Metallic oxide resistor disc appearance defect detection method and system

Abstract

The invention particularly relates to a metal oxide resistor disc appearance defect detection method and system, and relates to the technical field of electrical equipment nondestructive testing. A physical-image coupling model construction module; an inversion solving module based on a physical model; and a physical parameter visualization and quantitative evaluation module. According to the method, microsecond-level synchronization of excitation and acquisition is achieved through synchronous control, it is ensured that an infrared temperature sequence and a polarized light intensity image strictly correspond to physical excitation, and errors caused by data dislocation are avoided; high-quality data support is provided for detection in cooperation with high-precision equipment such as an infrared temperature measurement unit and a polarization control assembly; based on a heat conduction and photoelastic stress-light intensity forward model, a three-dimensional heat conductivity field and a stress field are inverted, accurate correlation between defects and physical parameters is established, traditional qualitative detection limitation is broken through, quantitative characterization of the geometric scale and the mechanical damage degree of the defects is achieved, and errors of detection results are reduced.

Description

Technical Field

[0001] This invention relates to the field of non-destructive testing technology for power equipment, and in particular to a method and system for detecting appearance defects in metal oxide resistors. Background Technology

[0002] Metal oxide resistors are core components of surge arresters, surge protectors, and other equipment in power systems. Their appearance and internal defects directly affect the insulation performance and service life of the equipment.

[0003] Traditional methods for detecting defects in resistive elements mainly include visual inspection, ultrasonic testing, and single thermal imaging testing. However, these methods have obvious limitations: visual inspection relies on human experience, is highly subjective, and is prone to missing tiny defects; ultrasonic testing has high requirements for surface flatness and is difficult to quantify the impact of defects on the physical properties of materials; single thermal imaging testing can only qualitatively determine the existence of defects and cannot accurately characterize the geometric scale and mechanical hazard of defects.

[0004] In existing technologies, some detection systems attempt to combine thermal excitation with image acquisition, but they suffer from problems such as low excitation energy control precision, poor synchronization of multiple sensors, and lack of accurate physical model support. This results in a weak correlation between detection data and the physical properties of defects, making it impossible to achieve quantitative assessment of defects.

[0005] Therefore, there is an urgent need for a defect detection system that can integrate multiple physical excitations, accurate synchronous acquisition, and physical model inversion to solve the problems of traditional detection methods, which are mainly qualitative, have large errors, and low reliability. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for detecting appearance defects in metal oxide resistors in order to solve the above-mentioned problems.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: The physical excitation and synchronous image acquisition module is configured to acquire infrared temperature sequences and polarized light intensity images corresponding to the physical excitation by applying controllable thermal excitation and polarized light field. The physics-image coupling model construction module is configured to establish a forward mathematical physics model of heat conduction and photoelastic stress-light intensity, and to perform simulation and prediction of internal physical properties to observable surface images through finite element discretization. The inversion solution module based on the physical model is configured to minimize the residual between measured and simulated data to inversely deduce the three-dimensional thermal conductivity field and stress field inside the material. The physical parameter visualization and quantitative evaluation module is configured to visualize the physical parameter field obtained by inversion, extract the geometric and physical quantitative indicators of defects, and obtain the defect evaluation coefficient after comprehensive processing, and match the corresponding appearance defect level based on the defect evaluation coefficient.

[0008] Preferably, the physical excitation and synchronous image acquisition module specifically includes: Controllable thermal excitation unit: Select a flash array; By controlling the pulse width, it is ensured that the excitation energy can penetrate to the subsurface of the material; Polarization field application unit: It consists of an electric rotary stage equipped with linear polarizers and waveplates; Three sets of standard polarization angle combinations are preset to obtain enough independent equations to solve the three independent components of the stress tensor; The multi-sensor synchronous image acquisition unit includes: Infrared thermal imager and polarization imaging camera; After the host computer sends the start command, the FPGA first triggers the thermal excitation unit to output a thermal pulse, and after a preset delay, it triggers the infrared camera to acquire data. The polarization camera triggers acquisition after the thermal excitation ends, and each frame of image is timestamped, corresponding one-to-one with the excitation signal timestamp.

[0009] Preferably, the physical-image coupling model construction module specifically includes: The three-dimensional unsteady heat conduction equation is: ; in, For material density, Specific heat capacity at constant pressure; It is a three-dimensional thermal conductivity distribution; For internal heat source items; For temperature; The surface boundary adopts a combined convection-radiation boundary condition; The internal boundary adopts a defect-matrix interface model using contact thermal resistance. At the initial moment, the temperature of the entire sample is uniformly distributed as follows: Equal to the uniform temperature of the sample at the initial moment ; The three-dimensional model of the sample was discretized using a tetrahedral unstructured mesh; The heat conduction equation was discretized into a system of linear equations using the Galerkin weighted residual method. The implicit Euler method is used to solve the transient equations. By inputting the hypothetical three-dimensional thermal conductivity distribution Solve the discrete equations to output the temperature values ​​of each pixel position on the sample surface at different times, forming a simulated temperature sequence consistent with the measured data format. .

