A printing band edge residue chemical detection system

Abstract

The application provides a printing tape edge residue chemical detection system and relates to the technical field of data processing, which comprises the following steps: identifying residue categories in samples, ranking the risk of each residue category to obtain a risk ranking result; correcting the risk ranking result by spatial hetero-direction weighting to obtain a key residue selection list; a judgment module is used to establish a quantitative detection model of each residue category according to the key residue selection list, analyze the samples, and obtain the measured content value of each residue category; the measured content value is corrected by spatial hetero-direction weighting to obtain a corrected content value; and the corrected content value is compared with the safety threshold of the corresponding residue category to obtain a judgment result. The application realizes effective identification, accurate quantification and reliable judgment of the uneven distribution of printing tape edge residues.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a chemical detection system for scraping residue on printed tape. Background Technology

[0002] In the production process of electronic components such as printed circuit boards, the spatial distribution of residues after the scraping process in the scraping zone is usually significantly non-uniform, especially near edges, corners, or areas affected by airflow, where the degree of residue aggregation and composition vary considerably. Existing detection methods for processing such samples often employ uniform grid sampling or select several feature points based on operator experience for data collection, and characterize the overall residue status by using simple arithmetic averaging or equal-weighted fusion of the detection results from each sampling point. However, in practical applications, due to the anisotropic spatial distribution of residues, the residue characteristics in key risk areas (such as areas with dense circuitry or near solder interfaces) often deviate significantly from those of uniformly sampled points. If the detection data from different spatial locations are processed with equal weights, the key information contained in densely distributed or high-risk areas is easily diluted by the detection values ​​from a large number of non-critical areas, leading to discrepancies between the final identified risk residue categories and content levels and the actual situation. This accuracy problem caused by the lack of effective modeling of spatial heterogeneity and the lack of differentiated weight correction for detection data is a shortcoming of existing detection methods when processing scraping residue samples. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a chemical detection system for scraping residue on printed tape, which realizes effective identification, accurate quantification and reliable determination of scraping residue on printed tape under uneven distribution.

[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0005] In a first aspect, a chemical detection system for scraping residue on printed tape includes:

[0006] The processing module is used to acquire samples of printing tape scraping residue, perform ultrasonic detection and analysis on the samples, collect ultrasonic echo signals and construct an acoustic reflection wave phase inversion basis; extract the distribution characteristic data of the residue based on the acoustic reflection wave phase inversion basis; set three spatial anchoring corner points on the sample, collect the corner point characteristic parameters at each spatial anchoring corner point, delineate and segment the test area framework based on the spatial distribution of the three spatial anchoring corner points, and obtain the cell eigenvalues; perform iterative calculations based on the cell eigenvalues ​​and distribution characteristic data to obtain a converged matrix perturbation progression conformation; calculate the cell eigenvalues ​​based on the matrix perturbation progression conformation to obtain spatial anisotropic weighting;

[0007] The parsing module is used to match and parse the corner feature parameters with the preset feature database to identify the residue categories in the sample, rank the risk of each residue category, and obtain the risk ranking result; the risk ranking result is corrected by spatial anisotropic weighting to obtain the key residue selection list.

[0008] The judgment module is used to establish quantitative detection models for each residue category based on the key residue selection list, analyze the samples, and obtain the measured content value of each residue category; correct each measured content value by spatial anisotropic weighting to obtain the corrected content value; and compare the corrected content value with the safety threshold of the corresponding residue category to obtain the judgment result.

[0009] In a second aspect, a computing device includes:

[0010] One or more processors;

[0011] A storage device for storing one or more programs that, when executed by one or more processors, enable the one or more processors to implement the system.

[0012] Thirdly, a computer-readable storage medium storing a program that, when executed by a processor, implements the system.

[0013] The above-described solution of the present invention has at least the following beneficial effects:

[0014] The method constructs an acoustic reflection wave phase inversion basis through a processing module, accurately extracts residue distribution feature data, delineates the test area framework by combining three spatial anchoring corner points, and obtains cellular eigenvalues ​​by segmentation. Through iterative calculation, a convergent matrix perturbation variation conformation is obtained, thereby achieving spatial anisotropic weighting. This method assigns differentiated weights to the distribution differences of residues in different areas such as edges, corners, and densely lined areas, avoiding the dilution of residue information in key risk areas by data from non-key areas, making the detection results more consistent with the actual residue situation. The corner point feature parameters are matched with a preset feature database to accurately identify residue categories and rank them by risk. At the same time, spatial anisotropic weighting is used to correct the ranking results, which can prioritize the screening of high-risk residues in key risk areas such as densely lined areas and welding interfaces, forming an accurate list of key residues. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a chemical detection system for scraping residue on printed tape provided in an embodiment of the present invention.

[0016] Figure 2This is a flowchart illustrating the process of matching and parsing corner feature parameters with a preset feature database to identify the types of residues in the sample, ranking each type of residue by risk to obtain a risk ranking result, and correcting the risk ranking result by using spatial anisotropic weighting to obtain a list of key residues. Detailed Implementation

[0017] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0018] like Figure 1 As shown, an embodiment of the present invention provides a chemical detection system for scraping residue on printed tape, comprising:

[0019] The processing module is used to acquire samples of printing tape scraping residue, perform ultrasonic detection and analysis on the samples, collect ultrasonic echo signals and construct an acoustic reflection wave phase inversion basis; extract the distribution characteristic data of the residue based on the acoustic reflection wave phase inversion basis; set three spatial anchoring corner points on the sample, collect the corner point characteristic parameters at each spatial anchoring corner point, delineate and segment the test area framework based on the spatial distribution of the three spatial anchoring corner points, and obtain the cell eigenvalues; perform iterative calculations based on the cell eigenvalues ​​and distribution characteristic data to obtain a converged matrix perturbation progression conformation; calculate the cell eigenvalues ​​based on the matrix perturbation progression conformation to obtain spatial anisotropic weighting;

[0020] The parsing module is used to match and parse the corner feature parameters with the preset feature database to identify the residue categories in the sample, rank the risk of each residue category, and obtain the risk ranking result; the risk ranking result is corrected by spatial anisotropic weighting to obtain the key residue selection list.

[0021] The judgment module is used to establish quantitative detection models for each residue category based on the key residue selection list, analyze the samples, and obtain the measured content value of each residue category; correct each measured content value by spatial anisotropic weighting to obtain the corrected content value; and compare the corrected content value with the safety threshold of the corresponding residue category to obtain the judgment result.

[0022] In this embodiment of the invention, a phase inversion basis for acoustic reflection waves is constructed through a processing module to accurately extract residual distribution feature data. The framework of the test area is defined by three spatial anchoring corner points and segmented to obtain cellular eigenvalues. A convergent matrix perturbation variation conformation is obtained through iterative calculation, thereby achieving spatial anisotropic weighting. This method assigns differentiated weights to the distribution differences of residuals in different areas such as edges, corners, and densely packed lines, avoiding the dilution of residual information in key risk areas by data from non-key areas, making the detection results more consistent with the actual residual situation. The corner point feature parameters are matched with a preset feature database to accurately identify residual categories and perform risk ranking. At the same time, spatial anisotropic weighting is used to correct the ranking results, which can prioritize the screening of high-risk residuals in key risk areas such as densely packed lines and welding interfaces, forming an accurate list of key residuals.

