A PCB molding detection method based on hole position scatter diagram and SPC analysis

Abstract

This invention relates to the technical field of printed circuit board manufacturing and quality inspection, particularly a PCB molding inspection method based on hole location scatter plots and SPC analysis. Actual hole location data for a PCB substrate is collected using an optical scanner or laser measurement device, generating a scatter plot containing coordinates, hole diameters, and tolerances. An improved density clustering algorithm is then used to identify abnormal areas. The scatter plot is then subjected to multi-dimensional analysis based on an SPC model, extracting characteristic parameters such as hole density deviation. Dynamic control limits are then established, and CPK (Computer Program Key) (CPK) is calculated to determine the machining status. When CPK falls below a threshold, optimization instructions such as drilling path correction and drill wear compensation are generated. These instructions are fed back to the equipment via an anti-saturation fuzzy PID controller to achieve closed-loop control. This method accurately detects hole machining quality, adaptively adjusts machining parameters, and effectively improves PCB molding accuracy and production stability.

Description

Technical Field

[0001] The present invention relates to the technical field of printed circuit board manufacturing and quality inspection, and in particular to a PCB forming inspection method based on hole position scatter diagram and SPC analysis. Background Art

[0002] In the electronics manufacturing sector, printed circuit boards (PCBs) are key components of electronic devices, and their manufacturing quality directly impacts their performance and reliability. As electronic products evolve toward miniaturization, higher density, and higher performance, the requirements for PCB manufacturing precision continue to increase. In particular, hole position accuracy and hole diameter consistency have become key factors in determining PCB quality.

[0003] During the PCB molding process, hole quality is impacted by a combination of factors. On one hand, process factors such as drill wear, equipment vibration, and substrate thermal expansion can lead to problems such as hole offset and diameter variation. Furthermore, differences in hole density and board material between different PCB types further complicate quality control. Traditional inspection methods rely on manual visual inspection or offline sampling. This approach is not only inefficient but also struggles to comprehensively and accurately detect subtle quality issues, failing to meet the real-time and high-precision requirements of modern PCB production.

[0004] Existing automated inspection technologies primarily include statistical process control (SPC) and intelligent algorithm-based inspection methods. However, traditional SPC models are typically based on fixed parameters and assumptions, such as data following a normal distribution and control limits remaining constant. These models struggle to adapt to non-steady-state changes in the PCB manufacturing process caused by equipment aging, process fluctuations, and other factors, and are prone to misjudgments or missed detections. While intelligent algorithm-based inspection methods, such as convolutional neural networks (CNNs), have advantages in image recognition, they lack the ability to analyze multi-scale features and lack effective integration with process parameters. Their reliability under complex working conditions needs to be improved. Furthermore, most existing inspection systems operate in open-loop mode. Even if quality issues are detected, rapid feedback and adjustment of processing parameters are impossible, making it difficult to achieve dynamic optimization and closed-loop control of the PCB forming process. Therefore, to address these issues, a PCB forming inspection method based on hole position scatter diagrams and SPC analysis is proposed. Summary of the Invention

[0005] The object of the present invention is to provide a PCB forming detection method based on hole position scatter diagram and SPC analysis to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A PCB molding detection method based on hole position scatter diagram and SPC analysis includes the following steps:

[0008] S1. Collect the actual hole position data of the PCB substrate and generate a hole position scatter diagram, which includes the coordinates, hole diameter and tolerance range of each hole position;

[0009] S2. Based on the statistical process control (SPC) model, a multi-dimensional analysis of the hole position scatter diagram was performed to extract the hole position distribution characteristic parameters, including hole position density deviation, hole group spacing standard deviation, and hole diameter variation coefficient;

[0010] S3. Build dynamic control limits based on characteristic parameters, calculate process capability index CPK in real time, and determine whether the hole machining process is in a statistically controlled state;

[0011] S4. When CPK is lower than a preset threshold, an optimization instruction is generated, the optimization instruction including a drilling path correction parameter, a drill wear compensation value, or an equipment calibration plan;

[0012] S5. Feedback the optimization instructions to the PCB molding equipment and adjust the processing parameters to achieve closed-loop control.

[0013] As a preferred solution, the generation of the hole position scatter diagram in step S1 includes:

[0014] Obtain the actual hole position data of the PCB substrate through an optical scanner or laser measuring device;

[0015] Map the actual hole position data to the design drawings to generate a scatter plot comparing the theoretical and actual positions;

[0016] An improved density clustering algorithm is used to identify abnormal clustered areas in the hole position distribution. The density threshold and minimum number of points of the improved density clustering algorithm are dynamically set according to the area of ​​the PCB substrate, the reference area, the total number of holes and the preset number of clusters. Among them, the density threshold is adjusted proportionally with the square root of the substrate area, and the minimum number of points is determined by rounding up the ratio of the total number of holes to the preset number of clusters.

[0017] As a preferred solution, the SPC model in step S2 includes:

[0018] An improved skewness-kurtosis joint test model for hole density distribution is established based on historical data, and the normality of the distribution is determined by calculating the joint statistics of sample skewness and kurtosis.

[0019] A multivariate T² control chart with dynamic covariance updating is used to jointly monitor the hole cluster spacing and hole diameter. The covariance matrix is ​​dynamically updated by the weighted average of historical data and real-time data, with the weights controlled by the forgetting factor.

[0020] An adaptive ARIMA model is introduced to predict processing trends, and its difference order is dynamically adjusted according to the ratio of the prediction error standard deviation to the reference standard deviation.

[0021] As a preferred solution, the method for constructing the dynamic control limit in step S3 is:

[0022] A nonlinear adaptive control limit adjustment strategy is adopted. The control limit bandwidth is dynamically adjusted according to the mean and standard deviation of the characteristic parameters, and the transition is smoothed by the hyperbolic tangent function.

[0023] The sliding window length is dynamically optimized based on the logarithmic relationship between the total number of samples in the current processing batch and the complexity index, and the upper limit of the window length is the preset maximum value.

[0024] As a preferred solution, the generation of the optimization instruction in step S4 includes:

[0025] If the hole density deviation exceeds the limit, the spatial autocorrelation weight is used to optimize the drilling path. The weight is calculated based on the Euclidean distance between the holes and the difference in local thermal expansion coefficient. The default values ​​of the attenuation factor are 0.1 and 0.05.

