Mobile phone mainboard visual detection method and system based on artificial intelligence

Abstract

The invention relates to the technical field of visual detection, and discloses a mobile phone mainboard visual detection method and system based on artificial intelligence, and the method comprises the steps: constructing an adaptive heat and optical environment, enabling a mobile phone mainboard to be in an auxiliary detection state, collecting a multi-dimensional visual angle image through a machine vision array, and carrying out the visual detection of the mobile phone mainboard; the light-weight visual detection model is used for key information identification, doubtful point information is obtained, the detection state and the visual array are adjusted according to the doubtful point information, supplementary detection is carried out, supplementary data are deeply analyzed through the depth visual detection model, and a detection conclusion is generated. The problems that the calculation burden is reduced, the detection efficiency is improved, the efficient production requirement is met, and the precise and complex detection requirement of the mobile phone mainboard is difficult to meet are solved.

Description

Technical Field

[0001] This invention relates to the technical field of visual inspection, and in particular to a visual inspection method and system for mobile phone motherboards based on artificial intelligence. Background Technology

[0002] With the popularization of smartphones and the continuous advancement of technology, the motherboard, as an important component of mobile phones, directly affects the user experience. However, as mobile phone motherboards become more precise and complex, traditional manual inspection methods can no longer meet the requirements of high efficiency and high precision. Furthermore, automatic inspection technology using conventional visible light machine vision technology is also difficult to meet the requirements. Summary of the Invention

[0003] The purpose of this invention is to provide a visual inspection method and system for mobile phone motherboards based on artificial intelligence, which aims to solve the problem that existing technologies cannot meet the inspection requirements of precise and complex mobile phone motherboards.

[0004] The present invention is implemented as follows: Firstly, the present invention provides a visual inspection method for mobile phone motherboards based on artificial intelligence, comprising: A thermal environment is constructed for the mobile phone motherboard to be tested, and an optical environment is applied to the mobile phone motherboard in the thermal environment so that the mobile phone motherboard is in an auxiliary testing state. The machine vision array is used to acquire images of the mobile phone motherboard in the auxiliary inspection state, and the key information of the acquired images is identified according to the lightweight visual inspection model to obtain the suspicious information of the motherboard. Based on the motherboard suspicious information, the auxiliary detection state and the machine vision array are adjusted accordingly to perform supplementary detection on the mobile phone motherboard. The supplementary detection data is then deeply analyzed according to the depth vision detection model to generate visual detection conclusion information for the mobile phone motherboard.

[0005] Secondly, the present invention provides an artificial intelligence-based mobile phone motherboard visual inspection system for implementing the artificial intelligence-based mobile phone motherboard visual inspection method described in any one of the first aspects, comprising: An environmental assistance module is used to construct a thermal environment for the mobile phone motherboard to be tested, and to apply an optical environment to the mobile phone motherboard in the thermal environment so that the mobile phone motherboard is in an auxiliary testing state. The preliminary inspection module is used to acquire images of the mobile phone motherboard in the auxiliary inspection state through a machine vision array, and to identify key information in the acquired images based on a lightweight visual inspection model to obtain information on suspicious points on the motherboard. The supplementary detection module is used to adjust the auxiliary detection state and the machine vision array accordingly based on the motherboard suspicious information, so as to perform supplementary detection on the mobile phone motherboard, and perform in-depth analysis of the supplementary detection data according to the depth vision detection model to generate visual detection conclusion information of the mobile phone motherboard.

[0006] This invention provides a visual inspection method for mobile phone motherboards based on artificial intelligence, which has the following beneficial effects: This invention constructs an adaptive thermal and optical environment to put the mobile phone motherboard in an auxiliary inspection state. Multi-dimensional images are acquired through a machine vision array, and a lightweight visual inspection model is used to identify key information and obtain suspicious information. Based on this suspicious information, the inspection state and visual array are adjusted for supplementary inspection. A depth visual inspection model is used to perform in-depth analysis of the supplementary data to generate inspection conclusions. This method can improve inspection accuracy, optimize environmental adaptability, reduce computational burden, and increase inspection efficiency, meeting the needs of high-efficiency production and addressing the problem of insufficient inspection requirements for the precision and complexity of mobile phone motherboards. Attached Figure Description

[0007] Figure 1 This is a schematic diagram illustrating the steps of a mobile phone motherboard visual inspection method based on artificial intelligence provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of a mobile phone motherboard visual inspection system based on artificial intelligence, provided in an embodiment of the present invention. Detailed Implementation

[0008] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail 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 invention.

[0009] The implementation of the present invention will be described in detail below with reference to specific embodiments.

[0010] Reference Figure 1 , Figure 2 The diagram shows a preferred embodiment of the present invention.

[0011] In a first aspect, the present invention provides a visual inspection method for mobile phone motherboards based on artificial intelligence, comprising: S1: Construct a thermal environment for the mobile phone motherboard to be tested, and apply an optical environment to the mobile phone motherboard in the thermal environment so that the mobile phone motherboard is in an auxiliary testing state. S2: The machine vision array is used to acquire images of the mobile phone motherboard in the auxiliary inspection state, and the key information of the acquired images is identified according to the lightweight visual inspection model to obtain the suspicious information of the motherboard. S3: Based on the motherboard suspicious information, adjust the auxiliary detection state and the machine vision array accordingly to perform supplementary detection on the mobile phone motherboard, and perform in-depth analysis of the supplementary detection data according to the depth vision detection model to generate visual detection conclusion information of the mobile phone motherboard.

[0012] Specifically, in step S1 of the embodiment provided by this invention, a suitable heating module (such as a heating plate, infrared heater, or other heat source device) is selected according to the size, material, and heat requirements of the mobile phone motherboard to ensure uniform heating of all areas of the motherboard. Based on the operating conditions of the mobile phone motherboard or the manufacturer's specifications, specific heat conditions that the motherboard needs to achieve are set, such as the temperature range of the motherboard during normal operation or the temperature differences that may occur in specific areas. The heat is controlled in real time by a temperature control system, and the temperature of each area of ​​the motherboard is monitored by a temperature sensor to ensure that the motherboard temperature reaches the expected heat environment, avoiding overheating or underheating. Once the motherboard reaches the target heat conditions, the heat stability is maintained to prevent temperature fluctuations from affecting the detection. The test results show that during this process, the system needs to monitor changes in heat and adjust for uneven heat distribution. By placing the motherboard in a specific thermal environment, the heat distribution of the motherboard during actual operation can be simulated. This is crucial for identifying defects caused by thermal expansion (such as cracks, connection problems, etc.). Some defects may only become apparent under temperature changes or thermal stress. Therefore, simulating the thermal environment can help reveal these hidden defects and ensure the comprehensiveness of the detection. This step can effectively identify hardware problems caused by thermal stress, thermal expansion, or excessively high local temperatures, such as poor circuit connections and loose solder joints. By precisely controlling the thermal environment, the sensitivity of the detection system to motherboard defects can be improved, especially for those defects that are difficult to detect under normal conditions.

[0013] More specifically, based on the motherboard's materials and characteristics, a suitable light source is selected for illumination, such as visible light, infrared light, or ultraviolet light, to enhance the visibility of different types of defects. The intensity and angle of the light source also need to be set according to the specific detection target. Under the set thermal environment, the ambient lighting conditions are adjusted through the ambient light module, including the brightness, color temperature, and illumination angle of the light source, to optimize the visibility of the motherboard and simulate different lighting environments that may be encountered in actual work, such as strong light, weak light, or light from different directions. This helps ensure that defects on the motherboard can still be detected in various practical application environments. Based on factors such as the reflectivity and gloss of the motherboard surface, the lens angle, focal length, and other optical parameters of the detection equipment are adjusted to ensure that the captured images are clear and rich in detail. Different types of defects may appear differently under different lighting conditions. By applying a suitable optical environment, the contrast of certain defects can be enhanced, making them easier to detect. In actual use, mobile phone motherboards face various lighting conditions (such as sunlight, low light environments, etc.). By simulating these environments, the detection system can ensure accurate identification of defects under any conditions. Optimizing the optical environment can improve image quality, making defects more conspicuous, which is helpful for subsequent image processing and defect identification. Under different lighting conditions, some minor defects (such as surface cracks, corrosion points, welding defects, etc.) become more visible, thereby improving the sensitivity and accuracy of the detection system.

