and specific process node, is as follows:

[0040] Extract the complete quality data archives corresponding to all root identifiers, associate the product destination data with the assembly and outbound node records, and build a data association index system based on the root identifier;

[0041] Design a bidirectional index mapping mechanism for the root identifier code, establish a sequential association link between the root identifier code and the processing data of each process and the product destination data in the forward direction, and establish a reverse mapping relationship between the characteristics of the problem product and the root identifier code and process nodes in the reverse direction.

[0042] A query interaction module is built. During forward queries, the root identifier code is received to call the sequential association link to output the trajectory and destination. During reverse queries, the characteristics of the problem product are matched to call the reverse mapping to locate the raw material batch and process node.

[0043] The beneficial effects of this invention are:

[0044] This invention effectively breaks down data barriers between various processes in the production of automotive aluminum trim panels, enabling precise quantitative tracing of the root causes of anomalies. By constructing a data collection system centered on root identification codes, process parameters that were originally discretely stored in stamping, anodizing, and other equipment are linked into a complete logical chain, solving the "data silo" problem. In particular, the introduction of a parameter causal mapping model between adjacent processes can mathematically quantify the correlation strength between preceding outputs and subsequent quality, allowing technicians to move beyond the limitations of manual experience-based troubleshooting and quickly locate key process factors causing defects through deterministic rules automatically identified by the system. Combined with a two-way traceability channel, it can not only accurately track the entire lifecycle of a single product but also pinpoint the specific batch of problematic raw materials, significantly improving the depth and accuracy of quality analysis.

[0045] This invention establishes a real-time dynamic interception mechanism based on parameter boundaries, significantly reducing ineffective losses and batch scrap risks during the production process. Unlike traditional post-production inspection, this invention sets stringent parameter monitoring boundaries at the front-end and intermediate stages, such as raw material warehousing and stamping. Once the system detects a hidden process deviation, it immediately triggers an alert and automatically locks the electronic circulation access for that batch of products, forcibly preventing unqualified semi-finished products from flowing into the subsequent high-cost anodizing or assembly stages. This shift from passive screening to proactive interception effectively prevents the accumulation and expansion of quality risks, minimizes the waste of raw materials and equipment time caused by continuous processing of defective products, and achieves optimized and closed-loop management of production resource allocation. Attached Figure Description

[0046] The invention will now be further described with reference to the accompanying drawings.

[0047] Figure 1 This is a flowchart illustrating a method for tracing the quality of automotive aluminum trim panels according to the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] Please see Figure 1 As shown, this invention provides a method for tracing the quality of automotive aluminum trim products, comprising the following steps:

[0050] S1: Generate a unique root identification code for each batch of aluminum decorative panel raw materials. This root identification code contains three basic information: supplier number, material batch number, and warehousing time.

[0051] S2: Deploy data acquisition terminals at five process nodes: raw material warehousing, stamping, anodizing, quality inspection, and assembly outbound. After the terminal is bound to the root identification code, it records the process parameters and inspection values ​​of each node in real time.

[0052] S3: Construct a quality data collection module to automatically match and integrate the parameter data collected from each process node through the root identifier code, and generate a complete quality data archive for a single aluminum decorative panel product in chronological order;

[0053] S4: Establish a parameter causal mapping model between adjacent processes, establish association rules between the output parameters of the preceding process and the quality performance of the following process, and form a cross-process parameter transmission link for anomaly tracing.

[0054] S5: Set parameter monitoring boundary values ​​for each process node. When the collected parameters exceed the boundary range, the system triggers an early warning signal and locks the current batch of products to prevent abnormal products from flowing to the downstream process.

[0055] S6: Construct a two-way traceability query channel. The forward channel traces the complete processing trajectory and final destination of the product from the root identification code, while the reverse channel locates the problematic product to the raw material batch and specific process node.