[0023] In a preferred embodiment of the present invention, a sample of printing tape scraping residue is obtained, and the sample is subjected to ultrasonic detection and analysis. Ultrasonic echo signals are collected, and a phase inversion basis for acoustic reflection waves is constructed. Distribution characteristic data of the residue are extracted based on the acoustic reflection wave phase inversion basis. Three spatial anchoring corner points are set on the sample, and corner point characteristic parameters at each spatial anchoring corner point are collected. The framework of the test area is delineated and segmented based on the spatial distribution of the three spatial anchoring corner points to obtain cellular eigenvalues. Iterative calculations are performed based on the cellular eigenvalues ​​and distribution characteristic data to obtain a converged matrix perturbation progression conformation. The cellular eigenvalues ​​are calculated based on the matrix perturbation progression conformation to obtain spatial anisotropic weighting, which may include:

[0024] The raw ultrasonic echo signal fed back by the sample under ultrasonic excitation was collected. The raw ultrasonic echo signal underwent time-domain filtering and frequency-domain transformation to extract characteristic reflection wave groups. Specifically, this included: collecting the raw ultrasonic echo signal fed back by the sample under ultrasonic excitation, which contained reflection information from different media interfaces within the sample, corresponding to the interface reflection signals between the printed tape scraping residue and the substrate, and between different components of the residue; performing time-domain filtering on the raw ultrasonic echo signal to remove environmental noise and interference signals; and then performing frequency-domain transformation on the filtered signal using the Fast Fourier Transform (FFT) method to convert the time-domain signal to the frequency-domain signal. During the transformation process, the filtered time-domain signal was first zero-padded to zeros. The length (u is a positive integer) is used to improve the transformation accuracy, and then the Fourier transform formula is applied. Convert the time signal in the time domain into a frequency signal in the frequency domain. Represents frequency domain signals, Represents angular frequency. The signal represents the time domain, where t represents time and j represents the imaginary unit. Signal components with frequencies in the range of 20 kHz to 500 kHz and amplitudes more than three times higher than the noise amplitude are selected to extract characteristic reflection wave groups that can characterize the residue.

[0025] Based on the characteristic reflection wave group, the location of the reflection interface corresponding to each reflection wave is identified, and the amplitude of the reflection wave at each reflection interface is extracted to obtain the reflection coefficient sequence. Specifically, this includes: based on the characteristic reflection wave group, identifying the location of the reflection interface corresponding to each reflection wave, each reflection interface corresponding to the interface of different media inside the sample, and extracting the amplitude of the reflection wave at each reflection interface. These amplitude values ​​are then arranged in the order of the reflection interfaces to obtain the reflection coefficient sequence.

[0026] The reflection coefficient sequence is weighted and averaged to decompose it into a homogenized reflection coefficient component and a micro-fluctuation correction component. The homogenized reflection coefficient component and the micro-fluctuation correction component are then superimposed and reconstructed to obtain a multi-scale homogenized reflection coefficient sequence. Specifically, this involves: weighting the reflection coefficient sequence, where the weights are determined based on the amplitude of the reflected wave at each reflection interface (larger amplitude, higher weight), with weights ranging from 0.1 to 1.0. This weighted averaging process decomposes the reflection coefficient sequence into a homogenized reflection coefficient component and a micro-fluctuation correction component. The homogenized reflection coefficient component characterizes the overall reflection properties of the sample, while the micro-fluctuation correction component corrects errors caused by localized minor fluctuations. Finally, the homogenized reflection coefficient component and the micro-fluctuation correction component are superimposed and reconstructed to obtain a multi-scale homogenized reflection coefficient sequence. The micro-wave correction components are superimposed and reconstructed. During reconstruction, the correspondence between the homogenized reflection coefficient components and the micro-wave correction components is first clarified to ensure that both correspond one-to-one with the reflection interface position. Based on the homogenized reflection coefficient components, the components corresponding to each reflection interface position are superimposed one by one. During superposition, the micro-wave correction component at that position is multiplied by a set superposition coefficient of 0.8. Then, the calculated correction component is superimposed point by point to the homogenized reflection coefficient components at the corresponding positions. During the superposition process, the changes in the reflection coefficient values ​​after superposition are monitored in real time to ensure that there are no abnormal fluctuations in the values ​​after superposition. After the superposition is completed, the superposition results of all reflection interface positions are integrated to finally obtain a multi-scale homogenized reflection coefficient sequence that can take into account both overall and local characteristics.

[0027] Based on the multi-scale homogenized reflection coefficient sequence, the acoustic impedance difference between adjacent reflection interfaces is calculated to obtain the acoustic impedance difference parameter. Then, based on the acoustic impedance difference parameter and its corresponding reflection interface position, a phase inversion basis for acoustic reflection waves is established. Specifically, this includes: calculating the acoustic impedance difference between adjacent reflection interfaces based on the multi-scale homogenized reflection coefficient sequence by dividing the difference in reflection coefficients between two adjacent reflection interfaces by the distance between the two reflection interfaces to obtain the acoustic impedance difference parameter; and then establishing a one-to-one correspondence between the acoustic impedance difference parameter and its corresponding reflection interface position to establish a phase inversion basis for acoustic reflection waves.

[0028] Based on the acoustic reflection wave phase inversion basis, the initial acoustic impedance value corresponding to each spatial point is calculated point by point within the sampling spatial domain of the sample to obtain the initial acoustic impedance distribution data. Specifically, this includes: calculating the initial acoustic impedance value corresponding to each spatial point within the sampling spatial domain of the sample, based on the acoustic reflection wave phase inversion basis. The sampling spatial domain covers the entire area of ​​the printed tape scraping residue sample. The point-by-point calculation is based on the acoustic impedance difference parameter and the position of the reflection interface in the acoustic reflection wave phase inversion basis, combined with the distance between each spatial point and the reflection interface, using the formula... = The initial acoustic impedance value at each spatial point was calculated, where Representing spatial points The initial acoustic impedance value, This represents the reference value of the acoustic impedance of the sample substrate. This represents the acoustic impedance difference parameter at the i-th reflecting interface. This represents the distance from the i-th reflective interface to the sample surface. Representing spatial points The distance to the i-th reflecting interface, where n represents the total number of reflecting interfaces, is used to summarize the initial acoustic impedance values ​​of all spatial points, thus obtaining the initial acoustic impedance distribution data.

[0029] Based on the initial acoustic impedance distribution data, the acoustic impedance difference between each spatial point and its adjacent points is iteratively corrected until the acoustic impedance difference between adjacent points converges to a preset threshold, generating spatial acoustic impedance distribution data. Specifically, this includes: iteratively correcting the acoustic impedance difference between each spatial point and its adjacent points based on the initial acoustic impedance distribution data. The preset threshold ranges from 0.03 to 0.08 Rayles. During each correction, the acoustic impedance difference between the current spatial point and its adjacent points is calculated using the formula... Calculation, where This represents the difference in acoustic impedance between the current spatial location and its adjacent locations. This represents the acoustic impedance value at the current spatial location. This represents the acoustic impedance values ​​of adjacent points in the current spatial location. The difference is compared to a preset threshold. If the difference is greater than the preset threshold, the acoustic impedance value of the current spatial location is adjusted by half the difference. The adjustment formula is as follows: ,in This represents the acoustic impedance value of the current spatial point after adjustment. The acoustic impedance difference between the adjusted spatial point and its adjacent points is calculated again. The above process is repeated until the acoustic impedance difference between all spatial points and their adjacent points converges to the preset threshold. At this point, spatial acoustic impedance distribution data that can accurately reflect the acoustic impedance distribution inside the sample is generated.

[0030] Spatial clustering analysis is performed based on spatial acoustic impedance distribution data to screen connected regions with consistent acoustic impedance. The geometric boundaries and distribution density of the connected regions are calculated to obtain the distribution characteristics data of the residues. Specifically, this includes: spatial clustering analysis based on spatial acoustic impedance distribution data using a density-based clustering algorithm, defining the core parameters of the algorithm and completing the specific clustering calculations, setting the cluster radius to a range of 0.1 to 0.3 mm, and the minimum number of neighboring points to 3 to 5. All spatial points in the spatial acoustic impedance distribution data are traversed, and each spatial point is used as an initial core point. The number of adjacent spatial points within the cluster radius of this initial core point is calculated. If the number of adjacent points is not less than the minimum number of neighboring points, then the initial core point is a valid core point, and it and all adjacent points within the cluster radius are grouped into the same initial cluster. Each adjacent point in the initial cluster is used as a new core point, and the above calculation process is repeated to continue searching for adjacent points within the cluster radius. Adjacent points that meet the conditions are merged into the new cluster. The process continues until no more points are added to the cluster, completing the construction of a cluster. A new initial core point is selected from the unclassified spatial points, and the above clustering process is repeated until all spatial points are classified or marked as isolated points. The criteria for an isolated point are that the number of adjacent points within the cluster radius is less than the minimum neighborhood number, and it does not belong to any cluster. Isolated points are considered error points and removed. During clustering, spatial points within each cluster are verified according to the criterion that the acoustic impedance difference is within the range of 0.02 to 0.05 Rayleigh. If the acoustic impedance difference between any two points within a cluster exceeds this range, the point exceeding the range is removed from the cluster, and the clustering is reassigned to ensure that all spatial points within each cluster have consistent acoustic impedance. Connected regions that meet the requirements are selected; these connected regions are the areas where the printing tape scraping residue is located. The geometric boundary of each connected region is calculated, and the boundary range is determined by connecting the outermost spatial points of the connected region. Simultaneously, the formula is used... Calculate the distribution density of the connected region, where The distribution density of the connected regions is represented by N, the number of spatial points within the connected regions is represented by S, and the area of ​​the connected regions is represented by S. By summing up the geometric boundaries and distribution densities of all connected regions, the distribution characteristic data of the residues are obtained.