[0026] If the standard deviation of the hole group spacing exceeds the limit, the drill feed speed is corrected according to the exponential function, the sensitivity coefficient range is 0.1-0.3, and the reference spacing standard deviation is 0.02mm;

[0027] If the coefficient of variation of the hole diameter exceeds the limit, the drill wear compensation amount is determined by the benchmark compensation amount, the process capability index deviation and the cumulative use time of the drill bit, and the time attenuation coefficient is 0.001 / mm.

[0028] As a preferred solution, the closed-loop control in step S5 includes:

[0029] An anti-saturation fuzzy PID controller is used to adjust the machining parameters. Its control law combines proportional, integral, differential terms and anti-chattering compensation terms. The integral term is limited by a threshold to prevent overshoot.

[0030] After the processing parameters are adjusted, the covariance matrix is ​​updated in real time based on the data in the current sliding window.

[0031] As a preferred embodiment, the method comprises:

[0032] The Multi-Head Attention Convolutional Neural Network (MA-CNN) is introduced to calculate attention weights of different dimensions through multiple sets of queries and key matrices, thus enhancing the ability to detect local anomalies.

[0033] The anomaly probability is determined by the fusion of SPC analysis results, MA-CNN output and historical anomaly probability mean. The fusion weight is normalized by the Sigmoid function, and the historical weight range is 0.2-0.4.

[0034] As a preferred solution, the method is suitable for the detection of high-density thermoelectric separation aluminum substrates, and the adjustment coefficient of the dynamic control limit is dynamically adjusted according to the thermal deformation, original size, temperature rise and thermal expansion coefficient of the substrate.

[0035] As a preferred solution, the calculation of the local thermal expansion coefficient includes:

[0036] Real-time acquisition of PCB substrate temperature distribution based on infrared thermal imaging data;

[0037] The thermal expansion coefficient is dynamically corrected based on the difference between the local temperature and the reference temperature, which is 25°C.

[0038] As a preferred solution, the historical weight is dynamically adjusted according to the harmonic mean of the precision of the SPC model and the recall of the MA-CNN model.

[0039] It can be seen from the technical solutions provided by the present invention that the PCB forming detection method based on hole position scatter diagram and SPC analysis provided by the present invention has the following beneficial effects:

[0040] High-precision data acquisition and visual analysis: Use optical scanners or laser measuring devices to obtain hole position data, generate a comparative scatter diagram containing theoretical and actual positions through coordinate mapping, and combine with an improved density clustering algorithm to dynamically identify abnormal clustering areas. This can accurately locate hole position distribution anomalies and provide an intuitive basis for quality analysis.

[0041] Multi-dimensional SPC analysis and dynamic monitoring: Based on an improved skewness-kurtosis joint test, a multivariate T² control chart with dynamic covariance updates, and an adaptive ARIMA model, this system enables real-time monitoring and trend prediction of parameters such as hole density, spacing, and hole diameter. Compared to single-model detection, this system significantly improves detection sensitivity and effectively captures subtle fluctuations in the machining process.

[0042] Adaptive control and closed-loop optimization: Nonlinear adaptive control limits are constructed and dynamically adjusted based on sample size and batch complexity to reduce the control limit misjudgment rate. When CPK is lower than the threshold, targeted optimization instructions are generated based on characteristic parameters such as hole density deviation, spacing standard deviation, and hole diameter variation coefficient. For example, drilling path optimization based on spatial autocorrelation weights and dynamic compensation for drill wear are implemented to effectively reduce the hole diameter variation coefficient and significantly improve processing stability.

[0043] Intelligent algorithm fusion enhances anomaly detection: The Multi-Head Attention Convolutional Neural Network (MA-CNN) is introduced to integrate anomaly probabilities with the SPC model and historical data, improving the detection accuracy of minor hole position deviations. Compared with traditional methods, the false alarm rate is significantly reduced, effectively avoiding unnecessary downtime.

[0044] Adaptability to special scenarios: For high-density thermoelectric separation aluminum substrates, the dynamic control limit adjustment coefficient is associated with the thermal deformation, and infrared thermal imaging is combined with real-time calculation of the local thermal expansion coefficient. This effectively improves the hole processing accuracy in heat-sensitive areas to meet the needs of high-end PCB manufacturing.

[0045] Dynamic model evolution and reliability: An anti-saturation fuzzy PID controller is used in closed-loop control to reduce parameter adjustment overshoot. The SPC model covariance matrix is ​​updated in real time after processing parameters are adjusted, and historical weights are dynamically adjusted based on model confidence, effectively shortening the system's adaptation cycle to process changes and continuously maintaining detection accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 The figure is a flow chart of a PCB forming detection method based on hole position scatter diagram and SPC analysis according to the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0048] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0049] like Figure 1 As shown, an embodiment of the present invention provides a PCB molding detection method based on hole position scatter diagram and SPC analysis, comprising the following steps:

[0050] S1. Collect the actual hole position data of the PCB substrate and generate a hole position scatter diagram, which includes the coordinates, hole diameter and tolerance range of each hole position;

[0051] S2. Based on the statistical process control (SPC) model, a multi-dimensional analysis of the hole position scatter diagram was performed to extract the hole position distribution characteristic parameters, including hole position density deviation, hole group spacing standard deviation, and hole diameter variation coefficient;

[0052] S3. Build dynamic control limits based on characteristic parameters, calculate process capability index CPK in real time, and determine whether the hole machining process is in a statistically controlled state;

[0053] S4. When CPK is lower than a preset threshold, an optimization instruction is generated, the optimization instruction including a drilling path correction parameter, a drill wear compensation value, or an equipment calibration plan;

[0054] S5. Feedback the optimization instructions to the PCB molding equipment and adjust the processing parameters to achieve closed-loop control.

[0055] In this embodiment, the generation of the hole position distribution diagram in step S1 includes:

[0056] Obtain the actual hole position data of the PCB substrate through an optical scanner or laser measuring device;

[0057] Map the actual hole position data to the design drawings to generate a scatter plot comparing the theoretical and actual positions;

[0058] An improved density clustering algorithm is used to identify abnormal clustered areas in the hole position distribution. The density threshold and minimum number of points of the improved density clustering algorithm are dynamically set according to the area of ​​the PCB substrate, the reference area, the total number of holes, and the preset number of clusters. The density threshold is adjusted proportionally with the square root of the substrate area, and the minimum number of points is determined by rounding up the ratio of the total number of holes to the preset number of clusters.

