Quality analysis method and system for circuit board and storage medium

Abstract

The invention relates to the technical field of image processing, and discloses a quality analysis method and system for a circuit board and a storage medium. The method comprises the following steps: extracting feature information according to a data set, and generating a defect distribution map; spectral reflection data is extracted based on the atlas, the material uniformity is calculated, and a welding quality data table is obtained; collecting quality characteristic parameters, establishing a quality grade index, and generating a graded data set; extracting process data, and performing cross validation and deviation analysis to form a process adjustment table; and combining the data set and the adjustment table, and outputting a quality evaluation report in combination with historical comparison. According to the invention, through combination of a multispectral imaging technology and a deep learning algorithm, comprehensive detection of surface and internal defects of the circuit board is realized; by establishing a data-driven analysis model, automatic diagnosis of quality problems and optimization and adjustment of process parameters are realized.

Description

Technical Field

[0001] The present application relates to the field of image processing, and in particular to a quality analysis method, system and storage medium for circuit boards. Background Art

[0002] Among the existing circuit board quality analysis technologies, optical inspection, electrical testing and X-ray inspection are mainly used. Optical inspection uses AOI (automatic optical inspection) equipment to collect circuit board surface images and compare them with standard images to identify defects; electrical testing uses flying probe testing or ICT (in-circuit testing) equipment to detect conductivity and insulation; X-ray inspection is used to check hidden defects such as BGA welding and inner layer interconnection. In terms of data processing, traditional image processing algorithms are used for defect identification, statistical analysis methods are used to evaluate quality parameters, and process parameters are adjusted and optimized through expert experience. The quality assessment process establishes a quality control system based on sampling inspection to comprehensively evaluate the appearance, performance and reliability of the product.

[0003] However, the existing circuit board quality analysis technology has the following shortcomings: First, various detection methods are independent of each other, lacking an effective data fusion mechanism, and unable to achieve collaborative analysis of detection data; second, traditional image processing algorithms mainly focus on defect detection, lack in-depth analysis of the causes of defects, and are difficult to provide effective guidance for process improvement; third, the quality analysis process is overly dependent on manual experience and lacks intelligent data analysis methods, which affects analysis efficiency and accuracy; finally, there is a lack of clear correlation between quality data and process parameters, and it is impossible to achieve accurate optimization of process parameters and continuous improvement of quality. These problems lead to unsatisfactory quality control effects in the circuit board manufacturing process, affecting product yield and reliability. Summary of the invention

[0004] The present application provides a quality analysis method, system and storage medium for circuit boards, which are used to achieve comprehensive detection of surface and internal defects of circuit boards by combining multispectral imaging technology with deep learning algorithms; and by establishing a data-driven analysis model, automatic diagnosis of quality problems and optimization adjustment of process parameters are achieved.

[0005] In a first aspect, the present application provides a quality analysis method for a circuit board, the quality analysis method for a circuit board comprising: obtaining original image data of the circuit board by multispectral imaging, performing signal noise reduction and image registration, forming a spatial feature database by fusing band data, and then obtaining a standardized multidimensional image data set by data standardization processing; extracting surface feature information of the circuit board according to the spectral band data in the standardized multidimensional image data set, comparing and analyzing feature point data with a preset threshold, and generating a defect distribution feature map after multi-dimensional defect feature marking; extracting spectral reflection data points on the surface of the circuit board according to the defect distribution feature map, and calculating material uniformity in combination with the light intensity attenuation curve of each band to obtain a welding quality data table; collecting quality feature parameters based on the defect distribution feature map and the welding quality data table, establishing a circuit board quality grade index through grading threshold interval division, and generating a quality grading data set; extracting key process data points according to the quality grading data set, cross-validating and analyzing process parameters, and forming a process adjustment parameter table; merging the quality grading data set and the process adjustment parameter table to construct a quality analysis data matrix, and outputting a circuit board quality assessment report in combination with historical parameter comparison.

[0006] In a second aspect, the present application provides a quality analysis system for a circuit board, the quality analysis system for a circuit board comprising: The registration module is used to obtain the original image data of the circuit board using multispectral imaging, and after signal noise reduction and image registration, the band data is fused to form a spatial feature database, and then a standardized multidimensional image data set is obtained through data standardization processing; A marking module, used to extract the surface feature information of the circuit board according to the spectral band data in the standardized multi-dimensional image data set, compare and analyze the feature point data with the preset threshold value, and generate a defect distribution feature map through multi-dimensional defect feature marking; A calculation module, used to extract the spectral reflection data points on the surface of the circuit board according to the defect distribution characteristic map, calculate the material uniformity in combination with the light intensity attenuation curve of each band, and obtain a welding quality data table; Establishing a module, for collecting quality characteristic parameters based on the defect distribution characteristic map and the welding quality data table, establishing a circuit board quality grade index through grading threshold interval division, and generating a quality grading data set; An analysis module, for extracting key process data points based on the quality classification data set, performing cross-validation and deviation analysis on process parameters, and forming a process adjustment parameter table; The merging module is used to merge the quality grading data set and the process adjustment parameter table, construct a quality analysis data matrix, and output a circuit board quality assessment report in combination with historical parameter comparison.

[0007] A third aspect of the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the above-mentioned circuit board quality analysis method.

[0008] In the technical solution provided by the present application, the original image data of the circuit board is obtained by using multispectral imaging, and after signal noise reduction and image registration processing, the high-precision acquisition of the surface features of the circuit board is realized, and the accuracy and reliability of data acquisition are improved; the spatial feature database is formed by band data fusion, and the effective integration of multi-band information is realized, and the comprehensiveness of feature extraction is enhanced; the standardized multidimensional image data set is obtained after data standardization processing, which provides a standardized data basis for subsequent analysis; by extracting the surface feature information of the circuit board and comparing and analyzing it with the preset threshold, the accurate identification of defects is realized; the defect distribution feature map is generated by multi-dimensional defect feature marking, and intuitive defect distribution information is provided; the material uniformity is calculated by combining the light intensity attenuation curve, and the quantitative evaluation of material quality is realized; by establishing the circuit board quality grade index, a scientific quality grading method is provided; the process adjustment parameter table is formed by cross-validation and deviation analysis, which provides a reliable basis for process optimization; finally, by comparing the quality analysis data matrix and historical parameters, a comprehensive circuit board quality assessment report is output, and the quality assessment process is systematized and standardized. This method organically combines a variety of analysis techniques, effectively improving the quality control level and process optimization effect of the circuit board manufacturing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0010] Figure 1 A schematic diagram of an embodiment of a quality analysis method for a circuit board in an embodiment of the present application; Figure 2 This is a schematic diagram of an embodiment of a quality analysis system for a circuit board in an embodiment of the present application. DETAILED DESCRIPTION

[0011] The embodiments of the present application provide a quality analysis method, system and storage medium for circuit boards. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0012] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 , an embodiment of the circuit board quality analysis method in the embodiment of the present application includes: Step S101, using multispectral imaging to obtain original image data of the circuit board, after signal noise reduction and image registration, the band data is fused to form a spatial feature database, and then through data standardization processing, a standardized multidimensional image data set is obtained; Step S102, extracting the circuit board surface feature information according to the spectral band data in the standardized multi-dimensional image data set, comparing and analyzing the feature point data with the preset threshold, and generating a defect distribution feature map through multi-dimensional defect feature marking; Step S103, extracting the spectral reflection data points on the circuit board surface according to the defect distribution characteristic spectrum, calculating the material uniformity in combination with the light intensity attenuation curve of each band, and obtaining a welding quality data table; Step S104, based on the defect distribution characteristic map and the welding quality data table, the quality characteristic parameters are collected, and the circuit board quality grade index is established through the classification threshold interval division to generate the quality classification data set; Step S105: extract key process data points based on the quality classification data set, perform cross-validation and deviation analysis on process parameters, and form a process adjustment parameter table; Step S106: Merge the quality grading data set and the process adjustment parameter table to construct a quality analysis data matrix, and output a circuit board quality assessment report in combination with historical parameter comparison.

