LED annular light source testing method, device and equipment and storage medium

Abstract

The invention provides an LED annular light source test method, device and equipment and a storage medium, and the method comprises the steps: carrying out the calibration initialization processing of an LED annular light source disposed on an angle-adjustable support and a standard silicon wafer simulation board, and obtaining a benchmark test sequence; an LED annular light source is controlled to irradiate the standard silicon wafer simulation board according to the benchmark test sequence, multi-area light field collection is carried out under different speed gears, and multi-dimensional test data are obtained; performing characteristic analysis on the multi-dimensional test data to obtain light field space characteristic data, light intensity fluctuation frequency characteristic data and a defect detection rate evaluation result, and further constructing a light field distortion model; and calculating a light intensity uniformity deviation value and a defect detection capability score according to the light field distortion model, and generating a test evaluation report. According to the method, a light field distortion model is constructed through multi-region acquisition and data analysis, and the uniformity of light intensity distribution and the defect detection capability can be accurately evaluated. And the effect of the LED annular light source in the high-precision detection of the silicon wafer is improved.

Description

Technical Field

[0001] The present invention relates to the field of semiconductor technology, and in particular to a method, device, equipment and storage medium for testing an LED ring light source. Background Art

[0002] In the field of semiconductor manufacturing, LED ring light sources are widely used as detection light source equipment in silicon wafer defect detection and quality assessment. LED ring light sources are gradually replacing traditional light source equipment due to their high efficiency, long life, energy saving and environmental protection. However, in practical applications, the uniformity of light intensity distribution and defect detection ability of LED ring light sources are key factors affecting detection accuracy and reliability.

[0003] In the prior art, simple light intensity measurement methods and empirical analysis methods are usually used to evaluate the performance of LED ring light sources. These methods usually only focus on light intensity data in a single dimension. At the same time, the prior art often lacks a systematic multi-region acquisition and analysis mechanism when evaluating light intensity distribution and defect detection capabilities, resulting in inaccurate evaluation of the degree of distortion of light intensity distribution. Especially in the semiconductor field, high-precision inspection of silicon wafers has higher requirements for the uniformity and stability of the light source. The problems of uneven light intensity and insufficient defect detection capabilities in the prior art are particularly prominent. This defect limits the effectiveness and reliability of LED ring light sources in high-precision inspection applications in the semiconductor field. Summary of the invention

[0004] The main purpose of the present invention is to solve the technical problem that the existing LED ring light source test lacks a systematic multi-region collection and analysis mechanism, resulting in inaccurate evaluation of the degree of distortion of the light intensity distribution; A first aspect of the present invention provides a method for testing an LED ring light source, the method comprising: Performing calibration initialization processing on an LED ring light source mounted on an angle-adjustable bracket and a standard silicon wafer simulation board, and obtaining a benchmark test sequence of the LED ring light source; Controlling the LED ring light source to illuminate the standard silicon wafer simulation board according to the benchmark test sequence, and collecting the light field illuminated by the LED ring light source on the standard silicon wafer simulation board in multiple regions at different speed gears to obtain multi-dimensional test data; Performing characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data, and defect detection rate evaluation results, and constructing a light field distortion model that characterizes the degree of light intensity distribution distortion based on the light field spatial characteristic data, light intensity fluctuation frequency characteristic data, and defect detection rate evaluation results; The light intensity uniformity deviation value and the defect detection capability score of the LED ring light source are calculated according to the light field distortion model, and a test evaluation report of the LED ring light source is generated according to the light intensity uniformity deviation value and the defect detection capability score.

[0005] Optionally, in a first implementation of the first aspect of the present invention, the calibration initialization processing of the LED ring light source mounted on the angle-adjustable bracket and the standard silicon wafer simulation board and obtaining a benchmark test sequence of the LED ring light source includes: Measuring the positional relationship between the LED ring light source and the standard silicon wafer simulation board to obtain position parameter data, and setting the motion parameters of the standard silicon wafer simulation board according to the position parameter data; According to the motion parameters, the lighting parameters of the LED ring light source are configured to obtain a parameter combination of lighting angle, working distance and light intensity level; The motion parameters and the parameter combinations are subjected to timing matching processing to obtain a motion-lighting linkage control scheme, and a benchmark test sequence is generated according to the motion-lighting linkage control scheme.