[0014] More specifically, after ensuring the motherboard is in the set thermal environment, further optimization of the optical environment is applied. This includes adjusting the reflected light on the motherboard surface, the shooting angle of the detection camera, and the light intensity. Based on the current thermal and optical environment, the operating parameters of the machine vision array (such as exposure time, focusing distance, image resolution, etc.) are adjusted. This process is to ensure that the vision array can obtain the best image data in the assisted detection state, and to ensure that all environmental conditions and machine vision array settings meet the expected goals. The motherboard is now in the assisted detection state, meaning that all conditions are most conducive to the manifestation of defects during the detection process. Through precise environmental control, the detection system can most clearly capture any potential defects in the motherboard under this state. This is to ensure the efficiency and accuracy of motherboard defect detection. Under the combined effect of thermal and optical environment, by optimizing the machine vision array, the quality of data acquisition is ensured, thereby enhancing the effectiveness of depth visual analysis. Precise environmental control ensures that the motherboard can be detected under the most suitable conditions, thereby improving the accuracy of defect detection and reducing the risk of missed detection. Under optimal environmental conditions, the machine vision array can more accurately identify minute defects, improving the overall performance of the system.

[0015] Understandably, by constructing a thermal environment and applying an optical environment, the physical conditions of a mobile phone motherboard in actual operation can be simulated, thereby enhancing the detection effect. Specific steps include: controlling the temperature through a heating module to ensure the motherboard is in a predetermined thermal environment; adjusting the optical environment to optimize factors such as light source, brightness, and illumination angle; and optimizing the machine vision array based on the above environment to ensure the motherboard is in an auxiliary detection state. The technical effect of this method is that by accurately simulating the actual working environment, it improves the visibility rate of motherboard defects and ensures that the depth vision detection model can identify potential problems under optimal conditions, thereby significantly improving the detection accuracy and quality of mobile phone motherboards.

[0016] Specifically, in step S2 of the embodiment provided by this invention, multiple high-definition cameras or image acquisition devices are used to capture images of the mobile phone motherboard from different angles and positions. The number and position of the cameras need to be configured according to the size, shape, and complexity of the motherboard to ensure that all areas of the motherboard are fully covered. Camera parameters (such as focal length, aperture, exposure time, etc.) are set to ensure clear images are obtained under different environmental conditions. The machine vision array simultaneously acquires images at multiple preset angles, forming a series of multi-dimensional image data. These images will show different perspectives of the motherboard, including the front, back, and sides. High-definition cameras are used to perform high-precision scanning of the motherboard surface, capturing surface details, component connections, and minor flaws or defects. Images are synchronized in time to ensure continuity and consistency. Using image stitching or 3D modeling techniques, images from different perspectives are combined into a complete, 3D motherboard model, further improving the accuracy of defect detection. By acquiring images from different perspectives, the details of the motherboard can be fully captured, avoiding problems that cannot be detected from a single perspective. For example, some defects are only visible on the side or back, which may be missed from a single perspective. Setting up high-resolution image acquisition equipment helps to capture tiny defects and ensures the accuracy of detection. Multi-dimensional image acquisition can provide more comprehensive visual information, ensuring that details from every angle are covered and reducing blind spots. By comprehensively acquiring images of the motherboard, it is possible to ensure that the motherboard is inspected from all dimensions, improving the defect detection rate.

[0017] More specifically, before inputting the images into the visual inspection model, the acquired images are first preprocessed, including denoising, image enhancement, and contrast adjustment, to improve image quality and clarity. The images are then standardized to ensure quality consistency across different images and reduce the impact of environmental factors on image quality. The preprocessed images are then input into a trained, lightweight visual inspection model. This optimized model can efficiently identify key information in the images. Based on deep learning technology, the visual inspection model uses convolutional neural networks (CNNs) or other suitable lightweight algorithms to identify defects, soldering problems, cracks, scratches, color deviations, etc., on the motherboard surface. The model analyzes details in the image to extract key information, such as the location, type, size, and possible causes of defects (e.g., short circuits, cold solder joints). The extracted data is labeled on the image, generating a suspicious information report, including the confidence level of the suspicious point and the defect type. The extracted suspicious information, such as the type and potential affected area, is categorized and further analyzed according to its type. For example, different treatments are applied to cracks, welding problems, or corrosion. The system generates feedback information based on the defect type, prompting the operator or subsequent processing system whether further operations (such as repair, replacement, or continued inspection) are needed. The lightweight visual inspection model, specially trained, can quickly and accurately identify defects in motherboard images without requiring high computing resources, making it suitable for practical inspection scenarios. Through deep learning model learning, it can extract precise details from images, automatically identify existing problems, and avoid errors and omissions in manual inspection. The use of the lightweight visual inspection model makes image processing fast and accurate, and can identify suspicious points on the motherboard in real time. The model can not only quickly identify defects, but also automatically classify and mark suspicious points, reducing manual intervention and improving the automation and intelligence level of the inspection process.

[0018] More specifically, based on the suspicious information identified by the model, a detailed inspection report is generated. The report includes the specific location, type, size, and inspection confidence level of the defect. In the report, defects in the image are highlighted to facilitate quick location of the problem area. If a serious defect or an unconfirmed suspicious point is found during the inspection process, the system will automatically send a real-time alarm and suggest further inspection or repair. If the defect seriously affects the function, the system will automatically repair it or notify the maintenance team. The generation of inspection reports can help to quickly locate problems and reduce the troubleshooting time in the later stages of inspection. The real-time feedback mechanism ensures that problems can be dealt with in a timely manner when they are discovered, avoiding major quality problems in the production process. By generating detailed reports and automatically marking suspicious points, potential problems on the motherboard can be located quickly and accurately. Real-time feedback and automated repair reduce the time and errors of manual intervention and improve the overall efficiency of inspection and repair.

[0019] Understandably, the entire process involves using a machine vision array to acquire images of the mobile phone motherboard from multiple perspectives, combining this with a lightweight visual inspection model to identify key information from the images, and finally extracting suspicious information from the motherboard. Specific steps include using multiple cameras to acquire high-quality images of the motherboard from different angles, ensuring that defects from every angle are captured; using a trained lightweight visual inspection model to preprocess and analyze the images, extracting key information such as defect location and type; generating a suspicious information report and providing real-time feedback; and ensuring that defects can be located and addressed promptly. This method can significantly improve the accuracy, speed, and intelligence level of mobile phone motherboard inspection, reduce manual intervention, and improve inspection efficiency and product quality.

