Torque sampling inspection method and sampling inspection system

Abstract

The invention relates to the technical field of torque sampling inspection, in particular to a torque sampling inspection method and system. Comprising the following steps: tool identity digitization: performing laser etching on an encrypted two-dimensional code in a tool stress concentration area, and associating and storing tool specification parameters, detection standards and use constraint conditions to a central database; intelligent task allocation: the system generates a dynamic detection plan according to a production line real-time state, an equipment service cycle and an operator skill level; multi-source data acquisition: acquiring tool identity information through a scanning device, and synchronously acquiring a physical signal of a torque detection device and a digital image of a tool identification file; grading judgment processing: performing process conformity verification on the measurement data, and generating a process stability evaluation report in combination with historical detection records; according to the closed-loop control strategy, the detection scheme is automatically adjusted according to the quality fluctuation characteristics. According to the scheme, the problem that a traditional method is seriously insufficient in the aspects of detection efficiency, abnormal traceability, process control and the like can be solved.

Description

Technical Field

[0001] The present invention relates to the technical field of torque spot checking, and in particular to a torque spot checking method and a spot checking system. Background Art

[0002] In the field of intelligent manufacturing, torque detection is a core component in ensuring equipment assembly quality. Traditional torque spot-checking systems often rely on a combination of fixed-station detection equipment and manual recording, which presents significant technical bottlenecks. First, each station's detection equipment operates independently, and detection data is dispersed in local storage units, making it impossible to establish a shop-level quality monitoring network, resulting in delayed response to exceptions. Second, sampling strategies rely on empirical fixed ratios, failing to establish a dynamic correlation between quality fluctuations and detection intensity. This results in both insufficient detection of high-risk links and wasted resources in low-risk areas. Third, quality determination is limited to a simple comparison of single measurements against standard thresholds, lacking statistical modeling of process stability parameters, making it difficult to identify potential systemic quality degradation. Fourth, multi-source data, such as tool authentication, physical testing, and image recording, lacks temporal correlation and spatial matching verification, leading to risks such as data disarray and equipment misuse. While existing quality spot-checking systems based on threshold alarms exist, their core flaws include: static sampling rules that cannot adapt to dynamic production line changes, discrete data analysis that struggles to predict trends, and single-dimensional decision logic that is prone to boundary misjudgments. Especially when faced with flexible production scenarios with multiple varieties and small batches, traditional methods have exposed serious deficiencies in detection efficiency, abnormality traceability, process control, etc.

[0003] There is an urgent need to build a new intelligent torque sampling inspection system that can solve the boundary misjudgment problem caused by the single-dimensional judgment logic mentioned above. Especially in the face of high-variety, small-batch flexible production scenarios, traditional methods have exposed serious deficiencies in detection efficiency, anomaly traceability, and process control. Summary of the Invention

[0004] Current static sampling rules are unable to adapt to the dynamic changes of production lines, discrete data analysis is difficult to support trend prediction, and single-dimensional judgment logic is prone to boundary misjudgments. Especially in the face of flexible production scenarios with multiple varieties and small batches, traditional methods have exposed serious deficiencies in detection efficiency, anomaly traceability, and process control. This application provides an automatic torque sampling inspection system and control method to solve the above problems.

[0005] To achieve the above object, the present invention is implemented through the following technical solutions: The present application discloses a torque sampling method, including: S1. Digitizing Tool Identity: Laser-etching an encrypted QR code in the stress concentration area of ​​the tool, and storing the tool specifications, testing standards, and usage constraints in a central database; S2. Intelligent Task Allocation: The system generates dynamic inspection plans based on the real-time status of the production line, equipment service life, and operator skill level. S3. Multi-source data acquisition: Acquires tool identity information through a scanning device, simultaneously collecting physical signals from the torque detection device and digital images of the tool identification file; S4. Grading determination processing: Verify process conformity of measurement data and generate a process stability assessment report based on historical test records. S5. Closed-loop control strategy: Automatically adjust the detection plan based on quality fluctuation characteristics, and trigger cross-station collaborative detection mechanism when abnormal data patterns are found.

