Etching paste material quality detection method and system based on spectral response difference

Abstract

The invention provides a spectral response difference-based etching paste material quality detection method and system. The method comprises the following steps of: decomposing and identifying an internal density abrupt change boundary position of a material through multiband spectral chromatography; applying a light beam with a specific frequency to a sudden change boundary to perform resonance excitation to reconstruct stress field distribution, determining a stress concentration point, performing reverse optical tracking to form a defect path, and establishing a self-organizing optical monitoring grid; performing density self-adaptive adjustment on the grids, and forming a graded spectrum fingerprint database through differential spectrum excitation to generate a material state boundary recognition standard; phase change critical point detection is carried out based on an identification standard, and a pulse light beam is applied to excite surface deformation to construct a multi-point linkage defect blocking network; and performing defect analysis on the network to form an interconnected optical energy field, monitoring the energy field to determine an abnormal position, and performing phase comparison marking through a phase difference spectrum group to complete quality detection of the etching paste material, so that accurate identification and intelligent quality classification of internal defects of the material are realized.

Description

Technical Field

[0001] This invention relates to the field of optical inspection technology, and in particular to a method and system for quality inspection of etching paste materials based on differences in spectral response. Background Technology

[0002] In the production of solar cells, etching paste is widely used as an important auxiliary printing material. Etching paste is gel-like and primarily used for etching transparent conductive metal coatings. It can create fine etched patterns through screen printing. Compared to traditional etching processes, etching paste printing eliminates the need for subsequent treatment with strong acids or alkalis, making the process convenient and the work site cleaner, thus meeting the requirements of modern green manufacturing.

[0003] However, etching paste materials face complex quality control challenges in practical use. Due to internal quality issues such as uneven density distribution, component segregation, and microscopic defects, these internal structural anomalies gradually evolve into macroscopic defects during material use, affecting the consistency and reliability of the etching effect. Traditional quality inspection methods mainly rely on surface observation and simple physical tests, making it difficult to deeply analyze the internal microstructure, stress distribution, and potential degradation risks of the material. In particular, existing testing technologies lack effective analytical tools for key quality indicators such as internal density gradient changes, stress concentration areas, and phase transformation characteristics. These blind spots mean that quality problems are often only discovered in the later stages of material use, hindering early warning and preventative quality control, severely impacting production stability and product quality consistency. Summary of the Invention

[0004] This invention provides a method and system for quality testing of etching paste materials based on spectral response differences. It aims to construct a multi-dimensional material characteristic analysis system, deeply explore the spectral phase response law of etching paste materials, establish a complete testing chain from material microstructure to macroscopic performance, and achieve accurate, comprehensive, and intelligent testing of etching paste material quality.

[0005] The first aspect of this invention proposes a method for quality testing of etching paste materials based on differences in spectral response, comprising the following steps: Acquire multi-band spectral signals and material thickness data on the surface of the etching paste, perform spectral tomography based on the multi-band spectral signals to form a layered spectral map of the material interior, and combine the material thickness data and the layered spectral map to determine the location of density abrupt boundary. A specific frequency beam is applied at the abrupt boundary location to resonate and generate material resonance response parameters. The resonance response parameters are then used to generate reconstructed data of the internal stress field of the material. Based on the stress field reconstruction data, the coordinates of stress concentration points are determined. A defect formation path is then formed by reverse optical tracing from the stress concentration points outwards, and a self-organizing optical monitoring grid is established through the defect formation path. The density of the self-organized optical monitoring grid is adaptively adjusted to determine an optimized grid distribution. Differential spectral excitation is applied to the optimized grid distribution to form a hierarchical spectral fingerprint database. Material state boundary identification criteria are generated based on the hierarchical spectral fingerprint database. Based on the material state boundary identification standard, the critical point of material phase transition is detected to obtain phase transition boundary data. A pulse beam is applied near the phase transition boundary to excite the material surface deformation and extract the surface deformation response features. The surface deformation response features are then used to construct a multi-point linkage defect blocking network. Defect analysis is performed on the multi-point linked defect blocking network to form an interconnected optical energy field. The interconnected optical energy field is monitored to determine the coordinates of abnormal locations. The phase adjustment of the hierarchical spectral fingerprint library generates a phase difference spectral group. The abnormal location coordinates are then marked by phase comparison using the phase difference spectral group to complete the quality inspection.

[0006] A second aspect of this invention provides a quality inspection system for etching paste materials based on spectral response differences, comprising: The spectral acquisition module is used to acquire multi-band spectral signals and material thickness data on the surface of the etching paste, perform spectral tomography based on the multi-band spectral signals to form a layered spectral map of the material's interior, and combine the material thickness data and the layered spectral map to determine the location of density abrupt boundary. The resonance excitation module is used to apply a specific frequency beam at the abrupt boundary position to resonate and generate material resonance response parameters, and use the resonance response parameters to generate material internal stress field reconstruction data. The tracking and analysis module is used to determine the coordinates of stress concentration points based on the stress field reconstruction data, perform reverse optical tracking outward from the stress concentration points to form a defect formation path, and establish a self-organizing optical monitoring grid through the defect formation path. The grid monitoring module is used to adaptively adjust the density of the self-organized optical monitoring grid to determine an optimized grid distribution, apply differential spectral excitation to the optimized grid distribution to form a hierarchical spectral fingerprint database, and generate a material state boundary identification standard based on the hierarchical spectral fingerprint database. The phase transition detection module is used to detect the critical point of material phase transition based on the material state boundary identification standard, obtain phase transition boundary data, apply a pulse beam near the phase transition boundary to excite the material surface deformation and extract surface deformation response features, and use the surface deformation response features to construct a multi-point linkage defect blocking network. The phase marking module is used to perform defect analysis on the multi-point linkage defect blocking network to form an interconnected optical energy field, monitor the interconnected optical energy field to determine the coordinates of abnormal positions, perform phase adjustment on the hierarchical spectral fingerprint library to generate a phase difference spectral group, and use the phase difference spectral group to perform phase comparison marking on the coordinates of the abnormal positions to complete the quality inspection.

[0007] The beneficial effects of this invention are reflected in the following points: First, by using material density gradient analysis and resonance excitation technology, the location of density abrupt change boundaries within the etching paste material can be identified and the stress field distribution reconstructed, thereby detecting internal defects and stress concentration areas that are difficult to detect using traditional methods, achieving in-depth analysis of the internal quality state of the material. Second, a self-organized optical monitoring grid and a hierarchical spectral fingerprint library are established. Differential spectral excitation strategies generate spectral characteristic standards for different quality levels. Combined with the collaborative response monitoring of a multi-point linked defect blocking network, precise grading and networked evaluation of material quality are achieved, improving the accuracy and reliability of quality judgment. Finally, by generating defect formation paths through reverse optical tracing, combined with phase transition critical point detection and surface deformation response characteristic analysis, the degradation trend and stability boundary of the material can be predicted. Ultimately, the precise marking of abnormal locations is achieved through differences in spectral phase response, forming a complete detection system from defect detection and degradation prediction to quality grading.

[0008] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0009] The accompanying drawings illustrate specific examples of the technical solutions described in this invention and, together with the detailed embodiments, form part of the specification, serving to explain the technical solutions, principles, and effects of this invention.

[0010] Unless otherwise specified, the same reference numerals in different figures represent the same or similar technical features, and different reference numerals may be used to represent the same or similar technical features.

[0011] Figure 1 This is a schematic flowchart of an etching paste material quality testing method based on spectral response differences according to the present invention.

[0012] Figure 2 This is a schematic diagram of the structure of a multi-band spectral acquisition system according to the present invention.

[0013] Figure 3 This is a structural block diagram of an etching paste material quality testing system based on spectral response differences according to the present invention.

[0014] Explanation of reference numerals in the attached diagram: 1. Area scan camera; 2. Front light source; 3. Side light source; 4. Object being measured. Detailed Implementation

[0015] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, circuits, and methods are omitted so as not to obscure the description of this application with unnecessary detail.

[0016] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0017] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0018] The technical solutions of the embodiments of this application will be described below.

[0019] like Figure 1 As shown, this embodiment of the invention provides a method for quality testing of etching paste materials based on differences in spectral response, including the following steps S110-S160: Step S110: Obtain multi-band spectral signals and material thickness data on the surface of the etching paste; perform spectral tomography based on the multi-band spectral signals to form a layered spectral map inside the material; and combine the material thickness data and the layered spectral map to determine the location of the density abrupt change boundary.

[0020] Specifically, multi-band spectral acquisition is performed on the surface of the etching paste, such as... Figure 2As shown, a three-dimensional structural design is adopted, including key components such as an area array camera 1, a front light source 2, a side light source 3, and the object under test 4. The area array camera 1 uses a high-sensitivity CMOS sensor with a pixel resolution of 2048×2048, ensuring that the etching paste occupies more than 80% of the field of view. The multi-band light source system includes ultraviolet light (300-400nm), visible light (400-700nm), near-infrared light (700-1100nm), short-wave infrared light (1100-1700nm), and mid-wave infrared light (1700-2500nm), each using an LED array configuration, with light intensity stability controlled within ±0.5%. The front light source 2 provides uniform diffuse illumination, while the side light source 3 is configured at a 45° tilt angle to eliminate shadows and reflection interference. The object under test 4 is the etching paste sample to be tested, accurately positioned using a precision positioning platform. The spectral resolution is set to 0.1nm, and the dynamic range reaches 16 bits. The material thickness measurement integrates a laser triangulation module with a measurement accuracy of 0.01mm. A hard trigger controller enables precise synchronization between the light source and the camera. When the object under test 4 moves to the detection position, the light source is automatically activated to illuminate the object and trigger the camera to acquire multi-band spectral signals and material thickness data.

