Stamping hardware quality management method and system

Abstract

The application relates to the technical field of quality management, in particular to a stamping hardware quality management method and system. Production batch information and production event context information are acquired; quality data uploaded by an automatic detection device and associated detection context information are acquired, the detection context information comprising detection time, detection device identification and batch identification; based on the production batch information, the production event context information and the detection context information, multidimensional consistency evaluation is carried out; according to the result of the multidimensional consistency evaluation, correlation judgment is carried out and a correlation between the quality data and the production batch information is established; quality data that cannot establish a correlation through correlation judgment is marked and manual checking is prompted. The application solves the technical problem of quality data and actual production batch correlation confusion caused by inconsistent or incorrect batch information entry in large-scale production of stamping hardware.

Description

Technical Field

[0001] This application relates to the field of quality management technology, and more specifically, to a method and system for quality management of stamped hardware parts. Background Technology

[0002] In the mass production of stamped hardware parts, ensuring product quality is paramount. However, in actual production environments, due to the inherent characteristics of manual operation and the accelerated production pace, inconsistencies or errors often occur in batch information entry. This directly leads to confusion in the correlation between quality data and actual production batches. This lack of or error in data correlation not only makes quality traceability difficult but also hinders the early detection of progressive problems such as mold wear, and makes it difficult to accurately pinpoint the cause of scrap, ultimately affecting overall production efficiency and the accuracy of quality management decisions.

[0003] For example, in a large-scale stamping hardware production facility, tens of thousands of various hardware parts are stamped daily. On the stamping production line, operators are responsible for entering key data such as raw material batch information, mold numbers, and production shifts into the quality data management system. However, in actual operation, due to the fast production pace and differences in the habits of different operators, input errors or inconsistent formats occasionally occur when entering batch information. For example, one batch number might be entered as "PROD-20231026-001," while another operator might enter it as "P231026-001," or even simple spelling errors, such as "PROD-20231026-0O1" (the number 0 is mistakenly entered as the letter O). Existing quality data management systems are designed to accept preset, strictly standardized input formats, lacking effective automatic correction or prompting mechanisms for these non-standard or slightly different inputs. The system often identifies them as different, independent batch identifiers rather than variations of the same batch, which results in the quality data of some stamped parts not being accurately associated with the corresponding production batch when they are received into the warehouse.

[0004] This confusion in linking quality data with production batches directly led to the failure of subsequent quality traceability functions. This prevented management from making decisions based on reliable data, hindering the effective optimization of stamping process parameters or adjustment of production schedules, thus impacting overall production efficiency and cost control.

[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0006] This application discloses a quality management method and system for stamped hardware parts, which aims to solve the technical problems in the large-scale production of stamped hardware parts, such as inconsistent or incorrect batch information entry caused by manual operation and accelerated production pace, which in turn leads to confusion between quality data and actual production batches, difficulty in quality traceability, difficulty in early detection of progressive problems such as mold wear, and difficulty in accurately locating the cause of scrap.

[0007] The technical solution of this application is as follows:

[0008] In a first aspect, this application discloses a method for quality management of stamped hardware parts, including:

[0009] Obtain the entered production batch information and its corresponding production event context information. The production event context information includes the entry time, operator identifier, production equipment identifier, and production status identifier.

[0010] Acquire the quality data uploaded by the automated testing equipment and its associated testing context information, including testing time, testing equipment identifier, and batch identifier;

[0011] Based on production batch information, production event context information, and detection context information, a multi-dimensional consistency assessment is conducted, which includes time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment.

[0012] Based on the results of the multi-dimensional consistency assessment, correlation judgments are made and the correlation between quality data and production batch information is established;

[0013] Quality data that cannot be correlated through association judgment is marked and prompted for manual verification.

[0014] Furthermore, based on production batch information, production event context information, and detection context information, a multi-dimensional consistency assessment is conducted, including:

[0015] When the multi-dimensional consistency assessment identifies multiple candidate production batches whose total matching scores differ from a preset difference threshold and all exceed a preset association threshold, the character variation pattern between the batch identifier in the detection context information and the candidate batch identifier in the production batch information is identified, and the source of the character variation is determined.

[0016] Identify inconsistencies in the production status identifiers within the time window used for time correlation assessment in the production event context information corresponding to each candidate production batch, determine the production status conflict pattern, and identify the source of the conflict.

[0017] Generate a diagnostic report containing character mutation patterns and their sources, production status conflict patterns and their sources;

[0018] The matching scores of multiple candidate production batches were adjusted and sorted based on the diagnostic reports;

[0019] Generate a manual verification interface, display the diagnostic report and sorting results, and obtain the corresponding returned information confirming the actual production batch.

[0020] In some preferred embodiments, inconsistencies in the production status identifiers within the time window used for time correlation assessment in the production event context information corresponding to each candidate production batch are identified, production status conflict patterns are determined, and conflict sources are identified, including:

[0021] Based on the characteristics of different information sources in the production event context information, assign corresponding state weights to the production status identifiers provided by each information source.

[0022] Query the most recent mold change records and product switch records, and adjust the status weights based on the mold change records and product switch records;

[0023] Based on the production status identifier and its corresponding status weight, as well as the mold change record and product switching record, calculate the comprehensive score for each possible production status;

[0024] The current actual production status is determined based on the overall score;

[0025] When the difference between the highest and second-highest comprehensive scores is less than a preset difference threshold, or when all comprehensive scores are lower than a preset reliability threshold, the actual production status will be marked as uncertain and an early warning will be triggered.

[0026] Furthermore, identifying character variation patterns between batch identifiers in the detection context information and candidate batch identifiers in the production batch information includes:

[0027] The character differences between the batch identifier in the context information and the candidate batch identifier in the production batch information are compared character by character, and the character variation patterns belonging to the preset similar deformed character pairs are identified in the character differences.

[0028] Based on the frequency and positional concentration of deformed characters in the character variation pattern, determine whether the source of the variation is transient local electromagnetic interference;

[0029] If the source of the variation is transient local electromagnetic interference, generate an electromagnetic interference probability score for the candidate production batch.

[0030] Furthermore, identifying character variation patterns between batch identifiers in the detection context information and candidate batch identifiers in the production batch information includes:

[0031] Identify character variation patterns that fall under preset common human error patterns in character differences;

[0032] Calculate the probability score of human input errors derived from common human input mistakes;

[0033] Based on the probability scores of electromagnetic interference and human input errors, the contribution of different sources to character variation is quantified.

[0034] Furthermore, the character differences between the batch identifier in the detection context information and the candidate batch identifier in the production batch information are compared character by character, and character variation patterns belonging to preset similar deformed character pairs are identified in the character differences, including:

[0035] A structured analysis is performed on the character differences between the batch identifier in the detection context information and the candidate batch identifier in the production batch information to extract the structured features of the character differences. The structured features include position, quantity, type and adjacent character relationship features.

[0036] The structured features are matched with preset pairs of similar deformed characters to identify the corresponding character variation patterns;

[0037] For character differences that fail to match the preset similar deformed character pairs, cluster analysis is performed based on their structural features to identify new character variation patterns;

[0038] The new character variation patterns are categorized and stored in the character variation pattern library for subsequent character variation pattern recognition.

[0039] Furthermore, based on the frequency and positional concentration of deformed characters in the character variation pattern, it is determined whether the source of the variation is transient local electromagnetic interference, including:

[0040] Real-time monitoring of electromagnetic field intensity fluctuations around production equipment, and recording of the instantaneous temporal characteristics of electromagnetic field intensity fluctuations;

[0041] Analyze the frequency and position concentration of deformed characters in character variation patterns, and extract the dynamic trajectory of the frequency and position concentration of deformed characters over time;

[0042] By performing time series correlation analysis on instantaneous changes and dynamic trajectories, the nonlinear correlation between the two can be identified.

