Circuit board detection method and system based on artificial intelligence

Abstract

The invention discloses a circuit board detection method and system based on artificial intelligence, and relates to the technical field of circuit board detection, and the method comprises the steps: recording the initial data of each circuit board sample; constructing a three-dimensional digital detection model of the circuit board sample; generating a corresponding loss image of the circuit board sample through a computer graphics technology; obtaining a circuit board sample loss feature image; identifying key factor data of circuit board sample loss; evaluating the severity of the loss of the circuit board sample; and judging whether the circuit board sample meets a preset performance standard or not. By constructing the three-dimensional digital model of the circuit board and performing automatic image analysis, the key factors of the loss of the circuit board sample can be quickly and accurately identified, the detection efficiency and accuracy are greatly improved, and the detection accuracy is greatly improved through automatic and intelligent processing. The labor cost required by manual detection is reduced, the rework or scrap cost caused by detection errors is reduced, and the total production cost can be remarkably reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of circuit board detection, and in particular to an artificial intelligence-based circuit board detection method and system. Background Art

[0002] Printed Circuit Board (PCB) is an indispensable component of electronic products and plays an important role in the electrical connection between electronic components. Because its production process uses special electronic printing technology, it is often called printed circuit board or printed circuit board. The design and quality of the circuit board directly affect the performance and reliability of the final product, so strict quality control during the production process is crucial. Traditionally, the quality inspection of circuit boards mainly relies on manual visual inspection, and the staff needs to carefully observe whether there are solder joint defects, short circuits, open circuits and other problems on the surface of the circuit board. However, this inspection method that relies on the human eye is not only inefficient and costly, but also easily affected by factors such as fatigue, experience and subjective judgment of the inspector, thereby reducing the accuracy and consistency of the inspection results. As electronic products develop towards high density and miniaturization, the structure of circuit boards is becoming increasingly complex, which further increases the difficulty of manual inspection. Therefore, in the field of circuit board inspection, it is particularly important to explore efficient and accurate methods combined with artificial intelligence technology. The use of artificial intelligence, especially machine learning and deep learning technology, combined with high-precision cameras and advanced image processing algorithms, can greatly improve the ability to identify and locate defects on circuit boards.

[0003] During the detection process, existing circuit board detection methods are not convenient for quickly and accurately identifying the key factors of circuit board loss, which greatly reduces the efficiency and accuracy of detection. They are also not convenient for adjustment and learning according to different types of circuit boards and loss conditions, and are not convenient for meeting changing production needs. At the same time, they are not convenient for diagnosing losses that have already occurred, and cannot predict potential loss risks, making it difficult to provide data support for circuit board maintenance and life extension.

[0004] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention proposes a circuit board detection method and system based on artificial intelligence, which solves the problems that the existing key factors for quickly and accurately identifying circuit board loss proposed in the above background technology greatly reduce the efficiency and accuracy of detection, and are not convenient for adjustment and learning according to different types of circuit boards and loss conditions, and are not convenient for meeting changing production needs. At the same time, it is not convenient to diagnose the loss that has already occurred, and it is impossible to predict potential loss risks and provide data support for the maintenance and life extension of circuit boards.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: According to one aspect of the present invention, a circuit board detection method based on artificial intelligence is provided, and the circuit board detection method based on artificial intelligence comprises the following steps: S1, obtaining initial data of the circuit board sample to be tested; S2, constructing a three-dimensional digital detection model of the circuit board sample, and inputting the acquired initial data of the circuit board sample to be tested into the three-dimensional digital detection model; S3. Simulate the loss of the circuit board during processing and use in the three-dimensional digital detection model, and generate the corresponding loss image of the circuit board sample through computer graphics technology; S4, preprocessing the generated circuit board sample loss image to obtain a circuit board sample loss feature image; S5. Based on the analysis algorithm, the obtained circuit board sample loss characteristic image is analyzed with the corresponding circuit board loss characteristic image in the database to identify the key factor data of the circuit board sample loss; S6. Based on the comparison algorithm, the key factor data of the circuit board sample loss identified is compared with the data of various factors affecting the circuit board loss, and the severity of the circuit board sample loss is evaluated; S7. Based on the evaluation results, determine whether the circuit board sample meets the preset performance standards.

[0007] Furthermore, simulating the loss of the circuit board during processing and use in the three-dimensional digital detection model and generating a corresponding loss image of the circuit board sample by computer graphics technology includes the following steps: S31, optimization tool and loss simulation tool for loading a three-dimensional digital inspection model of a circuit board sample; S32, performing simulation analysis on the three-dimensional digital detection model of the circuit board sample according to the initial data of the circuit board sample; S33. If the simulation analysis results show that the circuit board sample structure is unstable under the preset processing and use conditions, optimizing the layout or material properties of the circuit board sample in the three-dimensional digital detection model by using a structure optimization tool; S34, after optimization, re-simulate and analyze, and check whether the circuit board sample structure meets the preset processing and use conditions. If not, repeat S33 to S34 until the circuit board sample structure is stable; S35, performing loss detection on the optimized circuit board sample by using a loss simulation tool. If the detection result shows that the loss situation meets the preset standard, the optimized circuit board sample three-dimensional digital detection model is saved and exported to obtain an optimized model that simulates the loss situation of the circuit board during processing and use; S36. Based on the obtained optimization model, graphics software is used to simulate the loss effect of the circuit board sample under the preset processing and use conditions, and a circuit board sample loss image is generated according to the simulation results.

[0008] Furthermore, if the simulation analysis results show that the circuit board sample structure is unstable under the preset processing and use conditions, optimizing the layout or material properties of the circuit board sample in the three-dimensional digital detection model by the structure optimization tool includes the following steps: S331, running an optimization algorithm in a structural optimization tool for a circuit board sample; S332, determining the size of the design variable set according to the layout or material properties of the circuit board sample, and randomly selecting a set of layout or material properties from the design variable set as an initial design; S333, for each set of design parameters of the circuit board samples, randomly adjust the design variables and update each set of design parameters; S334, evaluating the performance of each group of circuit board sample designs by using simulation analysis tools, and calculating the performance fitness by comparing the performance of the designed circuit board samples under processing and use conditions with the differences under preset conditions; S335, evaluating the pros and cons of each group of circuit board sample designs according to the calculated performance fitness; S336, finding a circuit board design with the best performance, and recording the performance fitness and corresponding design parameters of the design; S337, adjusting the design parameters of other circuit board samples according to the design parameters with the best performance, repeating steps S333 to S336, and continuously optimizing until the performance of the circuit board sample design reaches the preset processing and use conditions, and then terminating the optimization process.

