Safety helmet capable of intelligently detecting assembly quality, detection system and detection method

Abstract

The invention belongs to the technical field of quality detection, and discloses a safety helmet capable of intelligently detecting assembly quality, a detection system and a detection method.The safety helmet comprises a safety helmet body, an image recognition module, an action recognition module, a timing module, a voice recognition module, an alarm module, a storage and communication module and a battery module, the storage and communication module is respectively connected with the image recognition module, the action recognition module, the timing module and the voice recognition module, and the battery module is respectively connected with the image recognition module, the action recognition module, the timing module, the voice recognition module, the alarm module and the storage and communication module. Multi-dimensional real-time detection and full-chain monitoring are achieved through wearable multi-module integration, software algorithm linkage and background data management, and data integrity and traceability are guaranteed through dual-mode alarm and data unified storage; and based on a self-optimization algorithm of historical data, continuous optimization of the system without manual intervention is realized.

Description

Technical Field

[0001] This invention belongs to the field of quality inspection technology, specifically relating to a safety helmet, inspection system, and inspection method for intelligently inspecting assembly quality. Background Technology

[0002] As new energy vehicles enter a stage of large-scale popularization, their production and assembly processes are undergoing significant changes: on the one hand, the variety of parts has increased by more than 30% compared to traditional fuel vehicles, and the assembly precision requirements for key workstations have been greatly improved; on the other hand, consumers' expectations for vehicle reliability and safety continue to rise. Therefore, automobile manufacturers need to find a balance between "efficient mass production" and "precise quality control".

[0003] Currently, the industry mainly relies on two types of methods for precise quality control of the new energy vehicle production process: one is manual visual inspection, where front-line employees and quality inspectors confirm whether the assembly result is qualified by visual observation and handheld tool measurement after the assembly worker completes a single process; the other is to rely on fixed equipment inspection, such as industrial cameras next to the production line to take static photos of the assembled parts for comparison, or independent motion sensors to monitor specific operation steps in a single dimension.

[0004] However, both of these testing methods have significant limitations and cannot fully meet the high-precision and high-efficiency requirements of new energy vehicle assembly. The main limitations are as follows:

[0005] 1. Insufficient real-time capability: Fixed equipment can only perform "post-assembly inspection" after the assembly process is completely completed. It cannot identify erroneous or redundant actions in real time during worker operation, which leads to the accumulation of errors and the need for rework, increasing time costs.

[0006] 2. Limited inspection scope: Manual inspection is susceptible to fatigue and experience, making it impossible to simultaneously judge the standardization of actions, compliance with time, and correctness of results; fixed equipment only focuses on the assembly result, ignoring the actions during the operation, resulting in the failure to discover the root cause of the problem;

[0007] 3. Inefficient data management: Problem records from manual inspections require manual filling out of forms, while inspection data from fixed equipment is only stored on local terminals. The two types of data are not interconnected, making it impossible to form a complete quality problem traceability chain and to optimize inspection standards based on historical data.

[0008] 4. Poor flexibility: Fixed equipment needs to be installed in a specific location on the production line and cannot be moved and operated by workers, while independent portable testing tools require additional operation, which affects assembly efficiency. Summary of the Invention

[0009] The purpose of this invention is to provide a smart safety helmet, detection system, and detection method for detecting assembly quality. It achieves multi-dimensional real-time detection through wearable multi-module integration, software algorithm linkage, and background data management, realizing full-chain monitoring of the assembly process and results. It features dual-mode alarms and unified data storage to ensure data integrity and traceability. Based on historical data, a self-optimizing algorithm analyzes abnormal events in the database and dynamically updates standard data, enabling continuous optimization of the system without human intervention.

[0010] The specific details of the plan are as follows:

[0011] A smart safety helmet for detecting assembly quality includes a helmet body and an image recognition module, a motion recognition module, a timing module, a voice recognition module, an alarm module, a storage and communication module, and a battery module disposed on the helmet body. The storage and communication module is connected to the image recognition module, motion recognition module, timing module, and voice recognition module, respectively. The battery module is connected to the image recognition module, motion recognition module, timing module, voice recognition module, alarm module, and storage and communication module, respectively.

[0012] The image recognition module is used to acquire image data of assembled parts in real time;

[0013] The motion recognition module is used to collect data on the operational actions of assembly workers;

[0014] The timing module is used to collect time data for each operation step;

[0015] The voice recognition module is used to pre-store descriptions of standard quality anomaly events and recognize the voice commands of quality inspectors;

[0016] The alarm module is used to receive alarm signals and issue audible and visual alarms;

[0017] The storage and communication module is used to locally store standard data and quality anomaly event data, and control data transmission. The standard data includes standard assembly result image data, standard action data, and standard time data.