[0031] The process involves collecting the boundary contour information of the sample, extracting the boundary point set based on this information, identifying all vertices constituting the convex hull of the sample boundary, and obtaining the convex hull vertex sequence. Specifically, this includes: collecting the sample's boundary contour information; scanning the sample edge point by point using a contact contour acquisition device; collecting the spatial coordinates of each point on the sample edge; removing abnormal coordinate points caused by equipment errors (the criterion for abnormal coordinate points is that the average deviation between the coordinates of that point and the coordinates of its 10 adjacent points is greater than 0.05 mm); extracting the sample's boundary point set, which constitutes the complete boundary contour of the sample; and then identifying all vertices constituting the convex hull of the sample boundary using a convex hull algorithm. The specific calculation process of the convex hull algorithm is as follows: taking any point in the boundary point set... As the initial convex hull vertices, traverse all boundary points in turn. Determine if the current boundary point is located outside the constructed convex hull edge, and select two adjacent points in the constructed convex hull. and Calculate vector sum vector The cross product, the formula for calculating the cross product is: If the cross product value is greater than 0, the point is located outside the convex hull and is included in the convex hull vertex set. If the cross product value is less than or equal to 0, the point is located inside the convex hull or on the edge of the convex hull and is removed. After traversal, all convex hull vertices are obtained. The convex hull vertices are the outermost vertices on the sample boundary. These vertices are then arranged in clockwise order to obtain the convex hull vertex sequence.

[0032] Based on the convex hull vertex sequence, select the first and second extreme points located at both ends of the sample's preset direction, as well as the third and fourth extreme points located at both ends perpendicular to the preset direction. Use these three extreme points as a candidate corner point set. Specifically, this involves selecting the first and second extreme points located at both ends of the sample's preset direction, where the preset direction is set as the length direction of the sample. The x-axis coordinates of all vertices in the convex hull vertex sequence are used as the criterion, with the first extreme point being the vertex corresponding to the minimum x-axis coordinate. That is, the outermost vertex at one end along the length direction; the second extreme point is the vertex corresponding to the maximum x-axis coordinate. The outermost vertex at the other end of the length direction is selected. Simultaneously, the third and fourth extreme points located at opposite ends of the preset direction are chosen. The preset direction is the width direction of the sample. The y-axis coordinates of all vertices in the convex hull vertex sequence are used as the criterion, with the third extreme point being the vertex corresponding to the minimum y-axis coordinate. That is, the outermost vertex at one end in the width direction, and the fourth extreme point is the vertex corresponding to the maximum value of the y-axis coordinate. That is, the outermost vertex at the other end of the width direction, and the first extreme point, the second extreme point, the third extreme point and the fourth extreme point are taken as the candidate corner point set.

[0033] Three spatial anchoring corner points are selected from the candidate corner point set. The first spatial anchoring corner point is the midpoint of the line connecting the first and second extreme points. The second spatial anchoring corner point is the third extreme point, and the third spatial anchoring corner point is the fourth extreme point. The spatial coordinates of the three spatial anchoring corner points are collected. Specifically, this includes: selecting three spatial anchoring corner points from the candidate corner point set, where the first spatial anchoring corner point is the midpoint of the line connecting the first and second extreme points. Let the coordinates of the first extreme point be... The coordinates of the second extreme point are The formula for calculating the midpoint coordinates is: That is, the coordinates of the first spatial anchoring corner point, and the second spatial anchoring corner point is the third extreme point. The third spatial anchoring corner is the fourth extreme point. The spatial coordinates of three spatial anchoring corner points were collected using a spatial coordinate acquisition device.

[0034] Based on the spatial coordinates of three spatial anchor points, the region enclosed by the three spatial anchor points is determined. Using this region as a basis, a predetermined offset distance is extended outwards to delineate the boundary of the framework of the region to be measured, thus obtaining the framework of the region to be measured. Specifically, this includes: determining the region enclosed by the three spatial anchor points based on their spatial coordinates, where the coordinates of the three anchor points are respectively... , , A triangular region is formed by connecting three spatial anchor points. This region is the core detection area of ​​the sample. The maximum circumcircle radius R of the core detection area is calculated using the formula R = ,in , , Let H be the lengths of the three sides of the triangular region, and let H be the area of ​​the triangular region. The lengths of the three sides are calculated using the distance formula between two points, and the area is calculated using Heron's formula: H = ... Calculate, where q is the half-perimeter of the triangular region. The preset offset distance is set to 0.5R to 1.0R. Based on this triangular area, the preset offset distance is extended outward to form a complete boundary of the test area.

[0035] Based on the boundary of the test area framework, the test area framework is divided into multiple cell blocks at equal intervals according to a preset division size, and the spatial position of each cell block is obtained. Based on the spatial position of each cell block, the corner feature parameters at the corresponding positions within each cell block are extracted, and the corner feature parameters are calculated to obtain the cellular eigenvalues. Specifically, this includes: dividing the test area framework into equal intervals according to a preset division size, the preset division size is determined based on the estimated size of the residue and the detection accuracy requirements, and the value ranges from 0.5 to 2.0 mm. After division, multiple uniformly sized cell blocks are obtained, and the side length of each cell block is consistent with the preset division size. The spatial position of each cell block is recorded. Coordinates are used, with the center coordinates of each cell block as the position identifier of the cell. Then, based on the spatial position of each cell block, corner feature parameters at the corresponding positions within each cell block are extracted. Corner feature parameters include acoustic impedance values, reflected wave amplitude values, and other feature parameters related to the residue at that position. The arithmetic mean of the corner feature parameters within each cell block is calculated. During the calculation, the total number of acquisition points for feature parameters within each cell is first counted. Then, the acoustic impedance values ​​of all acquisition points within the cell are added together and divided by the total number of acquisition points to obtain the average acoustic impedance value. Similarly, the reflected wave amplitude values ​​of all acquisition points are added together and divided by the total number of acquisition points to obtain the average reflected wave amplitude value. The average values ​​of each feature parameter are summed to obtain the cellular eigenvalues ​​of each cell block.

[0036] Based on the intrinsic eigenvalues ​​and distribution characteristic data of cells, the difference between the intrinsic eigenvalues ​​and distribution characteristic data of cells within each cell block at the same spatial location is calculated to obtain the initial disturbance deviation. Specifically, this includes: based on the intrinsic eigenvalues ​​and distribution characteristic data of cells, the difference between the intrinsic eigenvalues ​​and distribution characteristic data of cells within each cell block at the same spatial location is calculated. The calculation method is to subtract the acoustic impedance-related value in the distribution characteristic data corresponding to the same spatial location from the intrinsic eigenvalue of the cell at that location to obtain the difference value of each spatial location. Then, the difference values ​​of all spatial locations within the same cell block are arithmetically averaged to obtain the initial disturbance deviation of that cell block.