[0010] Preferably, the method further includes a photoelastic stress-light intensity positive model: The incident light is linearly polarized. After passing through the sample, a phase delay occurs. The intensity of the emitted light after passing through the analyzer is: ; in, The intensity of the incident polarized light; The polarization direction of the analyzer; The polarization direction of the incident polarized light; The phase delay of polarized light after passing through the sample; For phase shift; When the polarization direction of the incident light is orthogonal to the polarization direction of the analyzer, it simplifies to: ; in, This is the first principal stress of the sample; This is the second principal stress of the sample; The wavelength of the incident light; The material stress optical coefficient; For sample thickness; After phase delay calculation and stress field coupling; Input assumptions about the three-dimensional stress field distribution By combining material parameters and polarization parameters, simulated light intensity images at various polarization angles are calculated. .

[0011] Preferably, the inversion solution module based on the physical model specifically includes: Through observation data Inversely deduce unknown parameters Satisfying the positive model ,in, For positive mapping operators, To observe noise; The objective function is: ,in, It is a weighted matrix; For regularization parameters; This is a regularization term.

[0012] Preferably, the method further includes: Solving the forward model To obtain simulation data Calculate the residual ; adjoint equations based on the forward model , Given the Jacobian matrix transpose of the forward model, solve for the adjoint variables. ; Gradient calculation: , Let be the Jacobian matrix of the positive model, where ; Through iterative optimization, the three-dimensional physical parameter field and thermal conductivity field are output. and stress tensor field .

[0013] Preferably, the physical parameter visualization and quantitative evaluation module specifically includes: Develop custom visual interfaces; and Displayed in volume rendering format; Determine using an adaptive threshold segmentation algorithm Defect threshold ; Identify all that satisfy < The mesh cells are marked as defect regions; Thermal and mechanical defects in the defect region were extracted separately, and thermal and mechanical anomaly values ​​were obtained after analysis. After normalizing the thermal and mechanical anomalies, weighting factors for the thermal and mechanical anomalies are preset respectively. The thermal and mechanical anomalies are then multiplied by their corresponding weighting factors and summed to obtain the defect evaluation coefficient.

[0014] Preferably, the process of obtaining the thermal anomaly value includes: Obtain the defect volume corresponding to each defect region, extract the maximum defect volume, and preset the defect volume threshold. Divide the maximum defect volume by the defect volume threshold and subtract 1. If the resulting value is greater than 0, it is recorded as the defect degree. Obtain the thermal conductivity corresponding to each defect region, and take the average value to obtain the average thermal conductivity. Determine the thermal conductivity of the matrix material, and calculate the difference between the average thermal conductivity and the thermal conductivity of the matrix material, which is recorded as the thermal conductivity difference. Preset the maximum allowable value for thermal conductivity decrease, and divide the thermal conductivity difference by the maximum allowable value for thermal conductivity decrease to obtain the anomaly degree. After normalizing the defect degree and anomaly degree, a weighted summation is performed to obtain the thermal anomaly value.

[0015] Preferably, the process of obtaining the mechanical anomaly value includes: Stress field obtained from inversion In the middle, identify all that meet the requirements > The mesh elements representing the material's yield strength are marked as failure elements; the sum of the surface areas of all failure elements is obtained to get the total failure area; the maximum allowable failure area is preset, and the total failure area is divided by the maximum allowable failure area to obtain the area anisotropy; Obtain the stress tensor of each mesh element, calculate its principal stress, and iterate through the principal stresses of all elements to find the maximum principal stress. Using the material yield strength as a threshold, divide the maximum principal stress by the material yield strength to obtain the stress anisotropy. After normalizing the area anisotropy and stress anisotropy, a weighted summation is performed to obtain the mechanical anomaly value.

[0016] A method for detecting appearance defects in metal oxide resistors includes: By applying synchronous control through controllable thermal excitation and polarization light field, infrared temperature sequences and polarization light intensity images corresponding to physical excitation are acquired. A forward mathematical-physical model of heat conduction and photoelastic stress-light intensity is established, and the internal physical properties are simulated and predicted to the observable images of the surface through finite element discretization. With the goal of minimizing the residual between measured and simulated data, the three-dimensional thermal conductivity field and stress field inside the material are inversely derived. The physical parameter field obtained by inversion is visualized, and the geometric and physical quantitative indicators of defects are extracted. The defect evaluation coefficient is calculated through thermal and mechanical anomalies. The corresponding appearance defect level is matched based on the defect evaluation coefficient.