[0020] Specifically, in step S3 of the embodiment provided by this invention, suspicious information about the motherboard is extracted through the previous machine vision array and lightweight visual inspection model. These suspicious points include the location, type, and size of defects. The system first analyzes these suspicious points to identify the areas and questions that most need supplementary inspection. Based on this suspicious point information, the system automatically determines which areas may have missed or incompletely detected parts, or which areas have more serious defects that require further confirmation. According to the suspicious point information, the system adjusts the viewing angle and shooting area of ​​the machine vision array. The system will adjust the camera or image acquisition device to the area where the suspicious points are located to increase the visual scanning of these areas. For example, for the detected micro-welding defects, the system will automatically adjust the angle of the camera to ensure that the area is covered by more viewing angles and reduce blind spots. Based on suspicious information, the system appropriately enhances auxiliary detection settings, including increasing image acquisition resolution, adjusting camera focus mode, or increasing light source intensity to ensure the detection of more detailed information. The system can switch to high-precision mode or select specific lighting conditions to enhance detection in difficult-to-detect areas as needed. By dynamically adjusting the detection status and vision array based on suspicious information, the system ensures concentrated resources for high-precision supplementary detection. By adjusting the viewing angle and area, small-scale defects are avoided, improving the comprehensiveness and accuracy of detection. Adjusting the detection strategy based on actual suspicious information enables more efficient and targeted supplementary detection, reducing wasted time and improving defect detection accuracy. The system intelligently adjusts settings based on detection results to maximize resource utilization and avoid unnecessary duplicate detection.

[0021] More specifically, in the adjusted detection state, the machine vision array continues supplementary detection, capturing image data of areas not fully covered or suspected by previous detection. At this point, high-resolution images and precise light source settings are used to ensure that defects in these areas are clearly captured. The focus of supplementary detection is on detailed scanning of smaller or harder-to-observe defects, such as tiny cracks, poor solder joints, and exposed circuit boards. The images obtained from supplementary detection are input into a deep vision detection model for deep analysis. This model typically uses a convolutional neural network (CNN) or more advanced deep learning algorithms to extract more complex feature information from the image. The deep vision detection model performs multi-level feature extraction and classification on the supplementary detection data, analyzes every detail in the image, and identifies potential defect types (such as short circuits, component misalignment, etc.) based on training data. Through the output of the deep vision detection model, the system automatically generates a visual inspection report of the mobile phone motherboard. The conclusion information, including defect type, defect location, and defect severity, will be integrated into a detection report or conclusion data, provided to operators or subsequent processing systems. If serious defects are found in the detection results, the system will suggest further repair or replacement; if the defects are minor or do not affect functionality, the system will indicate that production can continue. The deep vision detection model can perform in-depth analysis of complex image features, thereby discovering minute defects that traditional methods cannot detect, improving the depth and accuracy of defect identification. By combining supplementary detection with the deep model, defects can be located and classified in detail, effectively avoiding omissions and providing detailed information for subsequent processing. Through deep feature extraction and analysis by the deep vision detection model, minute defects can be accurately identified and classified, improving the comprehensiveness and accuracy of detection. The system can automatically generate detailed detection conclusions, reducing manual analysis time, improving efficiency, and avoiding interference from human factors.

[0022] More specifically, based on the analysis results of the depth vision inspection model, the system generates a final inspection report. The report includes: inspection conclusion (whether there is a defect), defect type, defect severity, recommended handling measures, etc. The report displays suspicious information on the motherboard and highlights defective areas with image markings to facilitate subsequent manual review or automated repair. The system feeds back the inspection conclusion information to operators or the production control system in real time. If a serious defect is found, the system will suggest repair or scrapping; if the defect is minor, the system will suggest continuing production. The system adjusts the production process based on the actual inspection results. For example, for motherboards with many defects, it may be necessary to suspend the production line for manual handling or transfer the product to a later inspection stage. By providing real-time feedback on the inspection conclusions, corresponding handling measures can be taken quickly to prevent defective products from flowing to the next stage or the end consumer. Adjusting the production plan or process based on real-time inspection results can reduce ineffective production, optimize resource utilization, and the real-time feedback mechanism can effectively optimize the production plan, reduce the outflow of defective products, and improve overall production efficiency. Through accurate inspection and timely feedback, the quality of the final product is guaranteed, and later maintenance costs and customer complaints are reduced.

[0023] Understandably, by dynamically adjusting the detection strategy and configuration, targeted supplementary detection of suspicious areas on the mobile phone motherboard can improve the accuracy of detection. Supplementary detection data can be deeply analyzed through a deep vision detection model to extract complex features, accurately identify and locate defects, and intelligently adjust the production process or repair measures through real-time feedback of detection conclusions to ensure product quality. Through these steps, the efficiency, accuracy, and intelligence level of visual inspection of mobile phone motherboards can be significantly improved, while optimizing the production process and ensuring high-quality product output.

[0024] This invention provides a visual inspection method for mobile phone motherboards based on artificial intelligence, which has the following beneficial effects: This invention constructs an adaptive thermal and optical environment to put the mobile phone motherboard in an auxiliary inspection state. Multi-dimensional images are acquired through a machine vision array, and a lightweight visual inspection model is used to identify key information and obtain suspicious information. Based on this suspicious information, the inspection state and visual array are adjusted for supplementary inspection. A depth visual inspection model is used to perform in-depth analysis of the supplementary data to generate inspection conclusions. This method can improve inspection accuracy, optimize environmental adaptability, reduce computational burden, and increase inspection efficiency, meeting the needs of high-efficiency production and addressing the problem of insufficient inspection requirements for the precision and complexity of mobile phone motherboards.

[0025] Preferably, the steps of constructing a thermal environment for the mobile phone motherboard to be tested, and applying an optical environment to the mobile phone motherboard in the thermal environment to put the mobile phone motherboard into an auxiliary testing state include: S11: Deploy the mobile phone motherboard on the detection fixing structure and power on the mobile phone motherboard so that the mobile phone motherboard is in the internal heat environment brought about by power-on operation. S12: The area where the mobile phone motherboard is located is heated by the heating module so that the mobile phone motherboard is in an external heat environment brought about by the area heating. S13: Based on the internal and external thermal environments of the mobile phone motherboard, analyze the optical environment required by the mobile phone motherboard to generate optical environment configuration information; S14: Based on the optical environment configuration information, drive the ambient light module to apply ambient light conditions to the mobile phone motherboard, so that the mobile phone motherboard is in an optical environment that conforms to the optical environment configuration information, so that the mobile phone motherboard is in an auxiliary detection state.

[0026] Specifically, the mobile phone motherboard is installed onto a testing and fixing structure, ensuring it is securely placed in a predetermined position. This structure is an automated fixture or support device that can fix the motherboard and ensure its position remains unchanged. The motherboard is then powered on, causing its internal system to start working. This process simulates the motherboard's operating state during normal use. The generated current and power consumption will cause the internal circuits and components to generate heat. After power-on, the motherboard's circuits and components will be in normal working condition. The generated heat is an important step in simulating the heat generation of the device in real use. Power-on processing ensures the formation of the internal thermal environment of the motherboard, making the testing process closer to real working conditions and avoiding deviations in test results due to environmental differences. Ensuring the thermal environment of the motherboard under working conditions: By powering on, a thermal environment that conforms to actual usage conditions is generated, ensuring the reliability of the test results. Testing the motherboard in a real working environment can better detect potential faults or defects caused by temperature fluctuations or workload.

[0027] More specifically, a heating module is deployed around the phone's motherboard to ensure that it can evenly heat the external areas of the motherboard. The heating module can provide the required temperature conditions through electric heating, infrared radiation heating, or fan heating. Activating the heating module heats the area where the phone's motherboard is located in an external thermal environment, allowing the motherboard to reach the required operating temperature. This external heating environment simulates the phone's operation under high-temperature conditions. Unlike internal heat, external heat is primarily affected by changes in the external ambient temperature. By simulating this environment, the stability of the phone's motherboard in the face of ambient temperature fluctuations can be effectively assessed. This heating treatment can simulate potential thermal failures, overheating, or thermal expansion problems that may occur when the phone operates in a high-temperature environment, thereby ensuring detection accuracy at high temperatures. Through heating treatment, not only is the internal thermal environment of the motherboard simulated, but an external thermal environment is also created, comprehensively examining the temperature fluctuations the motherboard may face in actual use. This simulation of the external thermal environment helps evaluate the motherboard's performance and reliability under different temperature conditions, ensuring that the motherboard adapts to diverse usage scenarios.