[0021] In some embodiments, the step of forming a layered spectral map of the material by performing spectral tomography based on the multi-band spectral signal includes: determining the penetration depth of different bands based on the multi-band spectral signal; performing virtual layering of the material according to the penetration depth to obtain layer interface data; and extracting the spectral response features of each layer based on the layer interface data to obtain a layered spectral map of the material.

[0022] The effective penetration depth of different wavelengths in etching paste materials is determined based on the attenuation characteristics of multi-band spectral signals. This is achieved by comparing the attenuation patterns of spectral signal intensity in samples of different thicknesses. The attenuation coefficient α(λ) = -ln(I / I0) / t is used in the calculation, where I is the transmitted light intensity, I0 is the incident light intensity, and t is the sample thickness. The 1 / e attenuation standard is used to define the penetration depth, which is the thickness value corresponding to the light intensity attenuating to the initial value 1 / e. Based on penetration capability, the wavelengths are divided into five levels: ultra-shallow penetration (300-300nm, depth <0.1mm), shallow penetration (300-500nm, depth 0.1-0.8mm), medium penetration (500-1000nm, depth 0.8-3mm), deep penetration (1000-1800nm, depth 3-8mm), and ultra-deep penetration (1800-2500nm, depth >8mm). Material optical parameter corrections consider the influence of factors such as refractive index, scattering coefficient, and absorption coefficient, ultimately establishing an accurate database of wavelength penetration depths.

[0023] The etching paste material is virtually layered based on the penetration depth gradient to determine the boundary positions and thickness distribution of each layer. The layering strategy employs an arithmetic distribution of penetration depth, with 5-7 main layers corresponding to the number of main wavelength bands. The thickness of the i-th layer is calculated by the geometric mean of adjacent penetration depths: t_i = (D_i + 1 - D_i) / 2, where D_i is the characteristic value of the i-th penetration depth. Layer interface data includes geometric parameters such as interface location coordinates, tilt angle, and roughness; cubic spline interpolation ensures interface continuity. Layer quality is evaluated using intralayer spectral uniformity and interlayer variability indicators. A dynamic layering adjustment mechanism adaptively adjusts the number and thickness of layers based on changes in penetration depth, forming complete layer interface data.

[0024] Based on the layered interface data, the independent spectral response features of each virtual layer are extracted to reconstruct the spectral distribution pattern inside the material. The z-direction range of the i-th layer is determined by the coordinate difference between adjacent interfaces. The interface tilt angle θ_i provides a geometric correction to the light transmission path length, and the interface roughness σ_i affects the light scattering characteristics. The transmission matrix T is constructed based on the constraints of the interface geometric parameters. The matrix element T_ij = α_ij × sec(θ_i) × S_i × L_ij, where α_ij is the basic absorption coefficient, sec(θ_i) is the optical path correction coefficient, S_i is the scattering loss coefficient, and L_ij is the effective optical path length. In the spectral separation mathematical model S_layer = T^(-1) × S_measured, S_layer is the spectral vector of each layer, and S_measured is the measured composite spectral vector. The interface sharpness parameter R_i is used to constrain the degree of aliasing of the spectra of adjacent layers. Hard boundary separation is used when the interface is clear, and soft boundary processing is used when the interface is blurred. Finally, a layered spectral map reflecting the distribution of the optical properties inside the material is established.

[0025] The locations of density abrupt change boundaries are determined by combining material thickness data and layered spectral maps. First, absorption intensity is extracted based on the layered spectral maps and calculated through integration: A(x,λ)=∫[λ1,λ2]I_baseline(λ)-I_layer(λ)dλ, where A(x,λ) is the absorption intensity at wavelength λ at location x; I_baseline(λ) is the baseline spectral intensity; I_layer(λ) is the actual spectral intensity of that layer; and λ1 and λ2 are the lower and upper limits of the integration wavelength range, respectively. A continuous thickness field is constructed from the material thickness data using three-dimensional interpolation, and the actual thickness distribution of each layer is determined by combining the layer interface location information. Next, the modified Lambert-Beer law is used to calculate the layer density: ρ_layer(x,y,λ)=k(λ)×A(x,y,λ) / t_layer(x,y)+ρ0, where ρ_layer is the layer density; k(λ) is the wavelength-dependent density conversion coefficient; t_layer(x,y) is the layer thickness at position (x,y); and ρ0 is the baseline density. Multilayer density reconstruction is achieved by spatially superimposing the densities of each layer according to the layer interface data. The density gradient is calculated using a central difference scheme to obtain the three-dimensional spatial partial derivatives, yielding the gradient vector and gradient intensity. When the gradient intensity exceeds a dynamic threshold, it is identified as a density abrupt change point. The dynamic threshold is calculated as: Thr_gradient=μ_gradient+2×σ_gradient, where μ_gradient is the mean gradient intensity; and σ_gradient is the standard deviation of the gradient intensity. The mutation boundary is located through connected component analysis, which clusters adjacent mutation points to form mutation boundary regions. The boundary position coordinates are determined by centroid calculation, and the accurate density value ρ_boundary of the boundary position is obtained by three-dimensional interpolation. Finally, the location of the density mutation boundary inside the material is determined.

[0026] Step S120: Apply a specific frequency beam at the abrupt boundary location to resonate and generate material resonance response parameters, and use the material resonance response parameters to generate material internal stress field reconstruction data.

[0027] In some embodiments, the step of applying a specific frequency beam at the abrupt boundary location to generate material resonance response parameters for resonant excitation includes: analyzing material properties based on the abrupt boundary location to obtain an intrinsic resonance frequency; selecting an excitation beam frequency based on the intrinsic resonance frequency to obtain a frequency matching parameter; applying an excitation beam based on the frequency matching parameter to generate a resonance phenomenon; and monitoring the amplitude and phase changes of the resonance phenomenon to form material resonance response parameters.

[0028] The natural resonant frequency is obtained by analyzing the material properties at the abrupt boundary location. Based on the density-elasticity relationship model E=k_elastic×ρ_boundary^n, where k_elastic is a material-related constant and n is a power exponent, the local elastic modulus is calculated using the density value at the abrupt boundary location. The natural resonant frequency is calculated using a simplified beam vibration model: f_natural=(λ_n / (2π))×√(E×I / (ρ_boundary×A×L)). 4 The equation is: λ_n = I * A * L, where λ_n is the eigenvalue of the nth mode shape, I is the moment of inertia of the cross section, A is the cross-sectional area, and L is the characteristic length determined based on the location of the abrupt boundary. The characteristic length L is determined by analyzing the density distribution around the abrupt boundary, taking the distance where the density gradient decreases to 50% of the abrupt value as the characteristic length. Density non-uniformity correction is achieved by introducing a density variation coefficient CV_density = σ_local / ρ_boundary, where σ_local is the standard deviation of density in the local region around the abrupt boundary. Temperature compensation considers the influence of ambient temperature on the elastic modulus, with a correction coefficient k_temp = 1 - α_temp × (T_current - T_reference), where α_temp is the temperature coefficient. Multi-point sampling calculates the natural frequencies at multiple points along the abrupt boundary, and the average value is taken as the representative resonant frequency of that boundary, thus determining the natural resonant frequency characteristics of the material at that location.

[0029] For example, the step of selecting the excitation beam frequency based on the inherent resonant frequency to obtain the frequency matching parameters includes: determining the resonance peak position based on the inherent resonant frequency; setting a frequency scanning range near the resonance peak position; fine-tuning the frequency based on the frequency scanning range to find the maximum response point and obtain the optimal excitation frequency; and calculating the matching degree between the optimal excitation frequency and the inherent resonant frequency to form the frequency matching parameters.

[0030] The theoretical resonance peak position is determined based on the natural resonance frequency. The natural resonance frequency is used as the center frequency. Considering the influence of material damping and environmental factors, the theoretical resonance peak position is calculated as f_peak_theory = f_natural × (1 + δ_shift), where δ_shift is the frequency shift correction coefficient, calculated negatively using the material damping ratio. Preliminary verification of the resonance peak position is achieved through single-point excitation testing. A low-power excitation signal is applied at the f_natural frequency, and the measured response amplitude serves as a reference. The broadening effect of the resonance peak is estimated using the reciprocal relationship between the quality factor and the damping ratio. The peak width is calculated using the ratio of the center frequency to the quality factor, reflecting the sharpness of the resonance peak. The identification of multiple resonance modes involves theoretical calculations to determine possible higher-order resonance frequencies, using a square root function to consider the influence of modal order. For the detection of bubble defects in etching paste materials, the damping effect is significantly enhanced due to the rapid change in material density around the bubbles. Therefore, a larger negative correction is needed to the theoretical resonance peak position to ensure that the excitation frequency accurately matches the actual resonance characteristics. The effect of ambient temperature on the position of the resonance peak is linearly corrected by the temperature coefficient. Under high temperature conditions, the elastic modulus of the material decreases, which leads to a decrease in the resonance frequency.