[0056] In one embodiment of the present invention, step S1, the process of generating a unique root identifier for the batch of aluminum decorative panel raw materials, is as follows:

[0057] First, the initial collection and time anchoring of raw material information are performed. The server receives the original text data of the supplier number and material batch number uploaded by the raw material warehousing entry terminal through a dedicated encrypted communication protocol. This process ensures the authenticity of the source data. At the same time, the server calls its internal high-precision clock module or obtains the timestamp of the received data through a network time protocol interface and defines this timestamp as the warehousing time parameter. In this process, the server does not rely on the local time setting of the entry terminal but forces the use of the server's unified standard time. It should be noted that the use of the server-side timestamp instead of the terminal's local time is to eliminate possible clock drift errors between different entry devices and to prevent human tampering with the warehousing time, ensuring that all raw material warehousing records have an absolute unified benchmark and traceability in the time dimension. In addition, after receiving the text data, the server immediately establishes temporary storage space in the cache to await the next data cleaning and formatting operation. This real-time capture and locking mechanism can effectively avoid time recording deviations caused by network latency or data accumulation.

[0058] The collected basic information undergoes rigorous data standardization and format normalization. The server reads the entry time parameter from the cache and converts it from date and time format into a long integer pure numeric character sequence using an encoding conversion algorithm. This sequence typically represents the total duration calculated from a specific epoch with millisecond precision. Simultaneously, the server reads the supplier number and material batch number, and retrieves the pre-set standard character width configuration parameters to perform length verification on these two sets of data. When the actual character length is less than the preset fixed character width, the processing program automatically performs a loop padding operation on the leftmost side of the string, filling in 0s one by one until the total length of the string is strictly equal to the preset fixed character width. It should be noted that the zero-padded operation on the supplier number and material batch number is to solve the unstructured data problem caused by the different lengths of the coding rules of different suppliers and the variation in the number of digits of the material batch number. By forcibly unifying the data length, the misidentification rate of the barcode scanning equipment in the next process can be greatly reduced, and it can ensure that fixed-length fields can be used in the database to improve indexing efficiency and retrieval speed. Complex and variable text information is transformed into a standard combination of numbers and characters with a rigorous structure that is easy for machine vision to recognize.

[0059] After completing the final assembly and generation of the unique root identifier, the server extracts three sets of standardized data according to the pre-planned coding topology, strictly following the fixed arrangement order of supplier number first, material batch number in the middle, and warehousing time sequence last. A pre-selected specific meaningless separator is inserted between two adjacent sets of data. This separator is usually a special character not included in the character set of the business data itself. The three sets of data and the separator are then connected sequentially using a string concatenation function to form a complete long string, which is the unique root identifier of the aluminum decorative panel raw material batch.

[0060] In one embodiment of the present invention, in step S2, data acquisition terminals are deployed at five process nodes: raw material warehousing, stamping, anodizing, quality inspection, and assembly outbound. After the terminals are bound to the root identifier code, the process parameters and inspection values ​​of each node are recorded in real time as follows:

[0061] First, the physical deployment of hardware and the establishment of the communication network are carried out. Engineers need to fix and install industrial-grade data acquisition terminals at five key physical locations on the aluminum decorative panel production line: raw material warehousing, stamping and forming, anodizing, quality inspection, and assembly and warehousing. The installation location should be selected according to the principle of being as close as possible to the equipment controller and facilitating wiring. After the physical installation is completed, the network interface of each data acquisition terminal is physically hard-connected to the controller interface of the corresponding process equipment by laying industrial fieldbus cables to ensure the physical conduction of the data transmission link. It should be noted that the use of industrial fieldbus to establish physical communication connection instead of wireless transmission is based on the anti-interference requirements of the industrial environment. Because there is high-frequency stamping vibration and high-current electromagnetic field interference in the aluminum decorative panel processing workshop, wired connection can shield the impact of electromagnetic noise on signal transmission stability at the physical medium level.

[0062] To establish a logical link between the product's physical identity and its digital record, the control program inside the data acquisition terminal sends commands to drive the optical scanning module, which is connected to it via an external interface, into operation. When the transfer container carrying the aluminum trim moves into the field of view of the optical scanning module, the module quickly reads the root identification code affixed to the transfer container using optical imaging principles and sends the decoded character information back to the data acquisition terminal. Upon receiving the root identification code, the data acquisition terminal immediately allocates an independent time window variable area in its internal volatile memory and assigns the read root identification code to this variable area as the current value. The unique data index primary key within a time period is defined by locking the read root identifier as the data index primary key within the current process time window. This is based on a time-series logic binding mechanism. In assembly line operations, products flow continuously, and a specific process only processes one or a batch of products within a specific time period. By locking the root identifier and defining the time window, it can be ensured that all process data generated within that time period can logically belong only to the physical product corresponding to that root identifier. This prevents data ownership confusion or misattribution caused by multiple batches of products passing through continuously.

