Dynamic quality monitoring system for intelligent assembly detection of automobile door lock

Abstract

This application provides a dynamic quality monitoring system for intelligent assembly inspection of automotive door locks, relating to the field of quality monitoring technology. It includes: an intelligent tactile glove for recording worker assembly datasets; a multi-dimensional collaborative detection module for performing anomaly analysis and establishing a multi-dimensional linked dataset; a general detection module for performing general inspections of automotive door locks and establishing a detection dataset; and a collaborative analysis module for calling a collaborative analysis network to identify assembly quality anomalies based on the detection dataset, multi-dimensional linked dataset, and worker assembly dataset, and establishing anomaly identification results. The collaborative analysis network includes a batch collaborative analysis layer and a deep collaborative analysis layer. This application achieves the technical goals of multi-dimensional real-time monitoring, intelligent analysis, and accurate anomaly identification in the automotive door lock assembly process, thereby improving assembly quality, increasing production efficiency, reducing rework costs, and enhancing the intelligence level of the manufacturing process.

Description

Technical Field

[0001] This application relates to the field of quality monitoring technology, and in particular to a dynamic quality monitoring system for intelligent assembly inspection of automotive door locks. Background Technology

[0002] The automotive manufacturing industry is moving towards high automation and intelligence, and as a key component, the assembly quality of car door locks directly affects the safety of the entire vehicle and the user experience.

[0003] Currently, existing automotive door lock assembly inspection technologies lack real-time monitoring capabilities for the assembly process, failing to effectively capture the details of worker actions, such as pressure, angle, and trajectory. Due to individual differences in manual assembly, even the same worker may perform actions inconsistently at different times, making it difficult for traditional inspection methods to identify quality issues caused by variations in assembly methods. Secondly, data analysis capabilities are limited; existing inspection methods rely heavily on single-dimensional results, failing to integrate data from multiple sensors for comprehensive analysis, resulting in low accuracy and coverage of anomaly detection. Furthermore, the feedback mechanisms of existing technologies are relatively lagging, typically only initiating quality traceability after assembly defects are discovered, lacking proactive warnings and intelligent optimization mechanisms. This delays in the discovery and improvement of quality problems, impacting overall production efficiency.

[0004] In summary, existing technologies suffer from technical problems such as a lack of real-time monitoring of the assembly process, limited data analysis capabilities, and a delayed feedback mechanism, which make it difficult to detect and accurately identify assembly quality issues in a timely manner, further affecting production efficiency, quality stability, and manufacturing cost control. Summary of the Invention

[0005] The purpose of this application is to provide a dynamic quality monitoring system for intelligent assembly inspection of automotive door locks, in order to solve the technical problems in the prior art that make it difficult to detect and accurately identify assembly quality problems in a timely manner due to the lack of real-time monitoring of the assembly process, limited data analysis capabilities, and lagging feedback mechanisms, which further affect production efficiency, quality stability, and manufacturing cost control.

[0006] In view of the above problems, this application provides a dynamic quality monitoring system for intelligent assembly inspection of automotive door locks, comprising: an intelligent tactile glove for recording a worker assembly dataset after signal interaction with the automotive door lock, the worker assembly dataset being identified by the automotive door lock number, and the intelligent tactile glove integrating a flexible pressure sensor and an inertial measurement unit; a multi-dimensional collaborative detection module for receiving the worker assembly dataset, performing anomaly analysis, configuring linkage acquisition focus based on the anomaly analysis results, and performing assembly linkage acquisition based on the linkage acquisition focus after detecting the corresponding automotive door lock number, thereby establishing a multi-dimensional linkage dataset; a general detection module for performing general inspection of automotive door locks and establishing an inspection dataset; and a collaborative analysis module for calling a collaborative analysis network to identify assembly quality anomalies based on the inspection dataset, the multi-dimensional linkage dataset, and the worker assembly dataset, and establishing anomaly identification results, the collaborative analysis network including a batch collaborative analysis layer and a deep collaborative analysis layer.

[0007] The technical solution provided in this application has at least the following technical effects or advantages: by achieving the technical goal of multi-dimensional real-time monitoring, intelligent analysis and accurate anomaly identification of the automotive door lock assembly process, it achieves the technical effects of improving assembly quality, increasing production efficiency, reducing rework costs and enhancing the level of intelligence in the manufacturing process.

[0008] The above description is merely an overview of the technical solution of this application. To better understand the technical means of this application and to facilitate its implementation according to the description, and to make the above and other objects, features, and advantages of this application more apparent, specific embodiments of this application are described below. It should be understood that the content described in this section is not intended to identify key or important features of the embodiments of this application, nor is it intended to limit the scope of this application. Other features of this application will become readily apparent through the following description. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely exemplary. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0010] Figure 1 This is a schematic diagram of the dynamic quality monitoring system for intelligent assembly testing of automotive door locks, as described in this application.