[0043] By combining the operating status data of the production equipment, it is determined whether the time of character mutation matches the operation time indicated by the operating status data, and the judgment result is obtained;

[0044] Based on the results of the temporal correlation analysis and the judgment results, it is determined whether the source of the character variation is instantaneous local electromagnetic interference.

[0045] As an optional approach, inconsistencies in the production status identifiers within the time window used for time correlation assessment in the production event context information corresponding to each candidate production batch are identified to determine the production status conflict pattern and identify the source of the conflict, including:

[0046] The production status identifiers provided by each information source are timestamped to align the production status identifiers of different information sources in the time dimension.

[0047] Based on the calibrated timestamp, identify whether there are inconsistencies in the production status identifiers in the context information of different production events within a specific time window. If there are inconsistencies, identify the corresponding production status conflict mode based on the inconsistency type and duration.

[0048] Query the operation logs of production equipment that overlap with a specific time window;

[0049] Determine whether there are operation events in the production equipment operation log that correspond to the production status conflict mode, and determine the degree of consistency between the operation events and the production status conflict mode based on the type and occurrence time of the operation events.

[0050] Based on the degree of agreement, the conflict mode in the production status is determined to be either an instantaneous delay in system data acquisition or a specific operation triggered by human intervention.

[0051] Furthermore, when the degree of agreement indicates that the production state conflict pattern is a specific operation triggered by human intervention, the method also includes:

[0052] Query production work order information and production plan information that overlap with the time of a specific manually triggered operation;

[0053] Based on production work order information and production plan information, identify the degree of match between specific human-triggered operations and the preset routine operation types in the production work order or production plan;

[0054] Query the operator's historical operation records and authorization level;

[0055] Based on historical operation records and authorization levels, identify whether the operator has the authority to perform a specific operation triggered by the user, and whether the specific operation triggered by the user is within the scope of authorization;

[0056] Based on the degree of similarity between a specific human-triggered operation and a regular operation type, as well as the operator's authority and scope of operation, determine whether the specific human-triggered operation is a regular temporary operation or an illegal operation;

[0057] When a specific human-triggered operation is determined to be a violation, a violation warning is generated, and detailed information about the violation is recorded.

[0058] Secondly, this application also discloses a quality management system for stamped hardware parts, specifically including:

[0059] The batch information acquisition module is used to acquire the entered production batch information and its corresponding production event context information. The production event context information includes the entry time, operator identifier, production equipment identifier, and production status identifier.

[0060] The quality data acquisition module is used to acquire the quality data uploaded by the automated testing equipment and its associated testing context information, including testing time, testing equipment identifier, and batch identifier.

[0061] The consistency assessment module is used to perform multi-dimensional consistency assessments based on production batch information, production event context information, and detection context information. The multi-dimensional consistency assessment includes time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment.

[0062] The correlation judgment module is used to make correlation judgments and establish the correlation between quality data and production batch information based on the results of multi-dimensional consistency assessment.

[0063] The anomaly marking module is used to mark quality data that cannot be correlated through correlation judgment and prompt manual verification.

[0064] Beneficial effects

[0065] The quality management method for stamped hardware disclosed in this application acquires production batch information, production event context information, and quality data and testing context information from automated testing equipment. Based on this information, a multi-dimensional consistency assessment is performed, including time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment. This method can establish a correlation between quality data and production batch information based on the assessment results, and mark and manually verify data that cannot be correlated. The method of this application can significantly improve overall production efficiency and the accuracy of quality management decisions, overcoming the shortcomings of existing technologies such as difficulty in quality traceability, difficulty in locating the causes of defects, and inaccurate quality trend analysis. Attached Figure Description

[0066] Figure 1 This is a flowchart illustrating a quality management method for stamped hardware parts provided in this application.

[0067] Figure 2 This application provides a system structure block diagram of a quality management system for stamped hardware parts.

[0068] In the diagram: 1. Batch information acquisition module; 2. Quality data acquisition module; 3. Consistency assessment module; 4. Association judgment module; 5. Anomaly marking module. Detailed Implementation

[0069] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0070] Reference Figure 1 This application proposes a quality management method for stamped hardware parts, comprising the following steps:

[0071] S1000: Obtain the entered production batch information and its corresponding production event context information. The production event context information includes the entry time, operator identifier, production equipment identifier, and production status identifier.

[0072] S2000: Acquire the quality data uploaded by the automated testing equipment and its associated testing context information, including testing time, testing equipment identifier, and batch identifier;

[0073] S3000: Based on production batch information, production event context information, and detection context information, it performs multi-dimensional consistency assessment, which includes time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment.

[0074] S4000: Based on the results of the multi-dimensional consistency assessment, perform correlation judgment and establish the correlation between quality data and production batch information;

[0075] S5000: Mark quality data that cannot be correlated through association judgment and prompt manual verification.

[0076] This application, through multi-dimensional consistency assessment, can effectively identify and correct batch information inconsistencies caused by manual entry or system differences, thereby establishing an accurate correlation between quality data and production batch information, significantly improving the efficiency and accuracy of quality traceability, and providing a more reliable data foundation for the quality management of stamped hardware parts.

[0077] Specifically, production batch information refers to a set of data that identifies a specific quantity or time period of products produced during the stamping hardware manufacturing process. It typically includes batch number, product model, and production date. Production event context information refers to auxiliary data related to the production batch information, used to describe the specific context of the production process. For example, "entry time" refers to the time the batch information was recorded by the system; operator identifier refers to the unique identification code of the operator responsible for producing that batch; "production equipment identifier" refers to the unique identification code of the equipment used to produce that batch; and "production status identifier" refers to the status of that batch during the production process, such as "in production," "pending inspection," or "completed." Quality data refers to data generated after automated inspection equipment inspects the stamping hardware, such as dimensions, weight, and surface defects.

[0078] Testing context information refers to auxiliary data related to quality data, used to describe the specific context of the quality testing process. For example, "testing time" refers to the time when the quality data is recorded by the testing equipment; "testing equipment identifier" refers to the unique identification code of the equipment performing the quality testing; and "batch identifier" refers to the product batch number identified by the testing equipment during testing.

[0079] Multi-dimensional consistency assessment refers to comparing and analyzing production batch information, production event context information, and testing context information from multiple perspectives to determine whether there is logical consistency among them. This includes "time correlation assessment," which determines whether production events and testing events are reasonably related in time; production equipment consistency assessment, which determines whether production and testing use the same or related equipment; operator consistency assessment, which determines whether production and testing are handled by the same or related operators; and batch identifier similarity assessment, which determines the degree of similarity between batch identifiers from different sources at the character level.

[0080] Correlation determination refers to comprehensively judging whether there is a valid correlation between quality data and production batch information based on the results of multi-dimensional consistency assessment. A correlation is a logical connection established between quality data and production batch information, enabling traceability of corresponding quality data through batch information, and vice versa.

[0081] First, the system acquires the entered production batch information and its corresponding production event context information. This can be achieved in several ways. For example, operators can manually input production batch information through a human-machine interface. Upon receiving the input, the system automatically records the current input time, the operator's identification, and obtains the identification of the currently used production equipment and the production status based on sensor data from the production line. Alternatively, the production management system can interface with an Enterprise Resource Planning (ERP) system or a Manufacturing Execution System (MES) to automatically acquire preset production batch information and extract or generate corresponding production event context information from these systems.

[0082] Secondly, it is necessary to obtain the quality data uploaded by the automated inspection equipment and its associated inspection context information. After completing the inspection of the stamped hardware, the automated inspection equipment will upload the inspection results, i.e., the quality data, to the quality management system via a network interface. During the upload process, the inspection equipment will also include inspection context information, such as automatically recording the inspection completion time, its own equipment identification, and the batch identification read from the product through barcode scanning or optical character recognition (OCR) technology.

