Automobile production line ai visual inspection mistake proofing method and system

Abstract

The application relates to an automobile production line AI visual detection mistake-proofing method and device, which comprises the following steps: collecting a front wheel rim image of a to-be-detected vehicle body, performing standardization treatment on the front wheel rim image to obtain a standardized image; extracting corresponding key features from the standardized image; comparing the key features with a wheel rim visual template library within a preset comparison time interval to obtain a similar wheel rim visual template and a corresponding feature matching degree; and comparing the feature matching degree with a preset matching threshold value; when the feature matching degree is not less than the preset matching threshold value, taking a vehicle model matched by the similar wheel rim visual template as a vehicle model of the to-be-detected vehicle body, wherein the vehicle model comprises a pickup, a light truck, a three-type chassis and a non-vehicle. Through the application, the technical problems that a traditional automobile assembly production line cannot identify a three-type chassis without a cab by relying on grating and switch combination detection, leading to vehicle model misjudgment, automatic transfer failure, safety hazards of manual intervention and high modification cost are solved.

Description

Technical Field

[0001] This invention relates to the field of vehicle model recognition and transfer technology in vehicle manufacturing, and in particular to an AI visual inspection error prevention method and system for automobile production lines. Background Technology

[0002] In the context of large-scale, multi-model intelligent manufacturing in automobile final assembly, the final assembly conveyor line integrates the entire process of pre-assembly, assembly, and testing. Among these processes, the automated transfer of assembled car bodies and complete vehicles is a core and critical step to ensure continuous operation of the final assembly line and enable the co-production of multiple models. Currently, the industry's vehicle model identification solutions for vehicle transfer generally adopt a mechanical combination detection logic of limit switches, light truck gratings, and pickup truck gratings, which is a traditional and common detection solution.

[0003] Current traditional detection and judgment principles heavily rely on the external structure of the vehicle body obscuring the grating signal to determine the vehicle type: a prerequisite is that a limit switch detects the vehicle's presence, and the vehicle type is distinguished by the signal triggering combination of light truck and pickup truck gratings; that is, simultaneous sensing of both gratings indicates a light truck, sensing only the pickup truck grating indicates a pickup truck, and no signal from either grating indicates "no vehicle." This judgment logic is only designed for vehicles with a complete cab structure, and has inherent and structural flaws, making it unsuitable for identifying and detecting vehicles without a cab.

[0004] In actual mixed-line production, the three types of chassis have no cab and no upper shielding structure. When passing through the line, they cannot trigger the light truck and pickup truck light grating sensing signals. Traditional detection systems will misjudge the three types of chassis passing through the line normally as "no vehicle" mode. This identification defect will directly cause the conveyor belt to not stop at the stop point and directly perform the receiving and continuous conveying actions until the anti-fall switch at the end of the conveyor belt is triggered, causing the production line to stop and alarm. The production line must be manually reset and restarted, which will seriously affect the production line cycle. More importantly, once the anti-fall switch at the end of the conveyor belt fails mechanically, it will directly cause the entire three types of chassis to fall, resulting in major production accidents such as equipment damage, vehicle scrapping, and production line shutdown.

[0005] To prevent vehicle falls and ensure normal production line operation, the industry currently relies solely on long-term on-site maintenance personnel at vehicle transfer points to continuously intervene and assist in line control for three types of chassis crossing conditions. This solution has significant, unavoidable industry-wide drawbacks: First, the work area is a suspended pit structure for vehicle handling, requiring workers to enter the pit and walk on I-beams, posing a high risk of missteps and falls – a high-risk manual operation scenario. Second, 24 / 7 on-site maintenance significantly increases labor costs, and the limited response time and low error tolerance of manual intervention greatly reduce the automation rate and production stability of the vehicle transfer process, failing to meet the demands of large-scale, unmanned, and high-precision intelligent automotive manufacturing.