[0013] It is understandable that the execution subject of the present application may be a circuit board quality analysis system, or a terminal or a server, which is not limited here. The present application embodiment is described by taking a server as the execution subject as an example.

[0014] Specifically, a multispectral camera is used to acquire the circuit board image. The camera contains three bands of acquisition units: ultraviolet spectrum (200-400nm), visible spectrum (400-700nm) and near infrared spectrum (700-1000nm). The acquisition process adopts an adaptive exposure control algorithm to adjust the exposure parameters in real time according to the image grayscale histogram to ensure image quality. The spectral spectrometry system uses a dichroic mirror group to separate the incident light into different bands, and each band converts the optical signal into an electrical signal through an independent photoelectric converter. The image signal is processed by a wavelet transform denoising algorithm, decomposing the signal into wavelet coefficients of multiple scales, then thresholding the coefficients, and finally reconstructing the image by inverse transformation. Image registration adopts a method based on deep learning, using an improved U-Net network to extract feature points, and calculating the image transformation matrix through feature matching to achieve accurate alignment of images in different bands. The surface features of the circuit board are extracted based on a standardized multidimensional image data set. The improved SLIC algorithm is used for image segmentation, and the image is divided into multiple superpixel regions. Spectral band data is extracted for each region, and statistical features (mean, variance, skewness, kurtosis) are calculated. The feature point data is compared with the preset threshold using the adaptive threshold method, and the threshold is automatically calculated according to the image grayscale histogram using the OTSU algorithm. Defect feature marking uses a deep learning method, using an improved ResNet network to extract deep features, and combining the attention mechanism to enhance the perception of key features. For different types of defects (cold solder joints, short circuits, open circuits, etc.), a feature template library is established, and feature matching is performed through an improved twin network to generate a defect distribution feature map.

[0015] The spectral reflectance of the circuit board surface is measured using an integrating sphere spectrometer. The incident light source is a xenon lamp with a wavelength range of 200-1000nm and a spectral resolution of 1nm. The spectral reflectance curve is recorded for each measurement point to analyze the change of reflectance with wavelength. The light intensity attenuation analysis is based on the Lambert-Beer law, and the optical properties of the material are determined by analyzing the attenuation relationship between the reflected light intensity and the propagation distance. The material uniformity analysis uses a spectral angle mapping algorithm, which regards the spectral curve of each pixel as a high-dimensional vector, calculates the angle between the spectral vectors of adjacent pixels, and evaluates the surface uniformity of the material. The welding quality data table records the spectral characteristics of the welding area, including parameters such as spectral consistency and edge sharpness of the welding area. The quality feature vector is established by combining the defect distribution feature map and welding quality data. The feature vector contains parameters such as defect density, defect type distribution, and material uniformity index. The improved K-means algorithm is used to cluster the feature vectors to determine the boundaries of different quality levels. The quality grade index is established based on the clustering results, and the products are divided into multiple grades.

[0016] Extract key process parameters from the quality grading data set, use principal component analysis to reduce the dimension, and screen the main influencing factors. Use the k-fold method for cross-validation to evaluate the significance of the impact of process parameters on quality. Deviation analysis calculates the deviation of process parameters to determine the parameters that need to be adjusted and the adjustment direction. Perform correlation analysis on quality grading data and process adjustment parameters to establish a quality-process relationship model. Use tensor decomposition method to extract potential correlation patterns between quality features and process parameters. Evaluate quality change trends through time series analysis, predict potential quality risks, and generate circuit board quality assessment reports.

[0017] For example, in the quality analysis process of a multi-layer printed circuit board: first, the image data is collected by a multispectral camera, and a clear multi-band image is obtained after wavelet noise reduction and image registration processing. Then, the image is segmented into super-pixel areas of appropriate size based on the SLIC algorithm, and the spectral features of each area are extracted. These features are analyzed through a deep learning network to identify and classify the defective areas. For the defective areas found, an integrating sphere spectrometer is used to perform detailed spectral reflectance measurements to analyze material uniformity and welding quality. Based on the analysis results, the quality of the circuit board is graded, and the adjustment plan for the process parameters is determined through cross-validation and deviation analysis. The quality assessment report contains the analysis results and improvement suggestions.

[0018] In the embodiment of the present application, the original image data of the circuit board is obtained by using multispectral imaging, and after signal noise reduction and image registration processing, the high-precision collection of the surface features of the circuit board is realized, and the accuracy and reliability of data collection are improved; the spatial feature database is formed by band data fusion, and the effective integration of multi-band information is realized, and the comprehensiveness of feature extraction is enhanced; the standardized multidimensional image data set is obtained after data standardization processing, which provides a standardized data basis for subsequent analysis; by extracting the surface feature information of the circuit board and comparing and analyzing it with the preset threshold, the accurate identification of defects is realized; the defect distribution feature map is generated by multi-dimensional defect feature marking, and intuitive defect distribution information is provided; the material uniformity is calculated in combination with the light intensity attenuation curve, and the quantitative evaluation of material quality is realized; by establishing the circuit board quality grade index, a scientific quality grading method is provided; the process adjustment parameter table is formed by cross-validation and deviation analysis, which provides a reliable basis for process optimization; finally, by comparing the quality analysis data matrix and historical parameters, a comprehensive circuit board quality assessment report is output, and the quality assessment process is systematized and standardized. This method organically combines a variety of analysis techniques, effectively improving the quality control level and process optimization effect of the circuit board manufacturing process.

[0019] In a specific embodiment, the process of executing step S101 may specifically include the following steps: (1) Synchronously collect circuit board images in the ultraviolet, visible light, and near-infrared bands to form original image data; (2) Use the original image data to identify data interference items, separate the ambient light fluctuation data and sensor noise data, and generate a noise characteristic spectrum; (3) Based on the noise characteristic spectrum, the interference signal in the original image data is eliminated in a targeted manner, and the image signal is enhanced to obtain the noise-reduced image data; (4) Dividing the denoised image data into multiple sub-bands according to the wavelength range, mapping and transforming the position coordinates of each sub-band image to obtain position mapping data; (5) Convert the position mapping data into a unified coordinate system, superimpose and fuse the image data of different bands according to the spatial distribution law, and construct a spatial feature database; (6) Normalize the data samples in the spatial feature database, calculate the statistical feature parameters, remove abnormal data points, and output a standardized multidimensional image data set.