[0018] The battery module is used to power all modules.

[0019] The helmet body of this invention is an impact-resistant outer shell with a breathable inner liner and an adjustable headband. The helmet body secures all internal modules and provides installation space for the battery module. The helmet body houses all functional modules, ensuring both safety and comfort when worn.

[0020] The voice recognition module includes a voice recognition chip and microphone, pre-stores standard quality event descriptions (such as "missing bolt"), recognizes the voice commands of the quality inspector, and outputs the voice recognition signal to the processor module. The image recognition module includes a high-definition wide-angle camera, which acquires images of the assembly area in real time and outputs image data to the processor module. The motion recognition module preferably uses a microelectromechanical system (MEMS) gyroscope + accelerometer to acquire worker operation motion data in real time and outputs motion data to the processor module; an infrared motion capture sensor can also be used. The timing module includes a high-precision I²C real-time clock chip, which acquires the time of each operation step and outputs time data to the processor module. The processor module includes a microprocessor that receives data from each module, compares it with standard data, identifies anomalies and triggers alarms, controls data transmission, and bidirectionally connects to all modules, serving as the core control unit of the system. The alarm module includes a red LED light and a buzzer, which receive abnormal signals from the processor module and issue an audible and visual alarm; a vibration motor is also optional for high-noise workshops. The storage and communication module preferably uses a memory and a wireless LAN communication block (LoRa module can also be used) for local storage of standard data and abnormal event data, and uploads data to the background database system via wireless LAN; it receives storage / transmission commands from the processor and communicates with the background database system. The battery module of this invention preferably uses an 18650 lithium battery (5000mAh, fast charging for 2 hours) to power all modules, supporting 8 hours of continuous operation and providing a stable voltage (3.7V) for all modules. A solar charging panel can also be used for the battery module. The background database system is a MySQL database deployed on a workshop server. It stores standard data (standard actions, time, images) and quality abnormal event data, supports data querying and analysis, receives wireless data from the safety helmet of this invention, and provides a data source for the deep learning module. The deep learning module is a software module based on the CNN-LSTM fusion algorithm, capable of reading database data and outputting updated standard data to the processor module for periodic analysis of abnormal data in the database and updating the standard data (such as optimizing action judgment thresholds).

[0021] Furthermore, it also includes a matching indicator pen, which includes an indicator pen battery module and a positioning module for manually locating the location of quality anomaly events.

[0022] A system for intelligently detecting assembly quality includes the aforementioned intelligent safety helmet, as well as a processor module and a backend database system.

[0023] The processor module is connected to the storage and communication module. The processor module is used to receive real-time data collected by the image recognition module, motion recognition module and timing module transmitted by the storage and communication module, compare it with the pre-stored standard data, determine the deviation between the real-time data and the standard data, determine whether it is a quality abnormality event, and control the alarm module to issue an alarm signal when an abnormality is determined, and control the storage and communication module to transmit the data to the background database system.

[0024] The backend database system is connected to the storage and communication module. The backend database system includes a standard database and a quality anomaly event database, which are used to store standard data and quality anomaly event data, and support data query and analysis.

[0025] Furthermore, it also includes a deep learning module, which is a software module based on the CNN-LSTM fusion algorithm. It is used to read data from the background database system, periodically analyze abnormal data in the database, update standard data, and output the updated standard data to the processor module. The deep learning module is connected to both the background database system and the processor module.

[0026] The CNN algorithm optimizes image recognition accuracy, while the LSTM algorithm analyzes the patterns in action sequences and dynamically updates standard parameters, thus improving detection accuracy over time.

[0027] A method for intelligently detecting assembly quality, comprising using the aforementioned intelligent assembly quality detection system for detection, the method including:

[0028] S1. Pre-configured standardized data is stored in the backend database system and synchronized to the storage and communication module;

[0029] S2. The safety helmet is monitored in real time. If a quality abnormality event is detected, an alarm is triggered. The processor module transmits the quality abnormality event data to the storage and communication module. The storage and communication module saves the data and transmits it to the background database system via wireless local area network. The background database system saves the quality abnormality event data. The process leader reviews the assembled part. If it is indeed a quality abnormality event, the process jumps to step S4. If no quality abnormality event occurs, the assembly continues until the assembly is completed, and then the process continues to step S3.