[0006] In addition, traditional grating and limit switch combination detection solutions have rigid detection logic and a single recognition dimension, only able to identify two common vehicle models, resulting in extremely poor compatibility and inability to adapt to flexible mixed-model production lines. Furthermore, long-term operation of the equipment is prone to malfunctions such as grating dust accumulation, sensor misalignment, and mechanical jamming of limit switches. Lacking redundant error-proofing mechanisms, the detection error rate is high and stability is poor. In summary, traditional mechanical sensing detection structures have inherent flaws and cannot solve core industry problems such as recognition failures for three types of chassis without a driver's cab, high-risk manual intervention, and poor adaptability to multiple vehicle models. The industry urgently needs a new vehicle model recognition and detection solution to break through the technical barriers of traditional mechanical sensing, eliminate production safety hazards, and improve the automation and intelligence level of production lines. Summary of the Invention

[0007] Therefore, the purpose of this invention is to provide an AI visual inspection error prevention method and system for automobile production lines to address the shortcomings of existing technologies.

[0008] To achieve the above objectives, the present invention provides an AI visual inspection error prevention method for automotive production lines, the method comprising: The front wheel rim image of the vehicle body to be inspected is acquired by an AI smart camera, and the front wheel rim image is standardized to obtain a standardized image. Extract the corresponding key features from the standardized image, compare the key features with the wheel rim visual template library within the preset comparison time interval, obtain similar wheel rim visual templates and corresponding feature matching degree, and compare the feature matching degree with the preset matching threshold. When the feature matching degree is not less than the preset matching threshold, the vehicle model matched by the similar rim visual template is taken as the vehicle model to be detected. The vehicle model includes pickup trucks, light trucks, three types of chassis and no vehicle.

[0009] The beneficial effects of this invention are as follows: It acquires images of the front wheel rims of the vehicle to be inspected using an AI intelligent camera, standardizes these images to obtain standardized images, and compares them with a rim visual template library within a preset comparison time interval to obtain similar rim templates and corresponding feature matching degrees. The feature matching degree is then compared with a preset matching threshold. When the feature matching degree is not less than the preset matching threshold, the vehicle model matched by the similar rim template is taken as the vehicle model of the vehicle to be inspected. This solves the technical problem of traditional automobile assembly lines relying on gratings and switch combinations to detect chassis without driver's cabs, leading to model misjudgment, automatic transfer failure, and safety hazards and high modification costs associated with manual intervention. Simultaneously, it achieves accurate automatic identification of all vehicle models and safe and efficient automatic vehicle transfer, improving the automation and intelligence level of the production line while reducing modification costs and eliminating safety hazards.

[0010] Furthermore, the preset comparison time interval is 546ms-569ms.

[0011] Furthermore, the method also includes: When the feature matching degree is less than the preset matching threshold, it is determined that there is no matching template and an identification error is triggered.

[0012] Furthermore, the method also includes: When the duration of the comparison between the standardized image and the wheel rim visual template library is not within the preset comparison time interval, a comparison error is triggered.

[0013] Furthermore, the method also includes: Monitor the operating status of the AI ​​smart camera, and trigger a hardware fault alarm when the AI ​​smart camera has a communication failure; When the AI ​​smart camera has a standby timeout fault, it is determined to be a shutdown abnormality and a shutdown abnormality alarm is triggered.

[0014] To achieve the above objectives, the present invention also provides an AI visual inspection error prevention device for an automotive production line, used to implement the AI ​​visual inspection error prevention method for an automotive production line as described above. The device includes a photo trigger switch, an AI smart camera, an AI controller, a Rockwell PLC controller, an electronic Kanban board, a network switch, and a position detection switch. The photo trigger switch is located at the production line inspection point and is used to accurately capture the incoming vehicle signal to trigger a fixed-point photo operation. The AI ​​smart camera is an industrial PoE rugged camera, and the camera of the AI ​​smart camera is oriented to the front wheel rim of the vehicle body to be inspected; The AI ​​controller is responsible for storing wheel rim-specific templates, performing timed image comparison calculations, identifying vehicle status, monitoring camera hardware failures, and outputting accurate judgment signals. The Rockwell PLC controller is used to receive AI recognition signals and position switch signals, and to perform speed adjustment, fixed-point shutdown, data reset, and fault alarm linkage of the transfer equipment. The electronic dashboard is used to visualize the equipment operating status, vehicle model recognition results, fault type, and detection data, and the electronic dashboard supports manual data clearing. The network switch is used to ensure millisecond-level real-time communication between devices. The arrival detection switch is located at the transfer stop point and is used to assist in detecting the parking status of the vehicle body.