[0011] Figure 2 This is a schematic diagram of the collaborative analysis module in the dynamic quality monitoring system for intelligent assembly testing of automotive door locks, as described in this application.

[0012] Figure labeling: Smart tactile glove 10, Multi-dimensional collaborative detection module 20, General detection module 30, Collaborative analysis module 40, Batch analysis sub-module 41, Depth analysis sub-module 42, Output module 43. Detailed Implementation

[0013] This application provides a dynamic quality monitoring system for intelligent assembly and inspection of automotive door locks. This system addresses the technical problems in existing technologies where the lack of real-time monitoring of the assembly process, limited data analysis capabilities, and delayed feedback mechanisms make it difficult to promptly detect and accurately identify assembly quality issues, further impacting production efficiency, quality stability, and manufacturing cost control. The system achieves the technical goals of multi-dimensional real-time monitoring, intelligent analysis, and accurate anomaly identification of the automotive door lock assembly process, thereby improving assembly quality, increasing production efficiency, reducing rework costs, and enhancing the intelligence level of the manufacturing process.

[0014] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. It should be understood that this application is not limited to the exemplary embodiments described herein. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application. It should also be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all of them.

[0015] Please see the appendix Figure 1 This application provides a dynamic quality monitoring system for intelligent assembly inspection of automotive door locks, specifically including:

[0016] The smart haptic glove 10 is used to record a worker assembly dataset after signal interaction with a car door lock. The worker assembly dataset is identified by the car door lock number. The smart haptic glove 10 integrates a flexible pressure sensor and an inertial measurement unit.

[0017] Specifically, the smart haptic glove 10 is a wearable device capable of sensing the external environment and transmitting information to a dynamic quality monitoring system used for intelligent assembly inspection of automotive door locks. It not only provides hand protection but also collects information such as hand movements and contact pressure. After interacting with the automotive door lock, the glove's internal sensing system detects and records various data when a person operates the lock, such as pressure, contact time, and rotation angle. This data is then collected and stored as a dataset—the worker assembly dataset. The worker assembly dataset is identified by the automotive door lock number, indicating which lock the data is associated with, thus allowing for accurate tracking of the assembly status of each lock.

[0018] The smart haptic glove 10 integrates a flexible pressure sensor and an inertial measurement unit. The flexible pressure sensor is an electronic component that can detect the magnitude of pressure; it can bend and adapt to the curvature of the hand, and can measure the pressure applied when a person operates a door lock. The inertial measurement unit is a device that combines an accelerometer and a gyroscope, and can measure the glove's motion state in real time, including changes in speed, direction, and angle.

[0019] The multi-dimensional collaborative detection module 20 is used to receive the worker assembly dataset, perform anomaly analysis, configure linkage acquisition focus based on the anomaly analysis results, and perform assembly linkage acquisition based on the linkage acquisition focus after detecting the corresponding car door lock number, thereby establishing a multi-dimensional linkage dataset.

[0020] Specifically, the multi-dimensional collaborative detection module 20 receives and comprehensively analyzes data from multiple sources. It receives data generated by workers during the assembly process, calculates the degree of anomaly in various indicators, and determines whether a particular operation deviates from the standard range, thus obtaining anomaly analysis results. Specifically, it compares the current operation with standard data and calculates the degree of deviation to facilitate further optimization measures.

[0021] Based on the degree of anomaly in the anomaly analysis results, the data requiring further attention is determined, resulting in a configuration for coordinated data collection. Coordinated collection means not only analyzing a single data point but also examining multiple related data points. For example, if an anomaly is detected in a worker's pressure data, not only will pressure be monitored, but the worker's hand posture, fingertip trajectory, etc., will also be analyzed simultaneously to find the root cause of the problem. Configured attention refers to dynamically adjusting the focus of attention. For instance, if a worker is found to repeatedly exhibit abnormal hand angles while tightening screws, the monitoring frequency of that worker's hand posture will be increased, while pressure monitoring may remain at a normal level, ensuring accurate analysis of key issues.

[0022] Based on the configuration results of the linked data collection, the corresponding car door lock number is detected, and linked data collection is performed. The collected data is integrated to form a database containing multiple information dimensions, so that subsequent intelligent algorithms can use this data for deep learning and optimization.

[0023] The general detection module 30 is used to perform general detection on car door locks and establish a detection dataset.

[0024] Specifically, the universal testing module 30 is applicable to all automotive door locks, rather than being specific to a particular model or assembly. The universal testing module 30 is used to perform comprehensive testing on automotive door locks, such as checking whether they meet standard requirements, including dimensions, functionality, and durability, to ensure the quality of the door locks and establish a testing dataset.

[0025] The collaborative analysis module 40 is used to call the collaborative analysis network to identify assembly quality anomalies based on the detection dataset, the multidimensional linkage dataset, and the worker assembly dataset, and to establish anomaly identification results. The collaborative analysis network includes a batch collaborative analysis layer and a deep collaborative analysis layer.