[0030] S3. The safety helmet undergoes a secondary inspection of the assembly result. If a quality anomaly is detected, the safety helmet automatically alarms, and the anomaly data is saved to the storage and communication module and the backend database system. If the anomaly is not detected by the safety helmet but by a human, the anomaly is manually located, and the voice data describing the anomaly is automatically matched with the corresponding assembly result image and transmitted to the processor module, triggering an alarm. The anomaly data is saved to the storage and communication module and the backend database system. The assembled component is then manually reviewed. If it is confirmed to be a quality anomaly, step S4 continues. If no quality anomaly is found, the inspection ends, and step S5 continues.

[0031] S4. Abnormal handling: Return the assembled part to the production line for rework. After rework, repeat steps S2 and S3 until no quality abnormality events occur. The inspection ends, and step S5 is continued.

[0032] S5. System self-optimization: The deep learning module starts at a preset time, reads the quality anomaly event data of the day, analyzes and updates the standard data, and synchronizes it to the storage and communication module.

[0033] Furthermore, the method for pre-configuring standardized data in step S1 specifically includes:

[0034] S101. Configure standard assembly result image data: Based on vehicle model, assembly process, assembly station and corresponding standard assembly result image, configure standard assembly result image data.

[0035] S102. Configure standard motion data: Based on the vehicle model, assembly process, assembly station and corresponding assembly motion operation specifications, configure standard motion data;

[0036] S103. Configure standard time data: Configure standard time data based on the standard assembly time of vehicle model, assembly process, assembly station and corresponding assembly parts.

[0037] S104. Configure quality anomaly event data: Configure quality anomaly event data based on vehicle model, assembly process, assembly station and corresponding voice description and image features of quality anomaly events.

[0038] Furthermore, step S2 specifically includes:

[0039] S201. After the assembly worker puts on the safety helmet that enables the intelligent detection of assembly quality, he selects the corresponding assembly process, and the system automatically loads the standard data of the assembly station.

[0040] S202. The assembly worker begins the assembly operation. The motion recognition module, timing module, and image recognition module simultaneously collect data and transmit the collected data to the processor module through the storage and communication module.

[0041] S203. The processor module receives real-time data collected by the image recognition module, motion recognition module, and timing module, compares it with standard data, and determines whether there is a quality abnormality. If a quality abnormality event is determined, the alarm module is immediately triggered. The processor module transmits the quality abnormality event data to the storage and communication module for storage. The storage and communication module transmits the quality abnormality event data to the background database system according to the processor module's instructions and stores it in the quality abnormality event database. The process leader reviews the assembled part. If it is indeed a quality abnormality event, the process jumps to step S4; if no quality abnormality event occurs, the assembly continues until the assembly is completed, and then the process continues to step S3.

[0042] Furthermore, step S202 specifically includes:

[0043] S2021. The action recognition module collects operation action data in real time and transmits the operation action data to the processor module through the storage and communication module at a preset transmission speed.

[0044] S2022. The timing module starts when the assembly worker begins the operation, records the time of each operation action, updates the time data of the operation steps according to the preset update speed, and transmits it to the processor module through the storage and communication module.

[0045] S2023. The image recognition module captures images of the assembly parts at a preset shooting frequency and transmits them to the processor module through the storage and communication module.

[0046] Furthermore, step S3 specifically includes:

[0047] S301. The quality inspector selects the quality inspection mode of the safety helmet with intelligent detection of assembly quality, and selects the corresponding assembly process. The assembly parts completed in the previous step are re-inspected. The assembly result image is collected by the image recognition module and transmitted to the processor module through the storage and communication module. The processor module performs standard data comparison.

[0048] S302. If the safety helmet automatically identifies a quality anomaly, the quality anomaly event data is transmitted to the processor module and an alarm is triggered. The processor module transmits the quality anomaly event data to the storage and communication module, which saves it and transmits it to the background database system. The background database system saves it and continues to execute step S303. If the quality anomaly event is discovered manually by a quality inspector rather than automatically identified by the safety helmet, the quality inspector manually locates the problem location of the assembly component using a matching indicator pen and takes an image of the problem location using an image recognition module. At the same time, the quality inspector verbally describes the quality anomaly event and analyzes its cause. The voice recognition module recognizes the quality inspector's voice information and automatically matches it with the image data of the problem location. The quality anomaly event data is transmitted to the processor module, triggering an alarm. The processor module controls the storage and communication module to save the quality anomaly event data and transmits it to the background database system for storage. The process continues to execute step S303.

[0049] S303. The quality team leader reviews the assembled parts for quality anomalies. If an anomaly is identified, it is sent to the process team leader for review. If the process team leader confirms that it is a quality anomaly, proceed to step S4. If there is no quality anomaly, the inspection ends and step S5 continues.

[0050] Further, step S5 specifically includes: the deep learning module automatically starts at a preset time, reads and analyzes the quality anomaly event data of the day in the background database, supplements new standard data to the standard database, supplements new quality anomaly event data to the quality anomaly event database, and synchronizes to the storage and communication module to realize system optimization and update standard data.