[0083] Next, a multi-dimensional consistency assessment is performed based on production batch information, production event context information, and detection context information. This assessment includes time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment. In the time correlation assessment, the system compares the entry time in the production event context information with the detection time in the detection context information. For example, a reasonable time window can be set; if the detection time falls within a certain period after the production batch information is entered, it is considered to be time-related. In the production equipment consistency assessment, the system compares the production equipment identifier in the production event context information with the detection equipment identifier in the detection context information. For example, if the production equipment and detection equipment are physically adjacent or closely connected in the process flow, they are considered to be consistent in terms of equipment. In the operator consistency assessment, the system compares the operator identifier in the production event context information. For example, if the production batch information and detection data are handled by the same shift or the same operator within a similar time period, they are considered to be consistent in terms of operator. In the batch identifier similarity assessment, the system compares the batch number in the production batch information with the batch identifier in the detection context information. For example, string matching algorithms (such as Levenshtein distance) can be used to calculate the similarity between two batch identifiers to identify minor typographical errors or format differences.

[0084] Then, based on the results of the multi-dimensional consistency assessment, a correlation judgment is made and a relationship is established between quality data and production batch information. The system comprehensively considers the results of various assessments; for example, different weights can be assigned to each assessment to calculate a total matching score. If the total matching score exceeds a preset correlation threshold, a valid correlation is considered to exist between the quality data and the production batch information, and a relationship is established between the two. This relationship can be a foreign key connection in the database or a logical mapping relationship.

[0085] Finally, quality data that cannot be correlated through association judgment is marked and prompted for manual verification. If a quality data point fails to reach the preset association threshold after multi-dimensional consistency assessment, or if significant inconsistencies are found during the assessment process, the system will mark it as "not associated" or "abnormal" and generate corresponding prompts to notify quality management personnel for manual verification. Manual verification personnel can analyze the specific reasons based on the detailed assessment report provided by the system and make corrections or supplementary entries to ensure that all quality data can be accurately associated with production batch information.

[0086] In another embodiment of this application, a sub-step of S3000 is further proposed: based on production batch information, production event context information, and detection context information, a multi-dimensional consistency assessment is performed, including:

[0087] S3100: When the multi-dimensional consistency assessment identifies multiple candidate production batches whose total matching scores differ from a preset difference threshold and all exceed a preset association threshold, the character variation pattern between the batch identifier in the detection context information and the candidate batch identifier in the production batch information is identified, and the source of the character variation is determined.

[0088] S3200: Identify inconsistencies in the production status identifiers in the production event context information between candidate production batches within the time window used for time correlation assessment, determine the production status conflict pattern, and identify the source of the conflict.

[0089] S3300: Generates a diagnostic report containing character mutation patterns and their sources, production status conflict patterns and their sources;

[0090] S3400: Adjust and sort the matching scores of multiple candidate production batches based on the diagnostic report;

[0091] S3500: Generates a manual verification interface, displays diagnostic reports and sorting results, and retrieves the corresponding returned information confirming the actual production batch.

[0092] Specifically, when the results of the multi-dimensional consistency assessment indicate the existence of multiple candidate production batches whose total matching scores are close to each other (i.e., the difference is less than a preset difference threshold), and these scores all reach or exceed a preset association threshold, it indicates that these candidate batches have a high probability of association. In this case, deeper analysis is needed to eliminate ambiguity. Identifying the character variation patterns between the batch identifier in the detection context information and the candidate batch identifier in the production batch information, and determining the source of these character variations, involves a detailed comparison between the batch identifier in the detection context information and the batch identifier in each candidate production batch information, analyzing the character differences between them. These differences may manifest as additions, deletions, or modifications of characters, such as misprinting, omissions, or reversals of numbers or letters. By analyzing these variation patterns, the causes can be inferred, such as human input errors, system transmission failures, or transient electromagnetic interference.

[0093] Identifying inconsistencies in the production status identifiers within the time window used for time correlation assessment across different candidate production batches involves cross-referencing the production status identifiers in the production event context information of different candidate production batches within the time window set for the time correlation assessment. If inconsistencies or contradictions are found in these production status identifiers—for example, one batch is recorded as "in production" at a certain time point, while another batch is recorded as "stopped" at the same time point—it is necessary to identify the specific pattern of this inconsistency and analyze its possible causes, such as data entry errors, equipment status update delays, or abnormal situations in the actual production process.

[0094] This generates a diagnostic report containing character variation patterns and their sources, as well as production status conflict patterns and their sources. The purpose is to structure and visualize the aforementioned analysis results. This diagnostic report details the batch identifier differences, variation types, and probabilities of variation sources between each candidate batch and the quality data, as well as the conflict status, conflict patterns, and conflict source analysis for each candidate batch in terms of production status. Based on this, the matching scores of multiple candidate production batches are adjusted and ranked according to the diagnostic report. This involves correcting the matching scores obtained from the initial multi-dimensional consistency assessment based on the detailed information and source analysis of character variations and production status conflicts revealed in the diagnostic report. For example, if a candidate batch's character variation is judged to be a high-probability human input error, its matching score may be appropriately reduced; if a production status conflict indicates that a batch could not possibly be in production at the detection time point, its score will be significantly reduced. Through this adjustment, the true correlation between each candidate batch and the quality data can be more accurately reflected, and the batches can be re-ranked.

[0095] Finally, a manual verification interface is generated, displaying the diagnostic report and ranking results, and retrieving the corresponding confirmed production batch information. The purpose is to present the highly uncertain correlation results, which have undergone initial screening and analysis by the system, to human verification personnel. Based on the detailed information and ranking results in the diagnostic report, combined with their professional knowledge and experience, human verifiers can make final confirmations on the most likely candidate batches or conduct further investigations to determine the true production batch information.

[0096] In some preferred embodiments, assume that an automated inspection device uploads a batch of quality data, with the batch identifier in its inspection context information being "ABC-001". After a preliminary multi-dimensional consistency assessment, the system identifies two candidate production batches: production batch A, with production batch information "ABC-001", and production event context information showing that it is "in production" during the inspection period; and production batch B, with production batch information "ABG-001", and production event context information also showing that it is "in production" during the inspection period. At this point, the total matching scores of the two candidate batches are very close and both exceed the preset association threshold. According to the scheme of this application, the system will further identify character variation patterns. For production batch B, there is a character difference of "C" and "G" between the batch identifier "ABG-001" and the inspection batch identifier "ABC-001". The system will analyze whether this difference belongs to a preset similar deformed character pair and determine its variation source, such as whether it is instantaneous local electromagnetic interference or manual input error. At the same time, the system will identify whether there is an inconsistency in the production status identifiers in the production event context information corresponding to the two candidate batches. For example, if the production status identifier for production batch A comes from the MES system, while the production status identifier for production batch B comes from the SCADA system, and there is a slight difference at a certain point in time, the system will determine the production status conflict pattern and identify the source of the conflict. Subsequently, the system will generate a diagnostic report detailing the character variation patterns of "C" and "G" and their identified sources of variation (e.g., high probability of human input error), as well as potential conflict patterns of the production status identifiers and their sources. Based on this diagnostic report, the system will adjust the matching scores of production batches A and B. For example, if the variation between "C" and "G" is judged to be a high probability of human input error, the matching score of production batch B may be appropriately reduced. After adjustment, the matching score of production batch A may be significantly higher than that of production batch B, thus ranking first in the sorting. Finally, the system will generate a manual verification interface, displaying this diagnostic report and the adjusted sorting results. Manual inspectors can clearly see that there is a high probability of a human entry error in the batch identifier of production batch B, and there may also be minor conflicts in the production status information. This allows them to quickly and accurately confirm that production batch A is the real production batch, avoiding the difficulty in judgment caused by similar surface scores.