[0037] Based on the initial perturbation deviation, perturbation propagation calculations are performed sequentially along the spatial distribution direction of the sample. The calculation result of each cell block is used as the input to the adjacent cell blocks, and the perturbation deviation of each cell block is updated after each propagation calculation until the global rate of change of the perturbation deviation of all cell blocks converges to a preset threshold, generating a matrix perturbation progression conformation. Specifically, this includes: based on the initial perturbation deviation, perturbation propagation calculations are performed sequentially along the spatial distribution direction of the sample, ensuring that the spatial distribution direction only includes the length and width directions of the sample to ensure the relevance of the propagation calculation; during the propagation calculation, the initial perturbation deviation of each cell block is first recorded as the basic perturbation parameter of that cell, and then the initial perturbation deviation is directly used as the input parameter of its adjacent cell blocks. After receiving the input parameter, the adjacent cell blocks adjust their own initial perturbation deviation by adding one-third of the input perturbation deviation to their own initial perturbation deviation, thus calculating the adjusted perturbation deviation; After the adjustment is completed, the adjusted perturbation deviation is used as the new transfer parameter and passed to the next adjacent cell block of the cell, and so on, to complete the perturbation transfer in the entire sample space. After each transfer calculation, the current perturbation deviation of all cell blocks is updated in a timely manner to ensure the accuracy of subsequent transfer calculations. The above perturbation transfer calculation process is repeated, and the global rate of change of perturbation deviation of all cell blocks is calculated in real time. The specific calculation process of the global rate of change is as follows: first, calculate the average value of perturbation deviation of all cell blocks in the current calculation and the average value of perturbation deviation of all cell blocks in the previous calculation. Calculate the difference between the two and divide the difference by the average value of the previous calculation to obtain the current global rate of change. The calculated global rate of change is compared with a preset threshold, which ranges from 0.005 to 0.02. If the global rate of change is greater than or equal to the preset threshold, the perturbation transfer calculation is repeated. If the global rate of change is less than the preset threshold, the iterative calculation is stopped, and the matrix perturbation variation conformation is generated.

[0038] Based on the matrix perturbation progression conformation, the perturbation rate of change for each cell block in different spatial directions is extracted. The eigenvalues ​​of the cells are then calculated based on these perturbation rates to obtain the weighted values ​​for each cell block in different spatial directions. All weighted values ​​are then summarized to obtain spatially anisotropic weighting. Specifically, this involves: extracting the perturbation rate of change for each cell block one by one based on the generated matrix perturbation progression conformation, identifying the different spatial directions as including the length, width, and oblique directions of the sample; for each cell block, calculating its perturbation rate of change in the three different spatial directions. Specifically, the perturbation deviation of the cell block in two adjacent iterations is extracted, the difference between the two is calculated, and then this difference is divided by the number of iterations between the two iterations to obtain the perturbation rate of change for the cell block in the corresponding spatial direction. The calculations are performed separately for each of the three spatial directions. The perturbation rate ranges from 0.01 to 0.15, and the perturbation rate of each cell block is multiplied by the corresponding eigenvalue in different spatial directions based on the eigenvalue of the cell block. This yields the weighted value of the cell block in each spatial direction, with the weighted value ranging from 0.5 to 1.8. The weighted value of the cell block corresponding to the densely distributed residue area is between 1.2 and 1.8, while the weighted value of the cell block corresponding to the sparsely distributed residue area is between 0.5 and 1.1. After calculating the weighted values ​​of all cell blocks in the three spatial directions, the weighted values ​​of all cell blocks in all spatial directions are summarized and organized to obtain the spatial anisotropic weighting. The weight allocation of each direction in the spatial anisotropic weighting ranges from 0.2 to 0.4, and the sum of the weights in the three spatial directions is 1.0.

[0039] This embodiment, by combining ultrasonic detection with the construction of acoustic reflection wave phase inversion basis, can accurately extract the spatial distribution characteristics of residues, capture the distribution differences of residues in different regions, and avoid the omission of key information due to uniform sampling.

[0040] like Figure 2 As shown, in another preferred embodiment of the present invention, the corner feature parameters are matched and parsed with a preset feature database to identify the residue categories in the sample. The risk ranking of each residue category is then performed to obtain the risk ranking result. The risk ranking result is corrected using spatial anisotropic weighting to obtain a key residue selection list, which may include:

[0041] Feature identifier groups are extracted from the corner feature parameters. These feature identifier groups are then compared with standard feature identifier groups in a pre-defined feature database. Based on the similarity comparison results, the residue category corresponding to each cell block is identified, yielding the residue category identification result. Specifically, this involves a comprehensive review of the corner feature parameters of all cell blocks. These corner feature parameters are all characteristic parameters related to the residue, such as acoustic impedance values ​​and reflected wave amplitude values, collected within each cell block. The acoustic impedance value ranges from 1.5 × 10⁻⁶. 6 Up to 5.0×10 6 kg / (m²·s), with reflected wave amplitude ranging from 0.1 to 5.0 V; feature identifiers that can clearly distinguish different residue categories are extracted from the corner feature parameters of each cell block. These feature identifiers specifically include three core parameters: acoustic impedance range, reflected wave amplitude range, and the number of characteristic peaks. The acoustic impedance range is the interval between the maximum and minimum values ​​of all acoustic impedance values ​​within the cell block; the reflected wave amplitude range is the interval between the maximum and minimum values ​​of all reflected wave amplitudes within the cell block; and the number of characteristic peaks is the number of peaks in the reflected wave signal within the cell block. The total number of characteristic peaks is determined, ranging from 0 to 10. After extraction, a pre-set feature database is activated. This database is pre-constructed based on the characteristics of common printing tape scraping residues. The database stores standard feature identifier groups for various common printing tape scraping residues. Each standard feature identifier group uniquely corresponds to a type of residue, and each group clearly indicates the standard range and standard value of the acoustic impedance value, reflected wave amplitude value, and number of characteristic peaks for that residue type. The standard acoustic impedance value range is 1.5 × 10⁻⁶. 6 Up to 5.0×10 6 The standard reflected wave amplitude ranges from 0.1 to 5.0 V, and the number of standard characteristic peaks ranges from 0 to 10. The database also stores auxiliary information related to the chemical properties and physical characteristics of various residues.

[0042] Each cell is divided into blocks, and the extracted feature identifiers are compared with all standard feature identifiers in the feature database one by one. The comparison process is carried out sequentially according to the feature parameters. First, the acoustic impedance value range is compared, and the degree of overlap between the acoustic impedance value range in the cell's feature identifier group and the acoustic impedance value range in the standard feature identifier group is calculated. The degree of overlap ranges from 0 to 1.0. The higher the degree of overlap, the higher the similarity of the parameter. Next, the reflected wave amplitude range is compared. The same method is used to compare the acoustic impedance value range to calculate the degree of overlap between the two and obtain the similarity of the reflected wave amplitude range. The similarity value ranges from 0 to 1.0. Finally, the number of feature peaks is compared, and the difference between the number of feature peaks in the cell's feature identifier group and the number of feature peaks in the standard feature identifier group is calculated. The difference ranges from 0 to 10.

[0043] After calculating the similarity for each parameter, a corresponding weight is assigned based on the distinguishing power of each parameter for the residue category. The weight values ​​range from 0.1 to 1.0. The acoustic impedance value range has the highest distinguishing power and is assigned the largest weight, ranging from 0.4 to 0.5. The reflected wave amplitude value range has the second highest distinguishing power and a weight ranging from 0.3 to 0.4. The number of characteristic peaks has the lowest distinguishing power and a weight ranging from 0.1 to 0.2. The sum of the weights of the three is 1.0, ensuring that the weight allocation is consistent with the actual identification value of each parameter. The similarity weight calculation is performed by multiplying the similarity of the acoustic impedance value range by its corresponding weight, adding the similarity of the reflected wave amplitude value range by its corresponding weight, and adding the similarity of the number of characteristic peaks by its corresponding weight. This yields the overall similarity between the cell feature identifier group and the standard feature identifier group, with the overall similarity value ranging from 0 to 1.0.

[0044] A fixed similarity criterion is set, with a similarity threshold ranging from 0.7 to 0.8. When the overall similarity is greater than the threshold, the cell block corresponding to the feature identifier group can be clearly identified as the residue category corresponding to the standard feature identifier group. When the overall similarity is less than or equal to the threshold, the cell block is determined to have no clear residue or to be an interference area generated during the detection process, and it is not included in the residue category statistics. Following the above process, the feature identifier group comparison and category determination of all cell blocks are completed one by one, and the identification results of all cell blocks determined to have residue are summarized.