[0015] Furthermore, the Rockwell PLC controller, the AI ​​controller, the AI ​​smart camera, and the electronic signboard are configured in an Ethernet networking architecture.

[0016] Furthermore, the Rockwell PLC controller, the AI ​​controller, the AI ​​smart camera, and the electronic signboard are each configured with their own independent fixed IP address.

[0017] Furthermore, the Rockwell PLC controller contains four exclusive wheel rim visual templates for pickup trucks, light trucks, three types of chassis, and no-vehicle status. Attached Figure Description

[0018] Figure 1 This is a flowchart of an AI visual inspection error prevention method for an automobile production line according to an embodiment of the present invention.

[0019] The following detailed description, in conjunction with the accompanying drawings, will further illustrate the present invention. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of this application clearer, the application is described and illustrated below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments provided in this application without inventive effort are within the scope of protection of this application.

[0021] Obviously, the accompanying drawings described below are merely some examples or embodiments of this application. Those skilled in the art can apply this application to other similar scenarios based on these drawings without any inventive effort. Furthermore, it is understood that although the efforts made in this development process may be complex and lengthy, for those skilled in the art related to the content disclosed in this application, any changes to design, manufacturing, or production based on the technical content disclosed in this application are merely conventional technical means and should not be construed as insufficient disclosure of the content of this application.

[0022] In this application, the reference to "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment that is mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0023] Unless otherwise defined, the technical or scientific terms used in this application shall have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms “a,” “an,” “an,” “the,” and similar words used in this application do not indicate quantity limitation and may indicate singular or plural. The terms “comprising,” “including,” “having,” and any variations thereof used in this application are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units not listed, or may include other steps or units inherent to these processes, methods, products, or devices. The terms “connected,” “linked,” “coupled,” and similar words used in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. “Multiple” used in this application refers to two or more. “And / or” describes the relationship between related objects, indicating that three relationships may exist; for example, “A and / or B” can represent: A alone, A and B simultaneously, and B alone. The character " / " generally indicates that the preceding and following objects are in an "or" relationship. The terms "first," "second," and "third" used in this application are merely to distinguish similar objects and do not represent a specific ordering of the objects.

[0024] Example 1 Please see Figure 1 The flowchart below shows the AI ​​visual inspection error prevention method for automobile production lines according to the first embodiment of the present invention. The method includes the following steps: Step S101: Acquire the front wheel rim image of the vehicle body to be detected using an AI smart camera, and perform standardization processing on the front wheel rim image to obtain a standardized image; The system includes photo-trigger switches and AI smart cameras deployed around the inspection points on the production line, synchronized with the production line's transport rhythm. When the vehicle to be inspected is transported to the inspection point, the switch trigger signal is synchronously transmitted to the AI ​​controller and PLC controller used to control various equipment on the production line. The AI ​​controller and PLC controller then activate the AI ​​smart camera to capture a fixed-point image of the front wheel rim. The front wheel rim image is then standardized to complete standardized image acquisition, ensuring uniform recognition features and fixed acquisition positions, thus avoiding recognition deviations caused by random shooting.

[0025] It should be noted that AI visual inspection technology is well-suited to the industrial environment of automobile production workshops. The equipment is flexible in deployment, easy to modify, and highly compatible with existing equipment on the production line. At the same time, this technology can be extended to scenarios such as appearance inspection of key components in automobile production, detection of misassemblies and omissions, product counting, and detection of component segment differences, providing technical support for the intelligent upgrading of automobile production and possessing good value for promotion and application.

[0026] Step S102: Extract the corresponding key features from the standardized image, compare the key features with the wheel rim visual template library within the preset comparison time interval to obtain similar wheel rim templates and corresponding feature matching degree, and compare the feature matching degree with the preset matching threshold. The preset comparison time interval is 546ms-569ms. This preset comparison time interval is set based on the actual production cycle of the production line. By setting the preset comparison time interval of this special image, which is different from the unique process parameters of general visual inspection, it not only ensures the accuracy of image feature extraction, noise reduction and template matching, but also matches the production line conveying cycle, eliminating the problems of production line blockage and cycle lag.