[0097] In another embodiment of this application, S3200 further includes:

[0098] S3210: Assign corresponding state weights to the production status identifiers provided by each information source based on the characteristics of different information sources in the production event context information.

[0099] S3220: Query the most recent mold change records and product switch records, and adjust the status weights based on the mold change records and product switch records;

[0100] S3230: Calculate the comprehensive score for each possible production status based on the production status identifier and its corresponding status weight, as well as the mold change record and product switching record.

[0101] S3240: Determine the current actual production status based on the overall score;

[0102] S3250: When the difference between the highest and second-highest comprehensive scores is less than the preset difference threshold, or when all comprehensive scores are lower than the preset reliability threshold, the actual production status is marked as uncertain and an early warning is triggered.

[0103] Specifically, production event context information can originate from various sources, such as Manufacturing Execution System (MES), Supervisory Control and Data Acquisition (SCADA) systems, and manual data entry systems. Since these sources may differ in terms of data accuracy, real-time performance, and authority, assigning corresponding status weights to the production status identifiers provided by each source involves allocating different weight values ​​based on the inherent characteristics of these sources, such as their historical data accuracy, data update frequency, and the level or reliability of the data source. For example, data directly collected by automated systems (such as MES or SCADA) may typically be assigned a higher weight, while manually entered data may be assigned a relatively lower weight to reflect its potential human error risk.

[0104] The process of querying recent mold changeover and product switchover records and adjusting status weights based on these records refers to fully considering the decisive impact of key production events such as mold changes and product switchovers on production status when assessing production status. These records provide objective evidence of changes in production line configuration and product type, and can serve as important supplementary information to correct or verify the weights of production status indicators provided by different information sources. For example, if a mold change occurs at a certain point in time, the weight of production status indicators related to the new mold (such as "in production" or "in debugging") may be increased after that point, or the weight of status indicators related to the old mold may be decreased, to ensure that the judgment of production status is consistent with the key events of actual production operations.

[0105] In practical applications, calculating the comprehensive score for each possible production state based on the production state identifier and its corresponding weight, as well as mold change records and product switching records, involves comprehensively considering the production state identifiers and their corresponding weights from different information sources, and combining auxiliary information from mold change records and product switching records to calculate a comprehensive score for each possible production state (e.g., "in production," "stopped," "mold change," "standby," etc.). This comprehensive score quantifies the probability or confidence level of the production state within the current time window, thus providing data support for subsequent judgments. Determining the current true production state based on the comprehensive score means selecting the production state with the highest comprehensive score as the most likely true production state within the current time window. This helps to make more accurate and objective judgments through quantitative evaluation when there are multiple conflicting production state identifiers, avoiding subjective assumptions. Furthermore, when the difference between the highest and second-highest comprehensive scores is less than a preset difference threshold, or when all comprehensive scores are lower than a preset confidence threshold, the true production state is marked as an uncertain state and an early warning is triggered. This is to handle highly uncertain or ambiguous production state scenarios. If the highest and second-highest scores are very close, it indicates that there is a high probability of two or more production states, making it difficult for the system to make a clear judgment. If all scores are low, it indicates that the confidence level of all possible production states is insufficient, and the information may be seriously missing or contradictory. In both cases, the system will proactively identify the state as uncertain and trigger an alert, prompting manual intervention for verification, to avoid the system making incorrect automatic judgments under conditions of insufficient or highly ambiguous information.

[0106] In some preferred embodiments, it is assumed that within a specific time window, the production event context information of the production equipment shows a conflicting production status: the MES system reports the equipment as "in production," the SCADA system reports it as "stopped," while the manually entered system shows it as "mold change." The system assigns corresponding status weights to the production status identifiers provided by each information source according to preset rules. For example, MES system data, due to its high automation and real-time nature, has a weight of 0.5; SCADA system data, due to its direct monitoring of equipment operation status, has a weight of 0.3; and manually entered data, due to potential delays or errors, has a weight of 0.2. The system queries the most recent mold change records and product switchover records. If a clear mold change record is found within the time window, the system adjusts the status weights accordingly. For example, the weight of the "mold change" status is increased to 0.6, while the weights of the "in production" and "stopped" statuses are correspondingly decreased. Based on the adjusted status weights and the production status identifiers provided by each information source, the system calculates a comprehensive score for the three possible production statuses: "in production," "stopped," and "mold change." For example, if mold change records strongly support the "mold change" status, the overall score for "mold change" may be significantly higher than other statuses. The system determines the current actual production status based on the calculated overall score. If the overall score for "mold change" is the highest and far exceeds the second-highest score (e.g., the difference is greater than a preset difference threshold), the system determines the actual production status as "mold change." However, if the overall scores for "production in progress" and "downtime" are very close and not significantly different from the score for "mold change," or if all scores are below a preset confidence threshold, the system will mark the actual production status as uncertain and trigger an alert, prompting quality engineers to conduct manual verification to avoid misjudgments when information is highly ambiguous. In this way, this application can handle complex production status conflicts more accurately and intelligently, ensuring the accuracy of the correlation between quality data and production batches.

[0107] In another embodiment of this application, a sub-step of S3100 is further proposed: identifying a character variation pattern between the batch identifier in the detection context information and the candidate batch identifier in the production batch information, including:

[0108] S3110: Compare and detect the character differences between the batch identifier in the context information and the candidate batch identifier in the production batch information, and identify the character variation patterns that belong to the preset similar deformed character pairs in the character differences;

[0109] S3120: Based on the frequency and positional concentration of deformed characters in the character variation pattern, determine whether the source of the variation is transient local electromagnetic interference;

[0110] S3130: If the source of the variation is transient local electromagnetic interference, generate an electromagnetic interference probability score for the candidate production batch.

[0111] Specifically, comparing the character differences between the batch identifier in the context information and the candidate batch identifier in the production batch information means that the system compares the two batch identifier strings character by character from left to right or right to left to identify all mismatched character positions and their corresponding characters. For example, if the batch identifier is "ABC123X" and the candidate batch identifier is "ABC123Y", the system will identify a character difference at the last position, that is, "X" and "Y" do not match.

[0112] Character variation patterns identified through character difference recognition, specifically those belonging to predefined similar deformed character pairs, refer to the process where, after character differences are identified through character-by-character comparison, the system matches these differing characters with predefined "similar deformed character pairs." These predefined similar deformed character pairs can be understood as character combinations that are prone to deformation or confusion under specific conditions (such as during data transmission, display, or input), such as "O" and "0", "I" and "1", "S" and "5", "B" and "8", etc. When the identified character differences match these predefined pairs, a specific character variation pattern is considered to have been identified. The purpose is to systematically identify common, attributable character deformations, providing a basis for subsequent determination of the source of variation.

[0113] In some preferred embodiments, assume that the batch identifier in the quality data uploaded by the automated inspection equipment is "PN-A10B-007", and the system identifies two candidate production batch identifiers in the production batch information: "PN-A108-007" and "PN-A10B-007". The system performs a character-by-character comparison between the batch identifier "PN-A10B-007" in the inspection context information and the candidate batch identifier "PN-A108-007" in the production batch information. The comparison result shows that there is a character difference at the 6th position, that is, the "B" in the inspection batch identifier does not match the "8" in the candidate batch identifier. Next, the system matches this character difference "B" and "8" with a preset pair of similar deformed characters. Since "B" and "8" are easily confused visually or in some data transmission errors, they are preset as a pair of similar deformed characters. Therefore, the system recognizes this as a character variation pattern. The system analyzes the frequency and positional concentration of this deformed character "B / 8". In this example, there is only one deformed character, and its position is in the middle part of the string. By combining real-time electromagnetic field strength monitoring data around the production equipment (e.g., detecting a brief electromagnetic pulse near the detection time), the system determines that the variation is highly likely to originate from transient local electromagnetic interference based on the type and location of the deformed character and the electromagnetic field fluctuations. Ultimately, if the determination is transient local electromagnetic interference, the system generates a high electromagnetic interference probability score, such as 0.85, for the candidate production batch "PN-A108-007". This score will serve as part of the subsequent diagnostic report, guiding manual inspectors to prioritize electromagnetic interference as the cause of the batch identification difference, thereby more accurately determining the true production batch information.