[0028] More specifically, the system monitors and analyzes the internal and external thermal environment of the motherboard. Temperature data is collected using temperature sensors, and the differences between the internal and external thermal environments are calculated. Based on this analysis, the system determines the motherboard's optical environment requirements at specific temperatures. For example, a motherboard in a high-temperature environment may have different requirements for light intensity or spectral range. This step uses an algorithm to match the current environmental conditions with the motherboard's thermal response characteristics to determine the optimal optical environment. Temperature and light intensity interact, and the motherboard's thermal response may alter its surface or internal reflective properties. By analyzing the required optical environment based on the thermal environment, the system ensures that the lighting conditions during the detection process match the motherboard's current state, avoiding detection errors caused by insufficient or excessive light. At different temperatures, the motherboard surface exhibits different reflection or transmission characteristics. By accurately analyzing optical requirements, the system can better simulate and adapt to these changes. With appropriate optical environment configuration information, the system can automatically adjust the lighting conditions according to the thermal environment, ensuring that every detail of the motherboard is clearly captured under optimal lighting conditions. Dynamically adjusting lighting conditions based on the actual thermal environment improves the accuracy and reliability of image acquisition during the detection process.

[0029] More specifically, based on the previously analyzed optical environment configuration information, the system drives the ambient light module (such as LED light sources, laser illumination, etc.) to apply corresponding ambient lighting conditions to the mobile phone motherboard. This process ensures that the optical environment of the motherboard is completely matched with the analysis results. By adjusting the brightness, spectral characteristics, and illumination angle of the light source, the system ensures that the mobile phone motherboard is under suitable lighting conditions for detection. These lighting conditions are dynamically adjusted according to changes in motherboard temperature and the needs of suspicious areas. The internal and external temperature environments of the motherboard affect its lighting requirements, so the lighting conditions need to be precisely adjusted to match the current state of the motherboard. Through the driving of the ambient light module, it is ensured that the lighting always remains consistent with the working state and detection requirements of the motherboard. By dynamically applying appropriate ambient lighting, it is ensured that the motherboard can obtain accurate optical detection under different heat conditions, without being interfered with by excessively strong or weak lighting. The ambient lighting is automatically adjusted according to the actual temperature conditions of the motherboard to ensure that no errors occur during the detection process and improve the accuracy of the final detection results.

[0030] Understandably, deploying the mobile phone motherboard on a fixed structure and powering it on simulates actual working conditions, generating an internal thermal environment. A heating module then processes the external thermal environment of the motherboard, simulating high-temperature operating conditions. Based on the internal and external thermal environments, the required optical environment is analyzed to ensure detection accuracy. Based on the optical environment configuration information, an ambient light module is driven to apply appropriate lighting conditions so that the motherboard is in an ideal detection state. These steps, by accurately simulating the thermal and optical environment of the mobile phone motherboard, ensure efficient auxiliary detection under real working conditions, thereby improving the accuracy and reliability of the detection results and optimizing the product quality control process.

[0031] Preferably, the step of analyzing the optical environment required by the mobile phone motherboard based on the internal and external thermal environments to generate optical environment configuration information includes: S131: Based on the standard performance specifications and power-on processing parameters of the mobile phone motherboard, the internal thermal environment is simulated, and based on the module operating parameters of the heating module, the external thermal environment is simulated. Combining the simulation results of the internal and external thermal environments, thermal environment simulation information is obtained. S132: Based on the thermal environment simulation information, analyze the defect highlighting effect and image visual mapping relationship of the mobile phone motherboard under the thermal environment to obtain the defect highlighting effect features and image visual mapping features of the mobile phone motherboard under the current thermal environment; wherein, the defect highlighting effect features are used to describe the ease of display of defects of the mobile phone motherboard under the current thermal environment, and the image visual mapping features are used to describe the mapping relationship between the image visual information of the mobile phone motherboard under the current thermal environment and the normal state; S133: Based on the image visual mapping features, the defect highlighting effect features are subjected to optical environment adaptive processing to generate defect display optimization parameters and visual distortion compensation parameters for the optical environment of the mobile phone motherboard. S134: Perform factor synergy value analysis on the defect display optimization parameters and the visual distortion compensation parameters, and construct several factor synergy working groups based on the results of the factor synergy value analysis; S135: Perform optical environment simulation on each of the aforementioned factor collaborative working groups to evaluate the defect display optimization effect and visual distortion compensation effect of each of the aforementioned factor collaborative working groups under dual conditions. Based on the evaluation results, optimize the factor collaborative working groups to generate optical environment configuration information.

[0032] Specifically, based on the standard performance specifications and power-on processing parameters of the mobile phone motherboard, the internal thermal environment of the motherboard is simulated. This includes analyzing the heat generated by the motherboard during power-on operation, calculating the internal heat distribution of the motherboard using a thermal model, and simulating the external thermal environment of the area where the mobile phone motherboard is located based on the operating parameters of the heating module. This mainly involves providing an external heat source to the motherboard through the heating module, thereby analyzing the impact of the external environment on the motherboard temperature. Combining the simulation results of the internal and external thermal environments, comprehensive thermal environment simulation information is generated. This information will provide a basis for subsequent image visual mapping and defect highlighting effect analysis. Through the simulation of the internal and external thermal environments, a thermal environment close to actual usage conditions can be created, helping the system to better adapt to temperature fluctuations in real operation. Combining the influence of two heat sources (internal and external) allows for a more comprehensive evaluation of the performance of the mobile phone motherboard under different temperature conditions, ensuring the scientific nature and accuracy of the detection process. Through detailed thermal environment simulation, defects of the mobile phone motherboard under different thermal conditions can be accurately identified, avoiding errors caused by environmental changes. The dual simulation of internal and external environments provides more accurate temperature data, ensuring high precision in subsequent visual and image processing.

[0033] More specifically, based on thermal environment simulation information, this study analyzes the effect of highlighting defects on the mobile phone motherboard under the current thermal environment. This analysis simulates the impact of thermal changes on the motherboard surface material, identifying how factors such as thermal expansion and contraction affect the visibility of defects. By analyzing the impact of the thermal environment on the motherboard, it examines the mapping relationship between the motherboard's visual image information and its normal state. This primarily involves changes in the distortion, deflection, or reflection characteristics of the motherboard image under different thermal conditions, identifying the impact of heat on image quality. Different thermal environments affect the ease with which motherboard defects are displayed. Analyzing the defect highlighting effect helps to assess the impact of different thermal environments on defect display in advance, optimize detection schemes, and understand that thermal changes may cause visual distortion in images. Understanding this mapping relationship helps to more accurately optimize visual perception, ensuring that defects are clearly displayed under different thermal environments. Through defect highlighting effect analysis, the display characteristics of motherboard defects under different thermal environments can be identified, thereby optimizing the image capture and analysis process. Furthermore, through image visual mapping feature analysis, image capture technology can be further optimized, making images under different thermal environments more realistic and clear.

[0034] More specifically, the optical environment is adapted to enhance the highlighting effect of defects based on the visual mapping features of the image. This includes compensating for visual distortion caused by heat changes in the image and optimizing the display effect of defects. Based on the results of the adaptive processing, a set of defect display optimization parameters and visual distortion compensation parameters for the optical environment are generated. These factors are used to further optimize image quality and ensure that defects can be accurately displayed under actual working conditions. By generating defect display optimization factors and visual distortion compensation factors, it is ensured that the detection process of the mobile phone motherboard is not affected by environmental changes under different heat environments, and the display effect is optimized. When the heat environment changes, the image may be distorted. Generating compensation factors helps to correct these distortions and improve the accuracy of the image. The application of optimization factors can effectively compensate for visual distortion caused by environmental changes, ensuring that motherboard defects can be clearly displayed under different temperature conditions. By adjusting the optical environment parameters, the consistency and accuracy of the detection results under different environmental conditions can be ensured.