[0031] A systematic frequency scanning range is established near the resonance peak location. The lower and upper scan limits are determined by subtracting and adding the scan half-width from the peak position, where the scan half-width is the larger of three times the resonance peak width or 5% of the peak frequency. An adaptive adjustment mechanism for the scan range is optimized based on the initial scan results. When the response amplitude at the boundary exceeds 30% of the response in the central region, the scan range is extended by 50% in that direction. The frequency resolution is determined by dividing the scan range by the number of scan points, typically set to 300 scan points to ensure sufficient frequency accuracy. The scanning strategy combines coarse and fine scanning. Coarse scanning is used to quickly locate the response region, while fine scanning is used to accurately determine the peak position. When detecting adhesive breakage defects on the cell surface, the material density gradient at the breakage boundary creates a significant abrupt change in mechanical impedance. This abrupt change can cause the resonance peak to split or shift. Therefore, it is necessary to expand the scan range to capture all possible resonance modes and ensure that no key defect feature information is missed. Boundary condition checks are achieved by monitoring the response gradient at the scan boundary. When the response gradient is less than a set threshold, the scan range is confirmed to be sufficient.

[0032] Precise frequency fine-tuning is performed based on the frequency scanning range, and the optimal excitation frequency for maximum response is found through response monitoring. Frequency fine-tuning involves point-by-point excitation and response measurements within the scanning range, with excitation signal parameters remaining consistent at each frequency point. The excitation power is set to 10% of the device's maximum power, and the excitation time is 10 times the reciprocal of the current scanning frequency to ensure sufficient excitation effect. Response amplitude measurement is achieved through real-time imaging using area array camera 1, acquiring 50 frames of images at each frequency point. The response amplitude is calculated using the root mean square value of pixel displacement. Frequency stepping uses an equal interval method, increasing sequentially from the lower scan limit to the upper scan limit. Maximum response point identification is achieved by finding the global maximum value of the response amplitude array; the corresponding frequency is the optimal excitation frequency. Smoothing of the response curve uses a moving average filter with a window length of 5 frequency points to eliminate the influence of measurement noise. Multi-peak detection is achieved through local maximum search; when multiple local peaks exist, the frequency corresponding to the peak with the largest amplitude is selected. When inspecting defects with uneven etchant thickness, areas of drastic thickness variation generate complex stress distributions, leading to multiple resonance peaks. A systematic peak identification algorithm is needed to determine the strongest resonance peak corresponding to the primary defect characteristics, avoiding interference from secondary background responses. Frequency accuracy is improved through parabolic interpolation, using quadratic function fitting on three data points near the maximum response point, achieving an interpolation accuracy of up to 1 / 10 of the scan step size. Point-by-point scanning and multi-peak identification algorithms successfully identified the optimal excitation frequency producing the maximum resonance response.

[0033] The matching degree between the optimal excitation frequency and the natural resonant frequency is calculated, and a quantitative analysis is performed based on the numerical relationship between the optimal excitation frequency and the natural resonant frequency. Frequency deviation is calculated as the absolute difference between the two frequencies, and relative deviation is obtained by dividing the deviation by the natural frequency and then multiplying by 100%. The matching degree coefficient is calculated using a Gaussian function, with a matching tolerance of 2% of the natural frequency, and the coefficient varies within the range of 0-1. The response enhancement ratio is calculated by dividing the maximum response amplitude at the optimal frequency by the baseline response amplitude at the non-resonant frequency. The quality factor matching degree is assessed by comparing the measured quality factor with the theoretical quality factor, taking the smaller value and dividing it by the larger value. The phase matching degree is calculated by analyzing the phase response at the optimal frequency and the theoretically expected phase difference, and is quantified using a cosine function. The comprehensive matching parameter F_match = 0.4 × M_coefficient + 0.3 × R_enhancement / 10 + 0.2 × Q_match + 0.1 × φ_match, where the weighting coefficients are 0.4, 0.3, 0.2, and 0.1, respectively, and the response enhancement ratio is normalized by dividing by 10. For grout leakage defects, the resonance frequency will show a significant shift due to changes in local material properties caused by grout penetration. The degree of this shift can be quantified through matching degree analysis. When the comprehensive matching parameter is low, it indicates that there is a significant material anomaly in the area, requiring further defect confirmation and classification, similar to abnormal indicators in a physical examination suggesting the need for further investigation. The matching level is classified based on the comprehensive matching parameter value: excellent matching (>0.85), good matching (0.7-0.85), average matching (0.5-0.7), and poor matching (≤0.5).

[0034] A precisely tuned excitation beam, based on frequency matching parameters, generates a strong resonance at the abrupt boundary. The excitation beam parameters are set based on the optimal excitation frequency f_optimal from the frequency matching parameters. The laser frequency is tuned to f_optimal, and the power is set to P_excite = min(P_max, k_power × ρ_boundary × f_optimal) according to material properties and safety requirements, where P_max is the maximum output power of the device, and k_power is the power adjustment coefficient. The beam focusing system precisely focuses the laser beam to the coordinates (x_boundary, y_boundary) of the abrupt boundary, controlling the focused spot diameter within 0.1 mm. Continuous wave excitation is used, with the excitation time set to t_excite = 20 / f_optimal to ensure full resonance establishment. The resonance phenomenon is determined based on the time evolution characteristics of the response amplitude; resonance is confirmed when the response amplitude grows exponentially and the growth rate k_growth > f_optimal / 10. Resonance intensity was assessed using the quality factor Q = f_optimal / Δf_3dB, where Δf_3dB is the frequency bandwidth at which the response amplitude decreases by 3dB. Nonlinear effects were monitored by analyzing the harmonic components of the response signal; a significant nonlinear effect was considered to exist when the ratio of the second harmonic amplitude to the fundamental frequency amplitude exceeded 5%. Precise frequency control and power adjustment successfully established a stable resonance phenomenon.

[0035] The amplitude and phase changes of the resonance phenomenon are monitored through real-time data acquisition based on the resonance phenomenon generated during the excitation process. Amplitude change monitoring is achieved through high-speed imaging using area array camera 1, with a sampling frequency set to f_sample = 10 × f_optimal to ensure the capture of the complete time evolution of the resonance response. Pixel displacement calculation employs a sub-pixel precision correlation matching algorithm, achieving a displacement resolution of 0.01 pixels, corresponding to an actual displacement accuracy of approximately 0.1 μm. Phase change monitoring is achieved by comparing the phases of the excitation and response signals; the phase difference Δφ = phase(response) - phase(excitation) is calculated using Hilbert transform. The time-series data of the response amplitude A_response(t) and phase φ_response(t) are acquired synchronously, with a data length containing at least 20 complete resonance cycles. Frequency domain analysis is performed using Fast Fourier Transform (FFT) to extract the fundamental frequency response amplitude A_fundamental, phase φ_fundamental, and higher-order harmonic components. The damping ratio ζ is calculated by fitting the response envelope with exponential decay. After excitation stops, the response amplitude decays according to A(t) = A_0 × exp(-ζωt). The resonance quality factor Q_factor = 1 / (2ζ) reflects the energy dissipation characteristics of the material at that location. The nonlinear coefficient α_nonlinear is calculated by the ratio of the amplitude of the third harmonic to the fundamental frequency, α_nonlinear = A_3rd_harmonic / A_fundamental³. High-precision synchronous acquisition and frequency domain analysis ultimately established a complete set of material resonance response parameters containing four key pieces of information: amplitude, phase, damping, and quality factor.

[0036] Material resonance response parameters are used to generate reconstructed stress field data within the material. The stress-frequency relationship is established based on elasticity theory: the relationship between stress and resonance frequency is σ = k_stress × (f_response² - f_natural²), where k_stress is the stress-frequency coupling coefficient. The response amplitude A_response reflects the intensity of stress concentration; a larger amplitude indicates a higher degree of stress concentration at that location. Phase information φ_response is used to determine the direction and nature of stress; phase lead indicates compressive stress, and phase lag indicates tensile stress. The damping parameter ζ reflects the energy dissipation characteristics within the material; high-damping regions typically correspond to material defects or damage locations. The quality factor Q_factor = 1 / (2 × ζ) provides a quantitative assessment of material integrity; regions with a decreasing quality factor correspond to locations of high stress or material damage. The principal stress directions are determined by analyzing the phase response differences under different excitation directions; the direction of maximum principal stress corresponds to the excitation direction with the largest phase shift. The spatial distribution of the stress field is achieved by establishing a measurement grid around the abrupt boundary locations; the grid density is determined based on the geometry of the abrupt boundary. The three-dimensional stress tensor reconstruction uses amplitude and phase data from multi-directional excitation and solves for six independent stress components using the least squares method. When microcracks are detected inside the etchant, stress concentration at the crack tip leads to a significant increase in the resonance response amplitude, a sharp decrease in the quality factor, a noticeable phase jump, and an increase in the damping ratio.

[0037] Step S130: Determine the coordinates of the stress concentration point based on the stress field reconstruction data, perform reverse optical tracing from the stress concentration point outward to form a defect formation path, and establish a self-organizing optical monitoring grid through the defect formation path.