[0114] In another embodiment of this application, a sub-step of S3100 is further proposed: identifying a character variation pattern between the batch identifier in the detection context information and the candidate batch identifier in the production batch information, including:

[0115] S3140: Identify character variation patterns that belong to preset common human error patterns in character differences;

[0116] S3150: Calculate the probability score of human input errors derived from common human input errors;

[0117] S3160: Based on the probability scores of electromagnetic interference and human input errors, quantify the contribution of different sources to character variation.

[0118] Specifically, when analyzing the character differences between the batch identifier in the detection context information and the candidate batch identifier in the production batch information, in addition to identifying the deformed character patterns caused by electromagnetic interference, the system also identifies character variation patterns belonging to preset common human error patterns. Preset common human error patterns can be understood as error types that frequently occur during manual data entry, such as character misalignment, omission, redundancy, and substitution (e.g., confusion between the number "0" and the letter "O", or between the number "1" and the letter "l"). These patterns are typically statistically analyzed and summarized based on historical manual data entry, forming an error pattern library.

[0119] For identified common human error patterns, a human input error probability score is calculated. This score is calculated by considering factors such as the type of error pattern, the position of the erroneous character in the batch identifier, the frequency of the error pattern, and the overall complexity of the batch identifier. For example, a common misalignment error occurring in a critical position might have a higher human input error probability score.

[0120] Based on this, this application quantifies the contribution of different sources to character variation by using the aforementioned electromagnetic interference probability score and human input error probability score. This means that the system no longer simply determines whether the variation is caused by electromagnetic interference, but comprehensively evaluates the respective contributions of the two main sources—electromagnetic interference and human input error—to the observed character variation. This quantification can be achieved through various methods such as weighted averaging, Bayesian inference, and machine learning models, aiming to provide a more comprehensive variation source analysis for each candidate production batch.

[0121] As a specific implementation: Suppose that during the production of stamped hardware parts, a product with a batch identifier of "PN20230901-001" is produced. When the automated inspection equipment uploads quality data, due to momentary local electromagnetic interference, the batch identifier is recorded as "PN20230901-O01" (the number "0" is deformed into the letter "O"). Simultaneously, during the manual entry of production event context information, the operator, due to negligence, enters the batch identifier as "PN20230901-010" (the last two digits are reversed). At this point, when the system performs a multi-dimensional consistency assessment, it will identify multiple candidate production batches. Based on the character difference between "PN20230901-O01" in the inspection context information and "PN20230901-001" in the production batch information, the system will identify the deformation from "0" to "O" according to a preset similar deformed character pair, and combine this with information such as electromagnetic field strength fluctuations to generate a higher electromagnetic interference probability score. Meanwhile, regarding the character differences between "PN20230901-010" and "PN20230901-001" in the production batch information, the system identifies the character reversal pattern from "01" to "10" and categorizes it as a preset common human error pattern, thereby calculating a higher score for the probability of human input error. Ultimately, based on these two scores, the system quantifies the respective contributions of electromagnetic interference and human input errors to character variation, thus more accurately determining which candidate batch has a higher correlation with the actual production batch "PN20230901-001," or providing more detailed diagnostic information for manual verification. For example, it might indicate that the variation in "PN20230901-001" mainly stems from electromagnetic interference, while the variation in "PN20230901-010" mainly stems from human input errors.

[0122] In another embodiment of this application, S3110 further includes the following steps:

[0123] S3111: Perform structured analysis on the character differences between the batch identifier in the detection context information and the candidate batch identifier in the production batch information, and extract the structured features of the character differences. The structured features include position, quantity, type and adjacent character relationship features.

[0124] S3112: Match the structured features with preset similar deformed character pairs to identify the corresponding character variation patterns;

[0125] S3113: For character differences that fail to match preset similar deformed character pairs, perform cluster analysis based on their structural features to identify new character variation patterns;

[0126] S3114: Classify the new character variation patterns and store them in the character variation pattern library for subsequent character variation pattern recognition.

[0127] Structured analysis of character differences involves a detailed decomposition and description of the character differences detected in batch identifiers. The structured features include: "Position," which refers to the specific location of the character difference within the entire batch identifier string (e.g., first, last, or middle position); "Quantity," which refers to the number of differing characters (e.g., one character mutated or multiple consecutive characters mutated); "Type," which refers to the specific form of the character mutation (e.g., number to letter (e.g., "0" to "O"), letter case conversion (e.g., "a" to "A"), character loss, character addition, or character order reversal); and "Adjacent Character Relationship Features," which refers to the relationship between the mutated character and its preceding and following characters, helping to determine whether the mutation is influenced by the context. Extracting these structured features provides richer and more accurate input information for subsequent pattern matching and clustering analysis.

[0128] Matching structured features with pre-defined similar variant character pairs involves comparing the character difference features obtained through structured analysis with known and common character variation patterns pre-stored in the system. These pre-defined similar variant character pairs are typically based on historical data or expert experience, such as easily confused character pairs like "1" and "l", "0" and "O", and "B" and "8". Through matching, character variations that match known patterns can be quickly and accurately identified.

[0129] For character differences that fail to match preset similar deformed character pairs, cluster analysis is performed based on their structural features to identify new character variation patterns. This means that when a character difference does not belong to any known preset pattern, the system utilizes its structural features for unsupervised learning. Cluster analysis aims to group unknown character differences with similar structural features into one category, thereby discovering potential new variation patterns. For example, if a character is missing at a specific position multiple times, and the relationship features between adjacent characters are similar, it may be clustered and identified as a new "character missing at a specific position" pattern.

[0130] New character variation patterns are categorized and stored in a character variation pattern library for subsequent character variation pattern recognition. This means that once a new character variation pattern is identified through cluster analysis, the system names, describes, and adds it to the existing character variation pattern library. In this way, during future quality management processes, when encountering the same or similar character differences, the system can directly identify and apply these newly learned patterns without needing to perform complex cluster analysis again.

[0131] In some preferred embodiments:

[0132] Suppose that on a stamping hardware production line, the batch identifier uploaded by the automated inspection equipment is "P-12345," while the batch identifier in the candidate production batch information entered by the production management system is "P-1234S." First, the system performs a structured analysis on the character differences between "P-12345" and "P-1234S." It identifies the difference occurring at the sixth character position, where "5" changes to "S." Its structured features are extracted as follows: position = 6, quantity = 1, type = number changed to letter, adjacent character relationship feature = preceding "4," subsequent none. Next, the system matches these structured features with preset pairs of similar deformed characters. If a similar deformed character pair of "5-S" exists in the preset library (e.g., due to font or scanning blur), the system directly identifies this character variation pattern. However, if the "5-S" pattern is not found in the preset library, the system will perform cluster analysis on the structured features of the character difference (position = 6, quantity = 1, type = number to letter, adjacent character relationship feature = "4" before, none after). If the system finds multiple similar, unmatched character differences in historical data, such as "P-67890" becoming "P-6789O" or "P-ABCDE" becoming "P-ABCDL", and these differences show similarity in structured features (e.g., the last digit / letter becomes a similar-looking letter / digit), the clustering algorithm will group them into one category and identify a new character variation pattern, for example, named "last digit / letter deformation". Finally, the system will classify this newly identified "last digit / letter deformation" pattern and store it in the character variation pattern library. In this way, in subsequent quality management processes, when encountering similar character differences such as "5" becoming "S" or other last digit / letter changes, the system can directly identify and apply the pattern without having to perform complex cluster analysis again, thereby improving recognition efficiency and accuracy.