[0035] More specifically, a synergistic value analysis is conducted on defect display optimization parameters and visual distortion compensation parameters to assess the impact of these factors on defect display and visual effects under different conditions. This analysis helps determine which factors play a greater role in actual detection. Based on the synergistic value analysis results, multiple factor synergistic working groups are constructed and optimized. Factors in each working group will work synergistically under specific thermal environment conditions to achieve optimal visual effects and defect display. Optical environment simulation is performed on each factor synergistic working group to evaluate its dual role in optimizing defect display effects and compensating for visual distortion. Based on the evaluation results, the combinations of various factors are adjusted to achieve the best effect. By constructing factor synergistic working groups, the role of each factor under different conditions can be optimized to ensure that the final optical environment settings can maximize the detection effect. By simulating the optical environment, the visual effect of the motherboard under different thermal conditions can be optimized, external interference factors can be reduced, and the overall performance of the detection system can be improved. Through factor synergistic optimization and simulation, the optical environment settings can be dynamically adjusted according to the actual thermal environment conditions to ensure the best defect display effect. The optimized factor synergistic working groups can ensure the reliability and accuracy of detection results under different environmental conditions.

[0036] More specifically, based on all optimization and simulation results, optical environment configuration information is generated. This configuration information includes parameters such as optimal light intensity, wavelength, and light source distribution to ensure that the motherboard can receive optimal optical support under different thermal environments. The generated optical environment configuration information is the optimal solution after considering all environmental factors, which can provide stable and accurate detection results in actual use. By generating the most suitable optical environment configuration information, the adaptability of the system in various environments is ensured, further improving the accuracy of motherboard defect detection.

[0037] Understandably, by comprehensively simulating the internal and external thermal environment through these steps, analyzing the impact of the thermal environment on defect display and visual distortion, generating a series of optimization factors, and through factor synergistic analysis and optical environment simulation, the most suitable optical environment configuration information for the current environment is finally obtained. This method can effectively optimize the detection accuracy of the mobile phone motherboard and ensure the stability and accuracy of the detection results under different thermal environments.

[0038] Specifically, the step of performing optical environment adaptive processing on the defect highlighting effect features based on the image visual mapping features to generate defect display optimization parameters and visual distortion compensation parameters for the optical environment of the mobile phone motherboard includes: S1331: Vectorize the defect highlighting effect features to generate a display enhancement feature matrix; S1332: Analyze the potential optimization trend of the display enhancement feature matrix by combining the image visual mapping features, so as to generate a matrix form in which the display enhancement feature matrix theoretically has a display enhancement effect; S1333: Analyze the optical environment compensation of the display enhancement feature matrix according to the matrix form, so as to assist the display enhancement feature matrix to achieve the defect display optimization parameters required by the matrix form through the optical environment compensation method; S1334: Based on the image visual mapping features, perform optical environment compensation analysis on the visual distortion effect caused by the current thermal environment to obtain visual distortion compensation parameters.

[0039] Specifically, the features of the defect highlighting effect are vectorized to construct a display enhancement feature matrix. This matrix represents different attributes of defect highlighting, such as the display intensity and clarity of the display boundary. The matrix contains multi-dimensional feature data, covering the visual manifestation characteristics of defects under different thermal environments. This process is achieved by processing actual image data and quantifying the matrix form of defects. Through vectorization, the visual features of defect highlighting can be effectively described mathematically, facilitating subsequent analysis and optimization. Integrating multi-dimensional features into the matrix facilitates systematic processing and analysis, aiding in subsequent optimization and compensation. After vectorization, the defect highlighting effect can be presented more clearly, facilitating accurate analysis. The matrix form enables more efficient large-scale data analysis and reduces repetitive work.

[0040] More specifically, by combining image visual mapping features with a display enhancement feature matrix, potential optimization trends are analyzed. By analyzing the dynamic changes and visual distortion of defects in the image, the display enhancement effect of defects under different thermal environments is predicted. Based on the analysis of potential optimization trends, a matrix form with theoretically enhanced display effects is generated. This matrix form reflects the display effect that defects should achieve after optimization, such as clearer edges, more prominent or vivid display, etc. Combining image visual mapping features can accurately predict the impact of thermal environment changes on defect display effects, helping designers to make targeted optimizations. By analyzing potential trends, the best display effect can be predicted before actual optimization, avoiding blind adjustments. Through optimization trend analysis, the ideal effect that defect display should achieve under thermal changes can be predicted relatively accurately without actual experiments, which helps to theoretically optimize the display effect according to actual thermal environment changes.

[0041] More specifically, based on the generated theoretical matrix form, optical environment compensation analysis is performed on the display enhancement feature matrix. This process simulates changes in the optical environment and analyzes how to adjust optical parameters to achieve the desired effect in defect display. Through optical compensation measures (such as adjusting the light source, reflection angle, and display parameters), the display enhancement feature matrix is ​​assisted in achieving the required matrix form, thereby optimizing the display effect of defects. In environments with changing heat, the visual effect of images may be distorted. Optical environment compensation analysis can effectively correct these problems and achieve the expected display effect. Through compensation analysis, multiple parameters in the optical environment can be precisely controlled to ensure the optimization of image display effect. Compensation analysis helps correct display problems caused by environmental changes, improves the accuracy and visibility of defect detection, and enables the display system to perform consistently under different heat environments through optical environment compensation.

[0042] More specifically, based on the visual mapping features of the image, the system analyzes the visual distortion effects caused by the current thermal environment, including color distortion and blurred image boundaries. Based on the distortion analysis results, visual distortion compensation parameters are generated. These factors are mainly used to compensate for various visual deviations in the image caused by changes in the thermal environment, such as compensating for color differences and correcting image deformation. Changes in the thermal environment may cause image distortion. By generating a set of compensation factors, these distortions can be effectively corrected, ensuring the accuracy of the detection results. The generation of the compensation factor set enables the system to make precise adjustments for different visual distortions, avoiding display distortion. The compensation factor set can effectively repair image distortion caused by the thermal environment, improve image quality, and ensure the clarity and accuracy of image display under thermal changes, especially in defect detection, helping to improve recognition accuracy.

[0043] Understandably, by vectorizing the features of the defect highlighting effect, analyzing potential optimization trends, and performing optical environment compensation analysis, and generating visual distortion compensation parameters, an adaptive processing scheme for the optical environment is ultimately formed. This process not only optimizes the display enhancement effect but also effectively compensates for the visual distortion caused by the thermal environment, ensuring the defect detection accuracy and visual clarity of the mobile phone motherboard under different environments. This method helps to improve the image detection effect of the mobile phone motherboard, enabling it to work stably in complex environments.

[0044] Preferably, the steps of acquiring images of the mobile phone motherboard in an assisted inspection state using a machine vision array, and identifying key information from the acquired images using a lightweight visual inspection model to obtain suspicious information about the motherboard include: S21: Adjust the observation position and angle of each machine vision sensor in the machine vision array so that each machine vision sensor is at the specified observation angle. S22: Deploy corresponding operating parameters for each machine vision sensor in the machine vision array to drive each machine vision sensor to perform optical vision acquisition on the mobile phone motherboard in a specified form based on a specified observation angle, so as to generate multi-dimensional optical data; wherein, the specified form of optical vision acquisition performed by the machine vision sensor includes visible light form, infrared light form, 3D structured light form, and laser scanning form. S23: According to the specified form of optical vision acquisition corresponding to the multi-dimensional optical data, mark the data acquisition form of each data part of the multi-dimensional optical data, and divide the multi-dimensional optical data based on the marking to fill the pre-constructed information representation framework, so as to obtain the multi-dimensional optical data that meets the requirements of model analysis. S24: Substitute the multi-dimensional optical data into a pre-trained lightweight visual detection model, and let the lightweight visual detection model identify key information from the multi-dimensional optical data to obtain motherboard suspicious information.