[0017] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention achieves microsecond-level synchronization between excitation and acquisition through synchronous control, ensuring a strict correspondence between the infrared temperature sequence, polarized light intensity image, and physical excitation, thus avoiding errors caused by data misalignment. Combined with high-precision equipment such as infrared temperature measurement units and polarization control components, it provides high-quality data support for detection. Based on the thermal conduction and photoelastic stress-light intensity forward model, it inverts the three-dimensional thermal conductivity field and stress field, establishing a precise correlation between defects and physical parameters. This breaks through the limitations of traditional qualitative detection, achieving quantitative characterization of defect geometry and mechanical hazard, and reducing detection error.

[0018] 2. This invention comprehensively assesses the hazards of defects from the perspectives of thermal conductivity and structural stability by integrating thermal and mechanical parameter inversion, avoiding the one-sidedness of a single indicator; thermal anomalies quantify the impact of defects on heat dissipation, while mechanical anomalies characterize the risk of structural failure, and the weighted evaluation coefficients closely match the actual failure mechanisms; through adjustable excitation parameters, customizable polarization angle sequences, and settable evaluation thresholds, it adapts to the testing needs of resistor sheets of different specifications; quantitative indicators replace manual experience judgment, reducing subjective errors, and the test results are traceable and comparable, providing an objective basis for quality control and improving the consistency and reliability of testing. Attached Figure Description

[0019] Further details, features, and advantages of this application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 This is a system structure diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention. Detailed Implementation

[0020] Several embodiments of this application will now be described in more detail with reference to the accompanying drawings to enable those skilled in the art to implement this application. This application may be embodied in many different forms and for various purposes and should not be limited to the embodiments set forth herein. These embodiments are provided to make this application thorough and complete, and to fully convey the scope of this application to those skilled in the art. The embodiments described do not limit this application.

[0021] Unless otherwise defined, all terms used herein (including technical and scientific terms) shall have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It will be further understood that terms such as those defined in commonly used dictionaries shall be interpreted as having a meaning consistent with their meaning in the relevant field and / or the context of this specification, and shall not be interpreted in an idealized or overly formal sense unless expressly defined herein.

[0022] Example 1

[0023] Its specific implementation method is combined with the appendix Figure 1 and attached Figure 2 Please provide a detailed explanation.

[0024] Appendix Figure 1 This is a structural block diagram of a metal oxide resistor sheet appearance defect detection system provided in an embodiment of the present invention. It shows the connection relationship between the physical excitation and synchronous image acquisition module and the physical parameter visualization and quantitative evaluation module, and marks the main functional interaction flow of each module.

[0025] Appendix Figure 2 The flowchart of a method for detecting appearance defects in metal oxide resistive sheets provided in this embodiment of the invention shows the complete steps from acquiring infrared temperature sequences and polarized light intensity images corresponding to physical excitation, to visualizing the physical parameter field obtained by inversion, extracting defect geometric and physical quantitative indicators, and calculating defect evaluation coefficients through thermal and mechanical anomalies.

[0026] In this embodiment, it includes: The physical excitation and synchronous image acquisition module is configured to acquire infrared temperature sequences and polarized light intensity images that strictly correspond to the physical excitation through controllable thermal excitation, polarization light field application and FPGA synchronous control, so as to provide high-quality observation data for inversion. Specifically, it includes: Controllable thermal excitation unit: Based on the size of the resistor (e.g., 50mm×50mm×5mm) and the range of thermal conductivity of the material (e.g., 10-100W / (m・K)), select a high-energy xenon flash lamp array with 3-6 LEDs (single lamp power 500-1000W) or a fiber-coupled semiconductor laser source with a wavelength of 808nm (output power 0-500W continuously adjustable). The pulse width control is continuously adjustable from 0.1 to 10 ms by using IGBT power transistors, with an energy output error of ≤ ±5%, ensuring that the excitation energy can penetrate to the material subsurface of 1-5 mm and the surface temperature does not exceed the material safety threshold of 120℃ (to avoid oxidation or performance degradation). Temperature feedback mechanism: Equipped with an infrared temperature feedback unit with a response time ≤1ms, a temperature range of -20-300℃, and an accuracy of ±0.5℃, it monitors the surface temperature peak in real time and dynamically adjusts the excitation energy to avoid sample damage. Polarization field application unit: It consists of a high-precision electric rotary stage with an angular resolution of 0.01° and a repeatability of ±0.02°, equipped with a linear polarizer with an extinction ratio of ≥1000:1 (wavelength range 400-700nm) and a quarter-wave plate with a phase delay accuracy of λ / 100 (applicable wavelength 532nm), thus realizing programmable control of the polarization state of the incident light. Three sets of standard polarization angle combinations are preset (0° / 45° / 90°, 0° / 60° / 120°, 15° / 45° / 75°), and the host computer can customize the polarization angle sequence (step size 0.1°) to ensure that enough independent equations are obtained to solve the three independent components of the stress tensor. Polarization state calibration: Before each experiment, the polarization state of the incident light is automatically calibrated using a polarization analyzer based on a Wollaston prism to ensure that the polarization state accuracy is ≤ ±0.1°. The multi-sensor synchronous image acquisition unit includes: The infrared thermal imager and polarization imaging camera are coaxially mounted (angle ≤ 5°), with the lens center 50-100cm from the sample surface. It is equipped with an electric focusing unit with a focusing accuracy of 0.01mm to ensure that the image pixel resolution is ≤ 0.1mm / pixel. After the host computer sends the start command, the FPGA first triggers the thermal excitation unit to output a thermal pulse, and after a preset delay (0-100μs), it triggers the infrared camera to collect data (capture the rising edge of the thermal pulse); the polarization camera triggers data collection after the thermal excitation ends (or after the static load stabilizes), and each frame of image is stamped with an FPGA timestamp with a precision of 1μs, which corresponds one-to-one with the excitation signal timestamp; Data storage uses an NVMe solid-state drive array with a storage bandwidth of ≥2GB / s to store the original 16-bit grayscale image data in real time, and synchronously record metadata such as excitation parameters (energy, pulse width), sensor parameters (frame rate, exposure time), and environmental parameters (temperature, humidity). The data format is HDF5 (which supports large file block storage and facilitates parallel processing).