[0006] Adopting the above technical solution: The present invention can solve the problems in the traditional torque sampling method where tool identification relies on manual recording and is prone to information mismatch; the detection task allocation is rigid and cannot adapt to changes in production line status; the data collection link lacks multi-source verification and there is a risk of single point failure; quality judgment is disconnected from process control, and abnormality handling is delayed. In the existing technology, the various links of the detection process are linearly related and lack a feedback optimization mechanism. This solution uses laser encrypted QR codes to digitize tool identities, eliminate manual recording errors, and ensure information matching accuracy; the dynamic detection plan integrates parameters such as production line load, equipment aging, and personnel skills to achieve precise adaptation of detection resources to production needs; multi-source data collection integrates physical measurement, image verification, and timing verification to ensure data integrity; the hierarchical judgment mechanism breaks through the single threshold limit and combines real-time data with historical trends to make comprehensive decisions; the closed-loop control strategy establishes a positive cycle of detection-analysis-optimization to shorten the average processing time of quality anomalies.

[0007] Preferably, the determination process in S4 includes: Primary judgment stage: compare the absolute deviation of a single measurement value with the preset standard range to see if it is within the first safety interval; Advanced judgment stage: Analyze whether the statistical distribution characteristics of continuous detection data meet the process stability requirements; Comprehensive judgment stage: Combine the primary and advanced judgment results to generate the final quality conclusion and start the review and verification process for the boundary data.

[0008] Adopting the above technical solution: The above solution can solve the technical problems in traditional technical solutions that single detection judgment cannot identify systematic deviations, resulting in repeated abnormalities in critical qualified tools; fixed tolerance thresholds ignore the impact of process fluctuations, which is prone to misjudgment and missed judgments.

[0009] Further preferably, the dynamic fluctuation threshold calculation formula of the second determination layer is: , in, is the dynamic fluctuation threshold, is the standard deviation of recent historical data, is the sample size of the current test batch.

[0010] The above technical solution is adopted: Based on the statistical process control (SPC) theory, the standard deviation σ of historical data and the current sample size n are included in the calculation, so that the threshold setting can be adaptively adjusted according to the data distribution characteristics; the 3σ principle is inversely proportional to the square root of the sample size, ensuring that stricter control limits are used when testing small samples, and appropriately relaxed when testing large samples to improve efficiency.

[0011] Further preferably, the dynamic detection plan in S2 includes: The basic inspection frequency is set at 20%-30% of the completed quantity of each production unit; When the qualified rate of the last three batches dropped by more than 5 percentage points in a row, the inspection frequency was increased to 1.5 times the original plan; When two critical qualified data are detected for the same model tool, the full detection mode of the model is activated; The sampling distribution adopts a time-space balance strategy to ensure the detection coverage of each production period and equipment location.

[0012] Adopting the above technical solution: the intelligent inspection plan constructed by this claim realizes four-dimensional optimization: the basic inspection frequency is set in the range of 20%-30% to balance the inspection cost and quality risk; the continuous decline in the qualified rate triggers the inspection frequency doubling mechanism to quickly curb the trend of quality deterioration; the critical qualified data full inspection mode effectively identifies potential defect tools and prevents batch quality problems; the time and space balance strategy ensures each production period.

[0013] Further preferably, the data collection process in S3 includes: Image association verification: Verify the spatial position matching between the tool body and the identification file; Timing integrity check: ensures that the timing of identity recognition, physical inspection, and image recording operations conforms to process specifications; Equipment health monitoring: Real-time acquisition of calibration status information of detection devices, and blocking of detection processes at the critical value of calibration validity period.

[0014] Using this technical solution, the triple verification mechanism established by this claim provides comprehensive data assurance: Image correlation verification verifies the spatial consistency between the tool and the identification file using a feature point matching algorithm, preventing identity theft; Temporal integrity verification establishes a digital twin model of the operational process, ensuring strict compliance with specifications; and Equipment Health Monitoring automatically blocks the detection function for expired, uncalibrated equipment, ensuring data accuracy from the source. This mechanism reduces the data collection error rate to below 0.05% and increases the timely calibration execution rate to 98%.

[0015] Further preferably, the closed-loop control strategy in step S5 includes: When the process stability index drops to the warning threshold, the detection cycle is automatically shortened and the sampling sample size is expanded; Initiate a reverse tracing mechanism for production batches associated with abnormal data; Establish a linkage analysis model between test data and equipment maintenance records; Execute a gradual strategy callback after the quality indicators return to stability.