[0045] Specifically, the position and angle of each sensor in the machine vision array are adjusted to ensure that each vision sensor is located at a specified observation angle, avoiding visual overlap or blind spots. The size and shape of the mobile phone motherboard are taken into account during the adjustment to ensure that each sensor covers different parts of the motherboard, achieving all-round and multi-angle observation. The viewing angle of each machine vision sensor should be determined according to actual needs, such as front, side, top, bottom, etc., to ensure comprehensive inspection of the motherboard. It is ensured that sensors with different viewing angles can effectively cover all parts of the motherboard, avoiding the omission of key details. Reasonable sensor arrangement can improve the acquisition efficiency, reduce repetitive work, ensure that all key parts are monitored, improve the comprehensiveness of image data, and precise viewing angle adjustment makes the monitoring without blind spots, ensuring that all suspicious information can be captured.

[0046] More specifically, appropriate operating parameters, such as exposure time, aperture size, focal length, and image resolution, are configured for each machine vision sensor to ensure that the acquired image quality meets requirements. Corresponding parameters are deployed for different types of sensors (such as visible light, infrared light, 3D structured light, and laser scanning) to adapt to different optical vision acquisition methods. Based on the set operating parameters, each machine vision sensor performs optical vision acquisition at a specified observation angle. Acquisition methods include, but are not limited to: visible light: for detecting the color and structural features of the motherboard surface; infrared light: for detecting the temperature distribution and potential thermal faults of the motherboard; 3D structured light: for acquiring the three-dimensional morphology of the motherboard surface to help identify surface defects; and laser scanning: for high-precision surface morphology detection to identify minute flaws and deformations. Different optical methods are used for acquisition according to different working environments and detection needs to ensure detailed detection of all aspects of the motherboard. By precisely adjusting the sensor parameters, the best image acquisition effect is obtained. Through multi-dimensional optical data, the state of the motherboard can be analyzed from different angles and optical methods to obtain comprehensive detection information. Through parameter optimization, the quality of the acquired data is ensured to meet the analysis requirements, avoiding data errors caused by improper parameters.

[0047] More specifically, each data segment is labeled according to the form of the acquired optical visual data (such as visible light, infrared light, 3D structured light, etc.). These labels facilitate subsequent model analysis and processing. The labels include information such as the optical acquisition form and acquisition angle of each data segment. Based on the labeled data, the multi-dimensional optical data is segmented, and each data segment is filled into a pre-constructed information representation framework to ensure that the data meets the needs of subsequent model analysis. This process ensures the structural consistency of the data, facilitating further processing by the model. Through labeling and segmentation, each data segment can be clearly mapped to a specific acquisition form and angle, which is helpful for subsequent analysis. Filling the data into a predefined framework ensures that it meets the input requirements of the model, improving the efficiency of subsequent processing. Through segmentation and labeling, the raw image data is transformed into easily processed structured data. Through framework filling, the compatibility of the data with the lightweight visual inspection model is ensured, providing accurate input for model analysis.

[0048] More specifically, the labeled and segmented multi-dimensional optical data is fed into a pre-trained lightweight visual inspection model. This model can efficiently process image data and identify key features. Based on the training data, the model automatically identifies suspicious information on the motherboard, including defects, cracks, welding problems, component misalignment, etc. After training, the lightweight visual inspection model can quickly identify key information on the motherboard, especially when there are multiple suspicious points, it can accurately locate the problem area. The application of the model reduces manual intervention and improves inspection efficiency and accuracy. The model can accurately identify defects and suspicious points on the motherboard, providing higher inspection accuracy. Through automated key information identification, it can provide real-time feedback on suspicious information, reducing the workload of manual inspection.

[0049] Understandably, through the above steps, using machine vision arrays and lightweight visual inspection models, suspicious information on the mobile phone motherboard can be identified efficiently and accurately. The multi-dimensional perspective acquisition of sensors and the combination of different optical forms ensure the comprehensiveness and accuracy of the data, while data labeling and segmentation enable the input data to meet the requirements of model analysis. Finally, the lightweight visual inspection model completes the identification of key information, providing a reliable basis for subsequent quality inspection.

[0050] Preferably, during the process of identifying key information from the multi-dimensional optical data by the lightweight visual inspection model, a potential defect assessment mechanism is set up. The potential defect assessment mechanism is used to assess the potential defects of the data at various locations of the information representation framework obtained by the lightweight visual inspection model, and to perform interactive collaborative analysis on the various potential defect assessment results to obtain a collaborative assessment result. The collaborative assessment result is used to combine with the key information identified by the lightweight visual inspection model from the multi-dimensional optical data to determine the motherboard suspicious information used to report potential defects on the mobile phone motherboard.

[0051] Specifically, based on the lightweight visual inspection model, a potential defect assessment mechanism is established. This mechanism analyzes the key information output by the model and assesses the severity of each potential defect area. It comprehensively evaluates the data at each location, considering local features, global trends, and known defect patterns in historical data. By analyzing the key information of each part, the potential defect assessment mechanism evaluates the existence of potential defects such as cracks, poor contact, surface corrosion, overheating, and solder joint defects. Based on this mechanism, the model can further quantify the potential risk of each suspicious point and output the corresponding risk level (e.g., low, medium, high risk). Through the potential defect assessment mechanism, it is possible to more accurately identify those minor defects that may lead to subsequent failures, ensuring the depth and breadth of inspection. This mechanism provides a higher level of intelligence for the model's automated decision-making, enabling it not only to identify suspicious points but also to judge their potential impact, identify and predict potential hazards that may lead to product failures, and provide earlier warnings. Through the assessment results, it is possible to clearly indicate which parts have a high defect risk, which is helpful for subsequent maintenance and improvement.

[0052] More specifically, interactive collaborative analysis is performed on the assessment results of various potential defects. In this process, multiple assessment results are processed collaboratively, considering the correlation and cross-validation results of data from different sources. For example, if infrared, visible light, and 3D structured light data for a certain location all show anomalies, the potential defect risk in that area will be assessed more highly. Based on the results of the interactive collaborative analysis, the system generates a comprehensive assessment result. These collaborative assessment results are used to further determine whether the motherboard has higher-risk defects. In this process, the collaboration of data from different dimensions enhances the accuracy and reliability of identifying suspicious points. Cross-validation from multiple data sources can eliminate errors caused by a single data source, improving the accuracy of defect identification. Interactive collaborative analysis enables the model to make more accurate judgments based on the collaborative results of different data sources, avoiding errors from using only one data format. Through the collaborative analysis of multi-dimensional data, the accuracy of defect risk assessment is improved, providing a more precise basis for decision-making. Interactive collaborative analysis makes the results more reliable, thus providing strong support for subsequent testing and quality control.

[0053] More specifically, the key information from multi-dimensional optical data recognition using a lightweight visual inspection model is combined with the collaborative evaluation results of potential defects. Here, the key information includes not only the location of suspicious points but also the defect risk level of each suspicious point. For each identified suspicious point, combined with its potential defect evaluation results, the system assesses the potential failure risk or impact, generating final defect hazard information. Based on the above analysis, the system generates final motherboard suspicious point information, including location, defect type, and potential risk level. This information is fed back to subsequent algorithms for quality control or repair. Combining the identification and evaluation results, the system can comprehensively and accurately reflect the motherboard's status, especially in identifying potential defects before they fully manifest. The timely feedback mechanism helps to make targeted corrections or optimizations to the motherboard. By combining key information with potential defect evaluation results, a more comprehensive feedback is formed, helping to further judge the motherboard's quality. The system can provide early warnings of potential hazards, reducing product quality risks.