[0051] The deep learning module of the present invention is preferably a software module based on the CNN-LSTM fusion algorithm, but an image-action fusion algorithm based on Transformer can also be used.

[0052] Compared with the prior art, the present invention has the following advantages:

[0053] 1. This invention solves the problem of the lack of "real-time accompanying detection" in the automobile assembly process, realizes synchronous monitoring of operation actions, operation time and assembly results, improves the assembly qualification rate and reduces rework costs;

[0054] 2. Overcome the limitations of single-dimensional inspection, achieve multi-dimensional collaborative judgment of "action standardization + time compliance + result correctness", assist manual item-by-item inspection, and improve the efficiency of quality inspection;

[0055] 3. To solve the problems of scattered and difficult-to-trace test data, and to achieve unified storage and traceability of data automatically identified by equipment and data manually reported by quality inspectors;

[0056] 4. Solve the problem that the detection standard cannot be self-optimized, and achieve the improvement of long-term detection accuracy through a deep learning module that dynamically updates detection parameters based on historical data. Brief Description of the Drawings

[0057] Figure 1 It is a schematic structural diagram of a safety helmet for intelligently detecting assembly quality according to the present invention.

[0058] Figure 2 It is a schematic structural diagram of a supporting indicating pen according to the present invention.

[0059] Figure 3 It is a schematic diagram of data interaction of a system for intelligently detecting assembly quality according to the present invention.

[0060] Figure 4 It is a working flowchart of a method for intelligently detecting assembly quality according to the present invention.

[0061] Figure 5 It is a schematic diagram of the data structure of the background database according to the present invention.

[0062] In the figure:

[0063] 1. Safety helmet body; 2. Image recognition module; 3. Action recognition module; 4. Timing module; 5. Voice recognition module; 6. Alarm module; 7. Storage and communication module; 8. Battery module; 9. Indicator pen battery module; 10. Positioning module. Detailed Embodiments

[0064] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. For the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present invention.

[0065] The terms used in the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms of "a", "the" and "said" used in the present invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. "Plural" generally includes at least two.

[0066] It should be noted that the terms "front", "rear", "inner", "outer", "left", "right", etc., used in this invention refer to the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0067] The following examples illustrate this. Figure 1 , Figure 2 , Figure 3 , Figure 4 and Figure 5 The present invention will be described in detail below.

[0068] Example 1:

[0069] A smart safety helmet for detecting assembly quality, see Figure 1 As shown, the helmet includes a helmet body 1 and an image recognition module 2, a motion recognition module 3, a timing module 4, a voice recognition module 5, an alarm module 6, a storage and communication module 7, and a battery module 8, all mounted on the helmet body 1. The storage and communication module 7 is connected to the image recognition module 2, the motion recognition module 3, the timing module 4, and the voice recognition module 5, respectively. The battery module 8 is connected to the image recognition module 2, the motion recognition module 3, the timing module 4, the voice recognition module 5, the alarm module 6, and the storage and communication module 7, respectively.

[0070] Image recognition module 2 is used to acquire image data of assembled parts in real time;

[0071] The motion recognition module 3 is used to collect data on the operational actions of assembly workers;

[0072] Timing module 4 is used to collect time data for each operation step;

[0073] The voice recognition module 5 is used to pre-store descriptions of standard quality anomaly events and recognize the voice commands of quality inspectors;

[0074] Alarm module 6 is used to receive alarm signals and issue audible and visual alarms;

[0075] The storage and communication module 7 is used to locally store standard data and quality anomaly event data, and control data transmission. The standard data includes standard assembly result image data, standard action data, and standard time data.

[0076] Battery module 8 is used to power all modules.

[0077] Also includes a matching indicator pen, see Figure 2 As shown, the matching indicator pen includes an indicator pen battery module 9 and a positioning module 10 for manually locating the location where a quality abnormality event occurs.

[0078] Example 2:

[0079] This invention also provides a system for intelligently detecting assembly quality, see Figure 3 As shown, the safety helmet, which includes the aforementioned intelligent inspection and assembly quality testing, also includes a processor module and a backend database system.

[0080] The processor module is connected to the storage and communication module 7. The processor module is used to receive real-time data collected by the image recognition module 2, motion recognition module 3 and timing module 4 transmitted by the storage and communication module 7, compare it with the pre-stored standard data, determine the deviation between the real-time data and the standard data, determine whether it is a quality abnormality event, and control the alarm module 6 to issue an alarm signal when an abnormality is determined, and control the storage and communication module 7 to transmit the quality abnormality event data to the background database system.