[0027] The physics-image coupling model construction module is configured to establish a forward mathematical physics model of heat conduction and photoelastic stress-light intensity, and to perform accurate simulation and prediction of internal physical properties to observable surface images through finite element discretization.

[0028] Specifically, it includes: Three-dimensional forward heat conduction model: The three-dimensional unsteady heat conduction equation is: ; in, For material density, This refers to the specific heat capacity at constant pressure; both are considered constant when the temperature change is small. It is a three-dimensional thermal conductivity distribution; For internal heat source items; For temperature; The surface boundary adopts a combined convection-radiation boundary condition: ,in, The convective heat transfer coefficient; Surface emissivity (ε = 0.1-0.3 for metal surfaces, which can be obtained through blackbody calibration); It is the Stefan-Boltzmann constant; The ambient temperature; The internal boundary adopts a defect-matrix interface and a contact thermal resistance model. Contact thermal resistance Set according to the defect type; The thermal conductivity of the matrix material surrounding the defect; Thermal conductivity of defective regions (such as cracks and pores); The partial derivative of temperature with respect to the interface normal ( It refers to the normal direction of the interface, which usually points from the substrate to the defect. The temperature at the interface of the matrix; The temperature of the defect at the interface; At the initial moment ( When the temperature of the entire sample is uniformly distributed, the temperature is as follows: Equal to the uniform temperature of the sample at the initial moment ,Right now ; The three-dimensional model of the sample (based on CAD file import) was discretized using a tetrahedral unstructured mesh, and the mesh size was adaptively adjusted according to the defect size. The heat conduction equation is discretized into a system of linear equations using the Galerkin weighted residual method: The mass matrix is ​​as follows: Stiffness matrix: Load vector: ; For the shape function of the finite element method; For; for the integration region (i.e., the volume of the sample being analyzed); The thermal conductivity of the material; The gradient of the shape function; The integration region (i.e., the surface area of ​​the sample, corresponding to the boundary); The convective heat transfer coefficient (describes the intensity of convective heat transfer between the sample surface and the environment). The emissivity of the sample surface; It is the Stefan-Boltzmann constant; The temperature of the sample surface at the current moment; The ambient temperature; The intensity of the internal heat source inside the sample; The implicit Euler method is used to solve the transient equations, and the time step is adaptively adjusted to ensure numerical stability. By inputting the hypothetical three-dimensional thermal conductivity distribution Solve the discrete equations to output the temperature values ​​of each pixel position on the sample surface at different times, forming a simulated temperature sequence consistent with the measured data format. .