[0054] Understandably, by introducing a potential defect assessment mechanism into a lightweight visual inspection model and conducting interactive collaborative analysis, the accuracy and reliability of identifying motherboard defects can be further improved. The potential defect assessment mechanism can identify potential defects that have not yet fully manifested, while the interactive collaborative analysis enhances the comprehensive judgment of motherboard quality issues through the fusion and comparison of multi-dimensional data. Ultimately, by combining these results, the potential defects on the motherboard can be accurately reported, providing effective support for subsequent product optimization, quality control, and repair.

[0055] Preferably, the steps of adjusting the auxiliary detection state and the machine vision array based on the motherboard suspicious information to perform supplementary detection on the mobile phone motherboard, and performing in-depth analysis of the supplementary detection data according to the depth vision detection model to generate visual detection conclusion information of the mobile phone motherboard include: S31: Input the motherboard suspicious information into a pre-built digital twin model of the mobile phone motherboard to generate a digital twin that reflects potential motherboard defects; S32: Based on the digital twin, the detection effect of different thermal and optical environments is evaluated, and the thermal and optical environments are optimized in accordance with the trend analysis results through value evaluation of the detection effect and trend analysis of the value evaluation results. The optimization process is iterated until the optimal environmental configuration information is obtained. S33: Based on the optimal environment configuration information, the thermal environment and optical environment of the mobile phone motherboard are reconstructed so that the mobile phone motherboard is in an auxiliary detection state corresponding to the optimal environment configuration information, and the machine vision array is adjusted simultaneously to perform supplementary detection on the mobile phone motherboard and obtain supplementary detection data. S34: Assign attention weights to the digital twin, and analyze the form of defect and hidden danger verification of the assigned digital twin in combination with the optimal environment configuration information, so as to adjust the parameters of the pre-trained deep vision detection model for the detection task based on the analysis results. S35: Perform depth analysis on the supplementary detection data according to the depth vision detection model to generate visual detection conclusion information of the mobile phone motherboard.

[0056] Specifically, based on previously identified suspicious information about the mobile phone motherboard, a digital twin of the motherboard is created. This digital twin is a virtual copy of the motherboard and can be used to simulate and predict the motherboard's performance under different environmental conditions. The twin constructs a virtual model of the real-world motherboard by integrating multi-dimensional data (such as optical, thermal, and mechanical strength data). The digital twin will provide feedback on potential defects and hidden dangers on the motherboard, mainly by simulating the motherboard's response under various conditions to identify potential risks. Through digital twin modeling, the virtual status of the motherboard can be updated in real time, and its potential defects can be fed back to help identify problems in advance. Digital twins provide an efficient and non-intrusive way to monitor the status of the motherboard and assess risks.

[0057] More specifically, based on digital twins, the performance of mobile phone motherboards under different thermal and optical environments is simulated and tested. By simulating the potential response of the motherboard under different environmental conditions, such as the impact of temperature rise or light changes on motherboard performance, the test results are evaluated. The evaluation criteria can include multiple aspects such as motherboard performance, quality control, and test accuracy. Through tendency analysis, the most suitable combinations of thermal and optical environments for motherboard testing are determined. This process is repeated multiple times and iteratively to gradually optimize the conditions until the optimal environmental configuration is found. Through multiple iterations of optimization, the optimal thermal and optical environment configuration is obtained to ensure that the mobile phone motherboard can expose potential defects to the greatest extent under these environments. By reasonably optimizing the thermal and optical environments, the test accuracy of the motherboard can be improved to the greatest extent, ensuring the reliability of the test results. Iterative optimization helps to reduce unnecessary environmental configuration adjustments and speeds up the test process.

[0058] More specifically, the thermal and optical environments of the mobile phone motherboard are reconstructed based on optimal environmental configuration information. At this time, the motherboard will be in an optimized testing state, ensuring that all test conditions are most suitable for the motherboard's testing needs. While the motherboard environment is being reconstructed, the machine vision array is adjusted so that the vision system can acquire the clearest and most accurate image data under the new environmental conditions for supplementary testing. Supplementary testing is then performed in the adjusted environment to obtain new testing data, providing a basis for subsequent analysis. By synchronously adjusting the motherboard environment and the machine vision array, the best synergy between the two is ensured, thereby obtaining accurate testing data. By ensuring that the motherboard is tested under optimal environmental conditions, the quality and accuracy of the data are further improved.

[0059] More specifically, attention weights are allocated to the digital twin to determine which regions or features are more likely to have defects or hidden dangers. This step further refines the model output, ensuring that important regions or features receive sufficient attention. Combined with information on the optimal environmental configuration, the defects and hidden dangers of the digital twin are verified. By simulating the performance under different environmental conditions, it is verified which potential defects or hidden dangers are more significant under these conditions. The allocation of attention weights helps to conduct more targeted analysis of the twin, thereby improving the accuracy of defect identification. Through environmental reconstruction and verification, it is ensured that hidden dangers that may occur in the actual environment are effectively identified and corresponding preventive measures are taken.

[0060] More specifically, based on the analysis results of the digital twin and supplementary detection data, the parameters of the pre-trained deep vision detection model are adjusted. This includes modifying hyperparameters and loss functions in the detection algorithm to improve the model's sensitivity to defects and detection accuracy. The adjusted deep vision detection model is then used to perform deep analysis on the supplementary detection data. Through this process, the model can deeply mine potential defect information in the data and generate accurate detection conclusions. Based on the results of the deep analysis, visual inspection conclusions for the mobile phone motherboard are generated. These conclusions will be used for further decision-making, indicating whether defects exist, their nature, location, and potential impact. By adjusting the parameters of the deep model, it is possible to ensure that the model can more effectively identify potential defects in new environments. The detection conclusions after deep analysis can provide data support for subsequent quality control, repair, and optimization.

[0061] Understandably, the entire process combines multiple advanced technologies such as digital twin modeling, environmental optimization, visual array adjustment, and deep learning model tuning to form a comprehensive, automated, and refined motherboard defect detection solution. Through these steps, the motherboard can be deeply inspected in the most suitable environment, and accurate inspection conclusions can be generated through an efficient deep visual inspection model, thereby improving the product's quality assurance capabilities.

[0062] Reference Figure 2 As shown, in a second aspect, the present invention provides an artificial intelligence-based mobile phone motherboard visual inspection system for implementing the artificial intelligence-based mobile phone motherboard visual inspection method described in any one of the first aspects, comprising: An environmental assistance module is used to construct a thermal environment for the mobile phone motherboard to be tested, and to apply an optical environment to the mobile phone motherboard in the thermal environment so that the mobile phone motherboard is in an auxiliary testing state. The preliminary inspection module is used to acquire images of the mobile phone motherboard in the auxiliary inspection state through a machine vision array, and to identify key information in the acquired images based on a lightweight visual inspection model to obtain information on suspicious points on the motherboard. The supplementary detection module is used to adjust the auxiliary detection state and the machine vision array accordingly based on the motherboard suspicious information, so as to perform supplementary detection on the mobile phone motherboard, and perform in-depth analysis of the supplementary detection data according to the depth vision detection model to generate visual detection conclusion information of the mobile phone motherboard.

[0063] In this embodiment, the specific implementation of each module in the above system embodiment is described in the above method embodiment, and will not be repeated here.