[0009] Furthermore, the preprocessing of the generated circuit board sample loss image to obtain the circuit board sample loss characteristic image includes the following steps: S41, removing noise from the generated circuit board sample loss image by wavelet transform; S42, extracting an image sequence from the denoised circuit board sample loss image, and performing grayscale processing on each frame of the image sequence to obtain a grayscale image; S43, using the Prewitt operator to calculate the edge gradient of each pixel in the grayscale image; S44, using a morphological gradient method to locally screen and enhance edge gradients, and setting a threshold to filter edge gradients to obtain a gradient image; S45, thinning and binarizing the gradient image, and taking the pixel points whose gradient values ​​are greater than the threshold as edge points; S46, connecting adjacent edge points into a connected domain, and obtaining a final edge image as a circuit board sample loss feature image.

[0010] Furthermore, based on the analysis algorithm, the obtained circuit board sample loss characteristic image is analyzed with the corresponding circuit board loss characteristic image in the database to identify the key factor data of the circuit board sample loss, including the following steps: S51, obtaining a circuit board sample loss feature image, and processing the obtained circuit board sample loss feature image by normalization to calculate the normalized value of each feature; S52, assigning corresponding weights to the normalized features according to the importance of the loss features of each circuit board sample; S53, finding the ideal optimal value and the worst value of the loss characteristic of each circuit board sample through the weighted characteristic value, which respectively constitute the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S54, for each circuit board sample loss feature, calculating the grey correlation coefficient between the loss feature and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S55, based on the grey correlation coefficient, respectively calculating the grey correlation degree between each circuit board sample loss characteristic and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S56. Use grey correlation to construct the relative closeness of the ideal circuit board sample loss state to evaluate the advantages and disadvantages of each circuit board sample loss characteristic and identify the key factor data of circuit board sample loss.

[0011] Furthermore, the formula for respectively calculating the grey correlation between each circuit board sample loss feature and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state includes: ; ; in, H represents the grey relational degree; M represents the sum of the loss characteristics of the board samples; b Represents the loss characteristics of a circuit board sample; Expressed as aThe grey correlation between the loss characteristics of a circuit board sample and the ideal circuit board sample loss state; Expressed as a Grey correlation between the hair removal characteristics of towel samples and the loss status of non-ideal circuit board samples.

[0012] Furthermore, the formula for constructing the relative closeness of the ideal circuit board sample loss state is: ; in, Indicates a The relative closeness of the loss characteristics of a circuit board sample to the ideal circuit board sample loss state; T Indicates relative closeness.

[0013] Furthermore, based on the comparison algorithm, the key factor data of the identified circuit board sample loss is compared with the data of various factors affecting the circuit board loss, and the severity of the circuit board sample loss is evaluated, including the following steps: S61, comparing the identified key factor data of the circuit board sample loss with the data of various factors affecting the circuit board loss; S62. Analyze the key factors affecting circuit board loss and set corresponding evaluation levels; S63, training the key factor data of the identified circuit board sample loss through a comparison algorithm, calculating the loss risk score of the circuit board sample, and determining whether it has a high loss risk; S64, normalizing the loss risk scores of the circuit board samples obtained by the comparison algorithm, and calculating the fuzzy set; S65. Professionals evaluate the data of various factors that affect circuit board loss at different loss risk levels to form a loss risk assessment matrix; S66. Integrate and calculate the normalized loss risk score and the loss risk assessment matrix through a fuzzy logic method to obtain a comprehensive loss risk assessment result; S67, judging the relative position of the circuit board sample in different loss risk levels according to the component values ​​of the comprehensive loss risk assessment result, and determining the risk level of the circuit board sample loss; S68. Determine whether the loss risk level of the circuit board sample exceeds a set threshold. If so, mark it as a circuit board sample with a high loss risk level. If not, mark it as a circuit board sample with a low or medium risk level.

[0014] Furthermore, the analysis of the key factor data affecting the circuit board loss and setting the corresponding evaluation level includes the following steps: S621. Collect data on various factors affecting circuit board loss, screen and classify the collected data on various factors, remove duplicate factors, and summarize them into a measurable factor set; S622. Setting a weight for each factor based on actual testing experience, and setting an evaluation level for each factor; S623. Reflect the performance of each factor in the circuit board loss risk assessment according to the set evaluation level.

[0015] According to another aspect of the present invention, there is also provided a circuit board detection system based on artificial intelligence, the circuit board detection system based on artificial intelligence comprising: A test preparation module is used to obtain initial data of the circuit board sample to be tested; A model building module is used to build a three-dimensional digital detection model of a circuit board sample and input the acquired initial data of the circuit board sample to be tested into the three-dimensional digital detection model; An image generation module is used to simulate the wear and tear of the circuit board during processing and use in the three-dimensional digital detection model, and to generate a corresponding wear and tear image of the circuit board sample through computer graphics technology; An image processing module is used to preprocess the generated circuit board sample loss image to obtain a circuit board sample loss feature image; A data analysis module is used to analyze the obtained circuit board sample loss characteristic image with the corresponding circuit board loss characteristic image in the database based on the analysis algorithm, and identify the key factor data of the circuit board sample loss; A data comparison module is used to compare the key factor data of the circuit board sample loss identified with the data of various factors affecting the circuit board loss based on the comparison algorithm, and evaluate the severity of the circuit board sample loss; A performance judgment module, used to judge whether the circuit board sample meets the preset performance standard based on the evaluation result; Among them, the detection preparation module is connected to the image generation module through the model construction module, the image generation module is connected to the data analysis module through the image processing module, and the data analysis module is connected to the performance judgment module through the data comparison module.

[0016] The beneficial effects of the present invention are: 1. The present invention constructs a three-dimensional digital model of the circuit board and performs automated image analysis, so that the key factors of circuit board sample loss can be quickly and accurately identified, greatly improving the efficiency and accuracy of detection. In addition, through automated and intelligent processing, the labor cost required for manual detection and the rework or scrap cost caused by detection errors can be reduced, which can significantly reduce the total production cost. By using the digital model for testing, a large number of simulations can be performed on multiple samples in a short period of time, greatly improving the test efficiency. At the same time, by comparing the key factor data of the identified circuit board sample loss with the data of various factors affecting the circuit board loss, the severity of the circuit board sample loss can be evaluated and identified. This can not only identify the current loss problem, but also predict possible future loss trends, which will help optimize the design and production of circuit boards.