[0081] The backend database system and storage and communication module 7 are connected, see Figure 5 As shown, the backend database system includes a standard database and a quality anomaly event database, which are used to store standard data and quality anomaly event data, and support data query and analysis.

[0082] It also includes a deep learning module, which is a software module based on the CNN-LSTM fusion algorithm. It is used to read data from the background database system, periodically analyze abnormal data in the database, update standard data, and output the updated standard data to the processor module. The deep learning module is connected to the background database system and the processor module respectively.

[0083] Example 3:

[0084] This invention also provides a method for intelligently detecting assembly quality, see Figure 4 As shown, the system for intelligent assembly quality inspection is applied for inspection, and the method includes:

[0085] S1. Pre-configured standardized data is stored in the background database system and synchronized to the storage and communication module 7;

[0086] S2. The assembly worker wears the intelligent safety helmet that detects assembly quality, turns on the equipment, selects the corresponding assembly station, and begins assembling the component. The safety helmet performs real-time detection. If a quality abnormality event occurs, an alarm is triggered. The processor module transmits the quality abnormality event data to the storage and communication module 7. The storage and communication module 7 saves the data and transmits it to the background database system via wireless local area network. The background database system saves the quality abnormality event data. The process leader reviews the assembled component. If it is indeed a quality abnormality event, the process jumps to step S4. If no quality abnormality event occurs, the assembly continues until the assembly is completed, and then the process continues to step S3.

[0087] S3. The quality inspector, wearing the aforementioned intelligent safety helmet for detecting assembly quality, performs a secondary inspection of the assembly results. If the safety helmet detects a quality anomaly, it automatically alarms, and the anomaly data is saved to the storage and communication module 7 and the backend database system. If the anomaly is not detected by the safety helmet but manually by the quality inspector, the quality inspector manually locates the anomaly, and the voice data describing the anomaly and the corresponding assembly result image are automatically matched and transmitted to the processor module, triggering an alarm. The anomaly data is saved to the storage and communication module 7 and the backend database system. The quality team leader reviews the assembled component. If it is indeed a quality anomaly, proceed to step S4. If no quality anomaly is found, the inspection ends, and proceed to step S5.

[0088] S4. Abnormal handling: Return the assembled part to the production line for rework. After rework, repeat steps S2 and S3 until no quality abnormality events occur. The inspection ends, and step S5 is continued.

[0089] S5. System self-optimization: The deep learning module starts at a preset time, reads the quality anomaly event data of the day, analyzes and updates the standard data, and synchronizes it to the storage and communication module 7.

[0090] The method for pre-configuring standardized data in step S1 specifically includes:

[0091] S101. Configure standard assembly result image data: Based on vehicle model, assembly process, assembly station and corresponding standard assembly result image, configure standard assembly result image data.

[0092] S102. Configure standard motion data: Based on the vehicle model, assembly process, assembly station and corresponding assembly motion operation specifications, configure standard motion data;

[0093] S103. Configure standard time data: Configure standard time data based on the standard assembly time of vehicle model, assembly process, assembly station and corresponding assembly parts.

[0094] S104. Configure quality anomaly event data: Configure quality anomaly event data based on vehicle model, assembly process, assembly station and corresponding voice description and image features of quality anomaly events.

[0095] Specifically, step S2 includes:

[0096] S201. After the assembly worker puts on the safety helmet that enables the intelligent detection of assembly quality, he selects the corresponding assembly process, and the system automatically loads the standard data of the assembly station.

[0097] S202. The assembly worker begins the assembly operation. The motion recognition module 3, timing module 4, and image recognition module 2 simultaneously collect data and transmit the collected data to the processor module through the storage and communication module 7.

[0098] S203. The processor module receives real-time data collected by the image recognition module 2, motion recognition module 3, and timing module 4, compares it with standard data, and determines whether there is a quality abnormality. If a quality abnormality event is determined, the alarm module 6 is immediately triggered. The processor module transmits the quality abnormality event data to the storage and communication module 7 for storage. The storage and communication module 7 transmits the quality abnormality event data to the background database system according to the processor module's instructions and stores it in the quality abnormality event database. The process leader reviews the assembled part. If it is indeed a quality abnormality event, the process jumps to step S4; if no quality abnormality event occurs, the assembly continues until the assembly is completed, and then the process continues to step S3.

[0099] Specifically, step S202 includes:

[0100] S2021, The motion recognition module 3 collects the data of the operation motion in real time, and transmits the data of the operation motion to the processor module through the storage and communication module 7 at a preset transmission speed;

[0101] S2022, the timing module 4 starts when the assembly worker begins the operation, records the time of each operation action, updates the time data of the operation steps according to the preset update speed, and transmits it to the processor module through the storage and communication module 7.