[0059] Furthermore, the function of step S1 is to construct a visual map including hole coordinates, hole diameters, tolerances, and abnormal distribution through high-precision data collection and intelligent analysis, providing multi-dimensional basic data for subsequent SPC analysis. The following is a detailed description of step S1:

[0060] Step S1-1: Multi-device collaborative data collection:

[0061] Dual acquisition of hole position data using optical scanners and laser measuring devices:

[0062] Optical scanner: uses a linear array CCD camera to scan the PCB substrate line by line, identifies the hole contour through image edge detection algorithms (such as the Canny operator), combines perspective transformation to correct optical distortion, and outputs the two-dimensional coordinates (X, Y) of the hole position with an accuracy of up to ±5μm;

[0063] Laser measurement device: uses a semiconductor laser emitter to emit a thin beam of light. When the beam scans the edge of the hole, the hole center coordinates and aperture size are calculated using the triangular reflection principle. The specific process is as follows: the laser spot is diffusely reflected by the hole edge and then received by the CMOS sensor. The actual hole position (error ≤ ±2μm) and aperture size (resolution 0.1μm) are calculated based on the spot offset.

[0064] Data fusion: The coordinate and aperture data collected by the two types of equipment are checked for consistency, and abnormal points with deviations exceeding the tolerance range (such as ±10μm) are eliminated to generate the initial hole position data set;

[0065] Step S1-2: Coordinate mapping and comparative scatter plot construction:

[0066] Perform spatial registration and visual mapping between actual hole location data and design drawings:

[0067] Coordinate system alignment:

[0068] Extract fiducial marks (such as positioning holes and board edge contours) from the design drawings, and use the iterative closest point (ICP) algorithm to calculate the transformation parameters (translation vector (ΔX, ΔY), rotation angle θ) between the actual coordinate system and the design coordinate system to achieve rigid registration between the two.

[0069] Perform scaling correction on the actual hole coordinates after registration to compensate for thermal expansion or mechanical deformation of the PCB substrate during processing (the deformation coefficient is calculated based on the CTE parameter of the substrate material and the processing temperature difference);

[0070] Comparison chart generation:

[0071] Use professional plotting tools (such as MATLAB or Python's Matplotlib) to draw a scatter plot, where:

[0072] Theoretical hole position: represented by a hollow circle (2mm diameter), blue in color, and the coordinates of the circle center correspond to the design drawing data;

[0073] Actual hole position: represented by a solid circle in red. The coordinates of the circle center are the measured data after alignment. The size of the circle is scaled according to the hole diameter (e.g., a 1mm hole diameter corresponds to a 1mm diameter circle).

[0074] Deviation connection line: Use a black dotted line to connect the theoretical and actual center points of the same hole position. The line length directly reflects the position offset. Lines that exceed the tolerance range (such as ±50μm) are displayed in bold.

[0075] Legend annotation: Add aperture tolerance band (such as ±10μm), coordinate scale (unit: mm) and scale bar (1:1) to the figure to facilitate quantitative analysis;

[0076] Step S1-3: Abnormal area identification based on dynamic density clustering:

[0077] An improved DBSCAN algorithm is used to detect abnormal clusters or discrete points in hole distribution. The algorithm parameters dynamically adapt to the characteristics of the PCB layout:

[0078] Dynamic parameter calculation: , (in, The density threshold of the current PCB (unit: mm) reflects the density of the hole clustering. It increases nonlinearly with the substrate area to avoid misjudging large-area substrates as abnormal due to sparse holes. The default density threshold is 0.1mm, which is applicable to standard PCBs (area ≤ 100cm²). The measured PCB substrate area (unit: cm²) is calculated using the image edge detection algorithm. is the reference area, fixed at 100 cm², used for area normalization, is the minimum number of points; The total number of holes in the substrate is obtained by the data statistics in step S1-1; The number of clusters is preset, ranging from 3 to 10, and is manually set according to the complexity of the hole distribution (e.g. 8-10 for high-density boards and 3-5 for low-density boards); It is a rounding function to ensure that MinPts is a positive integer to avoid clustering failure due to insufficient number of holes);

[0079] Clustering execution:

[0080] Take each hole as the core point and calculate its The number of holes in the neighborhood (denoted as ):

[0081] like , it is determined as a core point, and all hole positions in the neighborhood are marked as the same cluster (such as cluster 1, cluster 2, ...);

[0082] like If a point does not belong to any core point neighborhood, it is considered a noise point (abnormal gathering area);

[0083] Visual annotation:

[0084] In the hole position scatter plot:

[0085] Normal clustered hole positions: represented by green diamonds, and the holes in the same cluster are connected by gray solid lines to highlight the distribution pattern;

[0086] Noise points: marked with a red five-pointed star, and their coordinates, aperture, and distance from the cluster center are listed next to the graph (e.g., "Noise point A: coordinates (10.2, 15.5) mm, 2.3 mm from the nearest cluster center");

[0087] Add analysis notes: such as "the abnormal cluster area may be caused by drill bit vibration or substrate clamping deviation and requires special monitoring";

[0088] Step S1-4: Tolerance range and attribute annotation:

[0089] Overlay tolerance information and physical properties of hole locations in a scatter plot:

[0090] Aperture tolerance: Mark the allowable deviation range of each hole position in the form of concentric circles. For example, if the theoretical aperture is 2mm and the tolerance is ±0.05mm, draw gray dotted circles with radii of 1.95mm (inner circle) and 2.05mm (outer circle);

[0091] Material properties: Different areas of the substrate (such as metal heat dissipation layer, FR-4 substrate) are distinguished by shading, and the shading color corresponds to the coefficient of thermal expansion (CTE) value (for example, red shading means CTE=18×10 -6 / ℃, blue shading indicates CTE=6×10 -6 / ℃), providing a visual reference for subsequent thermal deformation compensation;

[0092] Step S1 outputs the results:

[0093] Hole scatter plot file: An interactive PDF or vector diagram containing coordinates, hole diameters, tolerances, and clustering results, with support for zooming in and out to view details;

[0094] Data file: a hole position dataset in CSV format, with fields including hole position ID, theoretical coordinates, actual coordinates, measured aperture value, cluster label, noise marker, etc., for the SPC model in step S2 to call;

[0095] Technical advantages: Through multi-source data fusion, dynamic clustering and visual annotation, the "data-graphic" dual-dimensional representation of hole processing quality is achieved, laying a precise foundation for subsequent SPC-based process capability analysis.

[0096] In this embodiment, the SPC model in step S2 includes:

[0097] An improved skewness-kurtosis joint test model for hole density distribution is established based on historical data, and the normality of the distribution is determined by calculating the joint statistics of sample skewness and kurtosis.