[0026] Specifically, the collaborative analysis module 40 utilizes multiple data sources to identify anomalies in assembly quality, ensuring the stability and consistency of product quality. By invoking a collaborative analysis network, the module 40 comprehensively analyzes different types of data, ultimately establishing anomaly identification results to discover and resolve potential problems during the assembly process.

[0027] Invoking a collaborative analytics network refers to automatically or manually triggering collaborative analytics functions to analyze data collected during the current production process. A collaborative analytics network is a computational framework that combines multiple analytical methods, integrating data from multiple sources and performing comprehensive analysis. For example, on a production line, worker operation data, measurement data from inspection equipment, and operational data from automated systems are all collected and input into a collaborative analytics network for joint analysis.

[0028] Next, the collaborative analysis network needs to identify assembly quality anomalies based on the detection dataset, multi-dimensional linked dataset, and worker assembly dataset, identifying situations that deviate significantly from the normal assembly process. For example, if during the assembly of a batch of door locks, the pressing force used by the workers is significantly less than the standard value, and the detection equipment finds that the door locks have poor sealing performance, then this situation may be judged as an assembly anomaly.

[0029] The collaborative analysis network comprises a batch collaborative analysis layer and a deep collaborative analysis layer. The batch collaborative analysis layer performs comparative analysis on products within the same batch. The deep collaborative analysis layer, on the other hand, delves deeper into the root causes of anomalies, combining target detection data, multi-dimensional linked datasets, and worker assembly datasets for correlation analysis.

[0030] The dynamic quality monitoring system for intelligent assembly and testing of automotive door locks can achieve the technical goals of multi-dimensional real-time monitoring, intelligent analysis, and accurate anomaly identification of the automotive door lock assembly process, thereby improving assembly quality, increasing production efficiency, reducing rework costs, and enhancing the level of intelligence in the manufacturing process.

[0031] Furthermore, the smart haptic glove 10 is used for: performing monitoring activation after signal interaction with the car door lock; activating the flexible pressure sensors on the fingertips and palm of the smart haptic glove 10 to establish a pressure dataset; activating the inertial measurement unit to establish an inertial dataset, reconstructing the spatial trajectory based on the motion feature extraction channel, calculating the hand pose and performing fingertip motion modeling; matching standard motion postures based on the hand pose and fingertip motion modeling to establish posture anomalies; using the hand pose and fingertip motion modeling to perform collaborative assembly anomaly identification on the pressure dataset to establish assembly anomalies; and constructing a worker assembly dataset based on the posture anomalies and the assembly anomalies.

[0032] Specifically, after the smart haptic glove 10 interacts with the car door lock, it initiates monitoring. This means that when a person's hand touches the door lock and the contact meets the preset activation conditions, the monitoring function is automatically activated. For example, when a finger touches the door lock handle and maintains a certain pressure for more than one second, data recording is initiated; however, if the contact is only brief, monitoring will not be triggered.

[0033] Activate the flexible pressure sensors at the fingertips and palm of the smart haptic glove 10 to create a pressure dataset for subsequent analysis to determine whether the force applied by the person meets the assembly requirements.

[0034] An inertial measurement unit (IMU) consists of an accelerometer and a gyroscope, capable of detecting the hand's trajectory, angular velocity, and acceleration in space. Activating the IMU creates an inertial dataset, which is used to analyze the stability of the person's movements and to detect any abnormal shaking or tremors.

[0035] Based on the inertial dataset, the motion path of the hand in three-dimensional space is calculated, the hand pose is solved, and the fingertip motion model is performed. That is, the rotation angle and position of the hand are analyzed, and then the motion trajectory, speed, applied force and other information of the fingers are analyzed in a data manner and a mathematical model is established to determine whether the worker's operation meets the standards.

[0036] Based on hand pose and fingertip movement modeling, standard action pose matching is performed. If there is a significant deviation between a person's hand movements and the standard assembly movements, an abnormal pose is established to identify whether the person is performing an incorrect operation.

[0037] By modeling hand posture and fingertip movement, a collaborative assembly anomaly identification method is used to analyze pressure datasets. This method combines multiple data sources to analyze whether there are problems in the assembly process, thereby establishing an assembly anomaly identification. For example, if the hand posture of a person pressing a door lock component is correct, but the fingertip pressure is less than 30 Newtons, it indicates that the person may not be applying enough force to fix the component, and this is judged as an assembly anomaly.

[0038] Finally, based on the analysis results of posture anomalies and assembly anomalies, a worker assembly dataset was constructed, which contains all the key data of the worker throughout the entire assembly process.

[0039] Furthermore, the hand pose and fingertip movement modeling is used to perform collaborative assembly anomaly identification on the pressure dataset to establish assembly anomalies. This includes: aligning the hand pose and fingertip movement modeling with the pressure dataset in time series; extracting pressure features from the pressure dataset and establishing pressure anomaly nodes based on the extracted pressure features; calling the hand pose and fingertip movement modeling after time series alignment through the pressure anomaly nodes, and performing assembly pressing collaborative identification based on pose accuracy based on the calling results to establish assembly anomalies.