[0017] 2. The present invention simulates the wear and loss of the circuit board during processing and use in the three-dimensional digital model of the circuit board sample test, so that by using the circuit design optimization tool, the layout and material properties of the circuit board sample can be adjusted in detail to achieve more efficient and accurate design optimization. In addition, through simulation analysis tools and optimization algorithms, the performance of the circuit board sample can be continuously evaluated and improved, which not only improves the durability of the circuit board, but also ensures its stability under various environmental conditions. At the same time, by continuously adjusting the design parameters and evaluating their performance adaptability, iterative optimization is achieved, so that the best design solution can be quickly found to ensure that the circuit board sample reaches the best performance state. Not only does it enhance the performance and reliability of the circuit board, but it also promotes the efficiency of the product development process and improves the overall quality of circuit board design and production.

[0018] 3. The present invention uses normalization processing and weighting to comprehensively consider the influence of various loss characteristics, accurately identify and evaluate the key factors of circuit board sample loss through grey correlation analysis, and improve the accuracy of loss diagnosis. By identifying the key factors of loss and calculating the pros and cons of the loss characteristics of each circuit board sample, it can provide important data support for the design optimization of the circuit board, and by accurately identifying the key factors of loss, it can improve the production process and material selection in a targeted manner, reduce rework and scrap rates, thereby improving production efficiency and reducing costs.

[0019] 4. The present invention compares the identified key factors with the known loss factors, so that the main factors affecting the circuit board loss can be more accurately identified and evaluated, thereby improving the accuracy of loss risk identification. By setting different loss risk evaluation levels and using fuzzy logic methods, the loss risk level of the circuit board sample can be more carefully evaluated, providing a more detailed basis for subsequent risk management and decision-making. At the same time, by comprehensively analyzing and evaluating various factors affecting circuit board loss, the company's risk management capabilities are enhanced, which helps prevent potential problems and ensure production continuity and product stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0021] Figure 1 is a flow chart of a circuit board detection method based on artificial intelligence according to an embodiment of the present invention; Figure 2 The figure is a principle block diagram of a circuit board detection system based on artificial intelligence according to an embodiment of the present invention.

[0022] In the figure: 1. Detection preparation module; 2. Model building module; 3. Image generation module; 4. Image processing module; 5. Data analysis module; 6. Data comparison module; 7. Performance judgment module. DETAILED DESCRIPTION

[0023] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0024] In the description of the present invention, unless otherwise specified, the meaning of "plurality" is two or more. In addition, the terms "first", "second", "third", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0025] According to an embodiment of the present invention, a circuit board detection method and system based on artificial intelligence are provided.

[0026] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1As shown, according to an embodiment of the present invention, the circuit board detection method based on artificial intelligence includes the following steps: S1, obtaining initial data of the circuit board sample to be tested; Specifically, the initial data includes: Circuit board number or identification information: In order to ensure that each sample can be accurately tracked during the testing process, each circuit board sample should have a unique number or identification information.

[0027] Manufacturing date and batch: Recording the manufacturing date and batch can help analyze possible batch-related problems and facilitate traceability and quality control.

[0028] Material composition: Record the main materials used in the circuit board in detail, including substrate materials, conductive materials, insulating layer materials, etc., because these factors may affect the loss of the circuit board during use.

[0029] PCB size and thickness: The size and thickness of the PCB will affect its physical properties, including bending, fracture resistance, etc., so this is an important data to record before testing.

[0030] Layout and Component Layout: Documenting the layout and component layout of the board helps in understanding possible failure points and loss areas when analyzing losses.

[0031] Intended use environment: It is critical to record the environment and conditions (such as temperature, humidity, vibration, etc.) in which the circuit board is expected to be used, as these factors directly affect the performance and durability of the circuit board in actual application.

[0032] Functional and performance requirements: Record the functional requirements and performance indicators of the circuit board in detail, including but not limited to parameters such as current, voltage, frequency, etc., which will help to subsequently evaluate whether the circuit board meets the expected usage requirements.

[0033] Historical repair and replacement records: If the sample circuit board has a history of repair or component replacement, this information should also be recorded as it may have a significant impact on the current performance status of the circuit board.

[0034] Visual inspection results: Perform a visual inspection of the circuit board before testing and record any appearance defects that may affect performance, such as cracks, scratches or corrosion.

[0035] S2, constructing a three-dimensional digital detection model of the circuit board sample, and inputting the acquired initial data of the circuit board sample to be tested into the three-dimensional digital detection model; Specifically, a three-dimensional digital detection model of a circuit board sample is constructed by computer-aided design software (ie, CAD software).

[0036] Specifically, select appropriate CAD software according to your needs, such as AutoCAD, Altium Designer, Eagle, KiCad, etc., then set the basic parameters of the circuit board in the software (these parameters are based on the measured data of the actual circuit board sample), add appropriate wiring and component layout to the model to make it as close as possible to the design and appearance of the actual circuit board, including the number of layers, material type (such as FR-4, copper foil, etc.), size and shape, add more details such as pads, through holes, etc. as needed to improve the authenticity of the model, and input the recorded initial parameters of the circuit board sample into the model, convert all two-dimensional graphics into three-dimensional models, and perform various simulation tests on the three-dimensional model, such as electromagnetic compatibility (EMC) analysis and thermal analysis (simulation of heat dissipation performance), to evaluate the performance of the circuit board.

[0037] S3. Simulate the loss of the circuit board during processing and use in the three-dimensional digital detection model, and generate the corresponding loss image of the circuit board sample through computer graphics technology; Specifically, computer graphics technology refers to the technology of using computer technology to generate, process, analyze and display images, especially three-dimensional images. Computer graphics technology plays a key role in simulating the loss of circuit board samples and generating corresponding images.

[0038] Specifically, the use of computer graphics technology to generate a circuit board sample loss image includes: Create a detailed 3D model of the board in CAD software, ensuring that the model captures the wiring, component placement, and any microstructure of the board.

[0039] Use physical simulation software (such as ANSYS, Autodesk Simulation) to simulate the wear phenomenon of the circuit board under use and environmental stress (such as temperature change, vibration, humidity). This includes applying different environmental and mechanical parameters, as well as simulating chemical reactions or physical influences that may cause circuit board wear.