[0098] A multivariate T² control chart with dynamic covariance updating is used to jointly monitor the hole cluster spacing and hole diameter. The covariance matrix is ​​dynamically updated by the weighted average of historical data and real-time data, with the weights controlled by the forgetting factor.

[0099] An adaptive ARIMA model is introduced to predict processing trends, and its difference order is dynamically adjusted according to the ratio of the prediction error standard deviation to the reference standard deviation;

[0100] Furthermore, step S2 is to perform a multi-dimensional in-depth analysis of the data in the hole position scatter diagram based on the statistical process control (SPC) model, extract key characteristic parameters that can reflect the stability and quality characteristics of the hole position processing process, and provide a quantitative basis for subsequent judgment of whether the processing process is under control. The specific steps are as follows:

[0101] Step S2-1: Establish a normality test model for pore density distribution:

[0102] Data preprocessing: Extract the coordinate information of each hole position from the hole position scatter diagram data set generated in step S1, calculate the number of holes per unit area, and use this as the hole position density data; to ensure the validity of the data, eliminate extreme values ​​caused by measurement errors or abnormal factors;

[0103] Calculate sample skewness and kurtosis: For hole density sample data 、 、 、 , sample skewness and sample kurtosis The calculation formula is as follows:

[0104] , (in, Indicates the Sample values, Indicates the index of the sample ( ), is the sample size, is the sample mean, is the sample standard deviation);

[0105] Normality test: Use the improved skewness-kurtosis joint test formula:

[0106] (in, is the test statistic, is the sample skewness, is the sample kurtosis); the calculated The value is compared with the preset critical value. If the value is less than the critical value, it is considered that the hole density distribution conforms to the normal distribution; otherwise, it indicates that the distribution is abnormal and there may be problems such as unstable processing process.

[0107] Step S2-2: Construct a multivariate T² control chart with dynamic covariance updating:

[0108] Determine monitoring variables: Select hole coordinates ( 、 direction), aperture size as monitoring variables, and form a multivariate data vector (in, is the aperture);

[0109] Initialize the covariance matrix: Calculate the initial covariance matrix based on the hole position data in the historical stable production process ;

[0110] Dynamic covariance update: As new hole position data is continuously collected, the covariance matrix is ​​updated using the following formula:

[0111] (in, is the updated covariance matrix, is the historical covariance matrix, is the real-time covariance matrix calculated based on the current new sample data, is the forgetting factor, which ranges from 0.7 to 0.9 and is used to adjust the weights of historical data and new data in covariance updating);

[0112] Calculate the T² statistic: Based on the updated covariance matrix, calculate the statistics of the multivariate T² control chart:

[0113] (in, is a statistic, is the sample mean vector, is the covariance matrix The inverse matrix of Value and control limit to judge whether the current hole processing process is in a controlled state;

[0114] Step S2-3: Introduce the adaptive ARIMA model to predict processing trends:

[0115] Data stabilization processing: Perform time series analysis on hole feature data (such as hole density, hole size, etc.) to determine the stability of the data; if the data is not stable, stabilize it through differential operation;

[0116] Determine the differential order: The differential order d is dynamically adjusted according to the processing fluctuation, and the calculation formula is:

[0117] (in, is the difference order, is the standard deviation of the prediction error, which reflects the degree of deviation between the model prediction value and the actual value; is the reference standard deviation threshold, and its value is ; is the floor function);

[0118] Model parameter estimation and prediction: According to the determined difference order , combined with the autocorrelation function (ACF) and partial autocorrelation function (PACF) to determine other parameters of the ARIMA model ( 、 ), use historical data to estimate model parameters; use the trained adaptive ARIMA model to predict future hole processing data, discover changes in processing trends in advance, and provide a basis for timely adjustment of processing parameters;

[0119] Step S2-4: Extracting hole distribution characteristic parameters:

[0120] Calculating the hole density deviation: Compare the actual hole density obtained in step S2-1 with the standard hole density required by the design, and calculate the difference between the two, i.e., the hole density deviation; this deviation value reflects the degree of deviation between the actual hole distribution and the design requirement;

[0121] Calculate the standard deviation of hole group spacing: For each hole position in the hole group, calculate the spacing between each two hole positions to obtain a hole group spacing data set. Based on this data set, calculate its standard deviation. The hole group spacing standard deviation can reflect the degree of dispersion of the hole position distribution within the hole group. The larger the standard deviation, the greater the fluctuation of the hole position spacing and the lower the machining accuracy.

[0122] Calculate the coefficient of variation of the pore size: Based on the pore size data collected in step S1, calculate the average value and standard deviation of the pore size. The formula for calculating the coefficient of variation of the pore size is:

[0123] (in, is the coefficient of variation of pore size, is the standard deviation of the aperture, The coefficient of variation of pore size is used to measure the degree of dispersion of pore size relative to the average pore size and is an important indicator for evaluating the consistency of pore size processing.

[0124] Through the above steps, the hole scatter diagram data is comprehensively and deeply analyzed, and key characteristic parameters such as hole density deviation, hole group spacing standard deviation and hole diameter variation coefficient are extracted, which provides important data support for the subsequent construction of dynamic control limits, calculation of process capability index and judgment of the stability of the hole processing process.

[0125] In this embodiment, the method for constructing the dynamic control limit in step S3 is:

[0126] A nonlinear adaptive control limit adjustment strategy is adopted. The control limit bandwidth is dynamically adjusted according to the mean and standard deviation of the characteristic parameters, and the transition is smoothed by the hyperbolic tangent function.