[0045] To identify residue categories, initial risk weights are assigned to each category. A comprehensive risk score is calculated for each category based on its frequency of distribution in the sample and the initial risk weight. The categories are then sorted in descending order based on their comprehensive risk scores to obtain a risk ranking result. Specifically, after obtaining the residue category identification results, a scan-line algorithm is used to calculate the planar region and area of ​​each cell segment corresponding to each residue category. The calculation process involves determining the planar coordinate range of all cell segments. A Cartesian coordinate system is established with the sample's length as the x-axis and width as the y-axis. The x-axis ranges from 0 to 100 mm, and the y-axis ranges from 0 to 50 mm. The planar coordinates of all cell segments corresponding to each residue category are extracted. The planar coordinates of each cell block are based on its center coordinates. Combined with the side length of the cell block (ranging from 0.5 to 2.0 mm), the coordinates of the four vertices of each cell block are determined, thus obtaining the set of planar contour vertices of all cell blocks corresponding to each type of residue. First, the coordinates of all planar contour vertices corresponding to this type of residue are sorted in ascending order of x-axis coordinates. If the x-axis coordinates are the same, they are sorted in ascending order of y-axis coordinates. After sorting, an ordered set of vertices is obtained. A scan line parallel to the y-axis is set. Starting from the minimum x-axis coordinate, the scan line moves point by point along the positive x-axis. The step size of each movement is consistent with the side length of the cell block, ranging from 0.5 to 2.0 mm, to ensure that the scan line can cover all cell regions corresponding to this type of residue.

[0046] During the scan line movement, the intersection points of the scan line and the planar contour of the residue are detected in real time. Each time the scan line moves, all intersection points with the contour are recorded. The intersection points are sorted in ascending order of their y-axis coordinates. Two adjacent intersection points are considered a line segment, and the length of this line segment is calculated. The line segment length is the absolute value of the difference between the y-axis coordinates of the two intersection points, with a length range of 0.5 to 2.0 mm. This line segment length is multiplied by the scan line's movement step size to obtain the area covered by the scan line. The areas covered by all scan lines are then summed sequentially to obtain the total area of ​​the planar region corresponding to this type of residue, with an area range of 0.25 to 2000. Following the same procedure described above, calculate the total area of ​​the planar region corresponding to each residue category. Review all residue categories in the residue identification results to clarify the degree of impact of each type of residue on the performance of the printing belt and subsequent processing. The degree of impact is mainly determined by the chemical and physical properties of the residue. For example, if the residue is corrosive, it will seriously affect the material stability of the printing belt, and its impact will be high; if the residue is only a light dust substance, it will have a small impact on the performance of the printing belt, and its impact will be low.

[0047] Based on the impact of various residues, an initial risk weight is assigned to each residue category. The initial risk weight ranges from 0.1 to 1.0, with higher initial risk weights for greater impact. Specifically, the initial risk weight ranges from 0.7 to 1.0 for high-impact residues, from 0.4 to 0.6 for medium-impact residues, and from 0.1 to 0.3 for low-impact residues. The initial risk weights for all residue categories are reasonably allocated within their respective ranges to ensure clear differentiation between different residue categories, while also guaranteeing the standardization and comparability of weight allocation and avoiding risk scoring bias caused by unreasonable weight allocation. After weight matching, the distribution frequency of each residue category in the sample is counted. The distribution frequency is calculated by counting the total number of cell partitions corresponding to that residue category. Each cell partition identified as belonging to that residue category is counted as one occurrence, with the distribution frequency ranging from 1 to 1000 occurrences. The total number obtained after the statistics is the distribution frequency of that residue category.

[0048] Simultaneously, the distribution frequency is supplemented and corrected by combining the total planar area corresponding to each residue category calculated through the scan-line algorithm. The correction method is to multiply the distribution frequency by the ratio of the total planar area of ​​the residue category to the total planar area of ​​the sample. The total planar area of ​​the sample ranges from 500 to 5000 mm², resulting in a corrected distribution frequency ranging from 0.1 to 1000. Then, the comprehensive risk score of each residue category is calculated by multiplying the corrected distribution frequency by the initial risk weight corresponding to the residue category. The result of multiplying the two is the comprehensive risk score of the residue category, ranging from 0.01 to 1000. After completing the comprehensive risk score calculation for all residue categories, all residue categories are sorted in descending order of comprehensive risk score, with the residue category with the highest comprehensive risk score ranked first, the next highest ranked second, and so on, until all residue categories are sorted. After sorting, the risk ranking result is obtained.

[0049] Based on spatial anisotropic weighting, the weighting values ​​of the corresponding cell partitions for each residue category in the sample are extracted. The comprehensive risk score is then calculated using the weighting values ​​of the corresponding cell partitions to obtain a revised risk score. Based on the revised risk score, residue categories with risk levels exceeding a preset threshold are selected, generating a key residue selection list. Specifically, this involves invoking spatial anisotropic weighting, which is obtained through iterative calculation based on cell intrinsic values ​​and distribution characteristics. This allows for the assignment of differential weights to cell partitions in different regions based on the spatial heterogeneity of residue distribution. The weights for the differentiating directions range from 0.2 to 0.4, with a total weight of 1.0 for all three directions. Specifically, the weights for the length direction range from 0.3 to 0.4, the weights for the width direction range from 0.3 to 0.4, and the weights for the diagonal direction range from 0.2 to 0.3. The weight values ​​range from 0.5 to 1.8, with the weight values ​​for cell partitioning blocks corresponding to densely distributed residue areas ranging from 1.2 to 1.8, and the weight values ​​for cell partitioning blocks corresponding to sparsely distributed residue areas ranging from 0.5 to 1.1.

[0050] For each type of residue, the weighting values ​​of all cell blocks corresponding to that type of residue are extracted from the spatial anisotropic weighting. During extraction, the corresponding weighting values ​​are accurately matched based on the cell block positions recorded in the residue category identification results to ensure that the weighting value of each cell block accurately corresponds to the residue category, avoiding correction deviations caused by incorrect weighting value extraction. After extraction, the weighting value of each cell block is calculated with the comprehensive risk score of the residue category corresponding to that cell block. The calculation method is to multiply the comprehensive risk score corresponding to the cell block by the weighting value of the cell block. The result of the multiplication is the corrected risk score of that type of residue in that cell block. The corrected risk score ranges from 0.005 to 1800. During the calculation process, for each type of residue, the corrected risk score of all corresponding cell blocks is calculated one by one. At the same time, the corrected risk score and corresponding position of each cell block are recorded to ensure that the calculation process is traceable and the calculation results are accurate. After the calculation is completed, the corrected risk scores of all cell division blocks of the same type of residue are summarized to provide data support for the subsequent screening of key residues.

[0051] Based on the modified risk scores, residue categories with risk levels exceeding a preset threshold are selected, generating a critical residue selection list. Specifically, this involves: setting a preset risk level threshold, ranging from 0.6 to 0.8, determined by considering the actual application scenario of the printing tape, quality assurance requirements, and production process requirements. This threshold accurately distinguishes between high-risk and low-risk residues, ensuring that the selected critical residues are all categories that significantly impact the performance and safety of the printing tape. For each residue category, the modified risk scores of all corresponding cell blocks are calculated using an arithmetic mean. The calculation method involves adding the modified risk scores of all cell blocks for that residue category and then dividing by the total number of cell blocks corresponding to that residue category to obtain the final modified risk score for that residue category. The final modified risk score ranges from 0.005 to 1800, and this final modified risk score serves as the basis for the final risk level of that residue category.

[0052] The final revised risk score of each type of residue is compared with a preset threshold item by item. When the final revised risk score of a certain type of residue exceeds the preset threshold, the residue is determined to be a high-risk residue and needs to be included in the critical residue selection list. When the final revised risk score of a certain type of residue does not exceed the preset threshold, the residue is determined to be a low-risk residue and is not included in the critical residue selection list. After screening, all residue categories determined to be high-risk are compiled and summarized, arranged in descending order of final revised risk score. At the same time, key information such as the area and distribution location of the corresponding planar region for each type of high-risk residue is added to form a complete critical residue selection list.