[0016] Using this technical solution, the closed-loop control system constructed according to this claim achieves four key improvements: dynamic adjustment of testing cycles and sample sizes precisely matches resource inputs with quality risks; a reverse tracing mechanism establishes a correlation map between abnormal data and production batches, enabling rapid identification of responsible links; integrated analysis of testing data and maintenance records enables early identification of equipment performance degradation trends; and a policy callback mechanism prevents overtesting and automatically restores the baseline plan once quality stabilizes. This system has increased cross-departmental collaboration efficiency by 50% and raised the accuracy of equipment failure prediction to 85%.

[0017] Further preferably, the method includes abnormal data processing rules: The first abnormal data is recorded as an observation item and does not participate in the quality judgment; Continuous abnormal data triggers sound and light alarms and generates quality event codes; Correction of abnormal data requires two-factor authentication and retention of audit logs; All abnormal events are automatically linked to the traceability module of the production management system.

[0018] Adopting the above technical solution: The standardized exception handling mechanism established by this claim realizes four layers of protection: the first abnormal observation mechanism avoids overreaction and reduces more than 70% of invalid alarms; continuous abnormal sound and light alarms ensure timely intervention to prevent the situation from escalating; two-factor authentication and audit logs realize operation traces to meet ISO9001 quality management requirements; the quality event coding system connects the production management system to realize the full life cycle tracking of abnormalities.

[0019] Further preferably, the quality report output by the system includes: Quality heat map by time period and region; Time series evolution graph of process stability indicators; Abnormal event causal analysis tree diagram; Detection strategy optimization suggestion matrix; Equipment preventive maintenance forecast checklist.

[0020] This solution uses quality heat maps to intuitively display spatiotemporal distribution characteristics and quickly locate weak links; process stability evolution maps reveal quality trends and support forward-looking decision-making; cause-and-effect analysis tree diagrams identify the root causes of anomalies and shorten problem diagnosis time; and the optimization suggestion matrix provides quantitative improvement solutions to guide the precise allocation of resources.

[0021] A torque sampling inspection system, applied to any of the above sampling inspection methods, is characterized by comprising: Terminal equipment groups distributed across multiple production stations, each integrating a QR code scanning device, a torque detection device, an optical imaging device, and a data processing terminal; A central server cluster deployed in the workshop establishes real-time data channels with all terminal equipment groups; Adaptive sampling control module dynamically generates inspection task sequences based on equipment usage frequency and historical quality data; Data association engine, which establishes multi-dimensional binding relationships between test data and production batches, equipment identification, and operator information; The core module of quality analysis predicts the trend of test data based on process stability indicators; The data processing terminal comprises: Device identity verification unit, which parses the tool's identification information and matches it with pre-stored technical parameters; A real-time decision-making unit receives the physical quantity measurement values ​​of the detection device and executes the hierarchical decision logic; The image verification unit verifies the integrity and consistency of the tool identification file through feature extraction algorithm.

[0022] The above technical solution can address the three major flaws of traditional torque spot inspection systems: equipment silos, severe data discreteness, and static strategies. This patent establishes a real-time data channel between the workstation terminal equipment group and the central server to form a quality monitoring network covering the entire workshop, eliminating information silos. The data association engine deeply binds inspection data with production factors to achieve accurate traceability of quality issues. The adaptive sampling control module integrates multi-dimensional parameters such as equipment usage frequency and historical pass rate to generate a dynamic inspection task sequence and optimize inspection resource allocation. The core quality analysis module introduces a process capability index algorithm to overcome the limitations of single-shot inspections and achieve quality trend prediction. The overall inspection efficiency of the system is increased by more than 3 times, and the abnormal response speed is shortened to minutes.

[0023] Further preferably, the dynamic sampling module executes the following sampling probability model: , in, is the current sampling probability, , is the dynamic weight coefficient, is the target pass rate threshold, is the average pass rate of the last N times, is the frequency of the most recent M abnormalities, is the total number of historical anomalies, is the regulating factor.