[0127] The sliding window length is dynamically optimized based on the logarithmic relationship between the total number of samples in the current processing batch and the complexity index, and the upper limit of the window length is the preset maximum value;

[0128] Furthermore, the function of step S3 is to establish an adaptive control boundary and quantify the processing stability based on the characteristic parameters extracted in step S2, so as to realize real-time statistical monitoring of the hole processing process. The specific operation steps are as follows:

[0129] Step S3-1: Dynamic control limit construction:

[0130] A nonlinear adaptive control limit model is used to adjust the control limit range in combination with sample size changes and processing batch characteristics. The formula is as follows:

[0131] Nonlinear control limit adjustment formula:

[0132] (in, is the upper / lower control limit of the characteristic parameter; It is the sample mean of the hole characteristic parameters (such as density deviation and hole diameter variation coefficient), reflecting the central trend of the processing process; is the sample standard deviation, which measures the fluctuation range of the characteristic parameters; is the adjusted control limit coefficient, with a default value of 3 (corresponding to a 99.73% confidence interval). In the detection scenario of high-density thermoelectric separation aluminum substrates, it is dynamically adjusted according to the thermal deformation of the substrate ( ,in, is the baseline control limit coefficient (the default value is 3, corresponding to a 99.73% confidence interval), is the basic thermal deformation; The basic original size; is the thermal expansion coefficient of the aluminum substrate; The temperature rise is the difference between the current temperature of the substrate and the normal temperature (reference temperature 25°C), which is acquired in real time through infrared thermal imaging. It is updated in real time to the number of holes collected. is the baseline sample size (50), which is used to smooth the transition of control limits and avoid misjudgment caused by small sample size; is the hyperbolic tangent function, when When it approaches 1, the control limit converges to the traditional 3σ boundary; when When the value approaches -1, the control limits are temporarily expanded to accommodate data fluctuations);

[0133] Implementation process:

[0134] For each characteristic parameter (such as hole density deviation), the current mean is calculated in real time and standard deviation ;

[0135] Dynamically adjust the control limit width according to the current sample size n, and draw a control limit curve that changes over time (such as UCL is a red dotted line, LCL is a blue dotted line, and the mean is a black solid line);

[0136] Overlay the control limits on the well position scatter plot or SPC control chart to indicate whether the current sample point exceeds the limit (mark it with a red diamond if it exceeds the limit);

[0137] Dynamic optimization of sliding window length: (in, The maximum window length (200) is used to avoid delayed response to sudden abnormalities caused by excessive window length; The cumulative processing time of the current batch (unit: minutes), reflecting the processing progress; The batch complexity index (0-1) is comprehensively evaluated by factors such as the number of substrate layers, hole density, material type, etc. (such as high-density interconnection board , single-sided board ); It is a natural logarithmic function, which is used to suppress the linear influence of the complexity index on the window length and ensure smooth changes in the window length;

[0138] Application logic:

[0139] Complex batches (such as multilayer boards) use longer windows ( ), improve the stability of trend analysis;

[0140] Simple batches (such as single panels) use shorter windows ( ), speed up the response to exceptions;

[0141] Step S3-2: Real-time calculation of process capability index (CPK):

[0142] For the key quality characteristics of hole processing (such as hole diameter and hole coordinate deviation), the following formula is used to calculate CPK:

[0143] (CPK is the process capability index, which reflects the ability of the machining process to meet the specification requirements. The larger the value, the stronger the process capability. LSL is the lower specification limit, which is the minimum allowable value of the hole characteristic parameter (such as the minimum tolerance value of the hole diameter). USL is the upper specification limit, which is the maximum allowable value of the hole characteristic parameter (such as the maximum tolerance value of the hole diameter). is the current sample mean, is the current sample standard deviation, which is calculated based on the latest data in the sliding window);

[0144] Step S3-3: Statistical control state judgment:

[0145] Control limit determination rules:

[0146] If the real-time values ​​of the hole position characteristic parameters (such as density deviation, spacing standard deviation, and hole diameter coefficient of variation) all fall within the UCL / LCL range, and there are no abnormal patterns such as 7 consecutive points increasing / decreasing or 9 consecutive points being on the same side of the mean (following the SPC control chart abnormality criteria), the process is judged to be in a state of statistical control;

[0147] If any characteristic parameter exceeds the control limit or shows an abnormal pattern, an early warning signal is triggered, prompting the need for further analysis of the cause;

[0148] CPK threshold determination:

[0149] The default CPK threshold is usually 1.33 (corresponding to the "sufficient capability" level of the ISO9001 standard):

[0150] When CPK≥1.33, the process capability is determined to be sufficient and the processing quality is stable;

[0151] When CPK<1.33, the process capability is determined to be insufficient and the optimization mechanism of step S4 needs to be started;

[0152] Step S3-4: Visual monitoring of control limits and CPK:

[0153] Integrate dynamic control limits and CPK values ​​into the real-time monitoring system interface:

[0154] Control chart display: With time as the horizontal axis and hole characteristic parameters as the vertical axis, UCL, LCL, mean line (CL) and real-time data points are plotted. Abnormal points are automatically marked in red and a pop-up window prompts (such as "The coefficient of variation of the hole diameter exceeds the UCL, the current value is 0.08>0.05");

[0155] CPK trend curve: Draw the curve of CPK value changing with time in real time, set green (CPK ≥ 1.33), yellow (1.0 ≤ CPK < 1.33), and red (CPK < 1.0) warning intervals to intuitively reflect process capability fluctuations;

[0156] Data storage and traceability: Save control limit parameters, CPK values ​​and abnormal records by timestamp, support historical data query and trend analysis, and provide data support for process improvement;

[0157] Step S3 core technical points:

[0158] Dynamic adaptability: Through the hyperbolic tangent function and sliding window mechanism, the control limits can be automatically adjusted according to the sample size and batch complexity, solving the limitations of traditional fixed control limits in small samples or complex processes.

[0159] Multi-dimensional monitoring: Combining control limit deviation judgment and CPK quantitative evaluation, the processing status is judged from the dual dimensions of "process stability" and "capacity adequacy" to improve detection reliability;

[0160] Real-time response: Update control limits and CPK based on the latest data (within the window length W), ensuring that the monitoring system responds to process changes within seconds and promptly captures signs of quality fluctuations.

[0161] In this embodiment, the generation of the optimization instruction in step S4 includes:

[0162] If the hole density deviation exceeds the limit, the spatial autocorrelation weight is used to optimize the drilling path. The weight is calculated based on the Euclidean distance between the holes and the difference in local thermal expansion coefficient. The default values ​​of the attenuation factor are 0.1 and 0.05.