[0040] Specifically, hand pose and fingertip movement models from different data sources are synchronized with the pressure dataset in chronological order so that they can be analyzed on the same timeline.

[0041] Pressure features are extracted from the pressure dataset, such as the maximum and minimum pressure values, rate of change, and stability, to obtain the pressure feature extraction results. Pressure anomaly nodes are then established based on these results. These nodes represent abnormal pressure states detected at specific times or under specific conditions, i.e., nodes that characterize the applied pressure. These nodes are used for joint recognition of fingertip movements and hand postures based on the time of pressure application.

[0042] By using time-series aligned hand pose and fingertip movement modeling based on pressure anomaly nodes, and then performing pose-accuracy-based collaborative recognition of assembly presses to establish assembly anomalies, the system needs to...

[0043] By analyzing the time points of abnormal pressure, the movement state of the hand is analyzed, and by combining hand posture and fingertip movement modeling, abnormalities in the assembly process are identified. That is, the accuracy of hand posture and the pressure are combined to analyze the assembly quality. For example, if it is detected that the wrist angle deviates from the standard value by more than 5 degrees when a person presses the component, and the pressure is unstable, it can be determined that this operation may cause the component to fail to be properly fixed, and thus it is marked as an assembly abnormality.

[0044] Furthermore, such as Figure 2As shown, the collaborative analysis module 40 includes: a batch analysis submodule 41, used to call the batch collaborative analysis layer to perform assembly deviation fitting analysis within the same batch based on the detection dataset, and establish batch anomaly identification results; a deep analysis submodule 42, used to call the deep collaborative analysis layer to extract target detection data of car door lock numbers from the detection dataset, and perform joint assembly anomaly analysis based on the target detection data, the multidimensional linkage dataset, and the worker assembly dataset, and establish deep anomaly identification results; and an output module 43, used to receive the batch anomaly identification results and the deep anomaly identification results, and then construct the anomaly identification results.

[0045] Specifically, the batch analysis submodule 41's call to the batch collaborative analysis layer means that this submodule relies on a dedicated computational layer for batch data analysis, capable of processing large amounts of data simultaneously and identifying patterns. The inspection dataset refers to all inspection data collected during the assembly process, such as pressure data, hand trajectory data, and assembly time. The focus of batch analysis is on the assembly deviation fitting analysis within the same batch. Assembly deviation refers to the error between the actual assembly and the standard assembly, and fitting analysis refers to using mathematical methods to calculate the trend of these errors. Through batch analysis, these abnormal data can be identified, and batch anomaly identification results can be established. This effectively filters out assembly behaviors that deviate significantly from the standard, improving overall assembly quality.

[0046] The deep analysis submodule 42's call to the deep collaborative analysis layer signifies the adoption of higher-precision calculation methods to ensure the accuracy of anomaly detection. The target detection data for the car door lock number refers to all detection information related to a specific door lock. The multi-dimensional linked dataset combines information from multiple data sources, such as worker assembly data and inspection data. The worker assembly dataset specifically records worker operation data during the assembly process, including gestures and pressure. The purpose of joint assembly anomaly analysis is to synthesize this data to identify deeper-level assembly anomalies. For example, if a car door lock is found to have insufficient locking force during inspection, and deep analysis reveals that the assembler's hand trajectory deviated from the standard range, it can be inferred that the insufficient locking force may be due to incorrect hand posture during assembly. Deep analysis enables more accurate anomaly identification results, making detection more intelligent.

[0047] The output module 43 receives batch anomaly identification results and deep anomaly identification results, and ultimately constructs the anomaly identification results. Batch anomaly identification results are based on the overall anomaly situation derived from assembly data of the same batch, while deep anomaly identification results are a detailed analysis of a specific assembly process. Output module 43 combines these two types of data to form the final anomaly identification report. The final anomaly identification results can help the production line optimize the assembly process, reduce the defect rate, and improve overall assembly quality.

[0048] Furthermore, the output module 43 includes an additional analysis submodule, which is used to perform zero-point symmetrical detection dataset data interception with the detection node of the car door lock number as the zero point of time, establish an additional verification window, perform assembly verification based on the additional verification window, generate additional anomaly identification results, and construct the anomaly identification results by combining the additional anomaly identification results, the batch anomaly identification results and the deep anomaly identification results.

[0049] Specifically, the output module 43 integrates anomaly identification results from different levels to ensure the completeness and accuracy of anomaly detection. The additional analysis submodule further optimizes anomaly identification by using time-symmetric data extraction and assembly verification to make anomaly detection more precise. By establishing an additional verification window, the detected data is further analyzed, and combined with the results from other analysis modules, a complete anomaly identification report is ultimately generated.