[0029] It also includes the photoelastic stress-light intensity positive model: Based on the photoelastic effect of linear elastic isotropic materials, polarized light experiences a phase delay after passing through a stressed sample, and the relationship between this delay and stress follows the stress-optics law. ,in The difference in refractive index is due to birefringence. The material stress optical coefficient, , These are the first and second principal stresses at that point; The incident light is linearly polarized and experiences a phase delay after passing through the sample. , For sample thickness; The wavelength of the incident light; Given the birefringence refractive index difference (the difference in refractive index between the two perpendicularly polarized components of polarized light after passing through a stressed sample), the intensity of the emitted light after passing through the analyzer is: ; in, The intensity of the incident polarized light; The polarization direction of the analyzer; The polarization direction of the incident polarized light; This is the phase delay (phase difference between two perpendicular polarization components) of polarized light after it passes through the sample. For phase shift; When the polarization direction of the incident light is orthogonal to the polarization direction of the analyzer ( When ), it is simplified to: ; in, The first principal stress at a point within the sample; This represents the second principal stress at a point within the sample. The wavelength of the incident light; The material stress optical coefficient; For sample thickness; After phase delay calculation and stress field coupling: A four-step phase-shifting method was used to acquire four light intensity images at analyzer angles of 0°, 35°, 90°, and 135°. , , , Solve for phase delay : ; For plane stress problems (where the sample thickness is much smaller than its length and width dimensions), the principal stress difference... Combining the equilibrium equation, , The complete stress tensor field is solved using the finite element method, along with the geometric equations. ;in, For stress; Shear stress; Input assumptions about the three-dimensional stress field distribution By combining material parameters and polarization parameters, simulated light intensity images at various polarization angles are calculated. The image resolution is consistent with the measured data (1920×1080 pixels), and the light intensity value quantization range is 0-65535 (16-bit grayscale).

[0030] The inversion solution module based on the physical model is configured to minimize the residual between measured and simulated data. It combines regularization and L-BFGS iterative optimization algorithm to inversely deduce the three-dimensional thermal conductivity field and stress field inside the material, thus solving the ill-posedness of the inverse problem. Specifically, it includes: Through observation data (Measured temperature sequence and polarized light intensity image) to infer unknown parameters (Thermal conductivity field and stress field) satisfy the positive model ,in, For positive mapping operators, To observe noise; Using the weighted least squares criterion, combined with regularization to suppress ill-posedness, the objective function is: ,in, The weighting matrix is ​​a diagonal matrix, where each element is the reciprocal of the noise variance at each observation point. (Noise variance is calculated by repeatedly collecting measured data). The regularization parameter is 0 < α < 1, adaptively selected using the L-curve method to balance the weights of the data fitting term and the regularization term. For regularization terms, the appropriate type is selected for different physical properties: total variation regularization is used for thermal conductivity inversion; Tikhonov regularization is used for stress field inversion. Constraints: Physical constraints: thermal conductivity k > 0 (material thermal conductivity is positive), stress value σ satisfies the material yield strength constraint; Spatial constraints: The physical property field has spatial continuity, and the parameter difference between adjacent grid cells is less than or equal to the threshold.

[0031] The adjoint state method is used to efficiently calculate the gradient of the objective function with respect to unknown parameters. To avoid the high computational complexity of direct differentiation: Solving the forward model To obtain simulation data Calculate the residual ; adjoint equations based on the forward model , Given the Jacobian matrix transpose of the forward model, solve for the adjoint variables. ; Gradient calculation: , Let be the Jacobian matrix of the positive model, where (Obtained directly from the adjoint equation), without the need for explicit calculation of the Jacobian matrix; Through iterative optimization, the three-dimensional physical parameter field and thermal conductivity field are output. and stress tensor field The parameter field in the finite element mesh format is converted into a volume data format for easy reading by the subsequent visualization module; the 3σ criterion is used to remove isolated outliers in the parameter field.

[0032] Iteration process: Initialize the parameter fields: the thermal conductivity field is initialized with a uniform distribution, and the stress field is initialized with a zero-stress distribution; Forward simulation: Simulate the current parameter field Input a forward model and calculate the simulation data. ; Residual calculation: Calculate the residuals objective function value ; Convergence criterion: If <convergence threshold or residual norm ( If the expression is equal to 0.01, then the iteration converges, and the output is... Otherwise, proceed to the next step; Gradient calculation: calculated using the adjoint state method ; Step size selection: The Armijo criterion (Wolfe-Powell condition) is used to adaptively select the step size. This ensures that the objective function is monotonically decreasing; Parameter Update: Update the parameter field using the L-BFGS algorithm ; Constraint projection: updating the parameter field Projecting onto the constrained feasible region, we obtain ; Repeat the steps between the forward simulation constraint projections until convergence.

[0033] The physical parameter visualization and quantitative evaluation module is configured to visualize the physical parameter field obtained by inversion, extract the defect geometry and physical quantitative indicators, and obtain the defect evaluation coefficient after comprehensive processing, and match the corresponding appearance defect level based on the defect evaluation coefficient. Specifically, it includes: The three-dimensional physical parameter field obtained by inversion is transformed into intuitive visualization results, quantitative characteristic parameters of defects are extracted, rating rules are established based on physical failure mechanisms, and a structured report containing defect location, size, severity and handling suggestions is output, providing direct decision-making basis for quality control.