[0024] Using the above technical solution: Traditional sampling strategies use fixed ratios or periodic rules, which cannot distinguish the quality risk levels of different equipment, have a delayed response to quality fluctuations, and make it difficult to strengthen detection efforts in the early stages of anomalies. This solution uses the logistic function to construct a quality deviation response mechanism. When the real-time pass rate approaches the target threshold, the sampling intensity is automatically increased to form a quality safety buffer zone. The abnormality frequency influencing factor is introduced to implement targeted and enhanced detection of equipment types with frequent recent abnormalities. Dynamic weight coefficients α and β achieve a balanced regulation of quality stability and abnormality sensitivity, ensuring that the sampling strategy maintains both baseline efficiency and risk response capabilities. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0026] Figure 1 This is a flow chart of the torque sampling method for this application; Figure 2 This is the block diagram of the sampling inspection system for this application. DETAILED DESCRIPTION

[0027] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0028] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, operations, elements, components and / or groups thereof.

[0029] See also Figure 1 and Figure 2, for example, the traditional torque sampling inspection method has the following technical problems: tool identification relies on manual records, which is prone to information mismatch; the inspection task allocation is rigid and cannot adapt to changes in production line status; the data collection link lacks multi-source verification, which has the risk of single point failure; quality judgment is disconnected from process control, and abnormality handling is delayed. In the existing technology, the various links in the inspection process are linearly related and lack feedback optimization mechanism. Based on this, the embodiment of the present application discloses a torque sampling inspection method, including: S1. Digitizing Tool Identity: Laser-etching an encrypted QR code in the stress concentration area of ​​the tool, and storing the tool specifications, testing standards, and usage constraints in a central database; S2. Intelligent Task Allocation: The system generates dynamic inspection plans based on the real-time status of the production line, equipment service life, and operator skill level. S3. Multi-source data acquisition: Acquires tool identity information through a scanning device, simultaneously collecting physical signals from the torque detection device and digital images of the tool identification file; S4. Grading determination processing: Verify process conformity of measurement data and generate a process stability assessment report based on historical test records. S5. Closed-loop control strategy: Automatically adjust the detection plan based on quality fluctuation characteristics, and trigger cross-station collaborative detection mechanism when abnormal data patterns are found.

[0030] It is worth mentioning that the above claims require the construction of a closed-loop intelligent detection method: laser encrypted QR codes realize the digitization of tool identity, eliminate manual recording errors, and achieve 100% information matching accuracy; dynamic detection plans integrate production line load, equipment aging, personnel skills and other parameters to achieve precise adaptation of detection resources and production needs; multi-source data collection integrates physical measurement, image verification, and timing verification to provide triple guarantees, and data integrity is improved to 99.9%; the hierarchical judgment mechanism breaks through the single threshold limit and combines real-time data with historical trends to make comprehensive decisions; the closed-loop control strategy establishes a positive cycle of detection-analysis-optimization, which shortens the average processing time of quality anomalies to 30% of traditional methods.

[0031] Traditional quality determination methods have the following technical problems: a single test cannot identify systematic deviations, resulting in repeated abnormalities in critically qualified tools; fixed tolerance thresholds ignore the impact of process fluctuations, which can easily lead to misjudgments and missed judgments. In existing technologies, the determination logic lacks a hierarchical design, and the boundary data processing rules are vague. The determination process in S4 includes: Primary judgment stage: compare the absolute deviation of a single measurement value with the preset standard range to see if it is within the first safety interval; Advanced judgment stage: Analyze whether the statistical distribution characteristics of continuous detection data meet the process stability requirements; Comprehensive judgment stage: Combine the primary and advanced judgment results to generate the final quality conclusion and start the review and verification process for the boundary data.

[0032] Notably, this solution's three-tiered judgment system enables refined quality control. The primary judgment stage establishes a first safety interval to rapidly screen clearly qualified / unqualified data, completing over 80% of basic judgments. The advanced judgment stage analyzes statistical characteristics such as the moving range and standard deviation of continuous test data to identify potential process anomalies and provide early warning of systemic risks. The comprehensive judgment stage initiates manual review and verification of boundary data, incorporating auxiliary information such as tool usage history and environmental parameters for the final decision. This system has reduced the false positive rate to below 0.2% and increased the lead time for process anomaly warnings to over 8 hours.