[0006] Optionally, in a second implementation of the first aspect of the present invention, the LED ring light source is controlled to irradiate the standard silicon wafer simulation board according to the benchmark test sequence, and the light field irradiated by the LED ring light source on the standard silicon wafer simulation board is collected in multiple areas at different speed gears to obtain multi-dimensional test data including: According to the benchmark test sequence, determining the light source illumination control instruction of the LED ring light source, and performing speed gear switching control on the standard silicon wafer simulation board; Under the light source illumination control instruction and different speed gears of the standard silicon wafer simulation board, light field collection is performed on the inner ring, middle ring and outer ring of the LED ring light source through a photodetector array to obtain light intensity distribution data; The standard defects on the surface of the standard silicon wafer simulation board are imaged and processed by an industrial camera to obtain defect detection image data, and the spectrum of the inner ring, middle ring and outer ring regions is collected and processed by a spectrum analyzer according to the light intensity distribution data to obtain spectrum characteristic data; The light intensity distribution data, defect detection image data and spectral characteristic data are subjected to data integration processing to obtain multi-dimensional test data.

[0007] Optionally, in a third implementation of the first aspect of the present invention, the performing characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results includes: Performing Fourier transform processing on the light intensity distribution data to obtain a light intensity distribution spectrum feature vector, and discretizing and sampling the light intensity fluctuation frequency domain distribution of the LED ring light source according to the light intensity distribution spectrum feature vector to obtain light intensity fluctuation frequency characteristic data; Performing spatial mapping processing on the light intensity distribution of the inner ring, the middle ring and the outer ring according to the spectral characteristic data to obtain a light field spatial distribution characteristic matrix, and performing singular value decomposition operation on the light field spatial distribution characteristic matrix to obtain light field spatial characteristic data; The defect detection image data is subjected to edge detection and morphological processing to obtain a defect feature vector, and the standard defects are identified and classified according to the defect feature vector to obtain a defect detection rate evaluation result.

[0008] Optionally, in a fourth implementation manner of the first aspect of the present invention, the light intensity distribution of the inner ring, the middle ring and the outer ring is spatially mapped according to the spectral characteristic data to obtain a light field spatial distribution feature matrix, and a singular value decomposition operation is performed on the light field spatial distribution feature matrix to obtain the light field spatial characteristic data, including: Performing polar coordinate transformation processing on the spectral characteristic data to obtain angular light intensity distribution data, and evenly dividing the inner ring, middle ring and outer ring regions into three regions according to the angular light intensity distribution data to obtain annular region division parameters; Performing interpolation fitting processing on the angular light intensity distribution data according to the annular area division parameters to obtain a spatial sampling point sequence, and performing radial normalization processing on the spatial sampling point sequence to obtain standardized light intensity distribution data; Performing a matrix reconstruction operation on the standardized light intensity distribution data to obtain a light field spatial distribution characteristic matrix, and performing a covariance calculation process on the light field spatial distribution characteristic matrix to obtain a spatial correlation matrix; Performing eigenvalue calculation processing according to the spatial correlation matrix to obtain a singular value sequence, and performing a descending order operation on the singular value sequence to obtain a main eigencomponent; The main characteristic components are spatially reprojected to obtain reconstructed light field distribution data, and characteristic vector orthogonalization is performed on the reconstructed light field distribution data to obtain light field spatial characteristic data.

[0009] Optionally, in a fifth implementation of the first aspect of the present invention, constructing a light field distortion model characterizing the degree of light intensity distribution distortion based on the light field spatial characteristic data, the light intensity fluctuation frequency characteristic data, and the defect detection rate evaluation result includes: Standardizing the light field spatial characteristic data, the light intensity fluctuation frequency characteristic data and the defect detection rate evaluation result to obtain a normalized feature vector; Constructing a three-dimensional feature space according to the normalized feature vector, and performing a principal component analysis operation on the three-dimensional feature space to obtain a principal feature component; Performing feature weight calculation processing on the main feature components to obtain contribution coefficients of features in each dimension, and performing linear combination operations on the main feature components according to the contribution coefficients to obtain light field distortion feature parameters; A nonlinear regression equation is established according to the light field distortion characteristic parameters, and the coefficients of the nonlinear regression equation are iteratively solved by the least square method to obtain a light field distortion model that characterizes the degree of light intensity distribution distortion.

[0010] Optionally, in a sixth implementation of the first aspect of the present invention, calculating the light intensity uniformity deviation value and the defect detection capability score of the LED ring light source according to the light field distortion model, and generating a test evaluation report of the LED ring light source according to the light intensity uniformity deviation value and the defect detection capability score includes: Performing numerical calculation processing on the light field distortion model to obtain light intensity distortion coefficients of each region, and performing statistical distribution calculation based on the light intensity distortion coefficients to obtain a light intensity uniformity deviation curve; According to the light intensity uniformity deviation curve, weighted accumulation processing is performed on the light intensity distortion coefficients of each region to obtain a comprehensive distortion index, and normalization operation is performed on the comprehensive distortion index to obtain a light intensity uniformity deviation value; Performing classified statistical processing on the defect detection rate evaluation results to obtain the detection probability distribution of each type of defect, and performing error analysis calculation based on the detection probability distribution to obtain detection confidence data; Performing weight distribution processing on the detection probability distribution of the various defects according to the detection confidence data to obtain a detection capability weight matrix, and performing feature fusion operation on the detection capability weight matrix to obtain a defect detection capability score; The light intensity uniformity deviation value and the defect detection capability score are subjected to multi-dimensional indicator integration processing to obtain a comprehensive performance evaluation indicator, and a test evaluation report is generated based on the comprehensive performance evaluation indicator.