[0027] It should be noted that the PLC controller has four types of vehicle-specific wheel rim visual templates: pickup truck, light truck, three types of chassis, and vehicle-free state. The wheel rim visual template library is constructed using these four types of exclusive wheel rim visual templates. There are corresponding preset matching thresholds for matching the front wheel rim images of different vehicle models. When a similar wheel rim template is determined, the preset matching threshold for the vehicle model corresponding to the similar wheel rim template can be obtained. Then, the feature matching degree is compared with the preset matching threshold.

[0028] Furthermore, the preset matching threshold of 85% is obtained based on on-site calibration. This value setting satisfies the recognition requirements and does not place high demands on the imaging performance of the AI ​​smart camera. Understandably, when the preset matching threshold is below 85%, there is a significant risk of recognition errors. When the preset matching threshold is above 85%, the on-site scene is dynamic, and the captured images still have certain differences, making it difficult to meet the high imaging requirements. Step S103: When the feature matching degree is not less than the preset matching threshold, the vehicle model matched by the similar rim template is taken as the vehicle model to be detected. The vehicle model includes pickup trucks, light trucks, three types of chassis and no vehicle.

[0029] The PLC controller reads the AI ​​vehicle model determination signal in real time (the vehicle model matched with the similar rim template is used as the vehicle model to be tested, and the determination signal corresponding to this execution step), and outputs the corresponding transfer control command in combination with the production line transfer process rules. For the three types of chassis models that traditional equipment cannot identify, it can accurately identify and control the plate chain transfer equipment to stop at designated points, ensuring that the vehicle to be tested is stably stopped at the transfer stop point. From the control logic level, it prevents the vehicle from moving straight, triggering the anti-fall alarm, and preventing vehicle falls. It should be noted that the PLC controller supports timed automatic data reset (daily timed clearing of statistical data and determination cache).

[0030] It should be noted that when the trigger switch malfunctions (the PLC controller detects a valid vehicle arrival signal but the photo trigger signal is invalid), a switch failure alarm will be triggered.

[0031] Through the above steps, images of the front wheel rims of the vehicle to be inspected are acquired using an AI intelligent camera. These images are then standardized to obtain standardized images. Within a preset comparison timeframe, the standardized images are compared with a visual template library of wheel rims to obtain similar wheel rim templates and their corresponding feature matching degrees. The feature matching degrees are then compared with a preset matching threshold. When the feature matching degree is not less than the preset matching threshold, the vehicle model matched by the similar wheel rim template is taken as the vehicle model of the vehicle to be inspected. This solves the technical problem that traditional automobile assembly lines rely on gratings and switch combinations for detection, which cannot identify three types of chassis without a driver's cab, leading to misjudgment of vehicle models, failure of automatic transfer, and safety hazards and high modification costs associated with manual intervention. At the same time, it also achieves accurate automatic identification of all vehicle models and safe and efficient automatic transfer of the whole vehicle, improving the automation and intelligence level of the production line while reducing modification costs and eliminating safety hazards.

[0032] Furthermore, before comparing the key features with the wheel rim visual template library, the method includes: Obtain four types of exclusive wheel rim visual templates: pickup truck, light truck, three types of chassis, and vehicle-free status. Construct a wheel rim visual template library based on these four exclusive wheel rim visual templates.

[0033] Among them, the front wheel rim is a common structure inherent to all models of pickup trucks, light trucks, and Class III chassis. Class III chassis without cabs have no body or cab obstruction, but the front wheel rim structure is fully preserved. The front wheel rim is used as the only reference identification feature, which is different from conventional body appearance identification schemes. It abandons easily missing identification features such as cabs and body exteriors, effectively solving the pain point that traditional grating detection cannot identify bare chassis models.

[0034] Furthermore, the method also includes: When the feature matching degree is less than the preset matching threshold, it is determined that there is no matching template and an identification error is triggered.

[0035] Furthermore, the method also includes: When the duration of the comparison between the standardized image and the wheel rim visual template library is not within the preset comparison time interval, a comparison error is triggered.

[0036] Furthermore, the method also includes: Monitor the operating status of the AI ​​smart camera, and trigger a hardware fault alarm when the AI ​​smart camera has a communication failure; Among them, communication failures include abnormal camera operating parameters or communication interruptions, and abnormal device status codes.