[0033] For example, traditional process stability assessments have the following technical issues: using fixed thresholds to judge process capability ignores the impact of sample size and data distribution; and failing to establish a dynamic correlation between test data and process parameters, resulting in assessment results that deviate from actual operating conditions. In existing technologies, the setting of process control thresholds relies on empirical values ​​and lacks a scientific basis for calculation. The dynamic fluctuation threshold calculation formula for the second determination layer is: , in, is the dynamic fluctuation threshold, The standard deviation of recent historical data can reflect the actual fluctuation level of the process in real time and is more adaptable to gradual influencing factors such as equipment aging and environmental changes than fixed thresholds. is the sample size of the current test batch; To eliminate the impact of sample size differences between test batches on threshold calculation and ensure reliable judgments when testing small samples, this solution proposes a dynamic fluctuation threshold calculation model. Based on statistical process control (SPC) theory, this model incorporates the standard deviation σ of historical data and the current sample size n into the calculation, allowing threshold settings to be adaptively adjusted based on data distribution characteristics. The inverse relationship between the 3σ principle and the square root of the sample size ensures stricter control limits for small sample tests, while appropriately relaxing them for larger sample tests to improve efficiency. This model improves the accuracy of process stability assessment by 40% and increases the accuracy of pattern recognition to over 95%.

[0034] Traditional inspection plan development suffers from the following technical issues: fixed-frequency inspections waste resources and fail to focus on risk areas; response to quality trend changes is slow, with plan adjustments lagging behind actual needs; sampling distribution is irrational, and inspection coverage of some key equipment is insufficient. In existing technologies, inspection strategy optimization relies on manual experience and lacks quantitative decision support. Based on this, the dynamic inspection plan in S2 includes: The basic inspection frequency is set at 20%-30% of the completed quantity of each production unit; When the qualified rate of the last three batches dropped by more than 5 percentage points in a row, the inspection frequency was increased to 1.5 times the original plan; When two critical qualified data are detected for the same model tool, the full detection mode of the model is activated; The sampling distribution adopts a time-space balance strategy to ensure the detection coverage of each production period and equipment location.

[0035] It's worth noting that the intelligent inspection plan constructed in this solution achieves four-dimensional optimization: the basic inspection frequency is set between 20% and 30%, balancing inspection costs and quality risks; a continuous decline in the pass rate triggers a frequency doubling mechanism, rapidly curbing the trend of quality deterioration; a full inspection mode for critical pass data effectively identifies potential defect tools and prevents batch-wide quality issues; and a time-space balance strategy ensures inspection coverage of at least 85% across all production periods and equipment locations. This solution improves inspection resource utilization by 60% and increases coverage of high-risk links to 100%.

[0036] For example, the data collection process of traditional technical solutions has the following problems: the tool itself does not match the identification file; the detection operation sequence is chaotic, resulting in data disorder; and the detection equipment is out of use and causes errors. In the existing technology, data verification relies on manual verification, which is inefficient and prone to errors. The data collection process in S3 includes: Image association verification: Verify the spatial position matching between the tool body and the identification file; Timing integrity check: ensures that the timing of identity recognition, physical inspection, and image recording operations conforms to process specifications; Equipment health monitoring: Real-time acquisition of calibration status information of detection devices, and blocking of detection processes at the critical value of calibration validity period.

[0037] Notably, the triple verification mechanism established by this claim provides comprehensive data protection: Image correlation verification verifies the spatial consistency between the tool and the identification file through a feature point matching algorithm, preventing identity theft; Temporal integrity verification establishes a digital twin model of the operational process, ensuring strict compliance with specifications; and Equipment Health Monitoring automatically blocks the detection function of expired and uncalibrated equipment, ensuring data accuracy from the source. This mechanism has reduced the data collection error rate to below 0.05% and increased the timely calibration execution rate to 98%.

[0038] Traditional quality control strategies have the following technical issues: Exception handling is limited to the current workstation and lacks cross-departmental coordination mechanisms; strategy adjustments rely on manual decision-making, resulting in slow response and poor consistency. In existing technologies, inspection and maintenance systems are independent of each other, and the value of data is not fully explored. Based on this, the closed-loop control strategy in step S5 includes: When the process stability index drops to the warning threshold, the detection cycle is automatically shortened and the sampling sample size is expanded; Initiate a reverse tracing mechanism for production batches associated with abnormal data; Establish a linkage analysis model between test data and equipment maintenance records; Execute a gradual strategy callback after the quality indicators return to stability.