[0040] After obtaining the simulation data, the data is processed using computer graphics software such as Blender, Maya, and 3ds Max. These software can create high-quality visual effects to help visualize the losses of the circuit board.

[0041] To make the image more realistic, add material properties and appropriate lighting effects in computer graphics software. Material properties can help simulate the surface details and loss effects of the circuit board and its components, while lighting can enhance the three-dimensionality and realism of the image.

[0042] Use the software's rendering function to convert simulation results into high-resolution images. The rendering process generates a final image (i.e., a circuit board sample loss image) based on factors such as the model, material properties, lighting, and photography angle. This step is critical because it not only provides researchers and engineers with an intuitive visual reference to evaluate the circuit board's loss and potential failure areas, but also provides guidance for improvement plans during the design and testing phases. With this highly detailed and real-time visual display, potential weaknesses in circuit board design can be more effectively identified and addressed, thereby improving the reliability and performance of the final product.

[0043] S4, preprocessing the generated circuit board sample loss image to obtain a circuit board sample loss feature image; Specifically, the circuit board sample loss feature image includes: Identification of wear areas: Identify and highlight wear areas in the board using image processing techniques such as edge detection and segmentation. This can help identify areas of damage caused by overheating, corrosion or mechanical stress.

[0044] Quantification of loss: Quantify the loss by analyzing the size, density and distribution of the loss area. This can be achieved by calculating the ratio of the loss area to the entire board area, analyzing the number of broken wires or measuring the thickness of the corrosion layer.

[0045] Texture and Color Analysis: Analyze texture and color changes in lossy areas to better understand how losses affect the appearance and function of the board. For example, corrosion may change the color of metal traces, while thermal stress may cause texture changes in the substrate material.

[0046] Analysis of loss patterns: Identify loss patterns and trends, such as whether there are specific areas that tend to lose, or whether the losses show a certain shape or distribution. This helps identify potential causes of loss and provides guidance for design improvements to the board.

[0047] S5. Based on the analysis algorithm, the obtained circuit board sample loss characteristic image is analyzed with the corresponding circuit board loss characteristic image in the database to identify the key factor data of the circuit board sample loss; S6. Based on the comparison algorithm, the key factor data of the circuit board sample loss identified is compared with the data of various factors affecting the circuit board loss, and the severity of the circuit board sample loss is evaluated; Specifically, key factor data include: Circuit board material properties: such as substrate material type (FR-4, CEM, etc.), conductive material (thickness and type of copper layer), type and characteristics of insulating material, etc.

[0048] Circuit board structural parameters: including number of layers, line width, line spacing, through-hole size and layout, etc.

[0049] Circuit board manufacturing quality: solder joint quality, multi-layer board alignment accuracy, cutting and drilling accuracy, etc.

[0050] The board’s usage history: This includes the number of temperature cycles it has been subjected to, the history of mechanical stresses (such as bending or vibration), chemical corrosion or humidity exposure, etc.

[0051] Circuit design parameters: current density, thermal design, electromagnetic compatibility design, etc.

[0052] The factors affecting circuit board loss include: Environmental factors: temperature, humidity, chemically corrosive environment, salt spray environment, ultraviolet (UV) exposure, etc.

[0053] Mechanical factors: mechanical vibration, impact, bending stress, etc.

[0054] Electrical factors: overload, voltage fluctuation, excessive current density, electromagnetic interference (EMI), etc.

[0055] Time factors: the age of the circuit board, frequency of maintenance and overhaul, etc.

[0056] Human factors: installation errors, improper maintenance, use conditions that do not meet specifications, etc.

[0057] S7. Based on the evaluation results, determine whether the circuit board sample meets the preset performance standards.

[0058] Specifically, the evaluation results are obtained through testing and analysis, including but not limited to electrical testing (such as continuity testing, insulation resistance testing), thermal testing (such as thermal cycle testing), mechanical testing (such as bending testing), environmental stability testing (such as salt spray testing), etc. The data generated by these tests reflect the actual performance of the circuit board in various performance parameters, which can be in the form of numerical values ​​(such as resistance value, temperature rise value), images (such as loss characteristic images), and scores or grades obtained through algorithm processing.

[0059] Specifically, based on the evaluation results, judging whether the circuit board sample meets the preset performance standards includes: Numerical matching: For the evaluation results in numerical form, directly compare them with the numerical requirements in the performance standard to determine whether they are within the allowable range. For example, if the resistance value test result is within the allowable range, the item is judged to be qualified.

[0060] Image and feature matching: For evaluation results in image form, such as loss feature images, key features need to be extracted through image analysis techniques (such as edge detection and morphological analysis) and compared with standard features (such as feature images under ideal conditions) to determine whether the severity of the loss or defect exceeds the allowable threshold.

[0061] Comprehensive scoring and grade matching: For the scores or grades obtained through algorithm processing, these scores or grades are directly compared with the preset qualification standards to determine whether the circuit board meets the requirements.

[0062] Comprehensive judgment: Based on all the above test results, each performance parameter is judged. If all important performance parameters meet or exceed the performance standards, the circuit board sample is considered to meet the preset performance standards. On the contrary, if any key performance parameters do not meet the requirements, the circuit board sample may be judged as unqualified and requires further repair or optimization.

[0063] Specifically, the performance standards include: Electrical performance standards: including the circuit board's conductivity, insulation performance, resistance, capacitance, inductance characteristics, as well as signal integrity and noise level.

[0064] Thermal performance criteria: These involve the thermal stability and heat dissipation capabilities of the board, ensuring that heat loads can be effectively managed during operation to avoid overheating damage.

[0065] Mechanical performance standards: including the strength, hardness, flexibility, etc. of the circuit board to ensure that it will not be damaged under mechanical stress (such as vibration, impact, bending).

[0066] Chemical and Environmental Stability Standards: Evaluate how the board performs in corrosive environments to ensure that corrosion resistance is as expected, as well as resistance to environmental factors such as humidity, salt spray, and contaminants.

[0067] Reliability and life standards: Involving the performance of the circuit board during its expected life cycle, including durability and anti-aging performance.

[0068] Manufacturing and assembly standards: Ensure the manufacturing quality of the circuit board, including solder joint integrity, inter-layer alignment accuracy, defect-free surface treatment, etc.