[0163] If the standard deviation of the hole group spacing exceeds the limit, the drill feed speed is corrected according to the exponential function, the sensitivity coefficient range is 0.1-0.3, and the reference spacing standard deviation is 0.02mm;

[0164] If the coefficient of variation of the hole diameter exceeds the limit, the drill wear compensation amount is determined by the benchmark compensation amount, the process capability index deviation and the accumulated use time of the drill bit, and the time decay coefficient is 0.001 / mm;

[0165] Furthermore, the function of step S4 is to generate accurate process optimization instructions based on the insufficient machining process capability (CPK < preset threshold, such as 1.33) determined in step S3 through multi-dimensional analysis of the reasons for the exceeding of the hole position distribution characteristic parameters, thereby realizing intelligent adjustment of the machining parameters. The specific operation steps are as follows:

[0166] Step S4-1: Optimization strategy for hole density deviation exceeding the limit:

[0167] Trigger condition: Hole density deviation > design tolerance threshold (e.g., ±5%), indicating that hole distribution deviates from theoretical value, which may be caused by unreasonable drilling path or thermal deformation of substrate;

[0168] Spatial autocorrelation weights optimize drilling paths: (in, Hole and The spatial weight between them is used to measure the effect of drilling sequence on thermal deformation, and the smaller the weight, the higher the priority; Hole and Euclidean distance (unit: mm), the closer the distance, the more significant the thermal accumulation effect; 、 Hole 、 Thermal expansion coefficient of the area (unit: C), reflects the thermal deformation characteristics of the material; (Default 0.1mm -2 )、 (Default 0.05 C / mm) is the attenuation factor, which adjusts the ratio of the influence of distance and CTE difference on the weight);

[0169] Path planning implementation:

[0170] According to the weight matrix Construct a drilling sequence priority queue to prioritize processing of holes with low weights (such as holes at the edge of dense areas) to reduce local temperature rise caused by continuous drilling;

[0171] Insert cooling intervals (e.g., pausing for 2 seconds after drilling every 5 holes) in areas with high CTE differences (e.g., the interface between metal inserts and FR-4 substrate) to reduce thermal deformation accumulation.

[0172] Step S4-2: Optimization strategy for exceeding the standard deviation of hole group spacing:

[0173] Trigger condition: The standard deviation of hole group spacing is greater than the reference value (e.g., 0.02 mm), indicating that the hole spacing fluctuates too much, which may be caused by unstable drill feed speed or equipment vibration.

[0174] Drill feed speed correction: (in, is the feed speed after correction (unit: mm / s), is the original speed; is the sensitivity coefficient (0.1-0.3), which decreases as the drill diameter increases (e.g., 0.3 for a φ1mm drill and 0.1 for a φ3mm drill); is the deviation between the current standard deviation and the target value (0.02mm); is the reference spacing standard deviation, fixed at 0.02 mm);

[0175] Vibration suppression measures:

[0176] If the corrected speed is lower than the minimum process speed (e.g. 5 mm / s), the equipment vibration monitoring module is automatically started to detect the spindle vibration frequency through the acceleration sensor;

[0177] When the vibration frequency couples with the drill bit's natural frequency (e.g., a 100 Hz resonance peak is detected), adjust the spindle speed (e.g., from 20,000 rpm to 18,000 rpm) to avoid the resonance range;

[0178] Step S4-3: Optimization strategy for exceeding the limit of pore size variation coefficient:

[0179] Trigger condition: Coefficient of variation (CV) of hole diameter > 5%, indicating poor hole diameter consistency, mainly caused by drill bit wear or spindle runout;

[0180] Drill wear compensation calculation: (in, The drill wear compensation value (unit: mm) is used to adjust the drilling depth or hole diameter compensation parameters; The default compensation is 0.01mm, which is preset based on the drill bit material and the hardness of the processed plate. is the current process capability index (reflects only the aperture size), is the target value (1.33); is the time attenuation coefficient (0.001 mm / min), reflecting the cumulative effect of drill bit wear over time; The accumulated usage time of the drill bit (unit: minutes), collected in real time through the device sensor);

[0181] Spindle precision calibration:

[0182] If the CV is still greater than 5% after compensation, the spindle radial runout detection is triggered: a micrometer is used to measure the radial displacement of the spindle during rotation. If the runout is greater than 10μm, the device calibration command is automatically generated, including:

[0183] Adjust the spindle bearing preload (e.g. re-tighten with a torque wrench to the standard value of 15 N·m);

[0184] Clean and center the drill shank (use a shank cleaner to remove oil and dirt, and use a tool setter to calibrate the tool runout to <5μm);

[0185] Step S4-4: Multimodal generation of optimization instructions:

[0186] Instruction type and format:

[0187] Digital signal: Drilling path correction parameters (such as hole processing sequence list) and feed speed value (such as v=12mm / s) are transmitted to the machine tool controller in real time through the RS-232 interface;

[0188] Text work order: Equipment calibration plans (e.g., "Spindle runout is out of tolerance, precision calibration process required") are pushed to the mobile device of equipment maintenance personnel in PDF format;

[0189] Visual prompts: Use icons to mark optimization items on the machine tool HMI interface (such as a blue arrow for path adjustment and a yellow wrench for calibration requirements);

[0190] Instruction priority management:

[0191] Emergency instructions (such as drill wear causing hole diameter deviation) have the highest priority, immediately interrupting the current processing and executing compensation;

[0192] Optimization instructions (such as path adjustment) are automatically loaded after the current batch processing is completed to avoid interrupting continuous production;

[0193] After the calibration command is triggered three times cumulatively, the machine will be forced to shut down to ensure the accuracy of the equipment is restored;

[0194] Step S4 technical advantages:

[0195] Accurately locate the root cause of the fault: associate the specific process problem with the type of characteristic parameter exceeding the limit (density deviation / spacing fluctuation / aperture variation), avoiding the blindness of empirical adjustments;

[0196] Multi-parameter collaborative optimization: Integrates geometric parameters (hole coordinates), physical parameters (CTE, vibration frequency) and processing parameters (speed, compensation) to achieve thermal-mechanical-control multi-field coupled optimization;

[0197] Closed-loop control pre-adaptability: Optimization instructions include both immediate adjustments (such as speed correction) and long-term maintenance (such as equipment calibration), taking into account both production efficiency and quality stability.

[0198] In this embodiment, the closed-loop control in step S5 includes:

[0199] An anti-saturation fuzzy PID controller is used to adjust the machining parameters. Its control law combines proportional, integral, differential terms and anti-chattering compensation terms. The integral term is limited by a threshold to prevent overshoot.