[0039] Notably, the closed-loop control system established by this claim achieves four key improvements: dynamic adjustment of testing cycles and sample sizes precisely matches resource inputs with quality risks; a reverse tracing mechanism establishes a correlation map between abnormal data and production batches, enabling rapid identification of responsible links; integrated analysis of testing data and maintenance records enables early identification of equipment performance degradation trends; and a policy callback mechanism prevents overtesting and automatically restores the baseline plan once quality stabilizes. This system has increased cross-departmental collaboration efficiency by 50% and raised the accuracy of equipment failure prediction to 85%.

[0040] For example, the traditional exception handling process has three major loopholes: direct involvement in the judgment of the first exception is prone to false alarms; exception correction lacks authority control and audit tracking; and exception events are disconnected from the production management system, making tracing difficult. In the existing technology, the exception data management rules are vague and have poor operability. The method includes the following exception data processing rules: The first abnormal data is recorded as an observation item and does not participate in the quality judgment; Continuous abnormal data triggers sound and light alarms and generates quality event codes; Correction of abnormal data requires two-factor authentication and retention of audit logs; All abnormal events are automatically linked to the traceability module of the production management system.

[0041] Notably, the standardized exception handling mechanism established in this solution provides four layers of protection: a first-time exception observation mechanism prevents overreaction and reduces invalid alarms by over 70%; continuous abnormality sound and light alarms ensure timely intervention to prevent escalation; two-factor authentication and audit logs ensure operational traceability, meeting ISO9001 quality management requirements; and a quality event coding system connects to the production management system, enabling full lifecycle tracking of exceptions. This mechanism has increased the exception handling compliance rate to 99% and tripled traceability efficiency.

[0042] Traditional quality reports have the following technical issues: Data presentation is limited in form and lacks a multi-dimensional analytical perspective; a correlation model between quality indicators and production parameters is not established; predictive recommendations are missing, and decision support capabilities are weak. In existing technologies, report generation relies on fixed templates and has a low level of intelligence. The quality report output by the system includes: Quality heat map by time period and region; Time series evolution graph of process stability indicators; Abnormal event causal analysis tree diagram; Detection strategy optimization suggestion matrix; Equipment preventive maintenance forecast checklist.

[0043] Notably, the intelligent reporting system designed in this solution achieves five-dimensional value enhancement: quality heat maps intuitively display spatiotemporal distribution characteristics, quickly locating weak links; process stability evolution maps reveal quality trends, supporting forward-looking decision-making; cause-and-effect analysis tree diagrams pinpoint the root causes of anomalies, shortening problem diagnosis time; optimization suggestion matrices provide quantitative improvement solutions, guiding the precise allocation of resources; and maintenance prediction checklists establish equipment health records, extending tool life. This system has increased management decision-making efficiency by 60% and boosted the preventive maintenance execution rate to over 90%.

[0044] For example, the traditional torque sampling inspection system has three major defects: equipment islanding, serious data discreteness, and static strategy. Specifically, the following are: each station detection equipment operates independently, and it is impossible to form a workshop-level quality monitoring network; the detection data lacks effective correlation with production batches, equipment status, and operator information; the sampling strategy relies on fixed rules and cannot dynamically adjust the detection intensity according to quality fluctuations; the detection judgment is based only on a single measurement result, ignoring the process stability analysis. Based on the above problems, the present application provides a torque sampling inspection system, which is applied to the sampling inspection method as described in any of the above, and is characterized in that it includes: Terminal equipment groups distributed across multiple production stations, each integrating a QR code scanning device, a torque detection device, an optical imaging device, and a data processing terminal; A central server cluster deployed in the workshop establishes real-time data channels with all terminal equipment groups; Adaptive sampling control module dynamically generates inspection task sequences based on equipment usage frequency and historical quality data; Data association engine, which establishes multi-dimensional binding relationships between test data and production batches, equipment identification, and operator information; The core module of quality analysis predicts the trend of test data based on process stability indicators; The data processing terminal comprises: Device identity verification unit, which parses the tool's identification information and matches it with pre-stored technical parameters; A real-time decision-making unit receives the physical quantity measurement values ​​of the detection device and executes the hierarchical decision logic; The image verification unit verifies the integrity and consistency of the tool identification file through feature extraction algorithm.