[0133] In another embodiment of this application, S3120 further includes:

[0134] S3121: Real-time monitoring of electromagnetic field intensity fluctuations around production equipment, and recording the instantaneous temporal characteristics of electromagnetic field intensity fluctuations;

[0135] S3122: Analyze the frequency and position concentration of deformed characters in the character variation pattern, and extract the dynamic trajectory of the frequency and position concentration of deformed characters over time;

[0136] S3123: Perform time-series correlation analysis between instantaneously changing time-series characteristics and dynamic trajectories to identify whether there is a nonlinear correlation between the two;

[0137] S3124: Based on the operating status data of the production equipment, determine whether the occurrence time of the character mutation matches the operation time indicated by the operating status data, and obtain the judgment result;

[0138] S3125: Based on the results of the time-series correlation analysis and the judgment results, determine whether the source of the character variation is instantaneous local electromagnetic interference.

[0139] Specifically, real-time monitoring of electromagnetic field strength fluctuations around production equipment refers to continuously collecting electromagnetic field strength data in the environment by deploying electromagnetic sensors or electromagnetic field monitoring equipment near the production equipment. This data is recorded as instantaneous temporal characteristics, such as the peak value, duration, frequency range, and trend of electromagnetic field strength over time. The purpose is to obtain external electromagnetic environment information corresponding to the period when character variations occur.

[0140] Analyzing the frequency and positional concentration of deformed characters in character variation patterns, and extracting the dynamic trajectory of these frequencies and positional concentrations over time, can be understood as conducting in-depth time-series analysis of the identified character variation data. For example, it's possible to statistically analyze the frequency of specific deformed characters and their positional distribution within batch identifier strings across different time windows, and then plot curves or graphs showing how these statistics change over time. The aim is to reveal the evolutionary patterns of character variation phenomena over time.

[0141] In practical applications, temporal correlation analysis is performed between instantaneous changes in time series characteristics and dynamic trajectories to identify whether there is a nonlinear correlation between the two. Specifically, time series analysis methods such as cross-correlation analysis, Granger causality test, and dynamic time warping (DTW) are used to evaluate the synchronicity, lag, or lead-in of electromagnetic field intensity fluctuations and character variation dynamic trajectories in time, as well as whether there is a nonlinear mutual influence relationship between them. The purpose is to verify the potential link between electromagnetic interference and character variation from a statistical and temporal perspective.

[0142] By combining the operational status data of production equipment, it is determined whether the occurrence time of character mutations matches the operation time indicated by the operational status data. The judgment result is obtained by comparing the timestamp of the character mutation with the operational status records of production equipment (such as stamping machines, testing equipment, etc.) such as start-up, shutdown, mold change, and fault. For example, if the character mutations occur concentratedly at the moment of starting or stopping a certain piece of equipment, or during the operation of a high-power piece of equipment, a match is considered to exist. The purpose is to provide auxiliary evidence from the equipment operation level to further confirm the possibility of electromagnetic interference.

[0143] Determining whether the source of character variation is transient local electromagnetic interference, based on the results of time-series correlation analysis and judgment, involves comprehensively considering the results of the aforementioned analyses. For example, if time-series correlation analysis shows a significant nonlinear correlation between electromagnetic field strength fluctuations and the dynamic trajectory of character variation, and the occurrence time of character variation closely matches the operation time of a specific device, then the source of variation can be determined with greater certainty to be transient local electromagnetic interference. Conversely, if such strong evidence is lacking, other sources of variation may need to be considered.

[0144] In some preferred embodiments, a specific example is given below:

[0145] Suppose that on a stamping production line, in the quality data uploaded by the automated inspection equipment, the batch identifier "ABC12345" is detected as "ABG12345," where the character "C" is mutated to "G." The system first identifies this character mutation pattern and analyzes its frequency and location concentration. To more accurately determine the source of the mutation, the system initiates the following process: First, electromagnetic sensors deployed near the stamping machine monitor in real time a brief, high-amplitude fluctuation in electromagnetic field strength occurs at the moment the character mutation occurs (e.g., the instant the stamping machine performs a heavy-load stamping operation), and record its instantaneous temporal characteristics. Second, the system analyzes the character mutation pattern of "C" mutating to "G" in historical data, finding that this mutation has a high frequency within a specific time period, and usually occurs in the third position of the batch identifier, and extracts the dynamic trajectory of this frequency and location concentration over time. Next, the system performs a temporal correlation analysis between the instantaneous temporal characteristics of the electromagnetic field strength fluctuation and the dynamic trajectory of the character mutation. For example, by calculating the cross-correlation coefficient between the two, it was found that the frequency of character mutations increased significantly about 50 milliseconds after the electromagnetic field strength fluctuation reached its peak, indicating a significant nonlinear correlation between the two. Simultaneously, the system queried the operating status data of the stamping press and found that the time of character mutation highly coincided with the time of a high-power stamping operation. Ultimately, considering the strong temporal correlation between electromagnetic field strength fluctuations and character mutations, as well as the coincidence between character mutations and the high-power operation time of the stamping press, the system determined that the source of the batch identifier character "C" mutating to "G" was transient local electromagnetic interference. Based on this determination, the system can perform corresponding processing on the batch data, such as giving it higher weight in correlation judgments or triggering a check of the stamping press's electromagnetic shielding, thereby effectively improving the accuracy of quality data correlation.

[0146] In another embodiment of this application, S3200 further includes:

[0147] S3260: Perform timestamp calibration on the production status identifiers provided by each information source to align the production status identifiers of different information sources in the time dimension.

[0148] S3270: Based on the calibrated timestamp, identify whether there is inconsistency in the production status identifier in the context information of different production events within a specific time window. If there is inconsistency, identify the corresponding production status conflict mode based on the inconsistency type and duration.

[0149] S3280: Query the operation logs of production equipment that overlap with a specific time window;

[0150] S3290: Determine whether there are operation events in the production equipment operation log that correspond to the production state conflict mode, and determine the degree of consistency between the operation event and the production state conflict mode based on the type and occurrence time of the operation event.

[0151] S32100: Based on the degree of agreement, determine whether the production status conflict mode is due to instantaneous delay in system data acquisition or a specific operation triggered by human intervention.

[0152] Specifically, the production status identifiers provided by each information source refer to data on the status of production equipment or processes from various sensors, Manufacturing Execution Systems (MES), Supervisory Control and Data Acquisition Systems (SCADA), or manual data entry. These information sources may have timestamp discrepancies due to network latency, system synchronization strategies, or different data acquisition frequencies. Therefore, timestamp calibration of the production status identifiers provided by each information source involves adjusting these timestamps using a unified time base or synchronization protocol to ensure precise alignment of all production status identifiers in the time dimension, eliminating the illusion of data inconsistency caused by time deviations.

[0153] The specific time window can be understood as a time range set during production status consistency assessment, such as a few seconds, tens of seconds, or minutes. Its purpose is to focus on a relatively short period during which status changes may occur. Within this time window, the system identifies inconsistencies in the production status identifiers across different production event contexts. For example, if one information source shows the device as "running," while another source shows it as "stopped" within the same time window, this is identified as inconsistency. Inconsistency types can include conflicting status values ​​(e.g., running / stopped), discrepancies in status duration, etc., where duration refers to the length of time the inconsistent status persists. Through this information, specific production status conflict patterns can be identified, such as "short-term stoppage conflicts" or "state transition delay conflicts."