[0102] S2023, Image recognition module 2 captures images of the assembly parts at a preset shooting frequency and transmits them to the processor module through storage and communication module 7.

[0103] Step S3 specifically includes:

[0104] S301, the quality inspector selects the quality inspection mode of the safety helmet with intelligent detection of assembly quality, and selects the corresponding assembly process to re-inspect the assembly parts completed in the previous step. The assembly result image is collected by the image recognition module and transmitted to the processor module through the storage and communication module 7. The processor module performs standard data comparison.

[0105] S302. If the safety helmet automatically identifies a quality anomaly, the quality anomaly event data is transmitted to the processor module and an alarm is triggered. The processor module transmits the quality anomaly event data to the storage and communication module 7, which saves it and transmits it to the background database system. The background database system saves it and continues to execute step S303. If the quality anomaly event is discovered manually by a quality inspector rather than automatically identified by the safety helmet, the quality inspector manually locates the problem location of the assembly component using a matching indicator pen and takes an image of the problem location using the image recognition module 2. At the same time, the quality inspector verbally describes the quality anomaly event and analyzes its cause. The voice recognition module 5 recognizes the quality inspector's voice information and automatically matches it with the image data of the problem location. The quality anomaly event data is transmitted to the processor module, triggering an alarm. The processor module controls the storage and communication module 7 to save the quality anomaly event data and transmits it to the background database system for storage. The process continues to execute step S303.

[0106] S303. The quality team leader reviews the assembled parts for quality anomalies. If an anomaly is identified, it is sent to the process team leader for review. If the process team leader confirms that it is a quality anomaly, proceed to step S4. If there is no quality anomaly, the inspection ends and step S5 continues.

[0107] Step S5 specifically includes: the deep learning module automatically starts at a preset time, reads and analyzes the quality anomaly event data of the day in the background database, adds new standard data to the standard database, adds new quality anomaly event data to the quality anomaly event database, and synchronizes to the storage and communication module 7 to realize system optimization and update standard data.

[0108] Example 4:

[0109] This embodiment uses the brake pedal bolt assembly process in a car assembly workshop as an example. The data involved in this embodiment has been processed.

[0110] 1. Configure standard data:

[0111] a. The motion recognition module inputs standard motion data: acceleration threshold of 4.5-5.5g and angular velocity threshold of 100-120° / s when tightening bolts;

[0112] b. The timing module inputs standard time data: assembly time for a single bolt ≤ 30s, total time for 12 bolts ≤ 6min;

[0113] c. The image recognition module inputs standard image data: 50 high-definition images each of the front and side views of bolts with 2-3 exposed threads, forming a template library.

[0114] 2. Assembly and Inspection Worker ID: A01

[0115] a. Worker A01 wears a safety helmet and selects the "Automotive Assembly Workshop Brake Pedal Bolt Assembly" process; the system loads standard data.

[0116] b. When assembling the 5th bolt, the motion recognition module detected an acceleration of 3.8g, which is lower than 4.5g, and the timing module displayed a timeout of 2 seconds (32 seconds elapsed).

[0117] c. The processor module determines "insufficient torque + timeout", triggers LED flashing and buzzer alarm, and uploads abnormal data to the database event ID: 20250915A0101, process: brake pedal bolt assembly in automobile assembly workshop, abnormal type: insufficient torque, timeout.

[0118] 3. After confirmation by the process team leader, the brake pedal with quality problems in the automobile assembly workshop is returned to the production line for rework. After a second inspection without any abnormalities, it is inspected by the quality inspector.

[0119] 4. Quality inspector's inspection. Quality inspector ID: Q02:

[0120] a. Quality inspector Q02, wearing a safety helmet, inspects the 5th bolt. The image recognition module captures the image of the 5th bolt, showing that the 4 exposed threads do not meet the standard. An alarm is triggered again, and the abnormal data is uploaded to the database. Event ID: 20250915A0102, Process: Brake pedal bolt assembly in the automotive assembly workshop, Abnormal type: 4 exposed threads.

[0121] b. If the 8th bolt is found to be missing from the camera's original blind spot or not detected, the quality inspector can manually locate the quality problem using the matching indicator pen. The safety helmet will automatically take a picture of the problem area according to the quality inspector's instructions and upload it to the system. At the same time, the safety helmet can recognize the quality inspector's voice description of the quality problem, such as "the 8th bolt of the brake pedal cylinder block in the automobile assembly workshop is missing".

[0122] c. After the voice recognition module recognizes the information, the processor module uploads a manual alarm event with event ID: 20250915Q0203, problem description and picture: the 8th bolt is missing + picture, quality inspector: Q02.