[0020] Specifically, the original image data of the circuit board is obtained by a multispectral imaging acquisition device. The device uses three independent acquisition units: an ultraviolet band acquisition unit (wavelength range 200-400nm), a visible light band acquisition unit (wavelength range 400-700nm) and a near-infrared band acquisition unit (wavelength range 700-1000nm). Each acquisition unit contains a spectral spectrometer system, a photoelectric converter and an image sensor. The spectral spectrometer system uses a dichroic mirror group to separate the incident light into different bands, and the photoelectric converter converts the optical signal into an electrical signal. The image sensor adopts a back-illuminated CMOS structure with a pixel size of 2.5 microns and a resolution of 4096×3072 pixels. High-precision clock synchronization technology is used in the acquisition process to ensure that the three acquisition units trigger sampling synchronously. For the acquired original image data, the frequency domain analysis method is used to identify data interference items. First, the image signal is converted to the frequency domain space by two-dimensional Fourier transform. The ambient light fluctuation is mainly manifested as low-frequency interference (0-10Hz), while the sensor noise is manifested as high-frequency random fluctuations (above 1kHz). By designing a bandpass filter, the ambient light fluctuation signal and the sensor noise signal are separated. The bandwidth parameters of the bandpass filter are dynamically adjusted according to the actual test data to obtain the best separation effect. The separated noise characteristic spectrum records the spectral distribution, amplitude statistics and correlation characteristics of different types of noise.

[0021] Based on the obtained noise characteristic spectrum, the adaptive Wiener filtering algorithm is used to eliminate the interference signal in the original image data in a targeted manner. The algorithm dynamically adjusts the filter parameters according to the signal-to-noise ratio estimation of the local image area to achieve adaptive noise reduction in different areas. The denoised image is subjected to signal enhancement processing through methods such as histogram equalization and contrast enhancement to improve the signal-to-noise ratio and detail performance of the image. Each processing step is monitored by quality evaluation indicators to ensure that the processing effect meets the requirements of subsequent analysis.

[0022] The denoised image data is subdivided according to the spectral response characteristics. The ultraviolet band is divided into 4 sub-bands (200-250nm, 250-300nm, 300-350nm, 350-400nm), the visible light band is divided into 6 sub-bands (one every 100nm), and the near-infrared band is divided into 3 sub-bands (700-800nm, 800-900nm, 900-1000nm). The affine transformation algorithm is used to perform geometric correction on each sub-band image to eliminate the distortion caused by the optical system. The feature point matching algorithm is used in the transformation process to establish the spatial correspondence between different sub-bands.

[0023] The spatial transformation parameters are optimized by the least squares method, and the position mapping data of each band is converted into a unified standard coordinate system. The image fusion adopts a multi-resolution analysis method, including three steps: pyramid decomposition, coefficient fusion and image reconstruction. During the fusion process, the weight coefficient is dynamically adjusted according to the amount of information in different bands. The fused data is stored in the spatial feature database, which records information such as spatial coordinates, spectral response values ​​and feature descriptions. The samples in the spatial feature database are standardized. The statistical parameters (mean, standard deviation, skewness, kurtosis) of each feature dimension are calculated, and then the Z-score standardization method is used to convert the data to a distribution with a mean of 0 and a variance of 1. The identification of abnormal data points adopts the local outlier factor (LOF) algorithm, which identifies outliers by calculating the local density comparison of data points. After these processing steps, a standardized multidimensional image data set is finally obtained.

[0024] For example, the image data of the PCB board is collected by multispectral imaging equipment, and scanned simultaneously in three bands. Low-frequency fluctuations caused by changes in ambient light and high-frequency interference caused by sensor thermal noise are found in the original data. Through frequency domain analysis, it is confirmed that the ambient light fluctuations are concentrated in the 0-10Hz frequency band, and the sensor noise is distributed in the frequency band above 1kHz. According to these characteristics, the corresponding filter is designed for signal separation to obtain the noise characteristic spectrum. The noise characteristic spectrum is used for directional elimination, and the adaptive Wiener filter dynamically adjusts the parameters according to the signal-to-noise ratio of the local area. The denoised image data is subdivided according to the preset band division scheme, and the image of each sub-band is geometrically corrected. Accurate image registration is achieved through feature point matching, and then data fusion is performed in a unified coordinate system. The fusion process uses a 5-layer pyramid decomposition, and each layer uses Gaussian and Laplace operators to extract features, and the weight coefficient is adjusted according to the amount of information in different bands. Finally, the fused data is standardized, and the outliers are detected and eliminated by the LOF algorithm. The threshold for outlier determination is set to 3 times the standard deviation of the local density deviation. The processed standardized multidimensional image data set contains the complete spectral response characteristics of the PCB board surface.

[0025] In a specific embodiment, the process of executing step S102 may specifically include the following steps: (1) Segment the standardized multidimensional image data set according to the circuit board surface area, extract the spectral band data of each area, and generate a regional spectral data matrix; (2) Map the data points in the regional spectral data matrix to spatial coordinates, mark the feature points on the circuit board surface, and output the feature point data; (3) Perform density clustering calculation on feature point data, identify the distribution pattern of feature points, calculate the degree of clustering of feature points, and form a feature distribution statistics table; (4) Based on the feature distribution statistics table, the density value of the feature points is compared with the preset threshold, and the area exceeding the threshold range is marked to obtain the abnormal area data table; (5) Classify and count the data points in the abnormal area data table according to the defect type, calculate the spatial distribution characteristics of each type of defect, and generate a defect type distribution map; (6) Perform multi-dimensional feature extraction on the defect type distribution map, perform correlation analysis on parameters such as defect location, area, and type, and output a defect distribution feature map.

[0026] Specifically, the standardized multidimensional image data set is subjected to regional segmentation. The segmentation adopts the improved SLIC (Simple Linear Iterative Clustering) superpixel segmentation algorithm, which clusters adjacent similar pixels into superpixels through iterative optimization. The specific process includes: initializing uniformly distributed cluster centers, calculating the distance and color similarity between each pixel and the surrounding cluster centers, assigning pixels to the nearest cluster center according to the comprehensive distance, updating the location and color features of the cluster center, and repeating this process until convergence. The segmentation parameters include: the preset number of superpixels (determined according to the size of the circuit board), the compactness factor (controlling the weight ratio of spatial distance to color distance) and the maximum number of iterations. Spectral band data is extracted for each segmented area, including the spectral response values ​​of the three main bands of ultraviolet, visible light and near infrared and the subdivided bands, to form a regional spectral data matrix. The data points in the regional spectral data matrix are mapped to the actual spatial coordinate system through affine transformation. The transformation process takes into account the influence of factors such as camera distortion and perspective difference. The feature point detection adopts the Harris corner detection algorithm, which identifies the corner point features by calculating the grayscale change matrix of the local area of ​​the image. The core of the Harris algorithm is to calculate the grayscale changes in the window around each pixel and determine whether it is a corner point through eigenvalue analysis. The detected feature points are classified and marked, and the position coordinates, spectral response value and local gradient information of each feature point are recorded.

[0027] DBSCAN (Density-Based Spatial Clustering of Applications with Noise) density clustering calculation is performed on the feature point data. The DBSCAN algorithm automatically discovers clusters based on density connectivity, and there is no need to pre-specify the number of clusters. The core parameters of the algorithm include: neighborhood radius (neighborhood range of the definition point) and minimum number of points (threshold for determining core points). By analyzing the spatial distribution pattern of feature points, statistics such as local density, clustering radius and distance between points are calculated, the degree of aggregation of feature points is evaluated, and a feature distribution statistical table is formed. Based on the feature distribution statistical table, the dynamic threshold method is used to analyze the density value of feature points. The threshold setting is based on historical data statistics and expert experience, taking into account the feature differences of different types of circuit boards. The area beyond the threshold range is marked, the location, area and degree of excess of the abnormal area are recorded, and the abnormal area data table is generated. The threshold judgment adopts a multi-level threshold strategy, and different levels of thresholds correspond to abnormalities of different severity.