[0069] Safety and compliance standards: Ensure that the circuit board complies with relevant safety and environmental regulations, such as RoHS (lead-free), REACH, etc.

[0070] Preferably, simulating the wear and tear of the circuit board during processing and use in the three-dimensional digital detection model and generating a corresponding wear and tear image of the circuit board sample by computer graphics technology comprises the following steps: S31, optimization tool and loss simulation tool for loading a three-dimensional digital inspection model of a circuit board sample; Specifically, circuit layout optimization tools are used to simulate and optimize the wiring structure of the circuit board to achieve the predetermined performance goals. It involves calculating and adjusting the parameters such as the routing, density, width, and hierarchy of the wires, as well as the layout and spacing of components to improve the electrical performance of the circuit board, reduce interference, and enhance thermal management efficiency.

[0071] Specifically, the loss simulation tool is a tool specially designed to simulate the loss phenomena (such as corrosion, fracture, overheating, etc.) that occur during the use of circuit boards. It simulates the loss of circuit boards based on various assumptions and parameters, such as the temperature resistance, strength, thermal expansion coefficient, electrical conductivity, etc. of the material, as well as external conditions such as temperature and humidity, mechanical stress, chemical erosion and other factors.

[0072] S32, performing simulation analysis on the three-dimensional digital detection model of the circuit board sample according to the initial data of the circuit board sample; S33. If the simulation analysis results show that the circuit board sample structure is unstable under the preset processing and use conditions, optimizing the layout or material properties of the circuit board sample in the three-dimensional digital detection model by using a structure optimization tool; S34, after optimization, re-simulate and analyze, and check whether the circuit board sample structure meets the preset processing and use conditions. If not, repeat S33 to S34 until the circuit board sample structure is stable; S35, performing loss detection on the optimized circuit board sample by using a loss simulation tool. If the detection result shows that the loss situation meets the preset standard, the optimized circuit board sample three-dimensional digital detection model is saved and exported to obtain an optimized model that simulates the loss situation of the circuit board during processing and use; S36. Based on the obtained optimization model, graphics software is used to simulate the loss effect of the circuit board sample under the preset processing and use conditions, and a circuit board sample loss image is generated according to the simulation results.

[0073] Preferably, if the simulation analysis results show that the circuit board sample structure is unstable under the preset processing and use conditions, optimizing the layout or material properties of the circuit board sample in the three-dimensional digital detection model by the structure optimization tool includes the following steps: S331, running an optimization algorithm in a structural optimization tool for a circuit board sample; Specifically, the optimization algorithm is the Fruit Fly Optimization Algorithm (FOA), which is a new global optimization evolutionary algorithm derived from the evolution of fruit fly foraging behavior. Fruit flies have superior sense of smell and vision than other species. Fruit fly groups effectively collect various odors floating in the air through their sense of smell, and move toward food. Then they use their keen vision to distinguish the location of food and companions, gather in locations with high concentrations of food smell, and then use their sense of smell to fly out in random directions to find food, and repeat this cycle until they find food.

[0074] S332, determining the size of the design variable set (i.e., the size of the fruit fly population) according to the layout or material properties of the circuit board sample, and randomly selecting a set of layouts or material properties from the design variable set as an initial design (i.e., initialization of the fruit fly population); S333, for each set of design parameters of circuit board samples (i.e., individual positions of fruit flies), randomly adjust the design variables (i.e., random search of individual fruit flies), and update each set of design parameters (i.e., update individual positions of fruit flies); S334. Evaluate the performance of each group of circuit board sample designs by using simulation analysis tools (i.e., finite element analysis tools), and calculate the performance fitness (i.e., the individual fruit fly flavor concentration judgment value) by comparing the performance of the designed circuit board samples under processing and use conditions with the differences in preset conditions; S335, evaluating the pros and cons of each group of circuit board sample designs according to the calculated performance fitness; S336, finding the circuit board design with the best performance (i.e., the optimal fruit fly individual), and recording the performance fitness and corresponding design parameters of the design; S337, adjusting the design parameters of other circuit board samples according to the design parameters with the best performance, repeating steps S333 to S336, and continuously optimizing until the performance of the circuit board sample design reaches the preset processing and use conditions, and then terminating the optimization process.

[0075] Preferably, the preprocessing of the generated circuit board sample loss image to obtain the circuit board sample loss characteristic image comprises the following steps: S41, removing noise from the generated circuit board sample loss image by wavelet transform; Specifically, wavelet transform is a powerful mathematical tool for signal processing and image analysis, especially in the fields of noise removal, signal compression and feature extraction. It can provide a localized analysis of the time (or space) and frequency of a signal or image, and can analyze the local area of ​​the signal at different scales, thereby capturing the details of various frequency components.

[0076] S42, extracting an image sequence from the denoised circuit board sample loss image, and performing grayscale processing on each frame of the image sequence to obtain a grayscale image; S43, using the Prewitt operator to calculate the edge gradient of each pixel in the grayscale image; Specifically, the Prewitt operator is an operator in image processing technology used for edge detection. It is mainly used to calculate the edge gradient of each pixel in the image in order to identify the edge part of the image. It achieves this goal by performing a convolution operation on the image. Specifically, it uses two 3x3 kernels (or filters) to calculate the gradient of the image in the horizontal and vertical directions respectively.

[0077] S44, using a morphological gradient method to locally screen and enhance edge gradients, and setting a threshold to filter edge gradients to obtain a gradient image; Specifically, the morphological gradient method is a technique based on mathematical morphology in image processing, which is used to highlight the edges or object boundaries in an image. This method can enhance the edges in the image by processing the shape and structure of the image, making the edges clearer, which helps image analysis and object recognition. Morphological operations rely on structural elements (a predefined shape) to detect corresponding shapes in the image.

[0078] S45, thinning and binarizing the gradient image, and taking the pixel points whose gradient values ​​are greater than the threshold as edge points; S46, connecting adjacent edge points into a connected domain, and obtaining a final edge image as a circuit board sample loss feature image.

[0079] Preferably, the method of analyzing the obtained circuit board sample loss characteristic image with the corresponding circuit board loss characteristic image in the database based on the analysis algorithm to identify the key factor data of the circuit board sample loss includes the following steps: S51, obtaining a circuit board sample loss feature image, and processing the obtained circuit board sample loss feature image by normalization to calculate the normalized value of each feature; Specifically, normalization is a data preprocessing method that aims to adjust the range of numerical data so that it is within a specific interval, such as (0, 1) or (-1, 1), so that data of different dimensions or magnitudes can be compared and analyzed. In image processing and machine learning, normalization can help improve the convergence speed and accuracy of the algorithm, and also prevent certain features from having too much impact on the model due to their large numerical range.