[0123] 5. After the quality team leader and the team leader confirm that the brake pedal with quality problems in the automotive assembly workshop has been returned to the production line, the 5th bolt has been tightened again and the 8th bolt has been replaced; if the second inspection shows no abnormalities, the process ends.

[0124] 6. At 18:00 on the same day, the deep learning module analyzed event 20250915Q0203 and found that the position of the 8th bolt was a blind spot of the camera. The horizontal angle of the camera was updated and the image template of "no bolt at this position" was added to the quality anomaly event database.

[0125] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A safety helmet with intelligent assembly quality detection, characterized in that, The system includes a helmet body (1) and an image recognition module (2), an action recognition module (3), a timing module (4), a voice recognition module (5), an alarm module (6), a storage and communication module (7), and a battery module (8) mounted on the helmet body (1). The storage and communication module (7) is connected to the image recognition module (2), the action recognition module (3), the timing module (4), and the voice recognition module (5), respectively. The battery module (8) is connected to the image recognition module (2), the action recognition module (3), the timing module (4), the voice recognition module (5), the alarm module (6), and the storage and communication module (7), respectively. The image recognition module (2) is used to acquire image data of assembled parts in real time; The motion recognition module (3) is used to collect data on the operation actions of assembly workers; The timing module (4) is used to collect time data for each operation step; The voice recognition module (5) is used to pre-store descriptions of standard quality anomaly events and recognize the voice commands of quality inspectors; The alarm module (6) is used to receive alarm signals and issue audible and visual alarms; The storage and communication module (7) is used to store standard data and quality anomaly event data locally and control data transmission. The standard data includes standard assembly result image data, standard action data and standard time data. The battery module (8) is used to power all modules.

2. The safety helmet with intelligent assembly quality detection according to claim 1, characterized in that, It also includes a matching indicator pen, which includes an indicator pen battery module (9) and a positioning module (10) for manually locating the location of a quality anomaly event.

3. A system for intelligently detecting assembly quality, characterized in that, The safety helmet, including the intelligent assembly quality detection as described in claim 1 or 2, also includes a processor module and a backend database system. The processor module is connected to the storage and communication module (7). The processor module is used to receive real-time data collected by the image recognition module (2), motion recognition module (3) and timing module (4) transmitted by the storage and communication module (7), and compare it with the pre-stored standard data to determine the deviation between the real-time data and the standard data, determine whether it is a quality abnormality event, and control the alarm module (6) to issue an alarm signal when an abnormality is determined, and control the storage and communication module (7) to transmit the quality abnormality event data to the background database system. The background database system is connected to the storage and communication module (7). The background database system includes a standard database and a quality anomaly event database, which are used to store standard data and quality anomaly event data, and support data query and analysis.

4. The intelligent assembly quality detection system according to claim 3, characterized in that, It also includes a deep learning module, which is a software module based on the CNN-LSTM fusion algorithm. It is used to read data from the background database system, periodically analyze abnormal data in the database, update standard data, and output the updated standard data to the processor module. The deep learning module is connected to the background database system and the processor module respectively.

5. A method for intelligently detecting assembly quality, characterized in that, The system for intelligent assembly quality inspection as described in claim 3 or 4 is used for inspection, and the method includes: S1. Pre-configure standardized data, store it in the background database system, and synchronize it to the storage and communication module (7). S2. The safety helmet is monitored in real time. If a quality abnormality event is detected, an alarm is triggered. The processor module transmits the quality abnormality event data to the storage and communication module (7). The storage and communication module (7) saves the data and transmits the data to the background database system via wireless local area network. The background database system saves the quality abnormality event data. The assembled part is manually checked. If it is indeed a quality abnormality event, the process jumps to step S4. If no quality abnormality event occurs, the assembly continues until the assembly is completed, and step S3 is executed. S3. The safety helmet performs a secondary inspection on the assembly result. If the safety helmet detects a quality abnormality, the safety helmet will automatically alarm, and the quality abnormality event data will be saved to the storage and communication module (7) and the background database system. If the quality abnormality event is not detected by the safety helmet, but by a person, the voice data describing the quality abnormality event and the corresponding assembly result image will be automatically matched and transmitted to the processor module to trigger an alarm. The quality abnormality event data will be saved to the storage and communication module (7) and the background database system. The assembly component will be manually reviewed. If it is indeed a quality abnormality event, step S4 will continue. If no quality abnormality event is found, the inspection ends, and step S5 will continue. S4. Abnormal handling: Return the assembled part to the production line for rework. After rework, repeat steps S2 and S3 until no quality abnormality events occur. The inspection ends, and step S5 is continued. S5. System self-optimization: The deep learning module starts at a preset time, reads the quality anomaly event data of the day, analyzes and updates the standard data, and synchronizes it to the storage and communication module (7).