[0028] Deep learning classification is performed on abnormal areas. The improved ResNet network structure is used, and an attention mechanism module is added between traditional convolutional layers. The network training data includes characteristic samples of various typical defects, such as cold solder joints, short circuits, open circuits, and pad detachment. The spatial distribution characteristics of each type of defect are calculated, including distribution density, aggregation, and directionality. Finally, a defect type distribution map is generated to intuitively display the spatial distribution of various defects. Multi-dimensional feature extraction is performed on the defect type distribution map, including: defect location coordinates, area size, shape features, edge features, and texture features. The relationship between different feature parameters is studied through correlation analysis to discover potential quality problems. The defect distribution feature map finally output contains defect feature information.

[0029] Take the defect detection of a four-layer PCB board as an example: SLIC segmentation is performed on the surface of the circuit board, and the segmentation parameters are set so that the area of ​​each superpixel area is about 1 square millimeter. Such segmentation granularity not only meets the needs of detailed analysis, but also avoids excessive calculation. Full-band spectral data is extracted for each segmented area and recorded in the spectral data matrix. The matrix data is spatially mapped to convert the image coordinates into actual physical coordinates. Harris corner detection is used to extract feature points, and the response threshold is set to ensure that the detected feature points have significant corner point characteristics. DBSCAN clustering analysis is performed on the feature points, with the neighborhood radius set to the actual physical distance of 0.5 mm and the minimum number of points set to 5. Such parameter settings can effectively identify potential defect clustering areas. Based on the clustering results, the density of feature points in each area is calculated to generate a density distribution heat map. The heat map is compared with the preset density threshold to mark the area with abnormal density. A deep learning model is used to classify defects in these abnormal areas. The input of the model includes the multispectral image data and local feature descriptors of the area. The classification results show the spatial distribution of different types of defects. For example, welding defects tend to be locally concentrated, while substrate defects are dispersed over a larger area. Finally, multi-dimensional feature extraction and correlation analysis are performed on each type of defect to study the relationship between defect features. By analyzing the location distribution and area size of welding defects, it is found that the problem of cold solder joints in certain areas is related to the pad design.

[0030] In a specific embodiment, the process of executing step S103 may specifically include the following steps: (1) Perform spectral reflection sampling on the position information in the defect distribution characteristic map, record the reflection intensity of each sampling point in different bands, and form a spectral reflection data set; (2) Arrange the spectral reflectance data set in order of wavelength, calculate the difference in reflectance intensity between adjacent bands, and generate a light intensity change gradient table; (3) Normalize the data in the light intensity change gradient table, draw the light intensity attenuation curve of each sampling point, record the position of the key inflection point, and output the attenuation characteristic data; (4) Perform regional statistics on the attenuation characteristic data, calculate the variance of the attenuation curve in the local area, and form a material uniformity distribution map; (5) Based on the material uniformity distribution map, the welding point is located, the attenuation curve slope of the welding area is extracted, the welding quality parameters are counted, and the welding point quality matrix is ​​established; (6) Compare and analyze the parameters in the welding point quality matrix with the standard parameters, mark the abnormal welding points, and output the welding quality data table.

[0031] Specifically, spectral reflection sampling is performed on the marked positions in the defect distribution characteristic map. A high-precision integrating sphere spectrometer is used, and the wavelength range of the device covers 200-1000nm and the spectral resolution is 0.5nm. The incident light source uses a xenon lamp, and the light source is guided to the measurement point through an optical fiber light guide system. Spectral reflection data are recorded at each sampling point, including the reflection intensity values ​​of the ultraviolet band (200-400nm), the visible light band (400-700nm) and the near-infrared band (700-1000nm). An autofocus system is used during sampling to ensure measurement accuracy, and the data acquisition time for each measurement point is 100 milliseconds to ensure signal stability. The acquired spectral reflection data group is sorted from short to long according to wavelength to form a continuous spectral curve. The central difference method is used to calculate the reflection intensity change rate between adjacent wavelength points to obtain the light intensity change gradient. The gradient calculation adopts a five-point difference format to improve the calculation accuracy. The light intensity change gradient table records the reflectivity change characteristics of each sampling point in different wavelength ranges, reflecting the spectral response characteristics of the material. The data in the light intensity gradient table are normalized to the maximum and minimum values, and the data are mapped to the interval [0, 1]. The third-order spline interpolation method is used to fit the light intensity attenuation curve, and the key inflection point position is identified by analyzing the second-order derivative of the curve. The inflection point features include position coordinates and curvature values, which are directly related to the optical properties of the material. The attenuation characteristic data of each sampling point contains spectral response information. Regional statistical analysis is performed on the attenuation characteristic data. The sliding window method is used to calculate the variance of the attenuation curve in the local area, and the window size is dynamically adjusted according to the characteristic size of the circuit board. The variance calculation reflects the uniformity characteristics of the material surface, and the generated material uniformity distribution map intuitively shows the spatial distribution of the surface quality.

[0032] For the welding point location problem, the location formula based on spectral characteristics is as follows: , in: Indicates the welding point position index; is the weight factor of the i-th spectral feature; is the material uniformity weighting coefficient; is the spectral characteristic value of the i-th region; is the regional average spectral characteristic; is the local area variance; is the gradient weight coefficient; M is the material uniformity function, are the gradients of uniformity in the x and y directions, respectively.

[0033] Extract the spectral characteristics of the welding area based on the positioning results and analyze the change in the slope of the attenuation curve. Establish a welding point quality matrix and record parameters such as welding strength, uniformity and interface characteristics. Identify abnormal welding points by comparing with the standard parameter database and output a welding quality data table.

[0034] Take the analysis process of an actual circuit board as an example: grid sampling is performed on the surface of the circuit board, and spectral reflection data is recorded at each sampling point. Spectral data show that the normal welding area exhibits a smooth reflection curve in the 500-600nm band, while the virtual welding area shows obvious discontinuity. By analyzing the gradient of light intensity change, it is found that the gradient value of the virtual welding point in a specific wavelength range is significantly higher than that of the normal welding point. After normalization, the light intensity attenuation curve characteristics of each sampling point are extracted. The attenuation curve of the normal welding point shows characteristic inflection points near 350nm and 750nm, and the inflection point position is stable. Regional statistical analysis shows that the attenuation curve variance of the high-quality welding area is small, indicating that the material is evenly combined. Based on the material uniformity distribution map, the welding point is located, and the spectral feature score of the local area is calculated to accurately identify the welding point position. The welding quality analysis shows that the slope of the attenuation curve of the good welding point remains stable within a specific band, and the transition with the adjacent area is smooth. By comparing with the standard parameter library, the welding quality evaluation standard is established to achieve accurate identification and classification of abnormal welding points.

[0035] In a specific embodiment, the process of executing step S104 may specifically include the following steps: (1) Extract defect density parameters from the defect distribution feature map, divide the defect location distribution data into regions, and output the regional defect statistics table; (2) Classify the welding point data in the welding quality data table according to the quality level, perform numerical statistics on the welding quality parameters, and generate a welding quality distribution map; (3) Integrate the data of the regional defect statistics table and the welding quality distribution map, calculate the comprehensive quality score of each area, and form a quality characteristic parameter matrix; (4) Perform numerical analysis on the quality characteristic parameter matrix, statistically analyze the parameter distribution law, calculate the key interval nodes, and obtain the quality threshold distribution table; (5) Divide the quality threshold distribution table into grade intervals according to the parameter value range, assign quality grade labels to different intervals, and establish a circuit board quality grade index; (6) Classify the quality characteristic parameter matrix based on the circuit board quality grade index, organize and compile the classification results, and output the quality grading data set.