[0063] The system performs real-time capture and structured storage of process data. The data acquisition terminal uses the established physical communication connection and follows the preset communication protocol to read the process parameter values ​​and test values ​​from the output register of the process equipment controller in real time at a millisecond-level polling frequency. Each set of raw data packets is immediately stamped with the precise timestamp of the current reading time by the processing program. The processing program then calls the database write command, searches for the corresponding record row in the local database based on the previously locked data index primary key, and writes the timestamped process parameters and test values ​​into the local database field with the root identifier code as the data index primary key.

[0064] In one embodiment of the present invention, the process of constructing a quality data collection module in step S3, automatically matching and integrating the parameter data collected from each process node through the root identifier code, and generating a complete quality data archive for a single aluminum decorative panel product in chronological order is as follows:

[0065] First, a standardized access process for heterogeneous data is initiated by configuring a data access adaptation unit on the server side. This adaptation unit concurrently accesses the local databases within the data acquisition terminals distributed across various process sites through a pre-built database connection pool. The adaptation unit accurately extracts data packets containing specific root identifiers from massive amounts of raw records by executing structured query language commands. These heterogeneous data packets from controllers of different brands or models are then mapped to a predefined standard field model in the server's memory to complete the unified conversion of data formats. It should be noted that using a data access adaptation unit to interface with and uniformly convert data to a preset standard format is to solve the data silo problem caused by the diversity of communication protocols of industrial field equipment. Because controllers in different processes may use completely different data encoding methods and storage structures, direct logical operations will lead to serious compatibility errors. By forcibly executing format standardization operations at the entry point of data collection, the interference of underlying hardware differences on upper-level data analysis logic can be eliminated, ensuring that all data entering the collection module has strict consistency and parsability at the syntactic level.

[0066] To establish efficient data retrieval and association logic, the parsing program within the module receives the root identifier code to be processed and initiates a hierarchical index building program. This program, based on the encoding rules defined during root identifier code generation, decomposes the long string into three independent logical fields: supplier number sub-index, material batch number sub-index, and warehousing time sub-index. The processing program then uses these three sub-indexes to perform multi-level cascading queries in the database's B-tree index structure. First, it narrows down the supplier range; second, it locates the specific batch; and finally, it precisely hits the warehousing time point, thereby filtering out relevant records scattered across massive amounts of historical data. It's important to note that building a hierarchical index mechanism based on the root identifier code and filtering data hierarchically is based on the efficiency optimization requirements of big data retrieval. As the production cycle extends, the number of records in the database grows exponentially. A single full table scan would consume enormous computing resources and cause response delays. By decomposing the root identifier code to establish a multi-dimensional hierarchical index path, the time complexity of the search algorithm can be significantly reduced, enabling precise matching and association of the root identifier code with massive data rows distributed across different process tables within milliseconds.

[0067] Subsequently, the time sequence logic of the aggregated data undergoes rigorous cleaning and rearrangement. The timestamp calibration and sorting module traverses all successfully matched associated data packets and extracts the original millisecond-level timestamp values ​​recorded within them. The module obtains the average latency parameters of network transmission and the clock drift history of each acquisition terminal through the server's built-in time synchronization interface. It uses a compensation algorithm to perform weighted correction on the extracted original timestamps to eliminate time errors between physical devices. After correction, the module uses a fast sorting algorithm to sort all data packets in ascending order based on the calibrated time values, forming a preliminary quality data sequence that strictly increases unidirectionally in the time dimension. It should be noted that the construction of the timestamp calibration and sorting module and the calibration based on the interface are based on the requirement of authenticity in physical event restoration. Because network transmission has unavoidable random delays and the clock crystals of distributed hardware have physical deviations, direct sorting without unified benchmark calibration may lead to inconsistencies between the recorded order of process actions and the actual physical order of occurrence. By compensating and restoring the actual occurrence time through algorithms, it is possible to ensure that the data sequence faithfully reproduces the actual processing sequence of the product on the production line.