[0050] In the additional analysis submodule, each car door lock is assigned a unique number during the inspection process, and the time point corresponding to this number is used as the reference point. For example, if the assembly completion time of a car door lock is set as zero, all relevant data will be calculated relative to this time point. Next, a zero-point symmetrical data extraction process is performed, extracting inspection data within a certain time range before and after the zero point. This analysis covers data such as the door lock's pressure, torque, and position to ensure the integrity of the assembly process, avoid single-point data errors, and improve inspection accuracy.

[0051] After data capture, an additional verification window is established to provide more accurate assembly verification and generate additional anomaly identification results. For example, if the standard assembly time for a door lock is 45 to 55 seconds, but a worker completes the assembly in 60 seconds, this situation is marked as an anomaly. However, if the additional verification window reveals that the worker performed additional corrective operations in the last 5 seconds, bringing the assembly quality to the acceptable standard, then false alarms can be avoided.

[0052] Finally, by combining the additional anomaly identification results, batch anomaly identification results, and deep anomaly identification results, an anomaly identification result is constructed, indicating that the final anomaly identification result is determined by multiple data sources. Batch anomaly identification results are based on statistical analysis of the entire batch of data, enabling the detection of deviations in the overall assembly trend; deep anomaly identification results are based on the assembly condition of a specific door lock, using joint analysis of multi-dimensional data to identify more detailed anomalies; additional anomaly identification results are based on assembly process verification within a specific time range, ensuring that the detection results are not misjudged due to data fluctuations at a single point in time. By integrating these three types of anomaly identification results, a more comprehensive anomaly identification report is generated.

[0053] Furthermore, the multi-dimensional collaborative detection module 20 includes: an attention configuration module, used to reconstruct the attention of data indicators based on the anomaly analysis results, establish the indicator attention reconstruction results, use the indicator attention reconstruction results to perform principal component identification, reshape the indicators with the principal component identification results, and reallocate the indicator weights based on the indicator attention reconstruction results to complete the linkage collection attention configuration.

[0054] Specifically, the multi-dimensional collaborative detection module 20 includes a focus configuration module, used to dynamically adjust data indicators, enabling it to optimize data analysis strategies for different situations to improve detection accuracy. The main task of the focus configuration module is to reconstruct the focus of data indicators based on anomaly analysis results, thereby establishing the indicator focus reconstruction results. The focus of data indicators represents the degree of importance attached to different detection items. For example, in some cases, stress data may be more important than hand posture, while in others, hand trajectory may require more attention.

[0055] Principal component identification (PCI) is performed using the results of indicator attention reconstruction. This means that after attention adjustment, the key variables in the data are further analyzed to identify the main factors affecting the detection results, thus obtaining the PCI results. For example, when inspecting the quality of car door locks, multiple data dimensions may be involved, such as pressure, time, and hand movements. PCI can analyze this data to identify the key factors that most significantly affect assembly quality.

[0056] Reshaping indicators based on principal component analysis results means redefining the detection indicators according to the analysis results of principal component analysis, resulting in a reconstructed indicator focus. Reshaping indicators implies adjusting the configuration of detection parameters to better suit actual detection needs. For example, if it is found that the key factor affecting assembly anomalies is hand trajectory, rather than the traditionally considered pressing pressure, the detection accuracy of hand trajectory can be increased, and the reliance on pressure data can be reduced, making the detection more accurate, reducing the possibility of misjudgment, and improving the level of intelligence.

[0057] Based on the reconstructed indicator attention results, indicator weights are redistributed, and the attention weights of each indicator are readjusted to complete the linked data collection and attention configuration, ultimately determining a new detection strategy. Indicator weights refer to the importance of different detection parameters in the analysis. The completion of the linked data collection and attention configuration enables more accurate capture of key anomalies, improving the reliability and intelligence of detection.

[0058] Furthermore, this application also includes: an anomaly discrimination module, used to discriminate between group anomalies and individual anomalies based on the anomaly identification results, and to establish anomaly discrimination results; and an early warning module, used to perform anomaly level matching based on the anomaly discrimination results, and to issue an early warning based on the anomaly level matching results.

[0059] Specifically, the anomaly detection module is used to identify detected anomalies and issue warnings when necessary to ensure the stability and consistency of assembly quality. This is achieved by distinguishing the scope and severity of the anomalies, while the warning module is responsible for taking appropriate countermeasures based on the severity of the anomalies, distinguishing between group anomalies and individual anomalies based on the anomaly identification results. For example, in the assembly process of a batch of door locks, if multiple workers exhibit similar assembly deviations at a specific step, such as insufficient pressing pressure or torque, it may be judged as a group anomaly. This anomaly may be caused by factors such as equipment calibration problems, errors in process instructions, or insufficient training, thus requiring overall adjustments.