[0036] Specifically, the grid method is used to divide the surface of the circuit board into uniform rectangular areas, and the grid size is determined according to the actual size and characteristic size of the circuit board. The number and area ratio of defects per unit area are calculated for each grid area to form a defect density distribution map. The regional division adopts an adaptive grid subdivision strategy, and a smaller grid size is used in defect-intensive areas to improve the analysis accuracy. The regional defect statistics table records the type, number, distribution location and area proportion of defects in each grid area. The welding point data in the welding quality data table are classified and processed, and the welding points are graded according to the welding quality parameters (including welding strength, uniformity, interface characteristics, etc.). The quality grade division adopts a multi-parameter comprehensive evaluation method, which comprehensively considers the geometric characteristics, spectral characteristics and material characteristics of the welding points. The numerical statistics adopt a weighted scoring method, and the weights of different parameters are determined according to their influence on the welding quality. The welding quality distribution map intuitively shows the spatial distribution characteristics of welding points of different quality grades.

[0037] A multi-dimensional evaluation system is used to integrate the data of the regional defect statistics table and the welding quality distribution map. The comprehensive quality score of each region is calculated, taking into account factors including defect density, defect type weight, welding quality grade and regional importance coefficient. Each element in the quality characteristic parameter matrix contains quality characteristic information of multiple dimensions, reflecting the overall status of regional quality. A variety of statistical methods are used for the numerical analysis of the quality characteristic parameter matrix, including mean, standard deviation, skewness and kurtosis analysis of the parameters. The key interval nodes of the quality parameters are determined by cluster analysis, and the improved K-means algorithm is used to segment the parameter space. The quality threshold distribution table records the threshold distribution of each quality parameter.

[0038] The parameter space is divided into multiple grade intervals, and the interval division takes into account the distribution characteristics of the parameters and the actual quality requirements. Quality grade identification is assigned to different intervals, and a standardized circuit board quality grade index is established. The index system covers multiple grades from excellent to unqualified, and each grade has a clear parameter range definition. The grade division adopts a dynamic threshold strategy and is adjusted according to the product type and quality requirements. Based on the established quality grade index, the data in the quality characteristic parameter matrix is ​​classified. The classification results are systematically organized to record the data distribution of each grade. The quality grading data set contains complete information such as quality grade distribution, abnormal area identification and improvement suggestions.

[0039] For example: a 10mm×10mm grid is used to divide the surface of the circuit board, and a 5mm×5mm subdivision grid is used in the component-dense area. By counting the defects in each grid, it is found that defects are clustered in some grid areas, which are mainly distributed at the edge of the circuit board and in the high-density component installation area. According to the spectral characteristics and material uniformity parameters of the welding points, the welding points are divided into three grades: A, B, and C. Grade A indicates excellent welding quality and stable characteristic parameters; Grade B indicates qualified welding quality and parameter fluctuations within an acceptable range; Grade C indicates that there are hidden dangers in welding quality. Through spatial distribution analysis, it is found that Grade C welding points often show a positive correlation with the defect density of the surrounding area.

[0040] The integrated analysis shows that the welding quality is generally lower in areas with high defect density. This correlation indicates potential process problems. The comprehensive quality score calculation takes into account the defect density in the area, the distribution of solder joint quality grades and the regional importance coefficient. Areas with lower scores often have multiple quality problems superimposed. According to the parameter distribution law, the quality feature space is divided into five grade intervals: excellent, good, qualified, to be improved and unqualified. Each grade interval has a clear parameter range definition. For example, the excellent grade requires that the defect density is lower than the threshold and all solder joints meet the A-level standard. Through this grading method, a quality assessment system is established. Based on the established quality grade index, the quality of the entire circuit board is graded. The grading results not only show the overall quality grade of the circuit board, but also identify areas that need to be focused on.

[0041] In a specific embodiment, the process of executing step S105 may specifically include the following steps: (1) Extract key process data from the quality classification data set, classify the data according to the process flow nodes, and form a process data classification table; (2) Perform correlation analysis on the data in the process data classification table, calculate the correlation coefficient between process parameters, and generate a parameter correlation matrix; (3) Sort the data in the parameter correlation matrix according to the size of the correlation coefficient, screen the highly correlated process parameter combinations, and output the key parameter group; (4) Carry out cross-comparison verification for key parameter groups, statistically analyze parameter fluctuation ranges, calculate parameter deviation values, and obtain a process parameter deviation table; (5) Perform data analysis based on the process parameter deviation table, calculate the adjustment direction and adjustment range of each parameter, and form a parameter adjustment plan; (6) Verify the feasibility of the adjustment values ​​in the parameter adjustment plan, integrate the verification results, and output the process adjustment parameter table.

[0042] Specifically, key process data are extracted from the quality classification data set. The principal component analysis method is used to identify process parameters that have a significant impact on quality. The process parameters are classified according to the manufacturing process, including: pre-treatment parameters (cleanliness, surface treatment, etc.), board making parameters (pressing temperature, pressure, etc.), drilling parameters (rotation speed, feed speed, etc.), electroplating parameters (current density, time, etc.), solder mask parameters (exposure energy, development time, etc.) and surface treatment parameters (tinning temperature, time, etc.). The process data classification table records the key process parameters of each link and their value ranges.

[0043] The correlation between process parameters is calculated using the following formula: , in: is the correlation coefficient between parameters a and b; are the values ​​of parameters a and b in the kth sample; is the mean of parameters a and b; is the sample weight coefficient; is the parameter variance adjustment factor; is the process distance between parameters; is the scale parameter; n is the number of samples.

[0044] Correlation analysis uses a combination of Pearson correlation coefficient and Spearman rank correlation coefficient to evaluate the linear and nonlinear relationships between parameters. Each element in the parameter correlation matrix represents the strength of association between two process parameters. By setting the correlation coefficient threshold, highly correlated parameter combinations are screened out to form a key parameter group. The key parameter group contains information such as parameter name, correlation coefficient, and impact direction. The k-fold cross-validation method is used to verify and analyze the key parameter group. Record the changes of each parameter under different process conditions, and calculate the fluctuation range and standard deviation of the parameter. Compare with the standard process parameters, calculate the deviation degree of the actual parameters, and generate a process parameter deviation table. Deviation analysis takes into account the historical change trend of the parameters and the degree of quality impact.

[0045] Based on the deviation table data, calculate the direction and magnitude of each parameter adjustment. The formulation of the parameter adjustment plan takes into account the coupling relationship between parameters and uses the optimization algorithm to find the best adjustment combination. The feasibility of the adjustment plan is evaluated through historical data analysis and process expert knowledge base.

[0046] Taking the process parameter optimization of a batch of multi-layer PCB boards as an example, key process parameters are extracted from the quality grading data, including pressing temperature, pressure, drilling speed, electroplating time, etc. These parameters are classified and sorted according to the manufacturing process to form a process data classification table. Correlation analysis shows that there is a significant correlation between pressing temperature and interlayer bonding strength, and drilling speed is highly correlated with the uniformity of hole wall copper thickness. Through parameter screening, multiple highly correlated parameter combinations are found. The cross-validation results show that the correlation of these parameter combinations is consistent in different batches of products. Parameter fluctuation analysis found that the actual values ​​of some process parameters deviate from the standard values, such as the heating rate of pressing temperature is too fast, which affects the curing effect of resin. Based on these analysis results, specific adjustment plans are formulated: optimizing the pressing temperature curve, adjusting drilling parameters, improving electroplating process, etc. Through feasibility verification, it is confirmed that these adjustment plans meet the process requirements and are practically operable. The generated process adjustment parameter table provides clear guidance for parameter optimization of the production line. This systematic process parameter optimization method realizes precise control and quality improvement of the circuit board manufacturing process.