[0053] This embodiment, by combining spatially directional weighting to correct the risk ranking results, effectively eliminates the risk assessment bias caused by the spatial heterogeneity of residue distribution, ensuring that key high-risk residues can be accurately screened out.

[0054] In a preferred embodiment of the present invention, based on the key residue selection list, quantitative detection models for each residue category are established, and samples are analyzed to obtain the measured content values ​​for each residue category; spatial anisotropic weighting is used to correct each measured content value to obtain the corrected content value; the corrected content value is compared with the safety threshold of the corresponding residue category to obtain the judgment result, which may include:

[0055] For each residue category in the critical residue selection list, corresponding testing reagents, instruments, and parameters are configured to establish a quantitative detection model for each residue category. Specifically, this includes: retrieving the critical residue selection list, systematically reviewing each residue category, clarifying the chemical properties, physical characteristics, and causes of formation during the printing tape production process for each residue, determining the corresponding testing method based on the residue's characteristics, ensuring the method can accurately detect the residue's content without interference from the printing tape's material or other low-risk residues; and configuring corresponding testing reagents for each residue category, ensuring the reagent selection matches the residue's chemical properties. For acidic residues, alkaline reagents are used; for oily residues, lipophilic reagents are used. The concentration and dosage of the reagents are also determined, with the concentration set reasonably based on the residue's concentration range, and the dosage ensuring complete coverage of the detection point and sufficient reaction with the residue. This ensures the reagent reacts fully with the residue, resulting in a stable and observable reaction, avoiding deviations in test results due to improper reagent configuration.

[0056] Configure appropriate detection instruments, selecting devices capable of accurately capturing the reaction signals between residues and reagents. Based on the residue detection requirements, choose suitable instruments such as spectrophotometers and gas chromatographs to ensure the detection accuracy meets the needs of trace detection. Determine detection parameters, including detection temperature, detection time, reaction duration, and instrument detection wavelength. Each parameter is set based on the characteristics of the residue and the performance of the detection instrument. The detection temperature is determined based on the reactivity of the residue and reagent, typically controlled between room temperature and 50°C to ensure the reaction proceeds fully at a suitable temperature. The detection time is determined based on the reaction duration and instrument response speed, with the reaction duration controlled between 5 and 30 minutes. The detection time is extended by 5 to 10 minutes compared to the reaction duration to avoid insufficient reaction and low detection results due to too short a detection time, or reduced detection efficiency due to too long a detection time. The detection wavelength is determined based on the characteristic absorption wavelength after the residue reacts with the reagent. Accurately adjust the instrument's detection wavelength to ensure accurate capture of the reaction signal and reduce signal interference.

[0057] After configuring the testing reagents, instruments, and parameters, a quantitative detection model for each type of residue is established. The model construction process involves collecting standard samples of the residue type, selecting 5 to 8 different concentration gradients for each type of residue. The concentration gradients cover the common range of printed tape residue detection, increasing sequentially from trace amounts to possible exceedances to ensure the model can adapt to the detection of residues with different concentrations. Each set of standard samples is then tested according to the configured reagent concentration, dosage, instrument, and parameters. Each set of standard samples is tested three times, and the detection signal value is recorded for each test. The detection signal value is the signal intensity captured by the instrument after the residue reacts with the reagent. The known concentration values ​​of the standard samples are also recorded. The arithmetic mean of the three detection signal values ​​is taken as the final detection signal value for the set of standard samples. The calculation method is to add the three detection signal values ​​and then divide by 3 to reduce the error of a single test.

[0058] Using the known content values ​​of standard samples as the dependent variable and the corresponding final detection signal values ​​as the independent variables, a linear fitting method is employed for data fitting. The specific calculation process for linear fitting is as follows: The core principle of linear fitting is clarified, and the slope and intercept of the fitted line are calculated to minimize the sum of squared deviations between the detection signal values ​​of all standard samples and the known content values, thereby obtaining an accurate linear correspondence. Specifically, the calculation process involves first counting the number of all valid standard sample groups, assuming the number of valid standard sample groups is 1. The final detection signal value of each group of standard samples is the independent variable. The content value is known to be the dependent variable. Calculate separately Group The sum of values The sum of values and The sum of products The sum of the squares of the values, calculated as follows: The values ​​are added together to get ,all The values ​​are added together to get Each group and Multiply and then add to get Each group Square the values ​​and add them together to get Calculate the slope of the fitted line. and intercept slope The calculation method is as follows Multiply ,minus Multiply The difference is then divided by Multiply ,minus The difference is obtained by squaring the square of the two, and the specific formula is as follows: ;intercept The calculation method is as follows minus Multiply The difference is then divided by The specific formula is as follows The specific correspondence for linear fitting obtained through the above calculations is as follows: ,in This represents the content value of residues, and the unit is consistent with the known content value of the standard sample. Represents the slope of the fitted straight line, is dimensionless, and reflects the ratio of change between the detection signal value and the residue content value; This represents the final detection signal value captured by the detection instrument, expressed in units of signal strength. The intercept represents the fitted straight line, with units consistent with the residue content values, and is used to correct for systematic errors when the detection signal value is 0. During the fitting process, the deviation of each set of data is compared one by one; the deviation of a single data point is calculated as the deviation of that set of standard samples. The calculated content value obtained by substituting the values ​​into the corresponding formula is compared with the known content of the standard samples in this group. The absolute value of the difference in values, divided by the known values ​​of that set of standard samples. The value is used to obtain the percentage deviation of a single data point, and the specific formula is as follows: Data with a deviation percentage exceeding 5% are identified as outliers and removed to ensure the accuracy of the fitting results. After removing outliers, the corrected results are recalculated following the same procedure. and The linear correspondence is used, which is the quantitative detection model for this type of residue. After the model is built, it is trained by selecting another set of 4 to 6 standard samples with different content gradients as training samples. The content gradients do not overlap with the standard samples used in model building. The training samples are tested according to the set detection reagents, instruments and parameters to obtain the final detection signal value of each set of training samples. ,Will Substituting these values ​​into the quantitative detection model, we obtain the predicted content values ​​output by the model. ,Will Compared with the known content values ​​of the training sample The comparison is performed, and the prediction bias percentage of a single training sample is calculated using the following formula: If the prediction deviation percentage of a single sample exceeds 3%, the prediction deviation of that sample is deemed to be out of control. If there are out-of-control samples and the number of out-of-control samples exceeds 10% of the total number of training samples, the detection parameters and fitting relationship are adjusted, standard samples are reselected for detection, data fitting and model building, and the training samples are substituted into the newly built model to repeat the training process until the prediction deviation percentage of all training samples does not exceed 3%, thus ensuring the detection accuracy of the model.

[0059] After model training, model validation is performed. Three sets of standard samples with different contents are selected as validation samples: low content, medium content, and high content. Each set of validation samples is tested three times. The validation samples are substituted into the trained quantitative detection model to obtain the validation content value for each test. The arithmetic mean of the three validation content values ​​is taken as the final validation content value for the set of validation samples. The calculation method is to add the three validation content values ​​and then divide by 3. The final validation content value is compared with the known content value of the set of validation samples to calculate the average deviation of the three sets of validation samples. The calculation method is to add the validation deviation percentages of the three sets of samples and then divide by 3. If the average deviation percentage does not exceed 2%, the quantitative detection model is qualified and can be used for subsequent detection. If the average deviation percentage exceeds 2%, the model parameters are readjusted, and the model is trained and validated again until the average deviation percentage meets the requirements, ensuring that the quantitative detection model for each residue category can accurately and stably detect the residue content.

[0060] Based on the quantitative detection model, detection and analysis are performed separately within each cell division block of the sample to obtain the initial content value of each residue category in each cell division block; the initial content values ​​of the same residue category in each cell division block are summarized, the total distribution of each residue category in the sample is calculated, and the measured content value of each residue category is obtained. Specifically, this includes: reconfirming the specific location and boundary range of each cell division block according to the predetermined cell division method, marking the unique identifier of each cell division block, ensuring that each cell division block can be accurately located during the detection process, and avoiding detection omissions or positioning errors.