[0034] This module is the system's output window, and its core function is to transform abstract physical parameters into engineering-based quality assessment conclusions.

[0035] Based on the ParaView open-source visualization engine (which supports real-time rendering of large-scale volumetric data), a custom visualization interface was developed, supporting multi-view, multi-mode interactive viewing; and Displayed in volume rendering format; Determine using an adaptive threshold segmentation algorithm Defect threshold ( Defect area (Value below 50% of the matrix); identify all that meet the requirements. < The mesh cells are marked as defect regions; Thermal and mechanical defects in the defect region were extracted separately, and thermal and mechanical anomaly values ​​were obtained after analysis. After normalizing the thermal and mechanical anomalies, weighting factors for the thermal and mechanical anomalies are preset respectively. The thermal and mechanical anomalies are then multiplied by their corresponding weighting factors and summed to obtain the defect evaluation coefficient.

[0036] The process of obtaining thermal anomalies includes: Obtain the defect volume corresponding to each defect region, extract the maximum defect volume, and preset the defect volume threshold. Divide the maximum defect volume by the defect volume threshold and subtract 1. If the resulting value is greater than 0, it is recorded as the defect degree. Obtain the thermal conductivity corresponding to each defect region, and take the average value to obtain the average thermal conductivity. Determine the thermal conductivity of the matrix material, and calculate the difference between the average thermal conductivity and the thermal conductivity of the matrix material (which is a positive number), which is recorded as the thermal conductivity difference. Preset the maximum allowable value for thermal conductivity decrease, and divide the thermal conductivity difference by the maximum allowable value for thermal conductivity decrease to obtain the anomaly degree. After normalizing the defect degree and anomaly degree, a weighted summation is performed to obtain the thermal anomaly value; Preset weighting factors for defect degree and anomaly degree, multiply the defect degree and anomaly degree with their corresponding weighting factors respectively, and sum them to obtain the thermal anomaly value.

[0037] It achieves precise quantification of thermal defects, breaking through the limitations of traditional thermal testing, which can only qualitatively determine the existence of defects and is difficult to quantify the severity of defects.

[0038] The defect degree is obtained by extracting the maximum defect volume and comparing it with a preset threshold. The anomaly degree is calculated by combining the difference between the average thermal conductivity and the matrix thermal conductivity. Then, the thermal anomaly value is obtained by normalized weighted summation. This not only quantifies the geometric scale (volume) of the defect, but also characterizes the degree of damage to the thermal conductivity of the material (decrease in thermal conductivity). This provides an accurate and traceable thermal quantitative basis for subsequent comprehensive defect rating.

[0039] By presetting defect volume thresholds, maximum allowable decrease in thermal conductivity, and weighting factors for the two types of indicators, this system can adapt to the testing needs of metal oxide resistors of different specifications and application scenarios, avoiding the limitations of single-threshold judgment. Its quantitative logic aligns with the thermal failure mechanism of resistors, and thermal anomalies directly reflect the impact of defects on the heat dissipation performance of the resistor, providing objective and quantifiable decision support for quality control. This effectively reduces the subjective error of manual assessment and improves the consistency and reliability of defect detection.

[0040] The process of obtaining mechanical anomalies includes: Stress field obtained from inversion In the middle, identify all that meet the requirements > The mesh elements representing the material's yield strength are marked as failure elements; the sum of the surface areas of all failure elements is obtained to get the total failure area; the maximum allowable failure area is preset, and the total failure area is divided by the maximum allowable failure area to obtain the area anisotropy; Obtain the stress tensor of each mesh element, calculate its principal stress, and iterate through the principal stresses of all elements to find the maximum principal stress. Using the material yield strength as a threshold, divide the maximum principal stress by the material yield strength to obtain the stress anisotropy. After normalizing the area anisotropy and stress anisotropy, a weighted summation is performed to obtain the mechanical anomaly value. Preset weighting factors for area anisotropy and stress anisotropy, multiply area anisotropy and stress anisotropy by their corresponding weighting factors respectively, and sum them to obtain mechanical anomaly values.

[0041] This method achieves multi-dimensional and precise quantification of mechanical defects, overcoming the limitations of traditional mechanical testing that relies solely on a single stress threshold to determine failure. By identifying failure units where stress exceeds yield strength, calculating the ratio of the total failure area to the maximum allowable failure area (area anisotropy), and combining this with the ratio of the maximum principal stress to the material's yield strength (stress anisotropy), it characterizes the destructive impact of defects on the mechanical properties of resistor sheets from two key dimensions: the failure range and the degree of stress exceeding limits.

[0042] The mechanical anomaly values ​​obtained by normalized weighted summation not only conform to the essential laws of mechanical failure of resistor sheets, but also realize the upgrade of mechanical defects from qualitative judgment to quantitative assessment, providing an objective and accurate mechanical quantitative basis for comprehensive defect rating.