[0068] The final step is to complete the physical logical loop of the data chain and encapsulate the data. The data integration and output unit calls the pre-stored process flow topology diagram, which strictly defines the standard physical path from raw material warehousing through stamping, oxidation, and inspection to assembly and delivery. The output unit maps the time-sorted data sequence generated in the previous step to each node of this standard physical path, checking for any missing records or logical breaks in process nodes. After verifying the integrity of the data chain, all discrete parameter sets are encapsulated into an indivisible binary object or structured document, thereby generating a complete quality data archive for a single aluminum decorative panel with the root identifier as the unique index key. It should be noted that setting up the data integration and output unit and completing the association according to the process logic is to transform discrete points in the time dimension into continuous lines in the process dimension. Simple time sorting cannot reflect the process inheritance relationship between processes. Only by combining physical process logic with structured encapsulation can a bunch of cold parameters be transformed into a complete electronic history describing the quality status of the entire product life cycle, providing logically rigorous archive support for downstream quality traceability and problem analysis.

[0069] In one embodiment of the present invention, step S4, which involves establishing a parameter causal mapping model between adjacent processes, establishing association rules between the output parameters of the preceding process and the quality performance of the subsequent process, and forming a cross-process parameter transmission link for anomaly tracing, is as follows:

[0070] First, data extraction and alignment are performed. The quality data aggregation module obtains detailed parameter information from adjacent processes, including various monitoring values ​​during production and the final quality inspection results. Using the product's unique root identifier as the core index key, the output parameters from the preceding process are correlated one-to-one with the quality performance data from the immediately following process. During this process, the timestamps and batch information of the two types of data are carefully verified to ensure the consistency of the data source. The corresponding storage fields in the database for the two types of data are clarified, and physical data alignment is completed. It should be noted that the root identifier is used for correlation and field verification because industrial production involves massive amounts of data from diverse sources. Direct physical alignment eliminates interference from differences in data storage formats, ensuring that the process parameters of the preceding process are accurately mapped to the quality results of the specific product in the following process. This lays a noise-free data foundation for establishing accurate causal relationships and avoids biased analytical conclusions due to data mismatch.

[0071] Next, a gradient correlation screening mechanism for process parameters was designed and implemented. This mechanism deeply analyzes the process logic relationships between adjacent processes, automatically identifying and eliminating preceding parameters that are physically or chemically irrelevant to the quality performance of subsequent processes. For example, non-critical ambient lighting brightness in preceding processes usually does not affect the processing hardness of subsequent processes. After removing these invalid parameters, core parameters that have a substantial impact on product quality are retained, thus forming a candidate correlation set. It should be noted that setting up a gradient correlation screening mechanism and eliminating parameters based on process logic is to reduce the dimensionality of data processing and reduce the waste of computational resources. By pre-excluding non-causal interference variables, the convergence speed and accuracy of subsequent model calculations can be significantly improved, preventing spurious correlations from causing model misjudgments. This ensures that the final retained parameter set truly possesses the potential physical properties to affect the quality of subsequent processes, thereby allowing the analysis process to focus more on key process elements.

[0072] A causal inference algorithm is used to analyze the candidate association set, construct a parameter causal mapping model, quantify the causal association strength between preceding output parameters and subsequent quality performance, and generate deterministic association rules; the specific method is as follows:

[0073] First, a deep cleaning operation is performed on the preceding output parameters and subsequent quality performance data in the candidate association set. This operation uses statistical methods to scan the dataset to identify and remove null values ​​and outliers that significantly deviate from the normal distribution range. For some repairable missing data, linear interpolation is used to fill in the gaps to ensure data continuity. Simultaneously, data from different sources are standardized in format, and parameters with different physical dimensions are converted into dimensionless standardized values. This ensures that all data strictly meets the high standards of input quality required by causal inference algorithms. It is important to note that such rigorous data cleaning and standardization are necessary because raw industrial data often contains sensor noise and recording errors. Directly inputting data with missing values ​​or significant differences in dimensions into the model can lead to serious biases in the calculation results or even failure to converge. Standardizing the dimensions eliminates the influence of different units on weight calculations, ensuring that the algorithm objectively analyzes only the fluctuation trends of data changes, thus laying a solid digital foundation for accurate causal identification.