[0047] In a specific embodiment, the process of executing step S106 may specifically include the following steps: (1) Match the quality parameters in the quality grading data set with the process parameters in the process adjustment parameter table to generate a parameter association table; (2) Structuralize the data items in the parameter association table, group the data according to quality level and process type, and form a data grouping matrix; (3) Perform statistical analysis on each group of data in the data grouping matrix, calculate the corresponding relationship between quality parameters and process parameters, and output a comprehensive analysis data table; (4) Compare the data in the comprehensive analysis data table with the historical quality data, mark the abnormal parameter change points, and generate a quality trend analysis chart; (5) Conduct data mining based on quality trend analysis charts, extract quality change patterns, count key influencing factors, and form quality assessment conclusions; (6) Integrate the quality assessment conclusions with the specific parameter data, classify and organize them according to the assessment dimensions, and output the circuit board quality assessment report.

[0048] Specifically, parameter matching is performed. The quality grading data set contains quality parameters such as pass rate and defect rate, and the process adjustment parameter table contains process parameters such as temperature, pressure, and time. A multidimensional mapping algorithm is used to establish the correspondence between parameters, and the mapping process considers the physical meaning and influence mechanism of the parameters. The parameter association table records the matching relationship, including information such as parameter type, numerical range, and influence weight. The parameter association table is structured, and the data is grouped using a hierarchical clustering method. The quality grade grouping is based on the product quality standard, and the data is divided into four levels: good, medium, and poor. The process type grouping is based on the manufacturing process flow, including material parameters (substrate properties, copper foil thickness, etc.), process parameters (pressing temperature, drilling speed, etc.), and environmental parameters (temperature, humidity, cleanliness, etc.). The rows of the data grouping matrix represent the quality grade, and the columns represent the process type.

[0049] Statistical analysis uses a variety of methods to process the data grouping matrix, including variance analysis, regression analysis, and principal component analysis. Variance analysis evaluates the contribution of different process parameters to quality fluctuations, regression analysis establishes quantitative relationships between parameters, and principal component analysis reduces data dimensions and extracts main features. The comprehensive analysis data table records the mapping relationship between quality parameters and process parameters. The historical data comparison uses time series analysis to identify parameter change trends and outliers. The comparison process uses sliding window technology to calculate the moving average and standard deviation of the parameters. The 3σ criterion is used to identify outliers, mark data points with significant deviations, and record the time and specific parameters of the anomaly. The quality trend analysis chart shows the trajectory of parameter changes over time. The data mining process uses association rule mining algorithms to discover the law of quality changes. The algorithm includes three steps: data preprocessing, frequent item set generation, and rule evaluation. Effective rules are screened through support and confidence indicators to identify key factors affecting quality. The evaluation conclusion is based on the mining results and systematically describes the causes and effects of quality problems.

[0050] The quality assessment conclusions are integrated with the parameter data and classified according to dimensions such as product performance, process stability and quality reliability. The assessment report contains quality status analysis, problem diagnosis and improvement suggestions to provide guidance for production optimization. For example: match quality parameters (such as welding quality and line accuracy) with process parameters (such as welding temperature and exposure energy). Through analysis, it is found that there is a significant correlation between the welding quality of a batch of products and the welding temperature control. The parameter association table shows that when the welding temperature deviation exceeds the set range, the welding defect rate increases significantly. Data grouping analysis shows that the process parameter stability of the high-quality product group is significantly higher than that of other groups. The statistical analysis results show that the fluctuation range of process parameters is negatively correlated with the product quality grade. By comparing with historical data, it is found that the control accuracy of welding temperature has decreased recently, which is consistent with the fluctuation trend of quality level. Quality trend analysis found multiple key influencing factors: first, the stability of process parameters, especially the control accuracy of temperature and pressure; second, the consistency of material properties, including the quality of substrate and solder; and finally, the influence of environmental conditions, such as fluctuations in humidity and cleanliness. The combined effect of these factors determines the final product quality. The data mining results show important quality change rules: small fluctuations in process parameters will not significantly affect product quality within a certain range, but when the fluctuation exceeds the critical value, the quality will drop sharply. In addition, there are interactions between certain process parameters, which require coordinated optimization to achieve the best results. Based on these analysis results, the evaluation report puts forward specific improvement suggestions: optimize the process parameter control system to improve the control accuracy of temperature and pressure; strengthen raw material inspection to ensure the consistency of material properties; improve production environment control and reduce the interference of environmental factors. These suggestions are highly targeted and highly operational, providing clear guidance for improving the quality of circuit board manufacturing.

[0051] In a specific embodiment, the process of integrating the quality assessment conclusion with the specific parameter data may specifically include the following steps: (1) Extract the key evaluation points in the quality evaluation conclusion, classify them according to the defect type, and generate a statistical table of evaluation key points; (2) Conduct quantitative analysis on the data items in the evaluation key statistical table, calculate the deviation value of each evaluation item, and form a deviation distribution matrix; (3) Perform data clustering based on the deviation distribution matrix, divide the risk level according to the degree of deviation, and output a risk assessment list; (4) Conduct traceability analysis on each indicator in the risk assessment list, extract the root causes of quality problems, and generate a problem attribution table; (5) Conduct correlation analysis between the problem attribution table and process parameter data, statistically analyze the quality fluctuation pattern, and form a report on the direction of improvement; (6) Systematically organize the improvement direction reports, sort them according to the severity of the problems, and output the circuit board quality assessment report.

[0052] Specifically, the key evaluation points of the quality evaluation conclusion are extracted. The text mining algorithm is used to identify the important quality feature words and evaluation indicators in the conclusion. The feature words are selected based on the professional terminology library in the field of circuit board manufacturing. The defect type classification adopts the preset classification standards, including: process defects (such as cold solder joints, short circuits, open circuits), material defects (such as delamination, blistering, copper fading) and structural defects (such as line width deviation, aperture deviation). The evaluation key statistical table records the distribution characteristics, occurrence frequency and severity of various defects. The data in the evaluation key statistical table are quantitatively analyzed, and the qualitative description is converted into quantitative indicators based on the multi-dimensional evaluation system. Each evaluation item sets multiple evaluation dimensions, including defect area, depth, location and impact range. The deviation degree of each evaluation item relative to the standard value is calculated by the standard score method to form a deviation distribution matrix. The rows of the matrix represent different evaluation items, the columns represent different evaluation dimensions, and each element represents a specific deviation value. The improved K-means clustering algorithm is used to analyze the deviation distribution matrix. The clustering process takes into account the size and distribution characteristics of the deviation value and classifies similar deviation patterns into one category. The clustering results are used to determine the risk level, and the degree of deviation is divided into three risk levels: high, medium, and low. The risk assessment checklist records the risk level, impact, and potential consequences of each assessment item in detail.

[0053] Traceability analysis uses the cause-effect chain analysis method to track the occurrence mechanism of quality problems layer by layer. By establishing a problem tree, the root cause of quality deviation is identified. The problem attribution table systematically records the problem type, impact range, cause analysis and solution. During the attribution analysis process, the focus is on the influence of process parameters, material properties and environmental factors. The problem attribution table is associated with the process parameter data, and the association rule mining algorithm is used to discover the quality fluctuation pattern. The analysis content includes the change trend of process parameters, the interaction between parameters and the degree of influence on quality. The improvement direction report proposes specific optimization suggestions and improvement measures based on the analysis results.