[0038] Specifically, the coordinates of stress concentration points are determined based on the reconstructed stress field data, and a stress peak detection algorithm is used to identify stress concentration regions. The stress field data includes stress tensor components σ_xx, σ_yy, σ_zz, τ_xy, τ_yz, and τ_zx at each location in three-dimensional space. The degree of stress concentration at each location is assessed by calculating the equivalent stress σ_eq = √[(σ_xx-σ_yy)² + (σ_yy-σ_zz)² + (σ_zz-σ_xx)² + 6(τ_xy² + τ_yz² + τ_zx²)] / √2. The stress gradient is calculated by spatially differentiating the equivalent stress field, and the gradient vector is obtained. This reflects the severity of stress changes. Stress concentration point identification employs a local maximum search method. A stress concentration point is identified when the equivalent stress at a certain location simultaneously meets two conditions: the equivalent stress value exceeds twice the average stress, and the stress gradient magnitude exceeds a set threshold. Multi-scale analysis considers stress concentration phenomena at different scales, using a Gaussian filter to smooth the stress field and identify large-scale and small-scale stress concentration characteristics. Coordinate refinement improves coordinate accuracy by calculating the centroid position of the stress concentration region using the centroid method. When multiple stress concentration points are identified through multi-peak detection, they are prioritized according to stress intensity, with the main stress concentration points analyzed first. Peak detection and gradient analysis accurately identify the three-dimensional coordinate positions (x_center, y_center, z_center) of stress concentration points.

[0039] In some embodiments, the step of forming a defect formation path by reverse optical tracing outward from the stress concentration point includes: setting multiple tracing directions from the stress concentration point; detecting changes in optical properties along each tracing direction to obtain change gradient data; determining possible paths for defect propagation based on the change gradient data, and generating a defect formation path.

[0040] The tracking directions are set radially, starting from the coordinates of the stress concentration points. Radial tracking directions are defined using a spherical coordinate system, with azimuth angles φ ranging from 0° to 360° in 30° intervals (12 directions in total), and elevation angles θ ranging from 0° to 180° in 45° intervals, forming a five-layer tracking network. Tangential tracking directions consider the principal stress directions at stress concentration points, prioritizing tracking paths along the directions of maximum and minimum principal stresses. Axial tracking directions are set along the main geometric axes of the material, including length, width, and thickness. The tracking step size is determined based on the material's characteristic dimensions, with a typical step size of 0.1 mm to ensure sufficient spatial resolution. Directional weight allocation considers the anisotropic characteristics of stress concentration, assigning higher weights to principal stress directions and lower weights to secondary directions. For linear defects such as fractures in etchant materials, defects typically extend along the stress concentration direction; therefore, dense tracking directions need to be set perpendicular to the crack surface. A dynamic adjustment mechanism optimizes the direction settings based on initial tracking results; when a direction shows a strong change in optical properties, more tracking paths are added near that direction. Boundary limits set the maximum tracking distance to avoid ineffective long-distance tracking, typically limited to a 5mm range around stress concentration points.

[0041] Changes in optical properties are detected along each tracking direction, with point-by-point measurements performed based on the set tracking direction, and the measurement points in each direction are spaced 0.05 mm apart. Reflectivity changes are monitored and recorded to measure the light reflection intensity at different locations; reflectivity R(x,y,z) = I_reflected / I_incident reflects changes in the optical properties of the material's surface and interior. Transmittance changes are detected to measure the attenuation of light passing through the material; transmittance T(x,y,z) = I_transmitted / I_incident indicates changes in material density and structure. Scattering characteristic analysis measures the light scattering distribution through multi-angle illumination; changes in scattering intensity reflect the development of internal defects in the material. Polarization characteristic detection uses a polarization light source and polarization analyzer to measure changes in the material's birefringence effect, which is closely related to stress state and material structure. Spectral characteristic monitoring records changes in the response of light at different wavelengths; changes in material composition and structure can lead to changes in the absorption or emission characteristics of specific wavelengths. Phase change measurement is achieved through interferometry; changes in the material's optical thickness cause changes in the optical path difference, resulting in a phase shift. Gradient calculation employs the finite difference method, calculating the spatial derivatives of each optical parameter along the tracking direction. The gradient vector is G = (∂P / ∂s), where P represents the optical parameter and s is the distance along the tracking direction. Gradient intensity is calculated by determining the magnitude of the gradient vector: |G| = √(Gx² + Gy² + Gz²), reflecting the drastic change in optical properties. The gradient direction is determined by the unit vector of the gradient vector, with direction angles θ = arccos(Gz / |G|) and φ = arctan(Gy / Gx) representing the angle with the z-axis and the azimuth angle in the xy-plane, respectively. Multi-parameter fusion weights and combines the gradient intensity and direction information of different optical properties to form comprehensive gradient data containing both gradient intensity values ​​and direction angles.

[0042] Possible defect propagation paths are determined based on varying gradient data, calculated using gradient strength and direction information from the data. A gradient strength threshold is set to distinguish effective paths from noise interference; a valid defect propagation path is considered to exist in a direction when the gradient strength |G| exceeds the dynamic threshold Thr_gradient = μ_G + 1.5 × σ_G, where μ_G and σ_G are the mean and standard deviation of the gradient strength, respectively. Path continuity analysis is performed by checking the gradient continuity between adjacent measurement points; the continuity index C = |G_i+1-G_i| / |G_i| assesses the smoothness of gradient changes. Directional consistency assessment evaluates the stability of the gradient direction; good directional consistency is considered when the angle between gradient directions between consecutive measurement points is less than 30°. Path bifurcation detection identifies locations where a single path splits into multiple sub-paths, and path convergence analysis identifies situations where multiple paths converge at the same location. Path optimization employs the minimum energy principle, with defect propagation tending to develop along the path with the highest energy release rate. When analyzing stress corrosion cracking within the etching paste, cracks propagate along the path where chemical corrosion and mechanical stress are most intense, typically exhibiting a continuous gradient change in optical properties. Path geometric feature extraction includes path length statistics, density calculation, and morphological analysis. Path length is calculated by accumulating the distances between measurement points; path density is obtained by counting the number of paths per unit volume; and path morphology is determined by analyzing path curvature, branch number, and spatial distribution patterns. Development trend analysis is based on the spatiotemporal evolution characteristics of the path. The propagation rate is estimated using the gradient intensity change rate at path endpoints, directional stability is assessed using the coefficient of variation of the path tangent direction, and temporal evolution characteristics are identified by comparing path states at different times. Path ranking is performed based on a comprehensive score of length, density, and propagation rate to identify the most dangerous main paths. In-depth analysis and feature extraction of the changing gradient data form a complete defect formation path encompassing geometric features and development trends.

[0043] A self-organizing optical monitoring grid is established based on the defect formation path. The length statistics of the defect formation path reflect the defect propagation range, and the path density calculation counts the number of paths per unit volume to identify hazardous areas with concentrated defects. The propagation rate is estimated based on path geometry and stress driving force calculations, using Paris's law to describe the crack propagation rate: da / dN = C × (ΔK)^m, where a is the crack length; N is the number of cycles; C and m are material constants; and ΔK is the range of stress intensity factor variation. Critical path identification comprehensively considers factors such as path length, density, direction, and stress level. Degradation modes are classified into uniform degradation, locally concentrated degradation, directional degradation, and randomly distributed degradation. A time prediction model establishes the relationship between path development and time, predicting the time when the defect reaches a critical state, and considering the influence of external conditions such as temperature, humidity, and chemical environment on the degradation rate. Then, a self-organizing optical monitoring grid is established in the potential defect area based on the generated degradation prediction data. The initial grid construction uses path density data to determine the monitoring focus, setting high-density, medium-density, and low-density monitoring areas according to path density. The grid node coordinates are distributed based on path length data, with denser node spacing in long-path areas and sparser node spacing in short-path areas. The grid coverage is determined using degradation pattern data: locally concentrated degradation areas use rectangular grids, directional degradation areas use striped grids, and randomly distributed degradation areas use circular grids. The grid topology employs a hexagonal close-packed arrangement to improve space utilization efficiency, ultimately establishing a self-organizing optical monitoring grid covering all defect areas.

[0044] Step S140: The density of the self-organized optical monitoring grid is adaptively adjusted to determine the optimal grid distribution. Differential spectral excitation is applied to the optimized grid distribution to form a hierarchical spectral fingerprint library. Material state boundary identification criteria are generated based on the hierarchical spectral fingerprint library.

[0045] In some embodiments, the step of adaptively adjusting the density of the self-organized optical monitoring grid to determine an optimized grid distribution includes: calculating the importance weight of each region based on the self-organized optical monitoring grid; generating a grid density adjustment factor based on the importance weight; and increasing the grid point density in high-weight regions based on the adjustment factor to form an optimized grid distribution.

[0046] Importance weights for each region are calculated based on a self-organizing optical monitoring grid. Region division and feature analysis are performed based on the location coordinates and coverage area of ​​each node within the grid. Region division uses a Voronoi diagram method, generating corresponding Voronoi cells with each grid node as a seed point. Each cell represents the monitoring responsibility area of ​​that node. Node density statistics are achieved by calculating the ratio of the number of nodes to the cell area within each Voronoi cell. Coverage overlap analysis calculates the degree of overlap between adjacent Voronoi cells; the ratio of the overlapping area to the total area reflects monitoring redundancy. Node connectivity analysis uses graph theory to calculate the degree centrality and betweenness centrality of each node, assessing its importance in the network. Distance weights consider the distance from the node to the defect center; closer nodes have higher weights, calculated using an exponential decay function. The overall importance weight is obtained through a multi-indicator weighted calculation: node density 30%, overlap 20%, boundary complexity 20%, connectivity 15%, and distance 15%. When monitoring fracture propagation at the edge of the etching paste, the crack tip region has the highest importance weight and requires the densest distribution of monitoring points, while the stable region far from the crack has a lower weight. Weight normalization ensures that the sum of all weights is 1, facilitating subsequent calculation of adjustment factors. Weight grading divides continuous weight values ​​into three levels: high weight, medium weight, and low weight.