[0074] Next, a propensity score algorithm is used to conduct a core analysis on the preprocessed candidate association set. Specifically, a logistic regression model is introduced to calculate the propensity score value of the influence of the preceding output parameters on the subsequent quality performance, with the preceding output parameters set as independent variables. And set whether the subsequent quality performance is satisfactory as the dependent variable. The formula for calculating the propensity probability of each sample is as follows:

[0075]

[0076] In the formula, This represents the output parameters given the preceding sequence. Under the condition that the subsequent quality performance is qualified, The probability of a tendency For the intercept term The regression coefficient is used to quantify the conditional probability that changes in preceding output parameters lead to changes in subsequent quality performance, thereby identifying whether a statistically significant direct influence relationship exists between the two. It should be noted that the core purpose of introducing the propensity score algorithm is to address the problem of confounding variables commonly found in observational data. Conducting completely randomized controlled trials directly on industrial production lines is extremely difficult, but this algorithm can simulate the effect of randomized experiments through mathematical transformations, effectively balancing the differences in covariate distributions between different parameter groups. This allows for the isolation of quality changes solely caused by fluctuations in preceding parameters, ensuring that the identified causal relationship is a genuine physical connection rather than a spurious correlation.

[0077] Finally, based on the direct causal relationships identified in the above steps, a parametric causal mapping model is constructed. In this process, the calculated propensity score is further transformed into a specific causal association strength index, typically using the average treatment effect. To characterize this intensity, the calculation logic is to compare the average difference in quality performance between the high-parameter group and the low-parameter group. The calculation formula is expressed as follows:

[0078]

[0079] In the formula, This represents the expected quality of the subsequent sequence when the preceding parameters are at a high level. This represents the expected quality of the subsequent sequence when the preceding parameters are at a low level. Based on this intensity index... A threshold is set for the magnitude of the parameter values. Parameters exceeding the threshold are extracted and combined with actual process physical constraints to generate deterministic association rules that satisfy the process logic. It should be noted that converting propensity scores into causal association strength indicators and extracting rules is to transform abstract probability values ​​into operational instructions that frontline engineers can directly understand and execute. Simple probability values ​​are difficult to use directly to guide production adjustments. By setting thresholds and filtering in conjunction with process logic, weak associations that are statistically significant but have little practical engineering significance can be eliminated. This ensures that the final generated rules accurately reflect the specific quantitative impact of fluctuations in preceding parameters on subsequent quality, providing a definite logical basis for parameter optimization and anomaly interception in the production process.

[0080] The association rules generated by each adjacent process are linked together according to the actual processing sequence of the product. The root identifier code serves as the sole clue for data flow throughout the entire process, constructing a complete cross-process parameter transmission link. This link clearly defines the specific path of each core parameter from the preceding process to the following process, as well as their numerical dependencies. If a quality anomaly is detected in a subsequent process, it can be traced back along this link. It's important to note that constructing a cross-process parameter transmission link using the root identifier code breaks down the data silos between processes in traditional industrial production. This enables logical interconnection of quality data throughout the entire process, making quality problems no longer isolated points but traceable lines. When a quality deviation is detected at the terminal or intermediate stage, it can quickly pinpoint which parameter in which preceding process has deviated, thus achieving efficient and accurate anomaly tracing and significantly reducing the time cost required for troubleshooting.

[0081] In one embodiment of the present invention, step S5 involves setting parameter monitoring boundary values ​​for each process node. When the collected parameters exceed the boundary range, the system triggers an early warning signal and locks the current batch of products, preventing abnormal products from flowing to downstream processes.

[0082] First, the data backtracking and cleaning process is initiated. Detailed production records of all products judged to be of qualified quality in the past production cycle are retrieved from the industrial database. The focus is on extracting the physical parameter data recorded in real time during the processing of each process node, as well as the quality inspection data exhibited when the product is transferred to the next process. These complex data are finely classified and organized according to different processing technology types. The unique root identifier of the product is used as the core index key to connect the data fragments of the same product in different processes to form a complete end-to-end data chain.