[0054] The improvement direction reports are systematically collated and the issues are prioritized using a multi-criteria decision-making method. The ranking criteria take into account the severity, frequency, difficulty of resolution and economic impact of the issues. The quality assessment report contains the analysis results and improvement suggestions.

[0055] For example, key issues are extracted from the evaluation conclusions, such as insufficient bonding strength between the inner copper foil and the substrate, incomplete via filling, etc. These issues are classified into material defects and process defects through classification marking. Each defect item records specific feature descriptions and statistical data. In the quantitative analysis stage, each defect item is evaluated in multiple dimensions. For example, for the problem of insufficient bonding strength, the degree of deviation is evaluated through peel strength testing; for the problem of via filling, the deviation of the filling rate is evaluated through microscopic section testing. These deviation data constitute a multi-dimensional evaluation matrix.

[0056] Data cluster analysis found that some defects have similar deviation characteristics, such as inner layer bonding strength problems and via filling problems, which are related to the lamination process parameters. Through risk level classification, these defects are marked as high-risk items. The risk assessment list records the risk characteristics and impact of various defects in detail. Retrospective analysis reveals the root cause of the problem. For example, the problem of insufficient inner layer bonding strength is caused by improper control of the lamination temperature curve. By analyzing the historical data of lamination temperature, pressure and time parameters, it is found that the temperature rise rate is too fast, resulting in insufficient resin curing, which in turn affects the bonding strength.

[0057] Correlation analysis shows that there are interactions between multiple process parameters. The matching relationship between pressing temperature and pressure directly affects the bonding strength, while the exhaust time affects the quality of via filling. These findings provide a clear direction for process improvement. Improvement suggestions include: optimizing the pressing temperature curve, adjusting pressure parameters, improving the exhaust process, etc. The quality assessment report is sorted according to the severity of the problem. The inner layer bonding strength problem that affects product reliability is listed as the first improvement project, followed by the via filling problem that affects the yield rate. The report describes in detail the specific manifestations, cause analysis and improvement measures of each problem.

[0058] The above describes the circuit board quality analysis method in the embodiment of the present application. The following describes the circuit board quality analysis system in the embodiment of the present application. Figure 2 In the embodiment of the present application, one embodiment of the circuit board quality analysis system includes: The registration module 201 is used to obtain the original image data of the circuit board by using multispectral imaging, and after signal noise reduction and image registration, the band data is fused to form a spatial feature database, and then a standardized multidimensional image data set is obtained through data standardization processing; The marking module 202 is used to extract the circuit board surface feature information according to the spectral band data in the standardized multi-dimensional image data set, compare and analyze the feature point data with the preset threshold, and generate a defect distribution feature map through multi-dimensional defect feature marking; The calculation module 203 is used to extract the spectral reflection data points on the surface of the circuit board according to the defect distribution characteristic map, calculate the material uniformity in combination with the light intensity attenuation curve of each band, and obtain a welding quality data table; Establishing module 204, for collecting quality characteristic parameters based on the defect distribution characteristic map and the welding quality data table, establishing a circuit board quality grade index through grading threshold interval division, and generating a quality grading data set; An analysis module 205 is used to extract key process data points based on the quality classification data set, perform cross-validation and deviation analysis on process parameters, and form a process adjustment parameter table; The merging module 206 is used to merge the quality grading data set and the process adjustment parameter table, construct a quality analysis data matrix, and output a circuit board quality assessment report in combination with historical parameter comparison.

[0059] Through the cooperation of the above components, the original image data of the circuit board is obtained by multispectral imaging, and after signal noise reduction and image registration processing, the high-precision acquisition of the surface features of the circuit board is achieved, and the accuracy and reliability of data acquisition are improved; the spatial feature database is formed by band data fusion, which realizes the effective integration of multi-band information and enhances the comprehensiveness of feature extraction; the standardized multidimensional image data set is obtained after data standardization processing, which provides a standardized data basis for subsequent analysis; the defects are accurately identified by extracting the surface feature information of the circuit board and comparing and analyzing it with the preset threshold; the defect distribution feature map is generated by multi-dimensional defect feature marking, which provides intuitive defect distribution information; the material uniformity is calculated by combining the light intensity attenuation curve, and the quantitative evaluation of material quality is realized; by establishing the circuit board quality grade index, a scientific quality grading method is provided; the process adjustment parameter table is formed by cross-validation and deviation analysis, which provides a reliable basis for process optimization; finally, through the quality analysis data matrix and historical parameter comparison, a comprehensive circuit board quality assessment report is output, which realizes the systematization and standardization of the quality assessment process. This method organically combines multiple analysis techniques and effectively improves the quality control level and process optimization effect of the circuit board manufacturing process.

[0060] The present application also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the circuit board quality analysis method.

[0061] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A quality analysis method for a circuit board, characterized in that: The circuit board quality analysis method comprises: The original image data of the circuit board is obtained by multispectral imaging. After signal noise reduction and image registration, the band data is fused to form a spatial feature database. Then, a standardized multidimensional image data set is obtained through data standardization. According to the spectral band data in the standardized multi-dimensional image data set, the surface feature information of the circuit board is extracted, the feature point data is compared and analyzed with the preset threshold value, and a defect distribution feature map is generated through multi-dimensional defect feature marking; According to the defect distribution characteristic map, the spectral reflection data points on the circuit board surface are extracted, and the material uniformity is calculated in combination with the light intensity attenuation curve of each band to obtain a welding quality data table; Based on the defect distribution characteristic map and the welding quality data table, quality characteristic parameters are collected, and a circuit board quality grade index is established through grading threshold interval division to generate a quality grading data set; Extract key process data points based on the quality graded data set, perform cross-validation and deviation analysis on process parameters, and form a process adjustment parameter table; The quality grading data set and the process adjustment parameter table are merged to construct a quality analysis data matrix, and a circuit board quality assessment report is output in combination with historical parameter comparison.

2. The circuit board quality analysis method according to claim 1, characterized in that: The original image data of the circuit board is obtained by multispectral imaging, and after signal noise reduction and image registration, the band data is fused to form a spatial feature database, and then a standardized multidimensional image data set is obtained through data standardization processing, including: Synchronously collect circuit board images in three bands: ultraviolet, visible light, and near-infrared, to form original image data; Using the original image data to identify data interference items, separating ambient light fluctuation data and sensor noise data, and generating a noise characteristic spectrum; Directively eliminating interference signals in the original image data according to the noise characteristic spectrum, performing image signal enhancement, and obtaining noise-reduced image data; Dividing the noise reduction image data into a plurality of sub-bands according to the wavelength range, and performing mapping transformation on the position coordinates of each sub-band image to obtain position mapping data; The position mapping data is converted into a unified coordinate system, and image data of different bands are superimposed and merged according to the spatial distribution law to construct the spatial feature database; The data samples in the spatial feature database are normalized, statistical feature parameters are calculated, abnormal data points are removed, and the standardized multidimensional image data set is output.