[0047] A grid density adjustment factor is generated based on importance weights, establishing a quantitative mapping relationship between weights and adjustment intensity. The linear mapping relationship uses a piecewise function: for high-weight regions, the adjustment factor α_adjust = 2.0 + 1.0 × (W - 0.7) / 0.3, with a range of 2.0-3.0; for medium-weight regions, α_adjust = 1.0 + 1.0 × (W - 0.3) / 0.4, with a range of 1.0-2.0; and for low-weight regions, α_adjust = 0.5 + 0.5 × W / 0.3, with a range of 0.5-1.0. Nonlinear correction introduces a quadratic term in the weights to enhance the adjustment intensity in high-weight regions. Gaussian filtering is used for smoothing the adjustment factor to eliminate abrupt changes in adjustment factors between adjacent regions. A dynamic adjustment mechanism adjusts the mapping parameters based on real-time monitoring results, recalculating the adjustment factor when a significant change in weight distribution is detected. When a new defect initiation point appears in the etching paste, the importance weight of that region increases sharply, and the corresponding adjustment factor also increases rapidly. Discretization of the adjustment factor quantizes continuous adjustment factor values ​​into integer multiples of 0.1, facilitating practical grid adjustment operations.

[0048] Based on the adjustment factor, the grid density in high-weight regions is increased, and the grid distribution is intelligently adjusted through a spatial optimization algorithm to form an optimized grid distribution. The node addition algorithm employs an adaptive subdivision method; when the adjustment factor exceeds 1.5, new nodes are added to the existing grid. The addition strategy is determined based on the adjustment factor: four subdivisions are used for high adjustment factor regions, two subdivisions for medium adjustment factor regions, and one subdivision for low adjustment factor regions. The position of new nodes is determined using a centroid offset algorithm, with the offset direction pointing towards the direction of the largest weight gradient, and the offset distance being a certain proportion of the original node spacing. Position optimization uses the Lloyd algorithm iteratively, achieving uniform distribution by minimizing the coefficient of variation of the distance between nodes. The grid topology update uses Delaunay triangulation to reconstruct node connections, ensuring grid connectivity and stability. Geometric constraints consider physical boundaries and prohibited regions; newly added nodes cannot exceed material boundaries or enter unmonitored areas. When monitoring stress concentration in the central region of the etching paste, high-stress areas require denser monitoring points to capture subtle changes in the stress field; adaptively increasing node density significantly improves monitoring accuracy. Spatial optimization of the adjustment factor and intelligent node distribution form an optimized grid distribution covering all key regions.

[0049] Differentiated spectral excitations were applied to the optimized grid distribution, and a spectral response feature library for different grid regions was established using a multi-band excitation strategy. A partitioned excitation strategy was formulated based on the node density and spatial distribution within the optimized grid. The selected excitation bands covered the three main spectral regions: ultraviolet, visible, and near-infrared, with each region subdivided into multiple sub-bands. High-density regions used full-spectrum excitation with a power of 50-100 mW and an excitation time of 0.5 seconds per band. Medium-density regions used dual-band excitation, selecting the visible and near-infrared bands, with a power of 30-60 mW and an excitation time of 1.0 second per band. Low-density regions used single-band excitation, primarily selecting the visible band, with a power of 20-40 mW and an excitation time of 2.0 seconds. The excitation sequence employed time-division multiplexing to avoid interference between excitations of adjacent nodes. Response feature extraction included multi-dimensional optical parameters such as reflection intensity, transmission intensity, scattering mode, and polarization state. Spectral fingerprint construction was achieved by combining multi-band response data into a feature vector. The grading strategy is based on grid density and response intensity, establishing four levels: excellent, good, average, and poor. When establishing the spectral characteristics of different thicknesses of the etching paste, there are significant differences in the responses of thick and thin regions to the same wavelength excitation. By using differentiated excitation, a precise correspondence between thickness and spectral response can be established.

[0050] Based on a graded spectral fingerprint database, a material state boundary identification standard is generated, establishing a quantitative mapping relationship between spectral features and material states. This is achieved by calculating the cluster centers and distribution ranges of fingerprints at each level in the feature space. Boundary thresholds are determined by the midpoints of cluster centers at adjacent levels: T1 = (C1 + C2) / 2, T2 = (C2 + C3) / 2, T3 = (C3 + C4) / 2, where C1, C2, C3, and C4 are the cluster centers for the four levels, respectively. Confidence intervals are set considering the degree of distribution overlap; the uncertainty of the boundary threshold at a 95% confidence level is calculated using the standard deviation. The discriminant function employs linear discriminant analysis to establish a mapping relationship between feature vectors and state levels, with a required accuracy of over 90%. Multidimensional feature fusion reduces feature dimensionality through principal component analysis, retaining 95% of the information and simplifying the discrimination process. When identifying the aging degree of etching paste, the four states—fresh material, slightly aged, moderately aged, and severely aged—form distinct regional distributions in the spectral feature space. The aging state of any sample can be accurately determined using the boundary standard. Through statistical analysis and boundary optimization using a hierarchical spectral fingerprint database, a complete material state boundary identification standard was finally generated.

[0051] Step S150: Based on the material state boundary identification standard, the critical point of material phase transformation is detected to obtain phase transformation boundary data. A pulse beam is applied near the phase transformation boundary to excite the deformation of the material surface and extract the surface deformation response features. The surface deformation response features are used to construct a multi-point linkage defect blocking network.

[0052] Specifically, phase transition boundary data is obtained by detecting critical points of material phase transitions based on material state boundary identification standards. A boundary approximation method is used to identify critical points, searching for locations where material characteristics abruptly change near the boundary thresholds of adjacent state levels. The search algorithm establishes a scanning grid along the material surface with a grid spacing of 0.05 mm, measuring the spectral feature vector at each grid point. The distance between the feature vector and the boundary threshold is calculated using Euclidean distance; a distance less than 0.1 × σ_boundary is considered close to the critical phase transition state, where σ_boundary is the standard deviation of the boundary threshold. Critical points are classified based on the proximity of the boundary type: those close to the T1 threshold are excellent-good phase transition points, those close to the T2 threshold are good-average phase transition points, and those close to the T3 threshold are average-poor phase transition points. The phase transition boundary data includes critical point coordinates (x_critical, y_critical), phase transition type, critical intensity, and stability assessment. When detecting the temperature phase transition characteristics of etching pastes, materials undergo a transition from a solid to a semi-liquid state at specific temperatures; this phase transition critical point corresponds to a significant change in the material's spectral characteristics. The critical intensity is calculated using the magnitude of the gradient of the eigenvectors; a larger gradient indicates a more intense phase transition. Stability assessment considers the spatial distribution of features around the critical point; a more uniform distribution indicates a more stable phase transition. Phase transition boundary mapping connects all critical points to form a continuous boundary line, with mapping accuracy controlled within 0.02 mm. This ultimately results in phase transition boundary data containing critical point coordinates, phase transition type, and critical intensity.

[0053] In some embodiments, the step of applying a pulsed light beam near the phase transition boundary to excite surface deformation and extract surface deformation response features includes: setting multiple synchronous excitation points in the boundary region based on the phase transition boundary data; simultaneously applying pulsed light beams to the multiple synchronous excitation points to obtain superimposed excitation energy; generating synergistic deformation of the material surface through the superimposed excitation energy to obtain an enhanced deformation effect; and monitoring the propagation mode and recovery characteristics of the enhanced deformation effect to form surface deformation response features.

[0054] The layout of synchronous excitation points is designed based on the critical point coordinates and phase transition type in the phase transition boundary data. The distribution of excitation points uses the critical point coordinates as the reference positions, with 3-5 excitation points set within a 0.5mm range around each critical point. The spacing between excitation points is determined according to the phase transition type: 0.2mm for excellent-good phase transition regions, 0.3mm for good-moderate regions, and 0.5mm for moderate-poor regions. Excitation points are numbered using a combination of region index and serial number for easy synchronous control and data management. The excitation sequence design considers the continuity of the phase transition boundary, exciting points sequentially along the boundary direction to ensure the coherence of deformation propagation. Excitation intensity allocation is based on critical intensity data; higher critical intensity points are allocated greater excitation power. Time synchronization uses a master-slave control mode; the master controller sends a synchronization signal, and all excitation points start simultaneously upon receiving the signal. When analyzing the stress release characteristics of the etchant edge, multiple excitation points are set near the stress-concentrated phase transition boundary. Synchronous excitation can simulate multi-point stress under actual working conditions. Spatial analysis of phase transition boundary data and optimization of excitation point settings established a multi-point synchronous excitation network covering all phase transition regions.