[0083] Next, a process-related boundary value calibration mechanism was constructed and implemented. This mechanism deeply analyzes the physical processing logic and chemical reaction mechanism between each process, and uses multivariate statistical analysis to calculate the specific impact weight values ​​of the output parameter changes of the previous process on the processing quality of the current process. Based on these weight values, the parameter fluctuation range initially defined by historical data is dynamically corrected and refined. For highly sensitive parameters with high weights, the allowable fluctuation range is appropriately narrowed, ultimately determining the parameter monitoring boundary values ​​that each process node should strictly adhere to in actual production. It should be noted that the introduction of a process-related calibration mechanism and the adjustment of boundaries based on weights is because the processes in a complex industrial production line are not isolated. Small fluctuations in the previous stage are often amplified by the next stage. Static fixed boundaries are difficult to adapt to the dynamically changing production environment. Weight calibration allows the monitoring standards to better meet the continuity requirements of the actual process, ensuring that the generated boundary values ​​can tolerate normal process fluctuations while accurately intercepting small deviations that may lead to final quality defects.

[0084] Finally, the calibrated and determined monitoring boundary values ​​of each process parameter are written into the database of the real-time monitoring platform as the core judgment standard. The monitoring platform receives real-time production parameters uploaded by the field acquisition terminal at a frequency of milliseconds and immediately compares and analyzes these real-time data with the preset boundary values. Once a parameter value is found to exceed the set safety boundary range, the monitoring platform immediately activates the audible and visual alarm device to issue a warning signal. At the same time, it uses the root identification code to mark the batch of products as abnormal in the database and locks its electronic circulation permission. Then, it sends a mandatory prohibition of circulation command to the production terminal in the downstream process to physically block the continued processing of the batch of products.

[0085] In one embodiment of the present invention, in step S6, the process of constructing a two-way traceability query channel, where the forward channel traces the complete processing trajectory and final destination of the product from the root identifier code, and the reverse channel locates the problematic product to the raw material batch and specific process node, is as follows:

[0086] First, a full lifecycle data archive is constructed. Complete quality data archives for all coded products within a preset time period are extracted from the central database. This archive covers parameter records for every stage from raw material warehousing to finished product delivery, and further links logistics and customer delivery information recorded at assembly and shipment nodes. A unique root identifier is used as the core index key to seamlessly connect production-side process data with market-side flow data, thus constructing a data association index system containing several data dimensions. The preset time period mentioned here is typically determined based on the company's data storage capacity and the length of the product warranty period, such as 3 or 5 years. The various data dimensions mentioned here are determined based on the specific coverage requirements of product quality traceability, such as raw material batch number, processing equipment number, operator employee number, and logistics tracking number.

[0087] Next, a bidirectional indexing and mapping mechanism for the root identifier was designed and deployed. At the data logic level, a forward tracing link and a reverse tracing graph were constructed. The forward link starts with the root identifier and sequentially links the processing parameter data of each process to the final product destination data according to the timeline, forming a linear historical trajectory chain. The reverse graph establishes a reverse hash mapping relationship between several key feature tags of the problematic product and the root identifier and process nodes, enabling feature data to be indexed to the corresponding production batch. The key feature tags mentioned here are defined in a standardized manner based on frequently occurring fault phenomena in historical after-sales complaints, such as surface cracks, dimensional deviations, and insufficient hardness.

[0088] Finally, an efficient query interaction module is built to execute specific data retrieval tasks. When performing a forward query, this module receives the root identifier code command input by the user and directly calls the forward correlation link to output the entire processing trajectory and final market destination of the product. When performing a reverse query, this module receives the feature description of the problem product, matches the reverse mapping relationship through an algorithm, and calculates the feature matching weight. When the weight exceeds a preset judgment threshold, the corresponding raw material batch and specific process node are located. The feature matching weight mentioned here is set based on the correlation coefficient between the fault feature and the process parameter, using a weighted calculation, for example, a floating-point number between 0 and 1. The judgment threshold mentioned here is set based on the confidence requirement of the traceability results through statistical calculation, for example, 0.85 or 0.9. It should be noted that weight-based reverse matching queries are set because in actual quality analysis, a single feature often corresponds to multiple possible causes. Weight calculation can filter out the most likely source process or raw material batch, helping engineers quickly narrow down the investigation scope from massive amounts of production data, thereby achieving rapid location from vague fault descriptions to precise production nodes.