[0037] When the AI ​​smart camera has a standby timeout fault, it is determined to be a shutdown abnormality and a shutdown abnormality alarm is triggered.

[0038] Among them, the standby timeout fault is when the standby time of the AI ​​smart camera is less than the preset duration of not receiving a trigger signal.

[0039] Furthermore, the method also includes: View the equipment's operating status, vehicle model recognition results, fault types, and test data in real time through electronic dashboards.

[0040] The electronic Kanban system enables full visualization of the testing process, vehicle model data, and alarm information, facilitating on-site management and data traceability on the production line. It also enhances the skill level of the production team and accumulates valuable practical experience for automobile manufacturers to carry out more intelligent and automated transformation projects in the future. The electronic Kanban system supports manual data reset to eliminate recognition interference caused by the accumulation of historical data and ensure the independence of each test judgment. Example 2 The AI ​​visual inspection error prevention method for automobile production lines according to the second embodiment of the present invention is applied to the vehicle transfer process at the Jiangxi Isuzu final assembly plant, solving the problem of automatic identification and transfer of three types of chassis without driver's cabs. The specific steps are as follows: System network configuration Keyence IV4 series smart cameras were selected as AI smart cameras, paired with corresponding AI controllers. Rockwell Automation Allen-Bradley PLCs were selected as the device-side PLC controllers, with the PLC controllers configured with independent fixed IPs of 192.168.1.41, the AI ​​smart cameras with independent fixed IPs of 192.168.1.50, and the electronic display boards with independent fixed IPs of 192.168.1.51. The devices were networked via Ethernet through a network switch to achieve real-time communication and interconnection between the AI ​​controllers, PLC controllers, and electronic display boards. The electronic display boards are used to display camera status, vehicle model information, alarm information, etc.

[0041] Vehicle Model Visual Template Preset In the IV Smart Navigator software of the AI ​​controller, standard visual templates were established for three types of chassis models produced by Jiangxi Isuzu: pickup trucks, regular light trucks, and cab-less trucks, as well as the "no vehicle" state. The visual features of the front wheel rims of each model were learned, and the threshold for image comparison was adjusted to ensure that the rim features of the three types of chassis without cabs could be accurately identified. After the template learning was completed, the software displayed "7 images learned" and the comparison threshold was adjusted to 70%, which met the recognition accuracy requirements.

[0042] Oncoming vehicle triggers camera Photo trigger switches are installed at the inspection points during vehicle transfer at the final assembly plant, while the original light truck arrival switches and pickup truck arrival switches are retained as auxiliary verification. When a vehicle is transported to the inspection point along the conveyor line, the photo trigger switch detects the approaching vehicle and sends a switch trigger signal. After receiving the signal, the AI ​​controller triggers the AI ​​smart camera to take a picture of the front wheel rim of the vehicle to be inspected. After taking the picture, the AI ​​smart camera sends a "picture completed" signal to the PLC controller and the electronic display board. The electronic display board updates the "picture completed" status in real time.

[0043] Visual data comparison and judgment The AI ​​controller extracts features from the collected front wheel rim images and compares them with preset templates. The comparison processing time is 549ms, which is within the efficient processing range of 546ms to 569ms. Based on the comparison results, if a matching template for a cabless light truck chassis is found, the corresponding vehicle type determination signal is output. If a matching template for a pickup truck or a regular light truck is found, the corresponding vehicle type determination signal is output. If no matching template is found, it is determined to be a vehicle type recognition error and an NG signal is sent.

[0044] PLC instruction execution The PLC controller reads the vehicle type determination signal from the AI ​​controller through RSLogix 5000 software. According to the production line transfer rules, it sends action commands to the vehicle transfer equipment to control the three types of light truck chassis without cabs to accurately enter the transfer stop point and realize automatic transfer. At the same time, it verifies the vehicle parking position through light truck positioning switches and pickup truck positioning switches to ensure accuracy.