[0043] By presetting adjustable parameters such as the maximum permissible failure area and weighting factors, the mechanical testing requirements of metal oxide resistors of different specifications and application scenarios can be flexibly matched, avoiding the limitations of fixed standards. Its quantitative results can directly reflect the potential impact of defects on the structural stability and service life of the resistor, effectively reducing the subjective error of manual assessment. This provides traceable and comparable decision support for quality control and risk prediction, significantly improving the consistency and reliability of mechanical defect detection.

[0044] Example 2

[0045] Please see Figure 2 A method for detecting appearance defects in metal oxide resistors, comprising the following parts: By controlling thermal excitation, applying polarized light field and synchronous control of FPGA, infrared temperature sequence and polarized light intensity image corresponding to physical excitation are acquired to obtain high-quality observation data. A forward mathematical and physical model of heat conduction and photoelastic stress-light intensity is established, and accurate simulation and prediction of internal physical properties to observable surface images are achieved through finite element discretization. With the goal of minimizing the residual between measured and simulated data, the three-dimensional thermal conductivity field and stress field inside the material are inversely calculated by combining regularization and L-BFGS iterative optimization algorithm. The physical parameter field obtained by inversion is visualized, and the geometric and physical quantitative indicators of defects are extracted. The defect evaluation coefficient is calculated through thermal and mechanical anomalies. The corresponding appearance defect level is matched based on the defect evaluation coefficient.

[0046] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0047] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

[0048] It should be noted that, in this document, the use of relational terms such as "first" and "second" is merely for distinguishing one entity or operation from another, and does not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the element.

[0049] It should be understood that in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0050] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0051] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0052] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0053] In addition, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0054] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0055] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A system for detecting appearance defects in metal oxide resistors, characterized in that, include: The physical excitation and synchronous image acquisition module is configured to acquire infrared temperature sequences and polarized light intensity images corresponding to the physical excitation by applying controllable thermal excitation and polarized light field. The physics-image coupling model construction module is configured to establish a forward mathematical physics model of heat conduction and photoelastic stress-light intensity, and to perform simulation and prediction of internal physical properties to observable surface images through finite element discretization. The inversion solution module based on the physical model is configured to minimize the residual between measured and simulated data to inversely deduce the three-dimensional thermal conductivity field and stress field inside the material. The physical parameter visualization and quantitative evaluation module is configured to visualize the physical parameter field obtained by inversion, extract the geometric and physical quantitative indicators of defects, and obtain the defect evaluation coefficient after comprehensive processing, and match the corresponding appearance defect level based on the defect evaluation coefficient.

2. The metal oxide resistor sheet appearance defect detection system according to claim 1, characterized in that, The physical excitation and synchronous image acquisition module specifically includes: Controllable thermal excitation unit: uses a flash lamp array; through pulse width control, it ensures that the excitation energy can penetrate to the subsurface of the material; Polarization field application unit: consists of an electric rotary stage, equipped with linear polarizers and waveplates; three sets of standard polarization angle combinations are preset to obtain enough independent equations to solve the three independent components of the stress tensor; A multi-sensor synchronous image acquisition unit, including an infrared thermal imager and a polarization imaging camera; After the host computer sends the start command, the FPGA first triggers the thermal excitation unit to output a thermal pulse, and after a preset delay, it triggers the infrared camera to acquire data. The polarization camera triggers acquisition after the thermal excitation ends, and each frame of image is timestamped, corresponding one-to-one with the excitation signal timestamp.

3. The metal oxide resistor sheet appearance defect detection system according to claim 1, characterized in that, The physical-image coupling model construction module specifically includes: The three-dimensional unsteady heat conduction equation is: ; in, For material density, Specific heat capacity at constant pressure; It is a three-dimensional thermal conductivity distribution; For internal heat source items; For temperature; The surface boundary adopts a combined convection-radiation boundary condition; The internal boundary adopts a defect-matrix interface model using contact thermal resistance. At the initial moment, the temperature of the entire sample is uniformly distributed as follows: Equal to the uniform temperature of the sample at the initial moment ; The three-dimensional model of the sample was discretized using a tetrahedral unstructured mesh; The heat conduction equation was discretized into a system of linear equations using the Galerkin weighted residual method. The implicit Euler method is used to solve the transient equations. By inputting the hypothetical three-dimensional thermal conductivity distribution Solve the discrete equations to output the temperature values ​​of each pixel position on the sample surface at different times, forming a simulated temperature sequence consistent with the measured data format. .