[0200] After the processing parameters are adjusted, the covariance matrix is ​​updated in real time based on the data in the current sliding window;

[0201] Furthermore, the function of step S5 is to accurately map the optimization instructions generated in step S4 to the actuators of the PCB molding equipment through a reliable data transmission link and intelligent control algorithm, forming a complete closed loop of "detection-analysis-adjustment" to ensure that the processing parameters dynamically adapt to quality requirements. The specific operation steps are as follows:

[0202] Step S5-1: Optimize the protocol conversion and transmission of instructions:

[0203] Multi-device interface adaptation:

[0204] Develop dedicated communication protocol adapters for different brands of equipment (such as Germany's KohYoung and China's Han's Laser) to convert optimization instructions (such as drilling paths and compensation values) into control codes that can be recognized by the equipment:

[0205] Drilling path: Convert the hole priority queue into the processing sequence instructions in G code (such as G00X10Y15; G01Z-2F100);

[0206] Drill compensation: Set the wear compensation value Converted to machine tool radius compensation parameters (such as , assuming the drill diameter is 1mm, is the register number of tool radius compensation in the machine tool control system, used to store the compensation value of the drill radius);

[0207] Equipment calibration: Convert spindle runout calibration instructions into PLC control signals (e.g. M60 code triggers calibration procedure);

[0208] Adopting OPCUA industrial communication standard, it realizes real-time transmission of instruction data through Ethernet, with communication delay ≤50ms, ensuring synchronization of instructions and processing progress;

[0209] Transmission reliability guarantee:

[0210] The command data packet is appended with a CRC check code, which is executed after verification by the receiving end. If the verification fails, it will be automatically retransmitted (up to 3 times);

[0211] Establish an instruction execution log system to record the sending time, device reception status and execution result (success / failure) of each instruction for subsequent retrospective analysis;

[0212] Step S5-2: Parameter adjustment based on anti-saturation fuzzy PID:

[0213] The anti-saturation fuzzy PID controller is used to realize the dynamic adjustment of processing parameters. The control law formula is:

[0214] (in, It is the controller output (such as feed speed adjustment, spindle speed correction value); 、 、 They are proportional coefficient, integral coefficient and differential coefficient respectively, and are adjusted online through fuzzy reasoning (such as when the error m, automatically increases Speed ​​up response); Target value-actual value, such as hole coordinate deviation and hole size error; is the saturation function: ( The default value is the integral anti-saturation threshold. ,prevent the integral term from over-accumulating when the error is large, causing the system to overshoot); is a sign function used to determine the error direction; 1 is the anti-bounce coefficient ( ), suppressing fluctuations in control quantity when errors are small);

[0215] Application scenarios:

[0216] Feed speed adjustment: When the step S4-2 calculates When there is a deviation from the current speed, the controller Dynamically adjust the inverter output and adjust the accuracy ;

[0217] Drill compensation execution: The controller drives the servo motor to adjust the Z-axis feed depth, compensate for the hole diameter deviation caused by drill wear, and improve positioning accuracy. ;

[0218] Step S5-3: Real-time iterative update of SPC model:

[0219] After the processing parameters are adjusted, the SPC model parameters of step S2 are updated synchronously to ensure that the monitoring system reflects the latest process status:

[0220] Covariance matrix update:

[0221] (where, reflects the discrete degree of the adjusted hole position data; is the current sliding window length (dynamically determined by step S3-1); For the window A hole position data vector (including coordinates and aperture); is the mean vector of data in the window);

[0222] Adaptive correction of model parameters:

[0223] If the CPK value of 5 consecutive samples after adjustment is stable above 1.33, the optimization is determined to be effective and the current covariance matrix is Stored in the model library as historical data;

[0224] If CPK does not meet the standard (e.g. <1.0), a secondary optimization is automatically triggered: step S2-(4) is re-executed, and the number of optimizations is recorded in the system (a manual intervention prompt is triggered when the number of optimizations is ≥3).

[0225] Step S5-4: Verification and visualization of closed-loop control effect:

[0226] Real-time data feedback:

[0227] Embed a closed-loop control monitoring dashboard on the equipment HMI interface to display real-time:

[0228] Optimize the command execution status (such as "path adjustment loaded", "drill compensation value +0.005mm");

[0229] The trend of key quality parameters (such as pore diameter mean and pore position deviation standard deviation) is shown in green solid lines for adjusted data and red dashed lines for unadjusted data, allowing for intuitive comparison of optimization effects.

[0230] Statistical Process Validation:

[0231] After each processing batch (e.g. 50 substrates), a closed-loop control effect report is automatically generated, including:

[0232] The increase in CPK value (e.g., from 0.8 to 1.5);

[0233] Reduction of characteristic parameter fluctuations (e.g., the standard deviation of hole group spacing is reduced from 0.03mm to 0.015mm);

[0234] Optimize instruction execution efficiency (e.g., average instruction transmission delay is 23ms, and execution success rate is 99.7%);

[0235] Exception handling mechanism:

[0236] If the closed-loop control fails to meet the CPK standard for two consecutive times, the system will automatically generate a "Process Abnormality Analysis Report" that summarizes the hole position data, equipment status log, and optimization records, and pushes it to the process engineer's account, prompting manual intervention (such as changing the drill bit model or adjusting the substrate fixture);

[0237] Step S5 core technical points:

[0238] Heterogeneous system integration capability: Through protocol conversion and standardized interfaces, the detection system can be seamlessly connected with equipment of different brands, and is compatible with mainstream PCB processing equipment (such as drilling machines and gong machines);

[0239] Robust control algorithm: Anti-saturation fuzzy PID combined with feedforward compensation effectively suppresses mechanical system inertia delay and external interference (such as workshop temperature fluctuations), ensuring smooth parameter adjustment;

[0240] Dynamic model evolution: The SPC model is continuously iterated based on the latest processing data, so that the monitoring standards are gradually optimized as the process improves, forming an intelligent closed loop of "detection-adjustment-learning".

[0241] In this embodiment, the method includes:

[0242] The multi-head attention convolutional neural network MA-CNN is introduced to calculate the attention weights of different dimensions through multiple sets of queries and key matrices to enhance the detection ability of local anomalies. The attention weight is calculated as:

[0243] (in, For the Positions in the attention head and The attention weights between 、 For the The query and key matrices of the attention heads, is the dimension scaling factor, The position index in the sequence ranges from 1 to , is the sequence length);

[0244] The anomaly probability is determined by the fusion of SPC analysis results, MA-CNN output and historical anomaly probability mean. The fusion weight is normalized by the Sigmoid function. The historical weight range is 0.2-0.4. The anomaly probability fusion formula is improved to: (in, is the final abnormal probability, is the Sigmoid function, is the abnormal probability predicted by the SPC model, is the abnormal probability predicted by the MA-CNN model, is the mean of historical abnormal probability, is the SPC model weight (0.6-0.8), is the historical weight (0.2-0.4).