[0045] It's worth noting that this claim, through the construction of a distributed intelligent inspection system, can establish a real-time data channel between workstation terminal equipment groups and a central server, forming a quality monitoring network covering the entire workshop and eliminating information silos. The data association engine deeply binds inspection data with production factors to achieve precise traceability of quality issues. The adaptive sampling control module integrates multi-dimensional parameters such as equipment usage frequency and historical pass rates to generate dynamic inspection task sequences and optimize inspection resource allocation. The core quality analysis module introduces the process capability index (CPK) algorithm, breaking through the limitations of single-shot inspections and enabling quality trend prediction. The overall system inspection efficiency has increased by more than three times, and the response time to abnormalities has been shortened to minutes.

[0046] Traditional sampling strategies, which use fixed ratios or periodic rules, suffer from the following drawbacks: They are unable to distinguish between the quality risk levels of different devices, resulting in under-detection of high-risk devices and over-detection of low-risk devices; they also experience a lag in responding to quality fluctuations, making it difficult to enhance detection efforts in the early stages of anomalies. Existing sampling methods based on simple threshold adjustment lack the ability to dynamically model the effects of multiple coupled factors. To address these issues, the dynamic sampling module implements the following sampling probability model: , in, is the current sampling probability, , is the dynamic weight coefficient, is the target pass rate threshold, is the average pass rate of the last N times, is the frequency of the most recent M abnormalities, is the total number of historical anomalies, is the regulating factor.

[0047] The above solution is achieved by using the Sigmoid function Achieve a smooth transition of qualified rate deviation and avoid waste of testing resources caused by sudden change of sampling rate. near When the function output tends to be stable, it effectively controls the detection cost during the quality stabilization period. The α coefficient adjusts the sensitivity of the qualified rate deviation and is suitable for long-term control of quality trends. The β coefficient strengthens the immediate response to abnormal frequency and quickly intervenes in sudden quality problems. The weighted sum constraint α+β=1 ensures that the system strikes a balance between stability and sensitivity. Quantify the current abnormal situation and automatically increase the sampling intensity when abnormalities occur frequently in the near future, forming an indexed assessment model for quality risk; the design of the adjustment factor k controls the steepness of the function curve by taking the value range of 0.5-2.0 to adapt to the different requirements of different production processes for the tolerance of quality fluctuations. It's worth noting that the sampling probability model proposed in this claim achieves intelligent decision-making through mathematical modeling: It uses the logistic function to construct a quality deviation response mechanism, automatically increasing sampling intensity when the real-time pass rate approaches the target threshold, forming a quality safety buffer zone. It also introduces an abnormality frequency influencing factor to implement targeted enhanced detection of equipment types with recent frequent abnormalities. Dynamic weight coefficients α and β achieve a balanced regulation of quality stability and abnormality sensitivity, ensuring that the sampling strategy maintains baseline efficiency while also possessing risk response capabilities. This model increases detection coverage for high-risk equipment to 2.5 times the conventional level and reduces the anomaly missed detection rate to one-tenth of that of traditional methods.

[0048] Unless otherwise specified, the device components involved in the above embodiments are all conventional device components, and the connection methods and control methods involved are all conventional connection methods and control methods unless otherwise specified.

[0049] The present invention has been described in detail above with reference to the embodiments. However, those skilled in the art will appreciate that, without departing from the spirit of the present invention, the specific parameters in the above embodiments may be modified to form multiple specific embodiments, which are all within the common variation range of the present invention and will not be described in detail here.

Claims

1. A torque sampling method, characterized in that: include: S1. Digitize the tool identity by laser-etching an encrypted QR code in the stress concentration area of ​​the tool. Associate and store tool specifications, testing standards, and usage constraints in a central database. S2. Intelligent task allocation: The system generates dynamic inspection plans based on the real-time status of the production line, equipment service life, and operator skill level. S3 multi-source data acquisition, obtain tool identity information by scanning device, synchronous acquisition of physical signals and digital images of the torque detection device tool identification file; S4. Perform grading and determination, verify process conformity of measurement data, and generate a process stability assessment report based on historical inspection records. S5. Implement a closed-loop control strategy to automatically adjust the detection plan based on quality fluctuation characteristics, and trigger a cross-station collaborative detection mechanism when abnormal data patterns are found.