4. The metal oxide resistor sheet appearance defect detection system according to claim 3, characterized in that, It also includes the photoelastic stress-light intensity positive model: The incident light is linearly polarized. After passing through the sample, a phase delay occurs. The intensity of the emitted light after passing through the analyzer is: ； in, The intensity of the incident polarized light; The polarization direction of the analyzer; The polarization direction of the incident polarized light; The phase delay of polarized light after passing through the sample; For phase shift; When the polarization direction of the incident light is orthogonal to the polarization direction of the analyzer, it simplifies to: ； in, This is the first principal stress of the sample; This is the second principal stress of the sample; The wavelength of the incident light; The material stress optical coefficient; For sample thickness; After phase delay calculation and stress field coupling; Input assumptions about the three-dimensional stress field distribution By combining material parameters and polarization parameters, simulated light intensity images at various polarization angles are calculated. .

5. The metal oxide resistor sheet appearance defect detection system according to claim 1, characterized in that, The inversion solution module based on the physical model specifically includes: Through observation data Inversely deduce unknown parameters Satisfying the positive model ,in, For positive mapping operators, To observe noise; The objective function is: ,in, It is a weighted matrix; For regularization parameters; This is a regularization term.

6. The metal oxide resistor sheet appearance defect detection system according to claim 5, characterized in that, Also includes: Solving the forward model To obtain simulation data Calculate the residual ; adjoint equations based on the forward model , Given the Jacobian matrix transpose of the forward model, solve for the adjoint variables. ; Gradient calculation: , Let be the Jacobian matrix of the positive model, where ; Through iterative optimization, the three-dimensional physical parameter field and thermal conductivity field are output. and stress tensor field .

7. The metal oxide resistor sheet appearance defect detection system according to claim 1, characterized in that, The physical parameter visualization and quantitative evaluation module specifically includes: Develop custom visual interfaces; and Displayed in volume rendering format; Determine using an adaptive threshold segmentation algorithm Defect threshold ; Identify all that satisfy < The mesh cells are marked as defect regions; Thermal and mechanical defects in the defect region were extracted separately, and thermal and mechanical anomaly values ​​were obtained after analysis. After normalizing the thermal and mechanical anomalies, weighting factors for the thermal and mechanical anomalies are preset respectively. The thermal and mechanical anomalies are then multiplied by their corresponding weighting factors and summed to obtain the defect evaluation coefficient.

8. The metal oxide resistor sheet appearance defect detection system according to claim 7, characterized in that, The process of obtaining thermal anomalies includes: Obtain the defect volume corresponding to each defect region, extract the maximum defect volume, and preset the defect volume threshold. Divide the maximum defect volume by the defect volume threshold and subtract 1. If the resulting value is greater than 0, it is recorded as the defect degree. Obtain the thermal conductivity corresponding to each defect region, and take the average value to obtain the average thermal conductivity. Determine the thermal conductivity of the matrix material, and calculate the difference between the average thermal conductivity and the thermal conductivity of the matrix material, which is recorded as the thermal conductivity difference. Preset the maximum allowable value for thermal conductivity decrease, and divide the thermal conductivity difference by the maximum allowable value for thermal conductivity decrease to obtain the anomaly degree. After normalizing the defect degree and anomaly degree, a weighted summation is performed to obtain the thermal anomaly value.

9. The metal oxide resistor sheet appearance defect detection system according to claim 7, characterized in that, The process of obtaining mechanical anomalies includes: Stress field obtained from inversion In the middle, identify all that meet the requirements > The mesh elements representing the material's yield strength are marked as failure elements; the sum of the surface areas of all failure elements is obtained to get the total failure area; the maximum allowable failure area is preset, and the total failure area is divided by the maximum allowable failure area to obtain the area anisotropy; Obtain the stress tensor of each mesh element, calculate its principal stress, and iterate through the principal stresses of all elements to find the maximum principal stress. Using the material yield strength as a threshold, divide the maximum principal stress by the material yield strength to obtain the stress anisotropy. After normalizing the area anisotropy and stress anisotropy, a weighted summation is performed to obtain the mechanical anomaly value.

10. A method for detecting appearance defects in metal oxide resistors, comprising a metal oxide resistor appearance defect detection system according to any one of claims 1-9, characterized in that, include: By applying synchronous control through controllable thermal excitation and polarization light field, infrared temperature sequences and polarization light intensity images corresponding to physical excitation are acquired. A forward mathematical-physical model of heat conduction and photoelastic stress-light intensity is established, and the internal physical properties are simulated and predicted to the observable images of the surface through finite element discretization. With the goal of minimizing the residual between measured and simulated data, the three-dimensional thermal conductivity field and stress field inside the material are inversely derived. Visualize the physical parameter field obtained from the inversion, extract the geometric and physical quantitative indicators of the defects, and calculate the defect evaluation coefficient through thermal and mechanical anomalies. The corresponding appearance defect level is matched based on the defect assessment coefficient.

Images