[0061] For each cell segment, detection and analysis were performed according to the established quantitative detection models for each residue category. During detection, three evenly distributed detection points were first selected within the cell segment. These three detection points were located at the center and two diagonal positions of the cell segment, respectively, to avoid randomness in the detection results due to overly concentrated detection points and to ensure that the detection results could reflect the actual content of residues within the cell segment. The corresponding detection reagent for each residue category was added to each detection point according to the prepared dosage. The reaction was allowed to proceed at the set detection temperature, and the reaction time was strictly performed according to the predetermined parameters. After the reaction was completed, the detection point was placed on the corresponding detection instrument, and detection was performed according to the set detection parameters. The detection signal value of each detection point was captured. Each detection point was detected twice, and the arithmetic mean of the two detection signal values ​​was taken as the final detection signal value of the detection point. The calculation method was to add the two detection signal values ​​and then divide by 2.

[0062] Substitute the final detection signal value of each detection point into the corresponding quantitative detection model. Using the correspondence between the detection signal value and the content value in the model, calculate the initial content value of the residue category corresponding to that detection point. Then, calculate the arithmetic mean of the initial content values ​​of the three detection points within the cell division block. This is done by adding the initial content values ​​of the three detection points and dividing by three to obtain the initial content value of that residue category within the cell division block. This ensures the accuracy of the residue content detection results within the cell division block and avoids the influence of errors from a single detection point on the overall result. If the initial content values ​​of the three detection points deviate too much, i.e., the difference between the maximum and minimum values ​​exceeds 10% of the average, then add two more detection points, re-detect, and calculate the arithmetic mean of the five detection points as the initial content value of the cell division block.

[0063] Following the above procedure, the initial content values ​​of each residue category in all cell division blocks were detected and calculated one by one. At the same time, the unique identifier of each cell division block, the initial content value of each residue category, and the corresponding detection time were recorded. After the detection was completed, the initial content values ​​of the same residue category in each cell division block were summarized, and the total distribution of the residue category in the sample was calculated. The calculation method is to add the initial content values ​​of the residue category in all cell division blocks to obtain the total content of the residue category. Then, the total content is divided by the total number of cell division blocks in the sample to obtain the average content of the residue category in the sample. The total content and the average content together constitute the measured content value of the residue category.

[0064] Based on spatial anisotropic weighting, the weight values ​​of the corresponding cell partitions for each residue category in the sample are extracted to obtain a weight value set. Specifically, this includes: invoking spatial anisotropic weighting, which is obtained through iterative calculation based on cell eigenvalues ​​and distribution characteristics. This allows for the assignment of differentiated weights to cell partitions in different regions based on the spatial heterogeneity of residue distribution. The weight values ​​range from 0.5 to 1.8, with cell partitions corresponding to densely distributed residue areas receiving weights of 1.2 to 1.8, and cell partitions corresponding to sparsely distributed residue areas receiving weights of 0.5 to 1.1. The weights for each spatial direction are also considered. The allocation is reasonable and the total number of cell partitioning blocks is fixed at the total number of sample units; the key residue selection list is retrieved, and each residue category in the list is sorted out. At the same time, the unique identifier and location information of the cell partitioning blocks corresponding to each residue category are retrieved to clarify all cell partitioning blocks corresponding to each residue category; for each residue category, according to the unique identifier of its corresponding cell partitioning block, the weight value corresponding to each cell partitioning block is accurately extracted from the spatial anisotropic weighting. During the extraction process, the correspondence between cell partitioning block identifiers and weight values ​​is checked one by one to ensure that the cell partitioning block position corresponds to the weight value one by one.

[0065] The weighted values ​​of all cell partitions corresponding to each residue category are organized and arranged in order of the unique identifier of the cell partition. The cell partition identifier corresponding to each weighted value is labeled to form a subset of weighted values ​​for that residue category. After summarizing the subsets of weighted values ​​for all residue categories, the residue category, cell partition identifier, and location information corresponding to each weighted value are labeled to obtain a complete set of weighted values.

[0066] Based on the weighted value set and the measured content values ​​of each residue category, the measured content values ​​and corresponding weighted values ​​of the same residue category in each cell block are calculated to obtain the corrected content value of each residue category in each cell block. The corrected content values ​​of the same residue category in all cell blocks are then summarized to obtain the corrected content value of each residue category. Specifically, this includes: retrieving the measured content values ​​of each residue category, including the total content and average content of each residue category; simultaneously retrieving the weighted value set to clarify the weighted value subset, cell block identifier, and initial content value corresponding to each residue category; for each residue category, extracting each weighted value from its corresponding weighted value subset and the initial content value of the cell block corresponding to that weighted value; and calculating the initial content value of the cell block by multiplying the initial content value of the cell block by the corresponding weighted value. The result of multiplication is the corrected content value of the residue category in the cell block. Following the above calculation method, the corrected content value of all cell blocks corresponding to the residue category is calculated one by one. At the same time, the unique identifier of each cell block, the initial content value of the residue category, the weighting value, and the corrected content value are recorded. After the calculation is completed, the corrected content values ​​of all cell blocks corresponding to the residue category are summarized, and the corrected total content and corrected average content of the residue category are calculated. The corrected total content is calculated by adding the corrected content values ​​of all cell blocks. The corrected average content is calculated by dividing the corrected total content by the total number of cell blocks corresponding to the residue category. If no residue of this type is detected in a certain cell block, its corrected content value is recorded as 0 and included in the summary calculation. The corrected total content and the corrected average content together constitute the corrected content value of the residue category.

[0067] Each residue category is compared with its corresponding safety threshold. When the corrected content value exceeds the safety threshold, it is marked as exceeding the standard. All comparison results are summarized to obtain the judgment result. Specifically, based on the actual use scenario of the printing tape, subsequent processing requirements, and the degree of impact of residues on the performance and safety of the printing tape, a corresponding safety threshold is set for each residue category in the key residue selection list. The safety threshold is set in accordance with the characteristics of the residue, with a value range of 0.01 to 5.0 mg / kg. Among them, the safety threshold for high-risk residues is 0.01 to 0.1 mg / kg, for medium-risk residues it is 0.1 to 1.0 mg / kg, and for low-risk residues it is 1.0 to 5.0 mg / kg. This ensures that the safety threshold can accurately distinguish whether the residue content meets the standard. Different categories of residues correspond to different safety thresholds, which are reasonably set according to their degree of impact.

[0068] The corrected content values ​​for each residue category are retrieved, including the corrected total content and the corrected average content. Simultaneously, the corresponding safety threshold for each residue category is retrieved. For each residue category, the corrected total content and corrected average content are compared item by item with the corresponding safety threshold. A unified comparison standard is used during the comparison process, i.e., the corrected content value is directly compared with the safety threshold to avoid judgment bias caused by inconsistent comparison standards. When the corrected total content or the corrected average content of a residue category exceeds its corresponding safety threshold, the residue category is determined to be above the standard and marked as an excessive residue. The degree of exceeding the standard is calculated by dividing the exceeding value by the safety threshold to obtain the percentage exceeding the standard. The exceeding value is the difference between the corrected content value and the safety threshold (if both the corrected total content and the average content exceed the standard, the larger percentage exceeding the standard is taken as the degree of exceeding the standard for that residue category). When both the corrected total content and the corrected average content of a residue category do not exceed their corresponding safety thresholds, the content of that residue category is determined to be compliant and marked as compliant residue. During the comparison process, the comparison results of each residue category are recorded in detail, including the corrected total content, the corrected average content, the safety threshold, whether it exceeds the standard, and the degree of exceeding the standard (if it exceeds the standard).

[0069] After completing the comparison of all residue categories, summarize all comparison results, sort out the categories of residues exceeding the standard, their corresponding corrected content values, safety thresholds (0.01 to 5.0 mg / kg), and degree of exceeding the standard in descending order of the degree of exceeding the standard. At the same time, sort out the categories of residues that meet the standard, their corresponding corrected content values, and safety thresholds in order of residue category name, to form a complete judgment result.

[0070] This embodiment establishes a dedicated quantitative detection model for each key residue category, combines it with precisely configured detection reagents, instruments and parameters, and ensures the accuracy and specificity of residue content detection through repeated detection, data fitting, model training and validation. It also avoids interference between different residues and further improves the reliability of detection results.