[0060] On the other hand, if only a few workers exhibit abnormalities during assembly, while the operations of other workers are generally normal, this situation will be classified as an individual abnormality. Individual abnormalities are usually related to workers' operating habits, fatigue levels, or individual abilities. For example, if out of 100 workers, only 2 workers apply less than 50 Newtons of pressure when installing a door lock, while the other workers meet the standard, this will be identified as an individual abnormality, and it may be recommended that these 2 workers receive additional training or be reminded to pay attention to operational details. After the abnormality discrimination module completes the classification of group abnormalities and individual abnormalities, it establishes the abnormality discrimination results.

[0061] The early warning module matches anomaly levels based on the anomaly identification results. Anomaly level matching refers to classifying anomalies into different levels according to their severity, such as low, medium, and high-level anomalies. For example, if an anomaly is just a minor error occasionally made by a worker and does not affect the final assembly quality, it may be marked as a low-level anomaly; if an anomaly leads to a decrease in product quality but can be corrected through rework, it may be marked as a medium-level anomaly; if an anomaly seriously affects product safety or causes the entire batch of products to be defective, it will be marked as a high-level anomaly. For example, if a low door lock closing force is detected on a production line, and this anomaly affects most products, it may be determined as a high-level anomaly, requiring immediate action.

[0062] Once the anomaly level matching is complete, a pre-alarm is issued based on the matching results. Pre-alarm issuance refers to the system taking corresponding warnings and handling measures according to the severity of the anomaly. For example, if a minor anomaly is detected in a worker's assembly action, an indicator light or screen message may be used to remind the worker to adjust their operation; if a batch of products is detected to have potential quality risks, an alarm may be automatically sent to quality inspectors, requesting additional testing; if a large-scale anomaly is found in a piece of equipment or process parameter, an automatic production stoppage may be triggered, and engineers may be notified for emergency inspection. For example, if a large number of anomalies are detected in the motor installation of a production line, three alarms may be issued consecutively within 5 minutes, and the production line will be automatically suspended once the number of anomalies exceeds a set threshold.

[0063] Furthermore, this application also includes: in the general testing module 30, general testing of the car door lock includes component dimensional accuracy testing, fit clearance testing, micro motor parameter testing, sensor function testing, wiring connection quality testing, and locking / unlocking action testing.

[0064] Specifically, the general testing module 30 performs general testing on the car door lock to ensure it meets design standards and usage requirements. This includes testing several key parameters of the car door lock to identify potential assembly defects or performance anomalies.

[0065] First, the general inspection module 30 performs dimensional accuracy inspection of the components. These components refer to the various structural parts that make up the car door lock, such as the lock housing, lock hook, spring, and transmission mechanism. Dimensional accuracy inspection means measuring whether the specific dimensions of these components are within the specified tolerance range. If the dimensions exceed the tolerance, the door lock may fail to close properly or become loose, thus affecting safety.

[0066] Secondly, the general testing module 30 also needs to perform clearance testing. Clearance refers to the gap between two components, which affects the smoothness of the door lock's movement and its sealing performance. For example, if the gap between the bolt and the latch is too large, it may cause abnormal noises during operation; if the gap is too small, it may make it difficult to close the door. Therefore, it is necessary to measure the clearance of key components and ensure that it is within a reasonable range.

[0067] Furthermore, the general detection module 30 needs to perform micro-motor parameter detection. The micro-motor is responsible for driving the bolt to complete the locking or unlocking action. Micro-motor parameters include operating current, speed, torque, etc. For example, if the motor current of a door lock increases abnormally, it may mean that the motor has an overload problem, which may be caused by excessive assembly friction or internal motor failure. Therefore, detecting these parameters helps to determine whether the door lock can work properly and ensure its durability.

[0068] Next, sensor functionality testing is required. Car door locks are equipped with various sensors, such as position sensors and Hall effect sensors, to detect the lock's open / closed state and send signals to the vehicle's control system. Sensor functionality testing refers to checking whether the sensors can accurately perceive the lock's state. For example, a smart door lock needs to send a signal to the central control system immediately after unlocking to trigger the door unlocking function. If a sensor malfunctions, the vehicle may not be able to detect the unlocked state, thus affecting the normal opening of the door. Therefore, different operating scenarios are simulated, and the sensor response is verified.

[0069] In addition, the general testing module 30 also involves wiring connection quality testing. The automotive door lock's circuitry connects the micromotor, sensors, and control unit, and wiring connection quality testing primarily checks the reliability of electrical connections. For example, if the circuit soldering is not secure, the micromotor may experience a break in the circuit under vibration, leading to door lock malfunction. Therefore, methods such as resistance measurement and continuity testing are used to ensure stable wiring connections.

[0070] Finally, the general testing module 30 needs to perform locking and unlocking action testing. Locking and unlocking action refers to the smooth locking or unlocking process of the door lock after receiving a control signal or mechanical operation. For example, in automatic locking mode, the door lock should automatically lock when the vehicle reaches a certain speed, and immediately unlock when the owner presses the unlock button. Locking and unlocking action testing checks the door lock's response speed, operational smoothness, and the presence of any jamming or other problems through automated testing or manual operation.