[0080] S52, assigning corresponding weights to the normalized features according to the importance of the loss features of each circuit board sample; S53, finding the ideal optimal value and the worst value of the loss characteristic of each circuit board sample through the weighted characteristic value, which respectively constitute the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S54, for each circuit board sample loss feature, calculating the grey correlation coefficient between the loss feature and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S55, based on the grey correlation coefficient, respectively calculating the grey correlation degree between each circuit board sample loss characteristic and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S56. Use grey correlation to construct the relative closeness of the ideal circuit board sample loss state to evaluate the advantages and disadvantages of each circuit board sample loss characteristic and identify the key factor data of circuit board sample loss.

[0081] Specifically, the analysis algorithm is the TOP-SIS algorithm based on grey correlation, which is a decision analysis algorithm that combines grey correlation analysis and TOPSIS method. Grey correlation analysis is an analysis method used to process incomplete information and is used for system analysis of small samples and incomplete information. It determines the similarity between factors by calculating the grey correlation between each factor. TOPSIS method is a multi-attribute decision analysis method.

[0082] Specifically, the TOPSIS algorithm based on grey correlation combines the advantages of these two methods. It first uses grey correlation analysis to calculate the correlation between each health indicator (i.e., the loss characteristics of each circuit board sample) and the ideal health state (i.e., no loss) and the non-ideal health state (i.e., loss), and then uses the TOPSIS method to sort each health indicator according to the correlation.

[0083] Preferably, the formula for respectively calculating the grey correlation between each circuit board sample wear characteristic and the ideal circuit board sample wear state and the non-ideal circuit board sample wear state comprises: ; ; in, H represents the grey relational degree; M represents the sum of the loss characteristics of the board samples; b Represents the loss characteristics of a circuit board sample; Expressed as a The grey correlation between the loss characteristics of a circuit board sample and the ideal circuit board sample loss state; Expressed as aGrey correlation between the hair removal characteristics of towel samples and the loss status of non-ideal circuit board samples.

[0084] Preferably, the formula for constructing the relative closeness of the ideal circuit board sample wear state is: ; in, Indicates a The relative closeness of the loss characteristics of a circuit board sample to the ideal circuit board sample loss state; T Indicates relative closeness.

[0085] Preferably, the method of comparing the identified key factor data of the circuit board sample loss with the data of various factors affecting the circuit board loss based on the comparison algorithm and evaluating the severity of the circuit board sample loss includes the following steps: S61, comparing the identified key factor data of the circuit board sample loss with the data of various factors affecting the circuit board loss; S62. Analyze the key factors affecting circuit board loss and set corresponding evaluation levels; S63, training the key factor data of the identified circuit board sample loss through a comparison algorithm, calculating the loss risk score of the circuit board sample, and determining whether it has a high loss risk; Specifically, the following steps are included to train the key factor data of the identified circuit board sample loss through the comparison algorithm, calculate the loss risk score of the circuit board sample, and determine whether it has a high loss risk: S631, dividing the extracted key factor data of circuit board sample loss into a training set and a test set; S632, using the training set data, training the isolation forest model by randomly selecting split points within the range of features and feature values; S633, construct multiple isolated trees to form a complete isolation forest model; S63. Using the isolation forest model, calculate the average path length from the root node to the leaf node of each circuit board sample data point in the test set; S635. Calculate a risk score for each circuit board sample based on the average path length, so as to measure the loss risk of the circuit board sample; S636. Based on the set threshold, use the risk score to determine whether each circuit board sample is in a high-risk loss state.

[0086] Specifically, during the training process, the data points are divided into two subsets according to the selected eigenvalue (one containing data points less than or equal to the eigenvalue, and the other containing data points greater than the eigenvalue), and a split operation is performed; this process is recursively repeated for each subset until the stopping condition is met (such as the subset size reaches a predetermined threshold and the depth of the tree reaches the maximum value).

[0087] S64, normalizing the loss risk scores of the circuit board samples obtained by the comparison algorithm, and calculating the fuzzy set; S65. Professionals evaluate the data of various factors that affect circuit board loss at different loss risk levels to form a loss risk assessment matrix; S66. Integrate and calculate the normalized loss risk score (weight vector) and the loss risk assessment matrix (fuzzy relationship matrix) through a fuzzy logic method (fuzzy operation operator) to obtain a comprehensive loss risk assessment result; S67, judging the relative position of the circuit board sample in different loss risk levels according to the component values ​​of the comprehensive loss risk assessment result, and determining the risk level of the circuit board sample loss; S68. Determine whether the loss risk level of the circuit board sample exceeds a set threshold. If so, mark it as a circuit board sample with a high loss risk level. If not, mark it as a circuit board sample with a low or medium risk level.

[0088] Specifically, the comparison algorithm is a fuzzy isolation forest algorithm, which is a fuzzy improved algorithm based on the isolation forest. Isolation forest is a very effective anomaly detection method. Its core idea is to use a binary tree structure to isolate data points, in which the positions of normal data and abnormal data in the tree are significantly different, and abnormal data are usually isolated earlier. In the present invention, the purpose is to normalize the obtained anomaly scores and construct fuzzy sets to deal with the uncertainty and ambiguity in the circuit board data. This is particularly useful in circuit board defect identification and loss assessment, because fuzzy sets can be used to more accurately describe the loss level and possible abnormal state of the circuit board.

[0089] Preferably, the analyzing of the key factor data affecting the circuit board loss and setting the corresponding evaluation level comprises the following steps: S621. Collect data on various factors affecting circuit board loss, screen and classify the collected data on various factors, remove duplicate factors, and summarize them into a measurable factor set; S622. Setting a weight for each factor based on actual testing experience, and setting an evaluation level for each factor; S623. Reflect the performance of each factor in the circuit board loss risk assessment according to the set evaluation level.