[0045] Error prevention mechanism implementation Data cycle reset: Set a daily 22:00 timed data reset program in the PLC controller to automatically clear the judgment results and statistical data in the system. At the same time, the electronic dashboard supports manual reset by clicking "Clear Judgment Results" and "Clear Statistical Data". Multi-level fault alarms: If a vehicle model recognition error occurs, the electronic display will show "Vehicle model recognition error"; if there is an internal fault in the AI ​​camera, it will show "Internal camera fault"; if the camera stops working, it will show "Camera stopped"; if the camera switch is not triggered, it will show "Camera switch not triggered". All alarm information is highlighted in real time to facilitate troubleshooting by staff.

[0046] After being applied at the Jiangxi Isuzu assembly plant, the system successfully achieved precise automatic identification and transportation of three types of chassis without driver's cabs, thanks to its advantages such as compatibility with existing production lines, convenient modification, and low investment costs. This eliminated the safety hazards of manual intervention and freed up one maintenance worker. The project cost 27,000 yuan, saving 33,000 yuan compared to the 60,000 yuan quoted for outsourcing the modification. At the same time, it saves 100,000 yuan in labor costs annually, resulting in an annual comprehensive benefit of 133,000 yuan, demonstrating significant economic benefits.

[0047] Example 3 The structural block diagram of the AI ​​visual inspection error prevention device for automobile production line in the third embodiment of the present invention includes a photo trigger switch, an AI smart camera, an AI controller, a Rockwell PLC controller, an electronic Kanban board, a network switch, and an arrival detection switch. The photo trigger switch is located at the production line inspection point and is used to accurately capture the incoming vehicle signal to trigger a fixed-point photo operation. The AI ​​smart camera is an industrial PoE rugged camera. The AI ​​smart camera is deployed above the inspection point on the production line, and the camera of the AI ​​smart camera is oriented towards the front wheel rim of the vehicle body to be inspected. The AI ​​controller is responsible for storing wheel rim-specific templates, performing timed image comparison calculations, identifying vehicle status, monitoring camera hardware failures, and outputting accurate judgment signals. The Rockwell PLC controller is used to receive AI recognition signals and position switch signals, and to perform speed adjustment, fixed-point shutdown, data reset, and fault alarm linkage of the transfer equipment. Furthermore, the Rockwell PLC controller contains four exclusive wheel rim visual templates for pickup trucks, light trucks, three types of chassis, and no-vehicle status.

[0048] The electronic dashboard is used to visualize the equipment operating status, vehicle model recognition results, fault type, and detection data, and the electronic dashboard supports manual data clearing. The network switch is used to ensure millisecond-level real-time communication between devices. The arrival detection switch is located at the transfer stop point and is used to assist in detecting the parking status of the vehicle body.

[0049] In practical implementation, when the vehicle body to be inspected is transported to the inspection point, the switch trigger signal corresponding to the photo trigger switch is synchronously transmitted to the AI ​​controller and PLC controller. The AI ​​controller and PLC controller then activate the AI ​​smart camera to capture images of the front wheel rims at a fixed point. The front wheel rim images are then standardized to complete standardized image acquisition. The standardized images are then compared with the visual templates of each rim in the Rockwell PLC controller to obtain similar rim templates and corresponding feature matching degrees. The feature matching degree is compared with a preset matching threshold. When the feature matching degree is not less than the preset matching threshold, the vehicle model matched by the similar rim template is taken as the vehicle model of the vehicle body to be inspected. This solves the technical problem that traditional automobile assembly lines rely on gratings and switch combinations to detect chassis without cabs, which cannot identify the three types of chassis, leading to misjudgment of vehicle models, failure of automatic transfer, and safety hazards and high modification costs due to manual intervention. At the same time, it also realizes accurate automatic identification of all vehicle models and safe and efficient automatic transfer of whole vehicles, improves the automation and intelligence level of the production line, reduces modification costs, and eliminates safety hazards.

[0050] Furthermore, the Rockwell PLC controller, the AI ​​controller, the AI ​​smart camera, and the electronic signboard are configured in an Ethernet networking architecture.

[0051] The electronic dashboard is connected to an Ethernet network system, providing a real-time visual display of equipment operating status, vehicle model recognition results, fault alarm types, and OK / NG statistics. This enables traceability and monitoring of the entire inspection process, creating a closed-loop, visualized inspection system. It's worth noting that the integrated networking of AI visual inspection and the PLC control system automates the entire process of vehicle arrival triggering, photographing, comparing, judging, and transferring, eliminating the need for manual supervision and intervention. This removes the safety hazards of personnel entering the vehicle pit and improves production line safety.