[0245] In this embodiment, the method is applicable to the detection of high-density thermal and electrical separation aluminum substrates. The adjustment coefficient of the dynamic control limit is dynamically adjusted according to the thermal deformation, original size, temperature rise and thermal expansion coefficient of the substrate. The formula is:

[0246] (in, is the adjusted control limit coefficient, is the baseline control limit coefficient, is the thermal deformation of the substrate, is the original size of the substrate, is the thermal expansion coefficient of the aluminum substrate ( ), is the temperature rise).

[0247] In this embodiment, the calculation of the local thermal expansion coefficient includes:

[0248] Real-time acquisition of PCB substrate temperature distribution based on infrared thermal imaging data;

[0249] Dynamically correct the thermal expansion coefficient based on the difference between the local temperature and the reference temperature, with the reference temperature being 25°C;

[0250] Specifically:

[0251] Difference in thermal expansion coefficient The calculation includes:

[0252] Real-time acquisition of local area temperature distribution based on infrared thermal imaging data ;

[0253] Dynamic calculation of local thermal expansion coefficient: (in, For coordinates The local thermal expansion coefficient at is the thermal expansion coefficient at room temperature, is the real-time temperature of the location, is the reference temperature ( )).

[0254] In this embodiment, the historical weight is dynamically adjusted based on the harmonic mean of the precision of the SPC model and the recall of the MA-CNN model;

[0255] Historical Weight Dynamic adjustment based on model confidence:

[0256] (in, is the historical weight after dynamic adjustment, is the accuracy of the SPC model, is the recall rate of the MA-CNN model).

[0257] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A PCB molding detection method based on hole position scatter diagram and SPC analysis, characterized by: The following steps are involved: S1. Collecting actual hole position data of the PCB substrate and generating a hole position scatter diagram, wherein the hole position scatter diagram includes the coordinates, hole diameter and tolerance range of each hole position; S2. Based on the statistical process control (SPC) model, a multi-dimensional analysis is performed on the hole position scatter diagram to extract hole position distribution characteristic parameters, including hole position density deviation, hole group spacing standard deviation, and hole diameter variation coefficient. The SPC model in step S2 includes: An improved skewness-kurtosis joint test model for hole density distribution is established based on historical data, and the normality of the distribution is determined by calculating the joint statistics of sample skewness and kurtosis. A multivariate T² control chart with dynamic covariance updating is used to jointly monitor the hole cluster spacing and hole diameter. The covariance matrix is ​​dynamically updated by the weighted average of historical data and real-time data, with the weights controlled by the forgetting factor. An adaptive ARIMA model is introduced to predict processing trends, and its difference order is dynamically adjusted according to the ratio of the prediction error standard deviation to the reference standard deviation; S3. Construct dynamic control limits based on the characteristic parameters, calculate the process capability index CPK in real time, and determine whether the hole processing process is in a statistically controlled state; S4. When CPK is lower than a preset threshold, generating an optimization instruction, wherein the optimization instruction includes a drilling path correction parameter, a drill wear compensation value, or an equipment calibration solution; S5. Feedback the optimization instructions to the PCB molding equipment and adjust the processing parameters to achieve closed-loop control.

2. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 1, characterized in that: The generation of the hole position scatter diagram in step S1 includes: Obtain the actual hole position data of the PCB substrate through an optical scanner or laser measuring device; Map the actual hole position data to the design drawings to generate a scatter plot comparing the theoretical and actual positions; An improved density clustering algorithm is used to identify abnormal clustered areas in the hole position distribution. The density threshold and minimum number of points of the improved density clustering algorithm are dynamically set according to the area of ​​the PCB substrate, the reference area, the total number of holes and the preset number of clusters. The density threshold is adjusted proportionally with the square root of the substrate area, and the minimum number of points is determined by rounding up the ratio of the total number of holes to the preset number of clusters.

3. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 1, characterized in that: The method for constructing the dynamic control limit in step S3 is: A nonlinear adaptive control limit adjustment strategy is adopted. The control limit bandwidth is dynamically adjusted according to the mean and standard deviation of the characteristic parameters, and the transition is smoothed by the hyperbolic tangent function. The sliding window length is dynamically optimized based on the logarithmic relationship between the total number of samples in the current processing batch and the complexity index, and the upper limit of the window length is the preset maximum value.

4. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 1, characterized in that: The generation of the optimization instruction in step S4 includes: If the hole density deviation exceeds the limit, the spatial autocorrelation weight is used to optimize the drilling path. The weight is calculated based on the Euclidean distance between the holes and the difference in local thermal expansion coefficient. The default values ​​of the attenuation factor are 0.1 and 0.

05. If the standard deviation of the hole group spacing exceeds the limit, the drill feed speed is corrected according to the exponential function, the sensitivity coefficient range is 0.1-0.3, and the reference spacing standard deviation is 0.02mm; If the coefficient of variation of the hole diameter exceeds the limit, the drill wear compensation amount is determined by the benchmark compensation amount, the process capability index deviation and the cumulative use time of the drill bit, and the time attenuation coefficient is 0.001 / mm.

5. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 1, characterized in that: The closed-loop control in step S5 includes: An anti-saturation fuzzy PID controller is used to adjust the machining parameters. Its control law combines proportional, integral, differential terms and anti-chattering compensation terms. The integral term is limited by a threshold to prevent overshoot. After the processing parameters are adjusted, the covariance matrix is ​​updated in real time based on the data in the current sliding window.

6. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 1, characterized in that: The method comprises: The Multi-Head Attention Convolutional Neural Network (MA-CNN) is introduced to calculate attention weights of different dimensions through multiple sets of queries and key matrices, thus enhancing the ability to detect local anomalies. The anomaly probability is determined by the fusion of SPC analysis results, MA-CNN output and historical anomaly probability mean. The fusion weight is normalized by the Sigmoid function, and the historical weight range is 0.2-0.

4.

7. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 1, characterized in that: The method is applicable to the detection of high-density thermoelectric separation aluminum substrates, and the adjustment coefficient of the dynamic control limit is dynamically adjusted according to the thermal deformation, original size, temperature rise and thermal expansion coefficient of the substrate.

8. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 4, characterized in that: The calculation of the local thermal expansion coefficient includes: Real-time acquisition of PCB substrate temperature distribution based on infrared thermal imaging data; The thermal expansion coefficient is dynamically corrected based on the difference between the local temperature and the reference temperature, which is 25°C.

9. The PCB forming detection method based on hole position scatter diagram and SPC analysis according to claim 6, characterized in that: The historical weight is dynamically adjusted according to the harmonic mean of the precision of the SPC model and the recall of the MA-CNN model.

Images