3. The circuit board quality analysis method according to claim 1, characterized in that: The method extracts the circuit board surface feature information according to the spectral band data in the standardized multi-dimensional image data set, compares and analyzes the feature point data with a preset threshold, and generates a defect distribution feature map through multi-dimensional defect feature marking, including: Segmenting the standardized multidimensional image data set according to the circuit board surface area, extracting the spectral band data of each area, and generating a regional spectral data matrix; Mapping the data points in the regional spectrum data matrix to spatial coordinates, positioning and marking the characteristic points on the surface of the circuit board, and outputting the characteristic point data; Performing density clustering calculation on the feature point data, identifying the distribution law of the feature points, and counting the degree of clustering of the feature points to form a feature distribution statistics table; Based on the feature distribution statistics table, the density value of the feature point is compared with the preset threshold, and the area exceeding the threshold range is marked to obtain the abnormal area data table; Classify and count the data points in the abnormal area data table according to the defect type, calculate the spatial distribution characteristics of each type of defect, and generate a defect type distribution map; Multi-dimensional feature extraction is performed on the defect type distribution map, and correlation analysis is performed on parameters such as defect location, area, type, etc., to output the defect distribution feature map.

4. The circuit board quality analysis method according to claim 1, characterized in that: The method extracts the spectral reflection data points on the circuit board surface according to the defect distribution characteristic map, calculates the material uniformity in combination with the light intensity attenuation curve of each band, and obtains the welding quality data table, including: Performing spectral reflection sampling on the position information in the defect distribution characteristic map, recording the reflection intensity of each sampling point in different bands, and forming a spectral reflection data group; Arrange the spectral reflectance data set in order of wavelength, calculate the difference in reflectance intensity between adjacent bands, and generate a light intensity change gradient table; Normalize the data in the light intensity change gradient table, draw a light intensity attenuation curve for each sampling point, record the key inflection point position, and output attenuation characteristic data; Performing regional statistics on the attenuation characteristic data, calculating the variance of the attenuation curve in the local area, and forming a material uniformity distribution map; Based on the material uniformity distribution map, the welding point is located, the attenuation curve slope of the welding area is extracted, the welding quality parameters are counted, and the welding point quality matrix is ​​established; The parameters in the welding point quality matrix are compared and analyzed with the standard parameters, abnormal welding points are marked, and the welding quality data table is output.

5. The circuit board quality analysis method according to claim 1, characterized in that: The method of collecting quality characteristic parameters based on the defect distribution characteristic map and the welding quality data table, establishing a circuit board quality grade index through grading threshold interval division, and generating a quality grading data set includes: Extract defect density parameters from the defect distribution characteristic map, divide the defect location distribution data into regions, and output a regional defect statistics table; Classifying the welding point data in the welding quality data table according to the quality level, performing numerical statistics on the welding quality parameters, and generating a welding quality distribution map; Integrate the data of the regional defect statistics table and the welding quality distribution map, calculate the comprehensive quality score of each region, and form a quality characteristic parameter matrix; Performing numerical analysis on the quality characteristic parameter matrix, statistically analyzing the parameter distribution law, calculating the key interval nodes, and obtaining a quality threshold distribution table; Divide the quality threshold distribution table into grade intervals according to parameter value ranges, assign quality grade identifiers to different intervals, and establish the circuit board quality grade index; The quality characteristic parameter matrix is ​​classified based on the circuit board quality grade index, the classification results are sorted and counted, and the quality graded data set is output.

6. The circuit board quality analysis method according to claim 1, characterized in that: The method of extracting key process data points based on the quality classification data set, performing cross-validation and deviation analysis on process parameters, and forming a process adjustment parameter table includes: Extracting key process data from the quality classification data set, classifying the data according to process flow nodes, and forming a process data classification table; Performing correlation analysis on the data in the process data classification table, calculating the correlation coefficients between process parameters, and generating a parameter correlation matrix; Sorting the data in the parameter correlation matrix according to the size of the correlation coefficient, screening highly correlated process parameter combinations, and outputting a key parameter group; Cross-comparison verification is performed on the key parameter groups, parameter fluctuation ranges are statistically analyzed, parameter deviation values ​​are calculated, and a process parameter deviation table is obtained; Performing data analysis based on the process parameter deviation table, calculating the adjustment direction and adjustment range of each parameter, and forming a parameter adjustment plan; The feasibility of the adjustment values ​​in the parameter adjustment scheme is verified, the verification results are integrated, and the process adjustment parameter table is output.

7. The circuit board quality analysis method according to claim 1, characterized in that: The quality classification data set and the process adjustment parameter table are merged to construct a quality analysis data matrix, and a circuit board quality assessment report is output in combination with historical parameter comparison, including: Matching the quality parameters in the quality classification data set with the process parameters in the process adjustment parameter table to generate a parameter association table; Structuring the data items in the parameter association table, grouping the data according to quality level and process type, and forming a data grouping matrix; Performing statistical analysis on each group of data in the data grouping matrix, calculating the corresponding relationship between the quality parameters and the process parameters, and outputting a comprehensive analysis data table; Compare the data in the comprehensive analysis data table with the historical quality data, mark abnormal parameter change points, and generate a quality trend analysis chart; Perform data mining based on the quality trend analysis chart, extract the quality change rules, count the key influencing factors, and form a quality assessment conclusion; The quality assessment conclusion is integrated with the specific parameter data, classified and sorted according to the assessment dimensions, and the circuit board quality assessment report is output.

8. The circuit board quality analysis method according to claim 7, characterized in that: The quality assessment conclusion is integrated with the specific parameter data, classified and sorted according to the assessment dimensions, and the circuit board quality assessment report is output, including: Extract key evaluation points from the quality evaluation conclusion, classify them according to defect type labels, and generate an evaluation key statistical table; Quantitatively analyze the data items in the evaluation key statistical table, calculate the deviation value of each evaluation item, and form a deviation distribution matrix; Perform data clustering based on the deviation distribution matrix, divide risk levels according to the degree of deviation, and output a risk assessment list; Conduct traceability analysis on each indicator in the risk assessment list, extract the root causes of quality problems, and generate a problem attribution table; Conduct correlation analysis between the problem attribution table and process parameter data, statistically analyze the quality fluctuation pattern, and form a report on the direction of improvement; The improvement direction reports are systematically sorted, sorted according to the severity of the problems, and the circuit board quality assessment report is output.

9. A circuit board quality analysis system, used to implement the circuit board quality analysis method according to any one of claims 1 to 8, characterized in that: The circuit board quality analysis system comprises: The registration module is used to obtain the original image data of the circuit board using multispectral imaging, and after signal noise reduction and image registration, the band data is fused to form a spatial feature database, and then a standardized multidimensional image data set is obtained through data standardization processing; A marking module, used to extract the surface feature information of the circuit board according to the spectral band data in the standardized multi-dimensional image data set, compare and analyze the feature point data with the preset threshold value, and generate a defect distribution feature map through multi-dimensional defect feature marking; A calculation module, used to extract the spectral reflection data points on the surface of the circuit board according to the defect distribution characteristic map, calculate the material uniformity in combination with the light intensity attenuation curve of each band, and obtain a welding quality data table; Establishing a module, for collecting quality characteristic parameters based on the defect distribution characteristic map and the welding quality data table, establishing a circuit board quality grade index through grading threshold interval division, and generating a quality grading data set; An analysis module, for extracting key process data points based on the quality classification data set, performing cross-validation and deviation analysis on process parameters, and forming a process adjustment parameter table; The merging module is used to merge the quality grading data set and the process adjustment parameter table, construct a quality analysis data matrix, and output a circuit board quality assessment report in combination with historical parameter comparison.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the circuit board quality analysis method according to any one of claims 1 to 8 is implemented.

Images