[0090] According to another embodiment of the present invention, Figure 2 As shown, a circuit board detection system based on artificial intelligence, the circuit board detection system based on artificial intelligence includes: The test preparation module 1 is used to obtain the initial data of the circuit board sample to be tested; Model building module 2, used to build a three-dimensional digital detection model of the circuit board sample, and input the acquired initial data of the circuit board sample to be tested into the three-dimensional digital detection model; An image generation module 3 is used to simulate the wear and tear of the circuit board during processing and use in the three-dimensional digital detection model, and to generate a corresponding wear and tear image of the circuit board sample through computer graphics technology; An image processing module 4 is used to preprocess the generated circuit board sample loss image to obtain a circuit board sample loss feature image; The data analysis module 5 is used to analyze the obtained circuit board sample loss characteristic image with the corresponding circuit board loss characteristic image in the database based on the analysis algorithm, and identify the key factor data of the circuit board sample loss; The data comparison module 6 is used to compare the key factor data of the circuit board sample loss identified with the data of various factors affecting the circuit board loss based on the comparison algorithm, and evaluate the severity of the circuit board sample loss; The performance judgment module 7 is used to judge whether the circuit board sample meets the preset performance standard based on the evaluation result; Among them, the detection preparation module 1 is connected to the image generation module 3 through the model construction module 2, the image generation module 3 is connected to the data analysis module 5 through the image processing module 4, and the data analysis module 5 is connected to the performance judgment module 7 through the data comparison module 6.

[0091] In summary, with the help of the above technical solution of the present invention, the present invention simulates the wear and loss of the circuit board during processing and use in the three-dimensional digital model of the circuit board sample test, so that by using the circuit design optimization tool, the layout and material properties of the circuit board sample can be adjusted in detail to achieve more efficient and accurate design optimization. In addition, through simulation analysis tools and optimization algorithms, the performance of the circuit board sample can be continuously evaluated and improved, which not only improves the durability of the circuit board, but also ensures its stability under various environmental conditions. At the same time, by continuously adjusting the design parameters and evaluating their performance adaptability, iterative optimization is achieved, so that the best design solution can be quickly found to ensure that the circuit board sample reaches the best performance state. Not only does it enhance the performance and reliability of the circuit board, but it also promotes the efficiency of the product development process and improves the overall quality of circuit board design and production. The present invention, through normalization processing and weighting, can comprehensively consider the influence of each loss feature, accurately identify and evaluate the key factors of circuit board sample loss through grey correlation analysis, improve the accuracy of loss diagnosis, and provide important data support for the design optimization of the circuit board by identifying the key factors of loss and calculating the advantages and disadvantages of each circuit board sample loss feature. And by accurately identifying the key factors of loss, the production process and material selection can be improved in a targeted manner, rework and scrap rate can be reduced, thereby improving production efficiency and reducing costs. The present invention, by comparing the identified key factors with the known loss factors, can more accurately identify and evaluate the main factors affecting the loss of the circuit board, thereby improving the accuracy of loss risk identification, and by setting different loss risk evaluation levels and using fuzzy logic methods, it can more carefully evaluate the loss risk level of the circuit board sample, providing a more detailed basis for subsequent risk management and decision-making. At the same time, by comprehensively analyzing and evaluating various factors affecting the loss of the circuit board, the ability of the enterprise in risk management is enhanced, which helps to prevent potential problems and ensure the continuity of production and the stability of the product.

[0092] 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, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A circuit board detection method based on artificial intelligence, characterized in that: The circuit board detection method based on artificial intelligence includes the following steps: S1, obtaining initial data of the circuit board sample to be tested; S2, constructing a three-dimensional digital detection model of the circuit board sample, and inputting the acquired initial data of the circuit board sample to be tested into the three-dimensional digital detection model; S3. Simulate the loss of the circuit board during processing and use in the three-dimensional digital detection model, and generate the corresponding loss image of the circuit board sample through computer graphics technology; S4, preprocessing the generated circuit board sample loss image to obtain a circuit board sample loss feature image; S5. Based on the analysis algorithm, the obtained circuit board sample loss characteristic image is analyzed with the corresponding circuit board loss characteristic image in the database to identify the key factor data of the circuit board sample loss; S6. Based on the comparison algorithm, the key factor data of the circuit board sample loss identified is compared with the data of various factors affecting the circuit board loss, and the severity of the circuit board sample loss is evaluated; S7. Based on the evaluation results, determine whether the circuit board sample meets the preset performance standards.

2. The circuit board detection method based on artificial intelligence according to claim 1, characterized in that: The method of simulating the loss of a circuit board during processing and use in a three-dimensional digital detection model and generating a corresponding loss image of a circuit board sample by computer graphics technology comprises the following steps: S31, optimization tool and loss simulation tool for loading a three-dimensional digital inspection model of a circuit board sample; S32, performing simulation analysis on the three-dimensional digital detection model of the circuit board sample according to the initial data of the circuit board sample; S33. If the simulation analysis results show that the circuit board sample structure is unstable under the preset processing and use conditions, optimizing the layout or material properties of the circuit board sample in the three-dimensional digital detection model by using a structure optimization tool; S34, after optimization, re-simulate and analyze, and check whether the circuit board sample structure meets the preset processing and use conditions. If not, repeat S33 to S34 until the circuit board sample structure is stable; S35, performing loss detection on the optimized circuit board sample by using a loss simulation tool. If the detection result shows that the loss situation meets the preset standard, the optimized circuit board sample three-dimensional digital detection model is saved and exported to obtain an optimized model that simulates the loss situation of the circuit board during processing and use; S36. Based on the obtained optimization model, graphics software is used to simulate the loss effect of the circuit board sample under the preset processing and use conditions, and a circuit board sample loss image is generated according to the simulation results.

3. The circuit board detection method based on artificial intelligence according to claim 2 is characterized in that: If the simulation analysis results show that the circuit board sample structure is unstable under the preset processing and use conditions, optimizing the layout or material properties of the circuit board sample in the three-dimensional digital detection model by the structure optimization tool includes the following steps: S331, running an optimization algorithm in a structural optimization tool for a circuit board sample; S332, determining the size of the design variable set according to the layout or material properties of the circuit board sample, and randomly selecting a set of layout or material properties from the design variable set as an initial design; S333, for each set of design parameters of the circuit board samples, randomly adjust the design variables and update each set of design parameters; S334, evaluating the performance of each group of circuit board sample designs by using simulation analysis tools, and calculating the performance fitness by comparing the performance of the designed circuit board samples under processing and use conditions with the differences under preset conditions; S335, evaluating the pros and cons of each group of circuit board sample designs according to the calculated performance fitness; S336, finding the circuit board design with the best performance, and recording the performance fitness and corresponding design parameters of the design; S337, adjusting the design parameters of other circuit board samples according to the design parameters with the best performance, repeating steps S333 to S336, and continuously optimizing until the performance of the circuit board sample design reaches the preset processing and use conditions, and then terminating the optimization process.