[0154] In practical applications, querying production equipment operation logs that overlap with a specific time window refers to the system automatically accessing and retrieving stored production equipment operation records. These logs typically contain detailed operational events such as equipment startup, shutdown, parameter adjustments, mold changes, and fault alarms, along with their occurrence times. The purpose is to provide external, independent verification information for production status conflicts.

[0155] Determining whether a production equipment operation log contains an operation event corresponding to a conflicting production status pattern involves comparing the identified conflicting patterns with specific events in the operation log. For example, if a conflicting pattern indicates a short-term equipment downtime at a certain point in time, the system will check the operation log for downtime operation records near that point in time. The degree of match can be quantitatively assessed based on multiple dimensions, including the type of operation event (e.g., whether it was a planned downtime operation), the time of occurrence (whether it highly overlaps with the conflicting time), and the operator's identification. For example, if the log contains an "equipment maintenance" operation performed by an authorized person at the conflicting time, the degree of match is relatively high.

[0156] Therefore, based on the degree of agreement, the system can determine the source of production status conflict patterns. If the degree of agreement is high, and the operation log shows a clear operation event, the conflict pattern is likely a specific operation triggered by human intervention, such as manual intervention, equipment debugging, or unplanned maintenance. Conversely, if the degree of agreement is low, meaning there is no clear operation event corresponding to the conflict pattern in the operation log, or the correlation between the operation event and the conflict pattern is weak, the conflict pattern is more likely to be attributed to the instantaneous delay in system data acquisition, i.e., a brief lag occurs during data transmission or processing between different systems, leading to temporary inconsistencies.

[0157] In another embodiment of this application, when the degree of agreement indicates that the production state conflict mode is a specific operation triggered by human intervention, the method further includes:

[0158] S32110: Query production work order information and production plan information that overlap with the time of a specific manually triggered operation;

[0159] S32120: Based on production work order information and production plan information, identify the degree of match between a specific manually triggered operation and a pre-set routine operation type in the production work order or production plan;

[0160] S32130: Query the operator's historical operation records and authorization level;

[0161] S32140: Based on historical operation records and authorization levels, identify whether the operator has the authority to perform a specific operation triggered by the user, and whether the specific operation triggered by the user is within the scope of authorization;

[0162] S32150: Based on the degree of similarity between a specific operation triggered by human intervention and a regular operation type, as well as the operator's authority and scope of operation, determine whether the specific operation triggered by human intervention is a regular temporary operation or an illegal operation;

[0163] S32160: When a specific operation triggered by human intervention is determined to be a violation, a violation warning is generated and detailed information about the violation is recorded.

[0164] Specifically, production work order information and production plan information refer to formal documents or electronic records used to guide production activities. These include detailed information such as the expected product type, production quantity, production scheduling time, and permitted routine operation types. Routine operation types refer to standardized operations explicitly defined and permitted for operators in standard operating procedures (SOPs) or work instructions. Operator historical operation records and authorization levels refer to data on past operator behavior recorded by the system, and the scope of operational authority assigned to operators based on their job position, training, and qualifications. The degree of conformity can be understood as a quantitative assessment of the consistency or matching degree between a specific, human-triggered operation that actually occurs and the routine operation types preset in the production work order or production plan, as well as the operator's scope of authority.

[0165] Routine temporary operations refer to compliant, temporary operations performed by authorized personnel within their authorized scope during the production process to address unexpected but controllable production situations. Non-compliant operations refer to any non-standard operations that deviate from the production plan, exceed the operator's authorized scope, or violate operating procedures. When a non-compliant operation is detected, the system will generate a warning to promptly notify relevant management personnel for swift intervention. Simultaneously, the system will record detailed information about the non-compliant operation, including the operation time, operator identification, specific operation content, and the reason for determining it as a non-compliant operation, for subsequent traceability, analysis, and improvement.

[0166] As a specific implementation method, suppose that on a stamping hardware production line, the quality data uploaded by the automated inspection equipment is inconsistent with the production batch information in a time-related assessment. Based on the analysis in the aforementioned stamping hardware quality management method, the system determines that this inconsistency is caused by a specific, human-triggered operation performed by operator A at a specific time point: an "equipment parameter adjustment." The system queries production work order information and production plan information that overlap with the time of this "equipment parameter adjustment" operation. For example, the production plan shows that product X should be produced during this period, and the production specifications for product X do not include the regular operation type of this parameter adjustment. Next, the system queries operator A's historical operation records and authorization level. Assume that operator A's authorization level is limited to regular equipment start-up and shutdown, and does not include parameter adjustment permissions. Then, the system makes a judgment based on the above information. Since this "equipment parameter adjustment" operation does not match the preset regular operation type in the production plan, and operator A does not have the permission to perform this operation, the system will determine that the "equipment parameter adjustment" operation is a violation. Ultimately, the system will immediately generate a "violation warning" and record in detail the time of the operation, the operator A's identification, the adjusted parameters, and the reason for judging it as a violation, so that quality management personnel can intervene in time to verify and take appropriate measures.

[0167] Reference Figure 2 The specific embodiments of this application also disclose a quality management system for stamped hardware parts, including:

[0168] Batch information acquisition module 1 is used to acquire the input production batch information and its corresponding production event context information. The production event context information includes the input time, operator identifier, production equipment identifier, and production status identifier.

[0169] Quality data acquisition module 2 is used to acquire quality data uploaded by automated testing equipment and its associated testing context information, including testing time, testing equipment identifier, and batch identifier.

[0170] The consistency assessment module 3 is used to perform multi-dimensional consistency assessment based on production batch information, production event context information, and detection context information. The multi-dimensional consistency assessment includes time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment.

[0171] The association judgment module 4 is used to make association judgments and establish the association relationship between quality data and production batch information based on the results of the multi-dimensional consistency assessment.

[0172] The anomaly marking module 5 is used to mark quality data that cannot be correlated through correlation judgment and prompt manual verification.

[0173] This system, through its modular design, automates and intelligently manages the correlation between batch information and quality data during the production of stamped hardware parts. Batch information acquisition module 1 and quality data acquisition module 2 are responsible for collecting key data from the production and inspection stages, respectively, laying the foundation for subsequent data processing. Consistency assessment module 3 effectively identifies and corrects inconsistencies in the data through multi-dimensional analysis. Correlation judgment module 4 establishes accurate data correlations based on this, while anomaly marking module 5 ensures timely manual intervention for data that cannot be automatically correlated. This comprehensively improves the efficiency and accuracy of quality traceability and solves the problem of chaotic data correlations in traditional methods.

[0174] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for quality management of stamped hardware parts, characterized in that, include: Obtain the entered production batch information and its corresponding production event context information, wherein the production event context information includes the entry time, operator identifier, production equipment identifier, and production status identifier; Acquire the quality data uploaded by the automated testing equipment and its associated testing context information, including the testing time, testing equipment identifier, and batch identifier; Based on the production batch information, the production event context information, and the detection context information, a multi-dimensional consistency assessment is performed, which includes time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment. Based on the results of the multi-dimensional consistency assessment, an association judgment is made and the association between the quality data and the production batch information is established; Quality data that cannot be correlated through the aforementioned correlation judgment is marked and prompted for manual verification; The multi-dimensional consistency assessment based on the production batch information, the production event context information, and the detection context information includes: When the multi-dimensional consistency assessment identifies multiple candidate production batches whose total matching scores differ from a preset difference threshold and all exceed a preset association threshold, the character variation pattern between the batch identifier in the detection context information and the candidate batch identifier in the production batch information is identified, and the source of the character variation is determined. Identify inconsistencies in the production status identifiers within the time window used for time correlation assessment in the production event context information corresponding to each candidate production batch, determine the production status conflict mode, and identify the source of the conflict. Generate a diagnostic report containing the character mutation pattern and its mutation source, and the production status conflict pattern and its conflict source; The matching scores of the multiple candidate production batches are adjusted and sorted based on the diagnostic report; A manual verification interface is generated, displaying the diagnostic report and sorting results, and obtaining the corresponding returned information confirming the actual production batch.