[0089] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A method for tracing the quality of automotive aluminum trim panels, characterized in that, Includes the following steps: S1: Generate a unique root identification code for each batch of aluminum decorative panel raw materials. This root identification code contains three basic information: supplier number, material batch number, and warehousing time. S2: Deploy data acquisition terminals at five process nodes: raw material warehousing, stamping, anodizing, quality inspection, and assembly outbound. After the terminal is bound to the root identification code, it records the process parameters and inspection values ​​of each node in real time. S3: Construct a quality data collection module to automatically match and integrate the parameter data collected from each process node through the root identifier code, and generate a complete quality data archive for a single aluminum decorative panel product in chronological order; S4: Establish a parameter causal mapping model between adjacent processes, establish association rules between the output parameters of the preceding process and the quality performance of the following process, and form a cross-process parameter transmission link for anomaly tracing. S5: Set parameter monitoring boundary values ​​for each process node. When the collected parameters exceed the boundary range, the system triggers an early warning signal and locks the current batch of products to prevent abnormal products from flowing to the downstream process. S6: Construct a two-way traceability query channel. The forward channel traces the complete processing trajectory and final destination of the product from the root identification code, while the reverse channel locates the problematic product to the raw material batch and specific process node.

2. The method for tracing the quality of automotive aluminum trim panels according to claim 1, characterized in that, In step S1, the specific method for generating a unique root identifier code for the batch of aluminum decorative panel raw materials is as follows: It receives the original text data of supplier number and material batch number uploaded by the raw material storage entry terminal, and at the same time, it captures the current time stamp as the entry time parameter through the time synchronization interface on the server. The warehousing time parameter is converted into a long integer pure numeric character sequence. At the same time, a zero-digit padding operation is performed on the supplier number and material batch number so that the three sets of data reach the preset fixed character width. Following a fixed sequence of supplier number, material batch number, and warehousing time, three sets of standardized data are sequentially concatenated using specific meaningless delimiters to generate a unique root identifier.

3. The method for tracing the quality of automotive aluminum trim panels according to claim 1, characterized in that, In step S2, the process of deploying data acquisition terminals at five process nodes—raw material warehousing, stamping, anodizing, quality inspection, and assembly outbound—and recording the process parameters and inspection values ​​of each node in real time after the terminals are bound to the root identifier code is as follows: Data acquisition terminals are fixedly installed at the raw material warehousing, stamping, anodizing, quality inspection, and assembly and delivery processes. The industrial fieldbus is used to establish physical communication connections between each data acquisition terminal and the controller of the corresponding process equipment. The data acquisition terminal drives the external optical scanning module to read the root identification code pasted on the aluminum decorative panel transfer fixture; the data acquisition terminal locks the read root identification code as the data index primary key within the current process time window to complete the logical binding; The data acquisition terminal reads the process parameters and inspection values ​​output by the equipment controller in real time through a physical communication connection; it timestamps the read data packets and writes them into a local database field with the root identifier as the primary key of the data index.

4. The method for tracing the quality of automotive aluminum trim panels according to claim 1, characterized in that, In step S3, the process of constructing the quality data collection module, which automatically matches and integrates the parameter data collected from each process node through the root identifier code, and generates a complete quality data archive for a single aluminum decorative panel product in chronological order, is as follows: A data access adapter unit is used to connect to the local database of the data acquisition terminal of each process, extract data packets containing root identifiers, and convert them into a preset standard data format to ensure data structure consistency. A root identifier hierarchical indexing mechanism is constructed, which decomposes the root identifier into sub-indexes such as supplier number, material batch number, and warehousing time. Data is filtered according to the sub-index hierarchy to achieve accurate matching and association between the root identifier and the data of each process. A timestamp calibration and sorting module was built to extract the timestamps of each matching data, and time calibration was completed based on the server time synchronization interface. The associated data were then integrated according to the calibrated time order to form a preliminary quality data sequence. Set up an archive integration and output unit to complete and associate the data of each process after sorting by time according to the process logic from raw material warehousing to assembly and delivery, and generate a complete quality data archive of a single aluminum decorative panel with the root identifier code as the unique identifier.