4. The circuit board detection method based on artificial intelligence according to claim 1, characterized in that: The preprocessing of the generated circuit board sample loss image to obtain the circuit board sample loss characteristic image comprises the following steps: S41, removing noise from the generated circuit board sample loss image by wavelet transform; S42, extracting an image sequence from the denoised circuit board sample loss image, and graying each frame of the image sequence to obtain a grayscale image; S43, using the Prewitt operator to calculate the edge gradient of each pixel in the grayscale image; S44, using a morphological gradient method to locally screen and enhance edge gradients, and setting a threshold to filter edge gradients to obtain a gradient image; S45, thinning and binarizing the gradient image, and taking the pixel points whose gradient values ​​are greater than the threshold as edge points; S46, connecting adjacent edge points into a connected domain, and obtaining a final edge image as a circuit board sample loss feature image.

5. The circuit board detection method based on artificial intelligence according to claim 1 is characterized in that: Based on the analysis algorithm, the obtained circuit board sample loss characteristic image is analyzed with the corresponding circuit board loss characteristic image in the database to identify the key factor data of the circuit board sample loss, including the following steps: S51, obtaining a circuit board sample loss feature image, and processing the obtained circuit board sample loss feature image by normalization to calculate the normalized value of each feature; S52, assigning corresponding weights to the normalized features according to the importance of the loss features of each circuit board sample; S53, finding the ideal optimal value and the worst value of the loss characteristic of each circuit board sample through the weighted characteristic value, which respectively constitute the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S54, for each circuit board sample loss feature, calculating the grey correlation coefficient between the loss feature and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S55, based on the grey correlation coefficient, respectively calculating the grey correlation degree between each circuit board sample loss characteristic and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state; S56. Use grey correlation to construct the relative closeness of the ideal circuit board sample loss state to evaluate the advantages and disadvantages of each circuit board sample loss characteristic and identify the key factor data of circuit board sample loss.

6. The circuit board detection method based on artificial intelligence according to claim 5, characterized in that: The formula for respectively calculating the grey correlation between each circuit board sample loss feature and the ideal circuit board sample loss state and the non-ideal circuit board sample loss state includes: ； ； in, H represents the grey relational degree; M represents the sum of the loss characteristics of the board samples; b Represents the loss characteristics of a circuit board sample; Expressed as a The grey correlation between the loss characteristics of a circuit board sample and the ideal circuit board sample loss state; Expressed as a Grey correlation between the hair removal characteristics of towel samples and the loss status of non-ideal circuit board samples.

7. The circuit board detection method based on artificial intelligence according to claim 6 is characterized in that: The formula for constructing the relative closeness of the ideal circuit board sample loss state is: ； in, Indicates a The relative closeness of the loss characteristics of a circuit board sample to the ideal circuit board sample loss state; T Indicates relative closeness.

8. The circuit board detection method based on artificial intelligence according to claim 1, characterized in that: The method of comparing the key factor data of the circuit board sample loss identified with the data of various factors affecting the circuit board loss based on the comparison algorithm and evaluating the severity of the circuit board sample loss includes the following steps: S61, comparing the identified key factor data of the circuit board sample loss with the data of various factors affecting the circuit board loss; S62. Analyze the key factors affecting circuit board loss and set corresponding evaluation levels; S63, training the key factor data of the identified circuit board sample loss through a comparison algorithm, calculating the loss risk score of the circuit board sample, and determining whether it has a high loss risk; S64, normalizing the loss risk scores of the circuit board samples obtained by the comparison algorithm, and calculating the fuzzy set; S65. Professionals evaluate the data of various factors that affect circuit board loss at different loss risk levels to form a loss risk assessment matrix; S66. Integrate and calculate the normalized loss risk score and the loss risk assessment matrix through a fuzzy logic method to obtain a comprehensive loss risk assessment result; S67. Determine the relative position of the circuit board sample in different loss risk levels according to the component values ​​of the comprehensive loss risk assessment result, and determine the risk level of the circuit board sample loss; S68. Determine whether the loss risk level of the circuit board sample exceeds a set threshold. If so, mark it as a circuit board sample with a high loss risk level. If not, mark it as a circuit board sample with a low or medium risk level.

9. The circuit board detection method based on artificial intelligence according to claim 8, characterized in that: The analysis of the key factor data affecting the circuit board loss and setting the corresponding evaluation level includes the following steps: S621. Collect data on various factors affecting circuit board loss, screen and classify the collected data on various factors, remove duplicate factors, and summarize them into a measurable factor set; S622. Setting a weight for each factor based on actual testing experience, and setting an evaluation level for each factor; S623. Reflect the performance of each factor in the circuit board loss risk assessment according to the set evaluation level.

10. An artificial intelligence-based circuit board detection system, used to implement the artificial intelligence-based circuit board detection method according to any one of claims 1 to 9, characterized in that: The AI-based circuit board inspection system includes: A test preparation module is used to obtain initial data of the circuit board sample to be tested; A model building module is used to build a three-dimensional digital detection model of a circuit board sample and input the acquired initial data of the circuit board sample to be tested into the three-dimensional digital detection model; An image generation module is used to simulate the wear and tear of the circuit board during processing and use in the three-dimensional digital detection model, and to generate a corresponding wear and tear image of the circuit board sample through computer graphics technology; An image processing module is used to preprocess the generated circuit board sample loss image to obtain a circuit board sample loss feature image; A data analysis module is used to analyze the obtained circuit board sample loss characteristic image with the corresponding circuit board loss characteristic image in the database based on the analysis algorithm, and identify the key factor data of the circuit board sample loss; A data comparison module is used to compare the key factor data of the circuit board sample loss identified with the data of various factors affecting the circuit board loss based on the comparison algorithm, and evaluate the severity of the circuit board sample loss; A performance judgment module, used to judge whether the circuit board sample meets the preset performance standard based on the evaluation result; Among them, the detection preparation module is connected to the image generation module through the model construction module, the image generation module is connected to the data analysis module through the image processing module, and the data analysis module is connected to the performance judgment module through the data comparison module.

Images