6. The method for intelligently detecting assembly quality according to claim 5, characterized in that, The method for pre-configuring standardized data in step S1 specifically includes: S101. Configure standard assembly result image data: Based on vehicle model, assembly process, assembly station and corresponding standard assembly result image, configure standard assembly result image data. S102. Configure standard motion data: Based on the vehicle model, assembly process, assembly station and corresponding assembly motion operation specifications, configure standard motion data; S103. Configure standard time data: Configure standard time data based on the standard assembly time of vehicle model, assembly process, assembly station and corresponding assembly parts. S104. Configure quality anomaly event data: Configure quality anomaly event data based on vehicle model, assembly process, assembly station and corresponding voice description and image features of quality anomaly events.

7. The method for intelligently detecting assembly quality according to claim 5, characterized in that, Step S2 specifically includes: S201. After the assembly worker puts on the safety helmet that enables the intelligent detection of assembly quality, he selects the corresponding assembly process, and the system automatically loads the standard data of the assembly station. S202, the assembly worker begins the assembly operation. The motion recognition module (3), timing module (4) and image recognition module (2) simultaneously perform data acquisition and transmit the acquired data to the processor module through the storage and communication module (7): S203. The processor module receives real-time data collected by the image recognition module (2), motion recognition module (3) and timing module (4), compares it with standard data, and determines whether there is a quality abnormality. If it is determined to be a quality abnormality event, the alarm module (6) is triggered immediately. The processor module transmits the quality abnormality event data to the storage and communication module (7) for storage. The storage and communication module (7) transmits the quality abnormality event data to the background database system according to the processor module instruction and stores it in the quality abnormality event database. The process group leader reviews the assembly part. If it is indeed a quality abnormality event, the process group leader jumps to step S4. If no quality abnormality event occurs, the assembly continues until the assembly is completed, and the process group leader continues to execute step S3.

8. The method for intelligently detecting assembly quality according to claim 5, characterized in that, Step S202 specifically includes: S2021, The action recognition module (3) collects the data of the operation action in real time, and transmits the data of the operation action to the processor module through the storage and communication module (7) at a preset transmission speed; S2022, The timing module (4) starts when the assembly worker begins to operate, records the time of each operation action, updates the time data of the operation steps according to the preset update speed, and transmits it to the processor module through the storage and communication module (7); S2023, The image recognition module (2) takes pictures of the assembly parts at a preset shooting frequency and transmits them to the processor module through the storage and communication module (7).

9. The method for intelligently detecting assembly quality according to claim 5, characterized in that, Step S3 specifically includes: S301. The quality inspector selects the quality inspection mode of the safety helmet with intelligent detection of assembly quality and selects the corresponding assembly process. The assembly parts completed in the previous step are re-inspected. The assembly result image is collected by the image recognition module and transmitted to the processor module through the storage and communication module (7). The processor module performs standard data comparison. S302. If the safety helmet automatically identifies a quality abnormality event, the quality abnormality event data is transmitted to the processor module and an alarm is triggered. The processor module transmits the quality abnormality event data to the storage and communication module (7). The storage and communication module (7) saves it and transmits it to the background database system. The background database system saves it and continues to execute step S303. If the quality abnormality event is discovered manually by the quality inspector rather than automatically identified by the safety helmet, the quality inspector manually locates the problem location of the assembly part using the matching indicator pen and takes an image of the problem location using the image recognition module (2). At the same time, the quality inspector describes the quality abnormality event and analyzes the cause of the event. The voice recognition module (5) recognizes the voice information of the quality inspector and automatically matches it with the image data of the problem location. The quality abnormality event data is transmitted to the processor module and an alarm is triggered. The processor module controls the storage and communication module (7) to save the quality abnormality event data and controls the storage and communication module (7) to transmit the quality abnormality event data to the background database system for storage and continues to execute step S303. S303. The quality team leader reviews the assembled parts for quality anomalies. If an anomaly is identified, it is sent to the process team leader for review. If the process team leader confirms that it is a quality anomaly, proceed to step S4. If there is no quality anomaly, the inspection ends and step S5 continues.

10. The method for intelligently detecting assembly quality according to claim 5, characterized in that, Step S5 specifically includes: the deep learning module automatically starts at a preset time, reads and analyzes the quality anomaly event data of the day in the background database, supplements new standard data to the standard database, supplements new quality anomaly event data to the quality anomaly event database, and synchronizes to the storage and communication module (7) to realize system optimization and update standard data.

Images