5. The method for tracing the quality of automotive aluminum trim panels according to claim 1, characterized in that, In step S4, the process of establishing a parameter causal mapping model between adjacent processes, establishing association rules between the output parameters of the preceding process and the quality performance of the following process, and forming a cross-process parameter transmission link for anomaly tracing is as follows: Extract parameter data corresponding to adjacent processes from the quality data collection module, associate the preceding output parameters and subsequent quality performance data of the same product according to the root identifier code, clarify the corresponding fields of the two types of data and complete the data alignment; Design a gradient correlation screening mechanism for process parameters. Based on the process logic of adjacent processes, eliminate preceding parameters that have no process correlation with the quality performance of subsequent processes, and retain core influencing parameters to form a candidate correlation set. A causal inference algorithm is used to analyze the candidate association set, construct a parameter causal mapping model, quantify the causal association strength between the preceding output parameters and the subsequent quality performance, and generate deterministic association rules. By linking the association rules of each adjacent process according to the processing sequence, and using the root identifier code as the data thread, a cross-process parameter transmission link is constructed to clarify the transmission path and association relationship of each parameter for anomaly tracing.

6. The method for tracing the quality of automotive aluminum trim panels according to claim 5, characterized in that, The specific method for analyzing the candidate association set using a causal inference algorithm, constructing a parameter causal mapping model, quantifying the causal correlation strength between preceding output parameters and subsequent quality performance, and generating deterministic association rules is as follows: Clean the preceding output parameters and subsequent quality performance data in the candidate association set, remove missing values ​​and outliers, and unify the data format and units to ensure that the data meets the input requirements of the causal inference algorithm. The propensity score algorithm is used to analyze the preprocessed candidate association set. By calculating the propensity score values ​​between parameters, the direct causal relationship between the preceding output parameters and the subsequent quality performance is identified. Based on the identified direct causal relationships, a parametric causal mapping model is constructed, and the tendency score is transformed into a causal association strength index. Based on this index, deterministic association rules that satisfy the process logic are extracted.

7. The method for tracing the quality of automotive aluminum trim panels according to claim 1, characterized in that, In step S5, the process of setting parameter monitoring boundary values ​​for each process node, and triggering an early warning signal and locking the current batch of products when the collected parameters exceed the boundary range, preventing abnormal products from flowing to downstream processes, is as follows: Extract parameter data of historical qualified products from each process and corresponding quality performance data of subsequent processes, classify them by process, and associate them with the complete data chain corresponding to the same identification code; Design a process-related boundary value calibration mechanism, which combines the process logic of each process and, based on the influence weight of the previous output parameters on the current process, calibrates the historical parameter range to form the monitoring boundary values ​​of each process parameter; The monitoring boundary values ​​of parameters for each process are entered into the system monitoring module. The system compares the parameters uploaded by the acquisition terminal with the boundary values ​​in real time. If the values ​​exceed the boundary values, an early warning is triggered. The batch is locked by the root identifier code and a prohibition instruction is sent to the downstream terminal.

8. The method for quality traceability of automotive aluminum trim panels according to claim 1, characterized in that, In step S6, the process of constructing a two-way traceability query channel, where the forward channel tracks the complete processing trajectory and final destination of the product from the root identifier code, and the reverse channel locates the problematic product to the raw material batch and specific process node, is as follows: Extract the complete quality data archives corresponding to all root identifiers, associate the product destination data with the assembly and outbound node records, and build a data association index system based on the root identifier; Design a bidirectional index mapping mechanism for the root identifier code, establish a sequential association link between the root identifier code and the processing data of each process and the product destination data in the forward direction, and establish a reverse mapping relationship between the characteristics of the problem product and the root identifier code and process nodes in the reverse direction. A query interaction module is built. During forward queries, the root identifier code is received to call the sequential association link to output the trajectory and destination. During reverse queries, the characteristics of the problem product are matched to call the reverse mapping to locate the raw material batch and process node.

Images