[0064] 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 visual inspection method for mobile phone motherboards based on artificial intelligence, characterized in that, include: A thermal environment is constructed for the mobile phone motherboard to be tested, and an optical environment is applied to the mobile phone motherboard in the thermal environment so that the mobile phone motherboard is in an auxiliary testing state. The machine vision array is used to acquire images of the mobile phone motherboard in the auxiliary inspection state, and the key information of the acquired images is identified according to the lightweight visual inspection model to obtain the suspicious information of the motherboard. Based on the motherboard suspicious information, the auxiliary detection state and the machine vision array are adjusted accordingly to perform supplementary detection on the mobile phone motherboard. The supplementary detection data is then deeply analyzed according to the depth vision detection model to generate visual detection conclusion information for the mobile phone motherboard.

2. The artificial intelligence-based visual inspection method for mobile phone motherboards as described in claim 1, characterized in that, The steps of constructing a thermal environment for the mobile phone motherboard to be tested, and applying an optical environment to the motherboard in the thermal environment to put the motherboard into an auxiliary testing state include: The mobile phone motherboard is deployed on the detection and fixing structure, and the mobile phone motherboard is powered on so that the mobile phone motherboard is in the internal heat environment brought about by the power-on operation. The heating module heats the area where the mobile phone motherboard is located, so that the mobile phone motherboard is in an external heat environment brought about by the area heating. The optical environment required by the mobile phone motherboard is analyzed to generate optical environment configuration information; An optical environment conforming to the optical environment configuration information is applied to the mobile phone motherboard, which is in both internal and external thermal environments, so that the mobile phone motherboard is in an auxiliary detection state.

3. The artificial intelligence-based visual inspection method for mobile phone motherboards as described in claim 2, characterized in that, The steps for analyzing the required optical environment of the mobile phone motherboard based on its internal and external thermal environments to generate optical environment configuration information include: Based on the standard performance specifications and power-on processing parameters of the mobile phone motherboard, the internal thermal environment is simulated, and based on the module operating parameters of the heating module, the external thermal environment is simulated to obtain thermal environment simulation information. Based on the thermal environment simulation information, the defect highlighting effect of the mobile phone motherboard under the thermal environment and the image visual mapping relationship are analyzed to obtain the defect highlighting effect characteristics and image visual mapping characteristics of the mobile phone motherboard under the current thermal environment. Based on the image visual mapping features, the defect highlighting effect features are subjected to optical environment adaptive processing to generate defect display optimization parameters and visual distortion compensation parameters for the optical environment of the mobile phone motherboard. Factor synergy value analysis is performed on the defect display optimization parameters and the visual distortion compensation parameters, and several factor synergy working groups are constructed based on the results of the factor synergy value analysis. Optical environment simulation is performed on each of the aforementioned factor collaborative working groups to evaluate the defect display optimization effect and visual distortion compensation effect of each factor collaborative working group under dual conditions. Based on the evaluation results, the factor collaborative working groups are optimized to generate optical environment configuration information.

4. The artificial intelligence-based visual inspection method for mobile phone motherboards as described in claim 3, characterized in that, The steps of performing optical environment adaptive processing on the defect highlighting effect features based on the image visual mapping features to generate defect display optimization parameters and visual distortion compensation parameters for the optical environment of the mobile phone motherboard include: The defect highlighting effect features are vectorized to generate a display enhancement feature matrix; The potential optimization trend of the display enhancement feature matrix is ​​analyzed by combining the image visual mapping features to generate a matrix form that theoretically has a display enhancement effect; The optical environment compensation is analyzed based on the matrix form of the display enhancement feature matrix to achieve the defect display optimization parameters required by the matrix form through optical environment compensation. Based on the image visual mapping features, an optical environment compensation analysis is performed on the visual distortion effect caused by the current thermal environment to obtain visual distortion compensation parameters.

5. The artificial intelligence-based visual inspection method for mobile phone motherboards as described in claim 1, characterized in that, The steps involved in acquiring images of a mobile phone motherboard in an assisted inspection state using a machine vision array, and then identifying key information from the acquired images using a lightweight visual inspection model to obtain information about suspicious points on the motherboard include: The observation position and angle of each machine vision sensor in the machine vision array are adjusted so that each machine vision sensor is at a specified observation angle. Deploy corresponding operating parameters for each machine vision sensor in the machine vision array to drive each machine vision sensor to mark each data part of the multi-dimensional optical data in the form of data acquisition based on a specified observation angle; The multi-dimensional optical data is segmented based on the labels to fill the pre-constructed information representation framework, thereby obtaining multi-dimensional optical data that meets the requirements of model analysis. The multi-dimensional optical data is fed into a pre-trained lightweight visual detection model, which then identifies key information from the multi-dimensional optical data to obtain information about suspicious parts of the motherboard.

6. The artificial intelligence-based visual inspection method for mobile phone motherboards as described in claim 5, characterized in that, During the process of identifying key information from the multi-dimensional optical data, the lightweight visual inspection model incorporates a potential defect assessment mechanism. This mechanism assesses potential defects in the data at various locations within the information representation framework obtained through analysis by the lightweight visual inspection model. It also performs interactive collaborative analysis on the various potential defect assessment results to obtain a collaborative assessment result. This collaborative assessment result is then combined with the key information identified by the lightweight visual inspection model from the multi-dimensional optical data to determine and obtain motherboard-related information that may indicate potential defects on the mobile phone motherboard.

7. The artificial intelligence-based visual inspection method for mobile phone motherboards as described in claim 1, characterized in that, The steps of adjusting the auxiliary detection state and the machine vision array based on the motherboard suspicious information to perform supplementary detection on the mobile phone motherboard, and performing in-depth analysis of the supplementary detection data according to the depth vision detection model to generate visual detection conclusion information of the mobile phone motherboard include: The suspected motherboard information is input into a pre-built digital twin model of the mobile phone motherboard to generate a digital twin that reflects potential motherboard defects. Based on the digital twin, the detection effect under different thermal and optical environments is evaluated. The thermal and optical environments are then optimized in accordance with the trend analysis results through value evaluation of the detection effect and trend analysis of the value evaluation results. This iterative optimization process continues until the optimal environmental configuration information is obtained. Based on the optimal environmental configuration information, the thermal and optical environments of the mobile phone motherboard are reconstructed so that the mobile phone motherboard is in an auxiliary detection state corresponding to the optimal environmental configuration information, and the machine vision array is adjusted simultaneously to perform supplementary detection on the mobile phone motherboard and obtain supplementary detection data. Attention weights are assigned to the digital twin, and the assigned digital twin is analyzed for the form of verification of defects and hidden dangers in combination with the optimal environment configuration information, so as to adjust the parameters of the pre-trained deep vision detection model for the detection task based on the analysis results. The supplementary detection data is analyzed in depth using the depth vision detection model to generate visual detection conclusion information for the mobile phone motherboard.

8. A mobile phone motherboard visual inspection system based on artificial intelligence, characterized in that, A method for implementing the artificial intelligence-based visual inspection method for mobile phone motherboards as described in any one of claims 1-7, comprising: An environmental assistance module is used to construct a thermal environment for the mobile phone motherboard to be tested, and to apply an optical environment to the mobile phone motherboard in the thermal environment so that the mobile phone motherboard is in an auxiliary testing state. The preliminary inspection module is used to acquire images of the mobile phone motherboard in the auxiliary inspection state through a machine vision array, and to identify key information in the acquired images based on a lightweight visual inspection model to obtain information on suspicious points on the motherboard. The supplementary detection module is used to adjust the auxiliary detection state and the machine vision array accordingly based on the motherboard suspicious information, so as to perform supplementary detection on the mobile phone motherboard, and perform in-depth analysis of the supplementary detection data according to the depth vision detection model to generate visual detection conclusion information of the mobile phone motherboard.

Images