[0071] Furthermore, this application also includes: a recording module, used to record the anomaly identification results, and to establish a sensitive window based on the anomaly identification results, and to use the sensitive window for sensitive monitoring and management after anomaly warning.

[0072] Specifically, the recording module stores the anomaly identification results and further optimizes the system's monitoring capabilities to improve the accuracy of anomaly detection.

[0073] First, the recording module needs to record the anomaly identification results. This involves storing the detected anomalies in a database or log system to ensure subsequent analysis and tracking. Anomaly identification results refer to assembly deviations, functional abnormalities, or other quality issues discovered during the inspection process. Specific anomaly categories and inspection data are attached to facilitate subsequent analysis.

[0074] Then, a sensitivity window is established based on the anomaly identification results. The sensitivity window is a time or data range setting used to monitor time periods or product batches where anomalies may occur. The sensitivity window can be time-based, such as detecting an abnormal increase during a specific time period on a certain day; or it can be based on data characteristics, such as a worker's assembly actions deviating significantly from standard actions, leading to an increased anomaly rate.

[0075] Next, we will utilize the sensitive window for sensitive monitoring management following anomaly warnings. Anomaly warnings refer to the timely issuance of alerts to relevant personnel or equipment upon detecting an anomaly, enabling corrective actions to be taken. For example, if abnormal pressure data occurs during the assembly of a door lock product, an alert will be immediately sent to production line managers, reminding them to check the assembly process or equipment status. Sensitive monitoring management refers to further monitoring and analysis of the production situation within the sensitive window after the warning. For example, after identifying an anomaly, we will adjust the testing strategy, such as increasing the sampling rate of that batch of door locks or adding additional testing items, to ensure that the problem does not escalate.

[0076] Furthermore, this application also includes: an adaptive correction module, used to receive user detection feedback, identify and verify the anomaly identification result based on the detection feedback, establish correction feedback, and optimize the collaborative analysis module 40 based on the correction feedback.

[0077] Specifically, the adaptive correction module verifies the anomaly identification results based on user feedback and optimizes the collaborative analysis module 40 accordingly to improve detection accuracy and adaptability.

[0078] First, the adaptive correction module needs to receive user feedback. Users can provide data through manual input, sensor feedback, or automatic recording. Users refer to inspectors, assembly workers, or production managers who may discover errors in the anomaly identification results when using the system. For example, if the system detects an locking anomaly in a car door lock, but the inspector finds the lock functioning normally after inspection, the inspector can submit feedback indicating that the identification result may be a false alarm. Detection feedback refers to the information provided by users regarding the accuracy of the identification, including whether the anomaly is real, whether the anomaly category is correct, and whether the detection standard is reasonable.

[0079] Next, the anomaly identification results are verified based on the detection feedback. Verification refers to reviewing the anomaly identification results to confirm the accuracy of the judgment. For example, if the assembly defect rate of a batch of door locks is higher than normal, the batch is marked as abnormal. However, if user feedback indicates that some detection results may be erroneous, these results need to be re-analyzed to check whether false alarms are caused by sensor errors, detection standard deviations, or algorithm problems.

[0080] Next, a correction feedback mechanism is established. This refers to adjusting erroneous identification parts based on user-provided information and system self-test results. For example, if a high false alarm rate is found, the anomaly detection threshold can be adjusted to reduce misidentification of normal locks; if certain anomaly types are found to be unrecognized, new detection rules can be added to improve recognition capabilities. For instance, if inspectors report that the system easily misidentifies a certain type of lock during gap detection, historical data can be used to retrain the model to better adapt it to the characteristics of that lock type.

[0081] Finally, the collaborative analysis module 40 is optimized based on the correction feedback. Optimization of the collaborative analysis module 40 refers to further adjusting the anomaly detection algorithm using the corrected data to improve the overall system's intelligence level. For example, if the initial false alarm rate of the system is 5% in 10,000 detections, and the false alarm rate is reduced to 2% after optimization through correction feedback, it indicates that the anomaly detection capability of the collaborative analysis module 40 has been improved. Optimization methods may include adjusting data weights, optimizing the deep learning model, and resetting the parameter range for anomaly detection. For example, if a certain error pattern is found in the angular velocity data of the inertial measurement unit, new data samples can be used for model training to more accurately identify abnormal assembly operations.

[0082] In summary, the dynamic quality monitoring system for intelligent assembly and inspection of automotive door locks provided in this application has the following technical effects: by achieving the technical goals of multi-dimensional real-time monitoring, intelligent analysis, and accurate anomaly identification of the automotive door lock assembly process, it achieves the technical effects of improving assembly quality, increasing production efficiency, reducing rework costs, and enhancing the level of intelligence in the manufacturing process.