[0055] Superimposed excitation energy is obtained by simultaneously applying pulsed beams at multiple synchronous excitation points. Pulse parameters include pulse width, power density, and repetition frequency. The pulse width is set to 10-50 ns, the power density to 1-10 MW / cm², and the repetition frequency to 1-100 Hz. Synchronization accuracy is controlled using an atomic clock reference, with the time deviation of each excitation point controlled within 1 ns. The energy superposition effect is achieved through wavefront interference; when multiple pulsed beams converge on the material surface, constructive interference enhances the excitation energy. The energy density of the superposition region is calculated using the principle of vector superposition, with the total energy density being the coherent superposition of the energies at each excitation point. Phase control ensures that all pulsed beams maintain phase consistency in the superposition region, maximizing the energy superposition effect. Energy distribution monitoring is achieved using a high-speed spectrometer to measure the energy density distribution in the superposition region. Pulse shape optimization uses a Gaussian distribution to ensure energy concentration and smooth boundaries. When multiple excitation points simultaneously apply pulses to the microcrack region on the etchant surface, the superimposed excitation energy effectively excites crack propagation behavior, revealing the fracture characteristics of the material. Energy stability control is achieved through power feedback adjustment, ensuring consistent output power at each excitation point.

[0056] Synergistic deformation of the material surface is generated by superimposed excitation energy, and the deformation is enhanced by the spatial coupling effect of multi-point excitation. The deformation mechanism includes multiple physical processes such as thermal expansion, photoinduced strain, and phonon excitation, which couple under superimposed energy to produce a synergistic effect. Deformation amplitude is measured using laser interferometry with nanometer-level precision. Spatial deformation distribution is achieved through full-field measurement, recording the deformation amount and direction at each point on the material surface. Temporal deformation evolution is monitored to track the establishment, development, and decay of deformation at a sampling frequency of 1 MHz. The synergistic effect manifests as a nonlinear enhancement of deformation amplitude, with the total deformation exceeding the linear summation of deformations when each excitation point acts independently. The enhancement factor is defined as the ratio of synergistic deformation to independent deformation; an enhancement factor greater than 1 indicates the presence of a synergistic effect. Deformation mode analysis includes three components: radial deformation, tangential deformation, and perpendicular deformation; different modes reflect different response characteristics of the material. In the viscoelastic testing of etching paste, the surface deformation generated by multi-point synergistic excitation can reveal the time-dependent response of the material, providing important information for material performance evaluation. Deformation correlation analysis calculates the correlation coefficient between deformations generated at different excitation points; high correlation indicates a strong synergistic effect. Nonlinear effect identification is achieved through nonlinear analysis of the deformation-excitation energy relationship, with the nonlinear coefficient reflecting the constitutive properties of the material. The synergistic effect of superimposed excitation energy and the deformation coupling effect produce synergistic deformation of the material surface with enhanced characteristics.

[0057] The propagation mode and recovery characteristics of the enhanced deformation effect are monitored. Propagation mode monitoring includes the propagation velocity, propagation direction, and attenuation characteristics of the deformed wave. The propagation velocity is calculated using the time difference between adjacent measurement points, and the propagation direction is determined by the directional angle of the deformation gradient. Wavefront shape analysis describes the geometric characteristics of deformation propagation, including types such as circular, elliptical, and irregular wavefronts. Attenuation characteristics are described by the variation of deformation amplitude with distance, and an exponential attenuation model is used to fit the attenuation coefficient. Recovery characteristics monitoring includes recovery time, recovery degree, and recovery path. Recovery time is defined as the time it takes for the deformation amplitude to decrease to 10% of the initial value. The recovery degree is evaluated by the ratio of the final residual deformation to the initial deformation; a smaller ratio indicates more complete recovery. Recovery path analysis analyzes the temporal evolution trajectory of deformation during the recovery process, including monotonic recovery, oscillatory recovery, and multi-stage recovery modes. Frequency domain analysis is achieved by performing a Fourier transform on the deformation time series to extract the characteristic frequencies and damping characteristics of the deformation response. When detecting the dynamic modulus of the etching paste, the deformation propagation mode reflects the elastic properties of the material, and the recovery characteristics reflect the viscous properties of the material. Combining the two allows for a comprehensive evaluation of the viscoelastic properties of the material. Phase delay analysis calculates the phase difference between the deformation response and the excitation signal; this phase difference reflects the material's energy dissipation characteristics. Spatial correlation analysis reveals the correlation between deformation responses at different locations, demonstrating the material's spatial homogeneity. Dynamic monitoring and feature extraction of the enhanced deformation effect generate surface deformation response characteristics encompassing propagation modes, recovery properties, and frequency domain features.

[0058] A multi-point interconnected defect blocking network was constructed based on the propagation modes, recovery characteristics, and frequency domain features of surface deformation response. First, a stability assessment was conducted, including three dimensions: propagation consistency, recovery integrity, and frequency domain stability. Propagation consistency was assessed through the spatial uniformity of the propagation modes; the standard deviation of the propagation direction and the coefficient of variation of the propagation velocity reflected the degree of material isotropy. Recovery integrity was assessed through the temporal stability of the recovery characteristics; the monotonicity of the recovery path and the consistency of the recovery degree reflected the elastic stability of the material. Frequency domain stability was assessed through the reproducibility of frequency domain features; the degree of characteristic frequency shift and the amplitude of damping characteristic changes reflected the dynamic stability of the material. Based on the stability assessment results, network construction parameters were determined, including a stability threshold, a warning range, and a failure boundary. The stability threshold was determined based on the uniformity index of the propagation modes, the warning range was set based on the degree of deviation of the recovery characteristics, and the failure boundary was defined based on abnormal changes in frequency domain features. The stability threshold was used as the basis for node deployment, with main blocking nodes placed in areas close to the stability threshold. The primary blocking node is positioned at the weakest point of stability, with a node spacing of 0.5 times the warning range. Secondary blocking nodes are positioned at points of moderate stability, with a node spacing of 1.0 times the warning range. Auxiliary blocking nodes are positioned at points of relatively good stability, with a node spacing of 2.0 times the warning range. Network connections are designed based on failure boundary data. A strong connection is established when the stability difference between two nodes exceeds the failure boundary; a moderate connection is established when the difference is within the warning range; and a weak connection is established when the difference is within the stability threshold. The topology adopts a hierarchical network design, including a core layer, a convergence layer, and an access layer. Node functions include defect detection, signal transmission, and energy regulation. Ultimately, a multi-point, interconnected defect blocking network with multiple layers of nodes and multiple types of connections is constructed.

[0059] Step S160: Perform defect analysis on the multi-point linkage defect blocking network to form an interconnected optical energy field, monitor the interconnected optical energy field to determine the coordinates of abnormal positions, perform phase adjustment on the hierarchical spectral fingerprint database to generate a phase difference spectral group, and use the phase difference spectral group to perform phase comparison marking on the coordinates of abnormal positions to complete the quality inspection.

[0060] In some embodiments, the defect analysis of the multi-point linked defect blocking network to form an interconnected optical energy field includes: determining the energy demand of each blocking point based on the multi-point linked defect blocking network; establishing an energy transmission channel between each blocking point according to the energy demand; achieving energy balance between points through the energy transmission channel to obtain a distributed energy field; and performing networked integration processing on the distributed energy field to form an interconnected optical energy field.

[0061] First, the energy requirements of each blocking point are determined based on the functional level and coverage of each node in the multi-point linkage defect blocking network. The energy requirement of the main blocking node is determined based on its blocking capability requirements, which must be able to block the propagation of defects with the highest intensity. The energy requirement is E_main = k_main × A_coverage × I_max, where k_main is the main node coefficient, A_coverage is the coverage area, and I_max is the maximum defect intensity. The energy requirement of secondary blocking nodes is 60%-80% of that of the main node, and the energy requirement of auxiliary nodes is 30%-50% of that of the main node. Node load assessment considers the connectivity and centrality of nodes in the network. Nodes with higher connectivity undertake more collaborative tasks and require higher energy reserves. Blocking nodes in the stress concentration area at the edge of the etching paste have significantly higher energy requirements than monitoring nodes in stable areas because they need to handle the propagation of high-intensity stress. Next, energy transmission channels are established between each blockage point based on energy demand. The channel capacity design is determined based on the maximum energy transmission demand, with the transmission capacity of the main channel being P_main = 1.5 × max(E_demand) and the transmission capacity of the secondary channel being P_secondary = 1.2 × avg(E_demand). The transmission path is optimized using the shortest path algorithm, selecting the path with the shortest transmission distance while satisfying the capacity constraint. Transmission efficiency optimization considers transmission loss and delay; transmission loss is directly proportional to transmission distance, and delay is inversely proportional to channel capacity. Optical transmission is implemented using optical fiber or optical waveguides, with transmission loss controlled within the range of 0.1-0.5 dB / km. When the energy demand at a certain blockage point suddenly increases, the transmission power of adjacent channels is automatically adjusted to ensure timely energy supply. Then, energy balance between points is achieved through the energy transmission channels, using a diffusion model where high-energy nodes transmit energy to low-energy nodes until a balance is reached. The balance objective function is to minimize the variance of the energy at each node, with the objective function F = min(σ²_energy), where σ²_energy is the variance of the node energy. Energy flow calculation is based on the gradient descent principle, with the energy flow direction from high potential energy points to low potential energy points, and the flow rate proportional to the potential energy gradient. Distributed control avoids the risks of single-point control, with each node possessing a certain degree of autonomous adjustment capability. During etching paste inspection, when a large-area defect appears in a certain region, the energy consumption of the blocking nodes in that region increases sharply, automatically allocating energy from surrounding nodes to maintain the stability of the overall energy field. The balance control and distributed adjustment of the energy transmission channels form a stable distributed energy field with adaptive capabilities. Finally, the distributed energy field undergoes network-based integration processing. The integration architecture adopts a hierarchical design, including three layers: the physical layer, the network layer, and the application layer. The physical layer is responsible for energy generation, transmission, and distribution; the network layer is responsible for information transmission and coordination; and the application layer is responsible for function implementation and optimization. A cooperative control algorithm coordinates the work of each subsystem to ensure optimal overall performance.Information fusion technology comprehensively processes the state information of each node to form a global energy field situation. Spatial interpolation methods interpolate the energy field between nodes to form a continuous energy field distribution. Time synchronization ensures the coordinated operation of each node, with synchronization accuracy controlled at the millisecond level, ultimately forming an interconnected optical energy field with global coordination capabilities.