[0052] Furthermore, the Rockwell PLC controller, the AI ​​controller, the AI ​​smart camera, and the electronic signboard are each configured with their own independent fixed IP address.

[0053] This involves using network switches and configuring independent fixed IP addresses to enable millisecond-level data interaction between devices.

[0054] In the description of this specification, references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0055] Without causing conflict, those skilled in the art can freely combine and use the above-mentioned additional technical features.

[0056] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for error prevention in AI visual inspection of an automobile production line, characterized in that, The method includes: The front wheel rim image of the vehicle body to be inspected is acquired by an AI smart camera, and the front wheel rim image is standardized to obtain a standardized image. Extract the corresponding key features from the standardized image, compare the key features with the wheel rim visual template library within the preset comparison time interval, obtain similar wheel rim visual templates and corresponding feature matching degree, and compare the feature matching degree with the preset matching threshold. When the feature matching degree is not less than the preset matching threshold, the vehicle model matched by the similar rim visual template is taken as the vehicle model to be detected. The vehicle model includes pickup trucks, light trucks, three types of chassis and no vehicle.

2. The AI ​​visual inspection error prevention method for automobile production lines according to claim 1, characterized in that, The preset comparison time interval is 546ms-569ms.

3. The AI ​​visual inspection error prevention method for automobile production lines according to claim 1, characterized in that, The method further includes: When the feature matching degree is less than the preset matching threshold, it is determined that there is no matching template and an identification error is triggered.

4. The AI ​​visual inspection error prevention method for automobile production lines according to claim 1, characterized in that, The method further includes: When the duration of the comparison between the standardized image and the wheel rim visual template library is not within the preset comparison time interval, a comparison error is triggered.

5. The AI ​​visual inspection error prevention method for automobile production lines according to claim 1, characterized in that, The method further includes: Monitor the operating status of the AI ​​smart camera, and trigger a hardware fault alarm when the AI ​​smart camera has a communication failure; When the AI ​​smart camera has a standby timeout fault, it is determined to be a shutdown abnormality and a shutdown abnormality alarm is triggered.

6. An AI visual inspection error prevention device for an automobile production line, used to implement the AI ​​visual inspection error prevention method for an automobile production line as described in any one of claims 1-5, characterized in that, The device includes a photo trigger switch, an AI smart camera, an AI controller, a Rockwell PLC controller, an electronic signboard, a network switch, and an arrival detection switch. The photo trigger switch is located at the production line inspection point and is used to accurately capture the incoming vehicle signal to trigger a fixed-point photo operation. The AI ​​smart camera is an industrial PoE rugged camera, and the camera of the AI ​​smart camera is oriented to the front wheel rim of the vehicle body to be inspected; The AI ​​controller is responsible for storing wheel rim-specific templates, performing timed image comparison calculations, identifying vehicle status, monitoring camera hardware failures, and outputting accurate judgment signals. The Rockwell PLC controller is used to receive AI recognition signals and position switch signals, and to perform speed adjustment, fixed-point shutdown, data reset, and fault alarm linkage of the transfer equipment. The electronic dashboard is used to visualize the equipment operating status, vehicle model recognition results, fault type, and detection data, and the electronic dashboard supports manual data clearing. The network switch is used to ensure millisecond-level real-time communication between devices. The arrival detection switch is located at the transfer stop point and is used to assist in detecting the parking status of the vehicle body.

7. The AI ​​visual inspection error prevention device for automobile production lines according to claim 6, characterized in that, The Rockwell PLC controller, the AI ​​controller, the AI ​​smart camera, and the electronic signboard form an Ethernet network architecture.

8. The AI ​​visual inspection error prevention device for automobile production lines according to claim 6, characterized in that, The Rockwell PLC controller, the AI ​​controller, the AI ​​smart camera, and the electronic signboard are each configured with their own independent fixed IP address.

9. The AI ​​visual inspection error prevention device for automobile production lines according to claim 6, characterized in that, The Rockwell PLC controller contains four exclusive wheel rim visual templates: pickup truck, light truck, three types of chassis, and no-vehicle status.

Images