[0083] The above description of the disclosed embodiments enables those skilled in the art to make or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0084] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application also intends to include such modifications and variations.

Claims

1. A dynamic quality monitoring system for intelligent assembly inspection of automotive door locks, characterized in that, The system includes: A smart haptic glove is used to record a worker assembly dataset after signal interaction with a car door lock. The worker assembly dataset is identified by the car door lock number, and the smart haptic glove integrates a flexible pressure sensor and an inertial measurement unit. The multi-dimensional collaborative detection module is used to receive the worker assembly dataset, perform anomaly analysis, configure linkage acquisition focus based on the anomaly analysis results, and perform assembly linkage acquisition based on the linkage acquisition focus after detecting the corresponding car door lock number to establish a multi-dimensional linkage dataset. A general testing module is used to perform general testing on car door locks and establish a testing dataset. The collaborative analysis module is used to call the collaborative analysis network to identify assembly quality anomalies based on the detection dataset, the multidimensional linkage dataset, and the worker assembly dataset, and to establish anomaly identification results. The collaborative analysis network includes a batch collaborative analysis layer and a deep collaborative analysis layer. The smart tactile gloves are used for: After signal interaction with the car door lock, monitoring is initiated; Activate the flexible pressure sensors on the fingertips and palms of the smart haptic glove to build a pressure dataset; Activate the inertial measurement unit, establish an inertial dataset, reconstruct the spatial trajectory based on the motion feature extraction channel, calculate the hand pose, and model the fingertip motion. Standard action posture matching is performed based on hand pose and fingertip movement modeling to identify posture anomalies; The hand pose and fingertip movement modeling is used to perform collaborative assembly anomaly identification on the pressure dataset, and assembly anomalies are established. Construct a worker assembly dataset based on the posture anomalies and assembly anomalies; In the aforementioned smart haptic glove, the hand pose and fingertip movement modeling are used to collaboratively identify assembly anomalies in the pressure dataset, establishing assembly anomalies, including: The hand pose and fingertip movement models were aligned with the pressure dataset over time. Stress features are extracted from the stress dataset, and stress anomaly nodes are established based on the stress feature extraction results; The hand pose and fingertip movement modeling is called after time-series alignment through the pressure anomaly node. Based on the calling result, assembly pressing collaborative recognition is performed based on pose accuracy to establish assembly anomalies. The collaborative analysis module includes: The batch analysis submodule is used to call the batch collaborative analysis layer, perform assembly deviation fitting analysis within the same batch based on the detection dataset, and establish batch anomaly identification results. The deep analysis submodule is used to call the deep collaborative analysis layer, extract the target detection data of the car door lock number in the detection dataset, perform joint assembly anomaly analysis based on the target detection data, the multi-dimensional linkage dataset, and the worker assembly dataset, and establish a deep anomaly identification result. The output module is used to receive the batch anomaly identification results and the deep anomaly identification results and then construct the anomaly identification results; The output module includes: The additional analysis submodule is used to perform zero-point symmetrical detection dataset data interception with the detection node of the car door lock number as the zero point of time, establish an additional verification window, perform assembly verification based on the additional verification window, generate additional anomaly identification results, and construct the anomaly identification results by combining the additional anomaly identification results, the batch anomaly identification results, and the deep anomaly identification results. The multidimensional collaborative detection module includes: The attention configuration module is used to reconstruct the attention of data indicators based on the anomaly analysis results, establish the indicator attention reconstruction results, use the indicator attention reconstruction results to perform principal component identification, reshape the indicators with the principal component identification results, and reallocate the indicator weights based on the indicator attention reconstruction results to complete the linkage collection attention configuration.

2. The dynamic quality monitoring system for intelligent assembly inspection of automotive door locks as described in claim 1, characterized in that, The system includes: The anomaly detection module is used to detect group anomalies and individual anomalies based on the anomaly identification results, and to establish anomaly detection results; The early warning module is used to perform anomaly level matching based on the anomaly discrimination result and to issue an early warning based on the anomaly level matching result.

3. The dynamic quality monitoring system for intelligent assembly inspection of automotive door locks as described in claim 1, characterized in that, The general testing module performs general testing on car door locks, including component size accuracy testing, fit clearance testing, micro motor parameter testing, sensor function testing, wiring connection quality testing, and locking / unlocking action testing.

4. The dynamic quality monitoring system for intelligent assembly inspection of automotive door locks as described in claim 1, characterized in that, The system includes: The recording module is used to record the anomaly identification results and establish a sensitive window based on the anomaly identification results, and use the sensitive window to perform sensitive monitoring and management after anomaly warning.

5. The dynamic quality monitoring system for intelligent assembly inspection of automotive door locks as described in claim 1, characterized in that, The system includes: An adaptive correction module is used to receive user detection feedback, verify the anomaly identification results based on the detection feedback, establish correction feedback, and optimize the collaborative analysis module based on the correction feedback.

Images