[0071] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the system as described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0072] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the system as described above. All implementations in the above system embodiments are applicable to this embodiment and can achieve the same technical effects.

[0073] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A chemical detection system for scraping residue on printed tape, characterized in that, include: The processing module is used to acquire samples with scraped edge residue from printing, perform ultrasonic detection and analysis on the samples, collect the original ultrasonic echo signals fed back by the samples under ultrasonic excitation, perform time-domain filtering and frequency-domain transformation processing on the original ultrasonic echo signals, and extract characteristic reflection wave groups. Based on the characteristic reflection wave group, the location of the reflection interface corresponding to each reflection wave is identified, and the amplitude of the reflection wave at each reflection interface is extracted to obtain the reflection coefficient sequence. The reflection coefficient sequence is weighted and averaged to decompose it into homogenized reflection coefficient components and micro-wave correction components. The homogenized reflection coefficient components and micro-wave correction components are superimposed and reconstructed to obtain a multi-scale homogenized reflection coefficient sequence. Based on the multi-scale homogenized reflection coefficient sequence, the acoustic impedance difference between adjacent reflection interfaces is calculated to obtain the acoustic impedance difference parameter. A phase inversion basis for acoustic reflection waves is established based on the acoustic impedance difference parameter and its corresponding reflection interface position. Distribution characteristic data of residues are extracted based on the acoustic reflection wave phase inversion basis. Boundary contour information of the sample is collected, and a set of boundary points is extracted based on the boundary contour information. All vertices constituting the convex hull of the sample boundary are identified to obtain a convex hull vertex sequence. Based on the convex hull vertex sequence, the first and second extreme points located at both ends of the sample's preset direction, as well as the third and fourth extreme points located at both ends perpendicular to the preset direction, are selected. The first, second, third, and fourth extreme points are then used as a candidate corner point set. Three spatial anchoring corner points are selected from the candidate corner point set. The first spatial anchoring corner point is the midpoint of the line connecting the first and second extreme points. The second spatial anchoring corner point is the third extreme point, and the third spatial anchoring corner point is the fourth extreme point. The spatial coordinates of the three spatial anchoring corner points are collected. Based on the spatial coordinates of the three spatial anchoring corner points, the region enclosed by the three spatial anchoring corner points is determined. Using this region as a basis, a preset offset distance is extended outward to delineate the boundary of the test region framework, thus obtaining the test region framework. Based on the boundary of the test region framework, the test region framework is divided into multiple cell blocks at equal intervals according to a preset division size, obtaining the spatial position of each cell block. Based on the spatial position of each cell block, the corner point feature parameters at the corresponding positions within each cell block are extracted, and the corner point feature parameters are calculated to obtain... Cellular eigenvalues; Based on the cellular eigenvalues ​​and distribution characteristic data, calculate the difference between the cellular eigenvalues ​​and distribution characteristic data at the same spatial point within each cellular block to obtain the initial perturbation bias; Based on the initial perturbation bias, perform perturbation propagation calculations sequentially along the spatial distribution direction of the sample, using the calculation result of each cellular block as the input to the adjacent cellular blocks, and update the perturbation bias of each cellular block after each propagation calculation, until the global rate of change of the perturbation bias of all cellular blocks converges to a preset threshold, generating a matrix perturbation progression conformation; Based on the matrix perturbation progression conformation, extract the rate of change of perturbation in different spatial directions for each cellular block, calculate the cellular eigenvalues ​​based on the rate of change of perturbation, obtain the weighted values ​​of each cellular block in different spatial directions, and summarize all weighted values ​​to obtain spatially anisotropic weighting; The parsing module is used to match and parse the corner feature parameters with the preset feature database, identify the residue categories in the sample, rank the risk of each residue category, and obtain the risk ranking result. By using spatially directional weighting to correct the risk ranking results, a list of key residues was obtained. The judgment module is used to establish quantitative detection models for each residue category based on the key residue selection list, analyze the samples, and obtain the measured content value of each residue category; correct each measured content value by spatial anisotropic weighting to obtain the corrected content value; and compare the corrected content value with the safety threshold of the corresponding residue category to obtain the judgment result.

2. The chemical detection system for printing tape scraping residue according to claim 1, characterized in that, Samples of printing tape scraping residue were obtained, and the samples were subjected to ultrasonic detection and analysis. Ultrasonic echo signals were collected, and an acoustic reflection wave phase inversion basis was constructed. Based on the acoustic reflection wave phase inversion basis, the distribution characteristic data of the residue were extracted, including: The acoustic impedance difference parameters of each reflection interface are analyzed from the characteristic reflection wave group, and the acoustic reflection wave phase inversion basis is constructed based on the acoustic impedance difference parameters. Based on the acoustic reflection wave phase inversion basis, the initial acoustic impedance value corresponding to each spatial point is calculated point by point in the sampling spatial domain of the sample to obtain the initial acoustic impedance distribution data. Based on the initial acoustic impedance distribution data, the acoustic impedance difference between each spatial point and its adjacent points is iteratively corrected until the acoustic impedance difference between adjacent points converges to a preset threshold, thus generating spatial acoustic impedance distribution data. Spatial clustering analysis is performed based on spatial acoustic impedance distribution data to screen out connected regions with consistent acoustic impedance. The geometric boundaries and distribution density of the connected regions are calculated to obtain the distribution characteristic data of the residues.

3. The chemical detection system for printing tape scraping residue according to claim 2, characterized in that, The corner feature parameters are matched and parsed with the preset feature database to identify the residue categories in the sample. The risk ranking of each residue category is then performed to obtain the risk ranking results. By refining the risk ranking results using spatially oriented weighting, a list of key residues was obtained, including: Feature identifier groups are extracted from the corner feature parameters. The feature identifier groups are compared with the standard feature identifier groups in the preset feature database. Based on the similarity comparison results, the residue categories corresponding to each cell division block are identified, and the residue category identification results are obtained. For each residue category in the residue category identification results, an initial risk weight is matched. The comprehensive risk score of each residue category is calculated based on the distribution frequency of each residue category in the sample and the initial risk weight. The residue categories are then sorted in descending order based on the comprehensive risk score to obtain the risk ranking result. Based on spatial anisotropic weighting, the weighting value of the corresponding cell partition block of each residue category in the sample is extracted. The comprehensive risk score is calculated with the weighting value of the corresponding cell partition block to obtain the corrected risk score. Based on the corrected risk score, residue categories with risk levels exceeding the preset threshold are screened out to generate a key residue selection list.

4. The chemical detection system for printing tape scraping residue according to claim 3, characterized in that, Based on the key residue selection list, quantitative detection models were established for each residue category. Samples were analyzed to obtain the measured content values ​​for each residue category, including: For each residue category in the key residue selection list, corresponding detection reagents, detection instruments and detection parameters are configured to establish a quantitative detection model for the residue category. Based on the quantitative detection model, detection and analysis are performed separately in each cell block of the sample to obtain the initial content value of each residue category in each cell block; the initial content values ​​of the same residue category in each cell block are summarized, the total distribution of each residue category in the sample is calculated, and the measured content value of each residue category is obtained.

5. The chemical detection system for printing tape scraping residue according to claim 4, characterized in that, Spatial anisotropic weighting is used to correct each measured content value to obtain the corrected content value; the corrected content value is compared with the safety threshold of the corresponding residue category to obtain the judgment result, including: Based on spatial anisotropic weighting, the weighting values ​​of the corresponding cell partition blocks of each residue category in the sample are extracted to obtain the weighting value set; Based on the weighted value set and the measured content value of each residue category, the measured content value of the same residue category in each cell partition block and the corresponding weighted value are calculated to obtain the corrected content value of each residue category in each cell partition block. The corrected content values ​​of the same residue category in all cell partition blocks are summarized to obtain the corrected content value of each residue category. The corrected content value of each residue category is compared with the corresponding safety threshold item by item. When the corrected content value exceeds the safety threshold, it is marked as exceeding the standard. All comparison results are summarized to obtain the judgment result.

6. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the system as described in any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the system as described in any one of claims 1 to 5.

Images