[0062] The coordinates of anomaly locations are determined by monitoring the interconnected optical energy field. First, collaborative response monitoring involves real-time data acquisition based on the operating status and response characteristics of the interconnected optical energy field. Response characteristics include three dimensions: response speed, response intensity, and response consistency. Response speed is measured by the time difference between the excitation and response signals; response intensity is evaluated by the amplitude ratio of the output and input signals; and response consistency is determined by correlation analysis of the responses at different nodes. Spatial response distribution is obtained through full-field measurement; temporal response evolution is monitored to track the establishment, development, and stabilization of the response; and frequency domain response characteristics are identified through spectral analysis to determine the characteristic frequencies and resonant modes of the energy field. Finally, network stability data, including response characteristics, stability indices, and dynamic time parameters, is generated. Next, the material quality level is determined based on the network stability data, establishing a mapping relationship between network performance and material quality. The quality grading system adopts a four-level system: Excellent grade materials are judged by response speed <50ms, response intensity deviation <5%, response consistency >0.95, response settling time <30ms, and settling time >400ms; Good grade materials are judged by response speed <100ms, response intensity deviation <10%, response consistency >0.90, response settling time <60ms, and settling time >300ms; Qualified grade materials are judged by response speed <300ms, response intensity deviation <20%, response consistency >0.80, response settling time <120ms, and settling time >100ms; Unqualified grade materials are those that do not meet the qualified grade requirements. The comprehensive scoring algorithm uses a weighted average of multiple indicators, with weights allocated as follows: response characteristics 40%, stability index 35%, and dynamic time parameter 25%. Finally, the locations of quality anomalies are identified based on material quality grade and network stability data. A multi-level screening method is used, firstly, a coarse screening is performed based on material quality grade, marking unqualified and some qualified grade materials as potential anomaly areas. Specific indicators based on network stability data are carefully selected, and points with a response speed > 300ms, a response strength deviation > 20%, or a response consistency < 0.80 are marked as anomalous response points. Spatial clustering analysis groups adjacent anomalous response points into anomalous regions, and the coordinates of the anomalous locations are calculated using the centroid of the regions, ultimately forming an intuitive anomalous coordinate distribution map.

[0063] In some embodiments, the step of generating a phase difference spectral group by phase adjustment of the hierarchical spectral fingerprint database includes: extracting the real-time phase state of each level of spectrum from the hierarchical spectral fingerprint database; establishing a dynamic phase tracking mechanism based on the real-time phase state; and generating a phase difference spectral group by adaptive phase modulation of different levels of spectrum through the dynamic phase tracking mechanism.

[0064] Real-time phase states of spectra at each quality level were extracted from a graded spectral fingerprint database, and phase characteristic information corresponding to different quality levels was obtained through phase analysis technology. Phase state extraction was based on phase analysis of spectral fingerprint data at four levels: excellent, good, average, and poor. Phase analysis employed the Hilbert transform method to separate the real and imaginary parts of the spectral fingerprint, with the phase angle θ = arctan(Im / Re), where Im is the imaginary part and Re is the real part. The phase distribution characteristics of each quality level's spectra were obtained through statistical analysis, including the phase mean, phase standard deviation, and phase distribution range. The phase distribution of excellent-level spectra was relatively concentrated with a small phase standard deviation; the phase distribution of good and average-level spectra gradually diffused; and the phase distribution of poor-level spectra was the most dispersed. Phase stability was assessed through phase reproducibility analysis of repeated measurements; levels with good stability showed high phase reproducibility. Phase spectrum analysis expanded the phase information in the frequency domain to identify the characteristic phase frequencies of different quality levels. When analyzing the spectral characteristics of etching pastes at different aging degrees, the spectral phase of fresh materials was relatively stable, while the phase of aged materials showed significant shifts and dispersion. Real-time phase state includes three parameters: instantaneous phase value, phase change rate, and phase stability. The instantaneous phase value reflects the current phase state, the phase change rate reflects the dynamic characteristics of the phase, and the phase stability reflects the reliability of the phase. Through phase analysis and feature extraction using a hierarchical spectral fingerprint database, real-time phase state data corresponding to each spectral level is generated.

[0065] A dynamic phase tracking mechanism is established based on real-time phase state. A Kalman filter method is used to build a dynamic model of the phase state, predicting future phase states and making real-time corrections. The state equation describes the temporal evolution of the phase, while the observation equation describes the measurement relationship of the phase. The filter gain is adaptively adjusted according to the phase change rate and phase stability; the filter gain increases when phase changes drastically and decreases when the phase is stable. Tracking accuracy is evaluated through statistical analysis of prediction errors, and the tracking accuracy is required to be controlled within ±0.1 radians. Multi-level parallel tracking simultaneously monitors the phase state of four quality levels, with each level operating independently to avoid mutual interference. Phase jump detection identifies sudden phase changes; when the phase change rate exceeds a set threshold, jump processing is triggered. When the etching paste material undergoes a phase transition or new defects appear, the spectral phase of the corresponding level will jump, and the tracking mechanism can capture this change in a timely manner. The tracking results include three outputs: phase trajectory, phase prediction, and phase position confidence. The phase trajectory records the historical evolution of the phase, the phase prediction provides a phase estimate for future moments, and the phase position confidence reflects the reliability of the prediction.

[0066] A phase difference spectral set is generated by adaptively modulating the phases of different spectral grades using a dynamic phase tracking mechanism. The modulation strategy employs a phase correction method to eliminate phase shifts caused by factors such as environmental temperature changes, light source fluctuations, and device aging, maintaining the standard phase characteristics of each spectral grade. The excellent grade spectrum is selected as the reference phase, and the phases of other grades are corrected relative to the reference phase. The correction algorithm uses feedback adjustment based on the prediction error of the phase tracking mechanism, automatically compensating when a phase deviation from the standard value is detected. Good, average, and poor grade spectra are respectively phase-corrected, with correction accuracy controlled within ±0.01 radians. The adaptive mechanism adjusts the correction amplitude according to real-time environmental parameters to ensure the stability of the phase characteristics of each spectral grade. During long-term detection, light source aging causes an overall shift in the spectral phase; phase correction can eliminate this systematic error, maintaining the consistency and accuracy of the detection results. Correction verification is evaluated by comparing the phase stability before and after correction, generating a phase difference spectral set containing four standard spectra: excellent, good, average, and poor.

[0067] Phase comparison marking was performed on the coordinates of anomaly locations using a phase difference spectral group. The actual spectral phase was measured at each anomaly location coordinate, and the spatial characteristics of the phase distribution were obtained using a multi-point sampling method. The measured phase data were then compared and analyzed one by one with the four phase levels (excellent, good, average, and poor) in the phase difference spectral group. Phase similarity was calculated using the phase correlation coefficient method, with the formula: R=Σ(φ_m-φ_avg_m)(φ_s-φ_avg_s) / √[Σ(φ_m-φ_avg_m)²Σ(φ_s-φ_avg_s)²], where R is the phase correlation coefficient; φ_m is the measured phase; φ_s is the standard grade phase; φ_avg_m is the mean of the measured phase; and φ_avg_s is the mean of the standard phase. The grade with the highest correlation coefficient is the quality grade mark for that anomaly location. A correlation coefficient greater than 0.85 is considered a good match, and less than 0.7 is considered a low match. The marking accuracy was verified through statistical analysis of phase differences; the greater the phase difference, the higher the marking accuracy. Phase feature extraction at anomaly locations includes three key parameters: phase mean, phase variance, and phase shift. The phase mean reflects the average phase state at that location, the phase variance assesses the dispersion of the phase distribution, and the phase shift quantifies the deviation from the reference phase. These parameters are compared and analyzed with corresponding parameters of the standard grade to establish a feature vector matching matrix, and the optimal matching grade is solved using the least squares method. The quality inspection results include three dimensions: quality grade determination of the anomaly location, anomaly severity assessment, and phase deviation analysis, forming a complete quality inspection report. Quality grade determination is based on quantitative analysis of phase similarity, with anomaly severity categorized into three levels: slight, moderate, and severe. Phase deviation analysis calculates the deviation direction and magnitude, providing a preliminary basis for judging the defect type. When the phase feature of an anomaly location is highly similar to the spectrum of a poor-grade anomaly, it is marked as a severe quality anomaly, requiring focused attention and handling. After inspection, a quality distribution map of the anomaly location is generated, using a pseudo-color coding method to visually display the quality status of the material surface and the distribution of anomaly areas. Excellent areas are displayed in green, good in yellow, average in orange, and poor in red, with the severity of the anomaly distinguished by color intensity. The quality distribution map provides statistical information including key indicators such as the area ratio of each grade, the number of outliers, and the quality uniformity index, ultimately completing the quality inspection of the etching paste material based on the differences in spectral response.