2. A torque sampling method according to claim 1, characterized in that: The determination process in S4 includes: Primary judgment stage: compare the absolute deviation of a single measurement value with the preset standard range to see if it is within the first safety interval; Advanced judgment stage: Analyze whether the statistical distribution characteristics of continuous detection data meet the process stability requirements; Comprehensive judgment stage: Combine the primary and advanced judgment results to generate the final quality conclusion and start the review and verification process for the boundary data.

3. A torque sampling method according to claim 1, characterized in that: The calculation formula for the dynamic fluctuation threshold of the second determination layer is: ， in, is the dynamic fluctuation threshold, is the standard deviation of recent historical data, is the sample size of the current test batch.

4. A torque sampling method according to claim 1, characterized in that: The dynamic detection plan in S2 includes: The basic inspection frequency is set at 20%-30% of the completed quantity of each production unit; When the qualified rate of the last three batches dropped by more than 5 percentage points in a row, the inspection frequency was increased to 1.5 times the original plan; When two critical qualified data are detected for the same model tool, the full detection mode of the model is activated; The sampling distribution adopts a time-space balance strategy to ensure the detection coverage of each production period and equipment location.

5. The torque sampling method according to claim 1, characterized in that: The data collection process in S3 includes: Image association verification: Verify the spatial position matching between the tool body and the identification file; Timing integrity check: ensures that the operation timing of identity recognition, physical inspection, and image recording complies with process specifications; Equipment health monitoring: Real-time acquisition of calibration status information of detection devices, and blocking of detection processes at the critical value of calibration validity period.

6. A torque sampling method according to claim 1, characterized in that: The closed-loop control strategy in step S5 includes: When the process stability index drops to the warning threshold, the detection cycle is automatically shortened and the sampling sample size is expanded; Initiate a reverse tracing mechanism for production batches associated with abnormal data; Establish a linkage analysis model between test data and equipment maintenance records; Execute a gradual strategy callback after the quality indicators return to stability.

7. A torque sampling method according to claim 1, characterized in that: The method includes abnormal data processing rules: The first abnormal data is recorded as an observation item and does not participate in the quality judgment; Continuous abnormal data triggers sound and light alarms and generates quality event codes; Correction of abnormal data requires two-factor authentication and retention of audit logs; All abnormal events are automatically linked to the traceability module of the production management system.

8. The torque sampling method according to claim 1, characterized in that: The quality report output by the system includes: Quality heat map by time period and region; Time series evolution graph of process stability indicators; Abnormal event causal analysis tree diagram; Detection strategy optimization suggestion matrix; Equipment preventive maintenance forecast checklist.

9. A sampling inspection system, applied to a torque sampling inspection method according to any one of claims 1 to 8, characterized in that: include: Terminal equipment groups distributed across multiple production stations, each integrating a QR code scanning device, a torque detection device, an optical imaging device, and a data processing terminal; A central server cluster deployed in the workshop establishes real-time data channels with all terminal equipment groups; Adaptive sampling control module dynamically generates inspection task sequences based on equipment usage frequency and historical quality data; Data association engine, which establishes multi-dimensional binding relationships between test data and production batches, equipment identification, and operator information; The core module of quality analysis predicts the trend of test data based on process stability indicators; The data processing terminal comprises: Device identity verification unit, which parses the tool's identification information and matches it with pre-stored technical parameters; A real-time decision-making unit receives the physical quantity measurement values ​​of the detection device and executes the hierarchical decision logic; The image verification unit verifies the integrity and consistency of the tool identification file through feature extraction algorithm.

10. The torque sampling inspection system according to claim 9, characterized in that: The dynamic sampling module implements the following sampling probability model: ， in, is the current sampling probability, , is the dynamic weight coefficient, is the target pass rate threshold, is the average pass rate of the last N times, is the frequency of the most recent M abnormalities, is the total number of historical anomalies, is the regulating factor.

Images