2. The method for quality management of stamped hardware parts according to claim 1, characterized in that, The process of identifying inconsistencies in the production status identifiers within the time window used for time correlation assessment in the production event context information corresponding to each of the candidate production batches, determining the production status conflict pattern, and judging the source of the conflict includes: Based on the characteristics of different information sources in the production event context information, assign corresponding state weights to the production status identifiers provided by each information source. Query the most recent mold replacement records and product switching records, and adjust the state weights based on the mold replacement records and product switching records; Based on the production status identifier and its corresponding status weight, as well as the mold replacement record and the product switching record, calculate the comprehensive score for each possible production status; The current actual production status is determined based on the comprehensive score. When the difference between the highest and second-highest comprehensive scores is less than a preset difference threshold, or when all comprehensive scores are lower than a preset confidence threshold, the actual production status is marked as uncertain and an early warning is triggered.

3. The method for quality management of stamped hardware parts according to claim 1, characterized in that, The identification of character variation patterns between the batch identifier in the detection context information and the candidate batch identifier in the production batch information includes: The character differences between the batch identifier in the detection context information and the candidate batch identifier in the production batch information are compared character by character, and the character variation patterns belonging to the preset similar deformed character pairs in the character differences are identified. Based on the frequency and positional concentration of the deformed characters in the character mutation pattern, it is determined whether the mutation source is transient local electromagnetic interference; If the source of the variation is transient local electromagnetic interference, generate an electromagnetic interference probability score for the candidate production batch.

4. The method for quality management of stamped hardware parts according to claim 3, characterized in that, The identification of character variation patterns between the batch identifier in the detection context information and the candidate batch identifier in the production batch information includes: Identify character variation patterns that belong to preset common human error patterns among the character differences; Calculate the probability score of human input errors derived from common human input mistakes; Based on the electromagnetic interference probability score and the human input error probability score, the variation contribution of different sources to character variation is quantified.

5. The method for quality management of stamped hardware parts according to claim 3, characterized in that, The step of comparing the character differences between the batch identifier in the detection context information and the candidate batch identifier in the production batch information, and identifying character variation patterns belonging to preset similar deformed character pairs in the character differences, includes: A structured analysis is performed on the character differences between the batch identifier in the detection context information and the candidate batch identifier in the production batch information to extract the structured features of the character differences. The structured features include position, quantity, type, and adjacent character relationship features. The structured features are matched with the preset similar deformed character pairs to identify the corresponding character variation patterns; For character differences that fail to match the preset similar deformed character pairs, cluster analysis is performed based on their structural features to identify new character variation patterns; The new character variation patterns are categorized and stored in the character variation pattern library for subsequent character variation pattern recognition.

6. The method for quality management of stamped hardware parts according to claim 3, characterized in that, The step of determining whether the source of the mutation is transient local electromagnetic interference based on the frequency and positional concentration of the deformed characters in the character mutation pattern includes: Real-time monitoring of electromagnetic field intensity fluctuations around production equipment, and recording of the instantaneous temporal characteristics of the electromagnetic field intensity fluctuations; Analyze the frequency and position concentration of deformed characters in the character variation pattern, and extract the dynamic trajectory of the frequency and position concentration of deformed characters over time; The instantaneous change time series characteristics are analyzed in conjunction with the dynamic trajectory to identify whether there is a nonlinear correlation between the two. By combining the operating status data of the production equipment, it is determined whether the occurrence time of the character mutation matches the operation time indicated by the operating status data, and a judgment result is obtained; Based on the results of the temporal correlation analysis and the judgment results, it is determined whether the source of the character variation is instantaneous local electromagnetic interference.

7. The method for quality management of stamped hardware parts according to claim 1, characterized in that, The process of identifying inconsistencies in the production status identifiers within the time window used for time correlation assessment in the production event context information corresponding to each of the candidate production batches, determining the production status conflict pattern, and judging the source of the conflict includes: The production status identifiers provided by each information source are timestamped to align the production status identifiers of different information sources in the time dimension. Based on the calibrated timestamp, identify whether there are inconsistencies in the production status identifiers in the context information of different production events within a specific time window. If there are inconsistencies, identify the corresponding production status conflict mode based on the inconsistency type and duration. Query the operation logs of production equipment that overlap with the specific time window; Determine whether there is an operation event in the production equipment operation log that corresponds to the production state conflict mode, and determine the degree of consistency between the operation event and the production state conflict mode based on the type and occurrence time of the operation event. Based on the degree of agreement, the production status conflict mode is determined to be either an instantaneous delay in system data acquisition or a specific operation triggered by human intervention.

8. The method for quality management of stamped hardware parts according to claim 7, characterized in that, When the degree of agreement indicates that the production state conflict mode is a specific operation triggered by human intervention, the method further includes: Query production work order information and production plan information that overlap with the specific operation time triggered by the human; Based on the production work order information and the production plan information, identify the degree of consistency between the specific operation triggered by the user and the preset regular operation type in the production work order or production plan; Query the operator's historical operation records and authorization level; Based on the historical operation records and the authorization level, identify whether the operator has the authority to perform the specific operation triggered by the user, and whether the specific operation triggered by the user is within the scope of authorization; Based on the degree of similarity between the specific human-triggered operation and the regular operation type, as well as the operator's authority and scope of operation, it is determined whether the specific human-triggered operation is a regular temporary operation or an illegal operation. When a specific operation triggered by human intervention is determined to be a violation, a violation warning is generated, and detailed information about the violation is recorded.

9. A quality management system for stamped hardware parts, characterized in that, include: The batch information acquisition module is used to acquire the input production batch information and its corresponding production event context information. The production event context information includes the input time, operator identifier, production equipment identifier, and production status identifier. The quality data acquisition module is used to acquire the quality data uploaded by the automated testing equipment and its associated testing context information, including the testing time, testing equipment identifier, and batch identifier. The consistency assessment module is used to perform multi-dimensional consistency assessment based on the production batch information, the production event context information, and the detection context information. The multi-dimensional consistency assessment includes time correlation assessment, production equipment consistency assessment, operator consistency assessment, and batch identifier similarity assessment. The association judgment module is used to make association judgments based on the results of the multi-dimensional consistency assessment and establish the association relationship between the quality data and the production batch information; The anomaly marking module is used to mark quality data that cannot be associated through the association judgment and prompt manual verification. The multi-dimensional consistency assessment based on the production batch information, the production event context information, and the detection context information includes: When the multi-dimensional consistency assessment identifies multiple candidate production batches whose total matching scores differ from a preset difference threshold and all exceed a preset association threshold, the character variation pattern between the batch identifier in the detection context information and the candidate batch identifier in the production batch information is identified, and the source of the character variation is determined. Identify inconsistencies in the production status identifiers within the time window used for time correlation assessment in the production event context information corresponding to each candidate production batch, determine the production status conflict mode, and identify the source of the conflict. Generate a diagnostic report containing the character mutation pattern and its mutation source, and the production status conflict pattern and its conflict source; The matching scores of the multiple candidate production batches are adjusted and sorted based on the diagnostic report; A manual verification interface is generated, displaying the diagnostic report and sorting results, and obtaining the corresponding returned information confirming the actual production batch.

Images