[0068] To implement the etching paste material quality inspection method based on spectral response differences corresponding to the above method embodiments, and to achieve the corresponding functions and technical effects. See also Figure 3 , Figure 3This diagram illustrates a structural block diagram of an etching paste material quality inspection system 300 based on spectral response differences, according to an embodiment of this application. For ease of explanation, only the parts relevant to this embodiment are shown. The etching paste material quality inspection system 300 based on spectral response differences provided in this embodiment includes: The spectral acquisition module 301 is used to acquire multi-band spectral signals and material thickness data on the surface of the etching paste, perform spectral tomography decomposition based on the multi-band spectral signals to form a layered spectral map inside the material, and determine the location of density abrupt boundary by combining the material thickness data and the layered spectral map. The resonance excitation module 302 is used to apply a specific frequency beam at the abrupt boundary position to resonate and generate material resonance response parameters, and use the resonance response parameters to generate material internal stress field reconstruction data. Tracking and analysis module 303 is used to determine the coordinates of stress concentration points based on the stress field reconstruction data, perform reverse optical tracking from the stress concentration points outward to form a defect formation path, and establish a self-organizing optical monitoring grid through the defect formation path; The grid monitoring module 304 is used to adaptively adjust the density of the self-organized optical monitoring grid to determine an optimized grid distribution, apply differential spectral excitation to the optimized grid distribution to form a hierarchical spectral fingerprint library, and generate a material state boundary identification standard based on the hierarchical spectral fingerprint library. Phase transition detection module 305 is used to detect critical points of material phase transition based on the material state boundary identification standard, obtain phase transition boundary data, apply a pulse beam near the phase transition boundary to excite material surface deformation and extract surface deformation response features, and use the surface deformation response features to construct a multi-point linkage defect blocking network. The phase marking module 306 is used to perform defect analysis on the multi-point linkage defect blocking network to form an interconnected optical energy field, monitor the interconnected optical energy field to determine the coordinates of abnormal positions, perform phase adjustment on the hierarchical spectral fingerprint library to generate a phase difference spectral group, and perform phase comparison marking on the coordinates of abnormal positions through the phase difference spectral group to complete the quality detection.

[0069] The etching paste material quality inspection system 300 based on spectral response differences described above can implement the etching paste material quality inspection method based on spectral response differences described in the above method embodiments. The options in the above method embodiments are also applicable to this embodiment, and will not be detailed here. The remaining content of this application embodiment can be referred to the content of the above method embodiments, and will not be repeated in this embodiment.

[0070] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

[0071] The above embodiments are not an exhaustive list based on the present invention, and there may be many other embodiments not listed. Any substitutions and improvements made without departing from the concept of the present invention are within the protection scope of the present invention.

Claims

1. A method for quality testing of etching paste materials based on differences in spectral response, characterized in that, include: Acquire multi-band spectral signals and material thickness data on the surface of the etching paste, perform spectral tomography based on the multi-band spectral signals to form a layered spectral map of the material interior, and combine the material thickness data and the layered spectral map to determine the location of density abrupt boundary. A specific frequency beam is applied at the abrupt boundary location to resonate and generate material resonance response parameters. The resonance response parameters are then used to generate reconstructed data of the internal stress field of the material. Based on the stress field reconstruction data, the coordinates of stress concentration points are determined. A defect formation path is then formed by reverse optical tracing from the stress concentration points outwards, and a self-organizing optical monitoring grid is established through the defect formation path. The density of the self-organized optical monitoring grid is adaptively adjusted to determine an optimized grid distribution. Differential spectral excitation is applied to the optimized grid distribution to form a hierarchical spectral fingerprint database. Material state boundary identification criteria are generated based on the hierarchical spectral fingerprint database. Based on the material state boundary identification standard, the critical point of material phase transition is detected to obtain phase transition boundary data. A pulse beam is applied near the phase transition boundary to excite the material surface deformation and extract the surface deformation response features. The surface deformation response features are then used to construct a multi-point linkage defect blocking network. Defect analysis is performed on the multi-point linked defect blocking network to form an interconnected optical energy field. The interconnected optical energy field is monitored to determine the coordinates of abnormal locations. The phase adjustment of the hierarchical spectral fingerprint library generates a phase difference spectral group. The abnormal location coordinates are then marked by phase comparison using the phase difference spectral group to complete the quality inspection.

2. The method according to claim 1, characterized in that, The process of forming a layered spectral map of the material's interior based on the multi-band spectral signal through spectral tomography includes: The penetration depth of different bands is determined based on the multi-band spectral signals; Based on the penetration depth, the material is virtually layered to obtain layer interface data; Based on the layered interface data, the spectral response characteristics of each layer are extracted to obtain the layered spectral map of the material.

3. The method according to claim 1, characterized in that, The process of applying a specific frequency beam at the abrupt boundary location to resonate and excite the material to generate resonant response parameters includes: The inherent resonant frequency is obtained by analyzing the material properties based on the location of the abrupt boundary. The frequency matching parameters are obtained by selecting the excitation beam frequency based on the inherent resonant frequency. The resonance phenomenon is generated by applying an excitation beam based on the frequency matching parameters. The amplitude and phase changes of the resonance phenomenon are monitored to form the material's resonance response parameters.

4. The method according to claim 1, characterized in that, The process of forming a defect formation path by reverse optical tracing outward from the stress concentration point includes: Multiple tracking directions are set from the stress concentration point, including radial tracking direction, tangential tracking direction and axial tracking direction; Changes in optical properties are detected along each tracking direction to obtain gradient data. Based on the changing gradient data, possible paths for defect propagation are determined, and defect formation paths are generated.

5. The method according to claim 1, characterized in that, The step of adaptively adjusting the density of the self-organized optical monitoring grid to determine the optimal grid distribution includes: The importance weight of each region is calculated based on the self-organizing optical monitoring grid. A grid density adjustment factor is generated based on the aforementioned importance weights; Based on the adjustment factor, the grid point density in the high-weight region is increased to form an optimized grid distribution.

6. The method according to claim 1, characterized in that, The step of applying a pulsed light beam near the phase transition boundary to excite material surface deformation and extract surface deformation response characteristics includes: Multiple synchronous excitation points are set in the boundary region based on the phase transition boundary data; By simultaneously applying pulsed beams at the multiple synchronous excitation points, superimposed excitation energy is obtained; The enhanced deformation effect is achieved by generating synergistic deformation of the material surface through the superimposed excitation energy. The propagation mode and recovery characteristics of the enhanced deformation effect are monitored to form surface deformation response features.

7. The method according to claim 1, characterized in that, The step of performing defect analysis on the multi-point linked defect blocking network to form an interconnected optical energy field includes: The energy requirements of each blocking point are determined based on the multi-point linkage defect blocking network. Establish energy transmission channels between the blocking points according to the energy requirements; A distributed energy field is obtained by achieving energy balance between points through the energy transmission channel; The distributed energy field is networked and integrated to form an interconnected optical energy field.

8. The method according to claim 1, characterized in that, The step of generating a phase difference spectral group by phase adjustment of the hierarchical spectral fingerprint database includes: Extract the real-time phase state of each level of spectrum from the hierarchical spectral fingerprint database; A dynamic phase tracking mechanism is established based on the real-time phase state; The dynamic phase tracking mechanism is used to adaptively modulate the phase of different levels of spectra to generate a phase difference spectral group.

9. The method according to claim 3, characterized in that, The step of selecting the excitation beam frequency based on the inherent resonant frequency to obtain the frequency matching parameters includes: The location of the resonance peak is determined based on the inherent resonance frequency; A frequency scanning range is set near the location of the resonance peak; Based on the frequency scanning range, fine-tuning of the frequency is performed to find the maximum response point and obtain the optimal excitation frequency. The degree of matching between the optimal excitation frequency and the inherent resonance frequency is calculated to form a frequency matching parameter.

10. A quality inspection system for etching paste materials based on spectral response differences, characterized in that, include: The spectral acquisition module is used to acquire multi-band spectral signals and material thickness data on the surface of the etching paste, perform spectral tomography based on the multi-band spectral signals to form a layered spectral map of the material's interior, and combine the material thickness data and the layered spectral map to determine the location of density abrupt boundary. The resonance excitation module is used to apply a specific frequency beam at the abrupt boundary position to resonate and generate material resonance response parameters, and use the resonance response parameters to generate material internal stress field reconstruction data. The tracking and analysis module is used to determine the coordinates of stress concentration points based on the stress field reconstruction data, perform reverse optical tracking outward from the stress concentration points to form a defect formation path, and establish a self-organizing optical monitoring grid through the defect formation path. The grid monitoring module is used to adaptively adjust the density of the self-organized optical monitoring grid to determine an optimized grid distribution, apply differential spectral excitation to the optimized grid distribution to form a hierarchical spectral fingerprint database, and generate a material state boundary identification standard based on the hierarchical spectral fingerprint database. The phase transition detection module is used to detect the critical point of material phase transition based on the material state boundary identification standard, obtain phase transition boundary data, apply a pulse beam near the phase transition boundary to excite the material surface deformation and extract surface deformation response features, and use the surface deformation response features to construct a multi-point linkage defect blocking network. The phase marking module is used to perform defect analysis on the multi-point linkage defect blocking network to form an interconnected optical energy field, monitor the interconnected optical energy field to determine the coordinates of abnormal positions, perform phase adjustment on the hierarchical spectral fingerprint library to generate a phase difference spectral group, and use the phase difference spectral group to perform phase comparison marking on the coordinates of the abnormal positions to complete the quality inspection.

Images