[0011] A second aspect of the present invention provides a LED ring light source testing device, the LED ring light source testing device comprising: A preprocessing module, used for performing calibration initialization processing on the LED ring light source mounted on the angle-adjustable bracket and the standard silicon wafer simulation board, and obtaining a benchmark test sequence of the LED ring light source; A test module, used to control the LED ring light source to irradiate the standard silicon wafer simulation board according to the benchmark test sequence, and to collect multi-area light fields irradiated by the LED ring light source on the standard silicon wafer simulation board at different speed gears to obtain multi-dimensional test data; An analysis module is used to perform characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results, and to construct a light field distortion model that characterizes the degree of light intensity distribution distortion based on the light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results; An evaluation module is used to calculate the light intensity uniformity deviation value and defect detection capability score of the LED ring light source according to the light field distortion model, and generate a test evaluation report of the LED ring light source according to the light intensity uniformity deviation value and defect detection capability score.

[0012] A third aspect of the present invention provides an LED ring light source testing device, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via a line; the at least one processor calls the instructions in the memory so that the LED ring light source testing device performs the steps of the above-mentioned LED ring light source testing method.

[0013] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, which, when executed on a computer, enable the computer to execute the steps of the above-mentioned LED ring light source testing method.

[0014] The above-mentioned LED ring light source test method, device, equipment and storage medium obtain a benchmark test sequence by calibrating and initializing the LED ring light source and the standard silicon wafer simulation board installed on the angle-adjustable bracket; according to the benchmark test sequence, the LED ring light source is controlled to illuminate the standard silicon wafer simulation board, and multi-region light field acquisition is performed at different speed gears to obtain multi-dimensional test data; the multi-dimensional test data is analyzed for characteristics to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results, and then a light field distortion model is constructed; according to the light field distortion model, the light intensity uniformity deviation value and the defect detection capability score are calculated, and a test evaluation report is generated. This method constructs a light field distortion model through multi-region acquisition and data analysis, and can accurately evaluate the uniformity of light intensity distribution and defect detection capability. Improve the effect of LED ring light sources in high-precision detection of silicon wafers.

[0015] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the description, claims and drawings.

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 This is a schematic diagram of a first embodiment of a method for testing an LED ring light source in an embodiment of the present invention; Figure 2 A schematic diagram of an embodiment of a LED ring light source testing device in an embodiment of the present invention; Figure 3 It is a schematic diagram of an embodiment of an LED ring light source testing device in an embodiment of the present invention. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0019] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device end including a series of steps or units is not limited to the listed steps or units, but may optionally include other steps or units that are not listed, or may optionally include other steps or units that are inherent to these processes, methods, products or device ends.

[0020] To facilitate understanding of this embodiment, a method for testing an LED ring light source disclosed in an embodiment of the present invention is first described in detail. Figure 1 As shown, the method comprises the following steps: 101. Perform calibration initialization processing on the LED ring light source and the standard silicon wafer simulation board installed on the angle-adjustable bracket, and obtain a benchmark test sequence of the LED ring light source; In one embodiment of the present invention, the calibration initialization processing of the LED ring light source and the standard silicon wafer simulation board installed on the angle-adjustable bracket and obtaining the benchmark test sequence of the LED ring light source includes: measuring the position relationship between the LED ring light source and the standard silicon wafer simulation board to obtain position parameter data, and setting the motion parameters of the standard silicon wafer simulation board according to the position parameter data; configuring the lighting parameters of the LED ring light source according to the motion parameters to obtain a parameter combination of lighting angle, working distance and light intensity level; performing timing matching processing on the motion parameters and the parameter combination to obtain a motion-lighting linkage control scheme, and generating a benchmark test sequence according to the motion-lighting linkage control scheme.

[0021] Specifically, since the LED ring light source has a special ring structure feature, the measurement process requires the use of a laser rangefinder to perform three-dimensional coordinate positioning of the feature points of the inner ring, middle ring and outer ring of the ring light source. The laser rangefinder is installed on a three-axis slide, and the rangefinder is driven by a stepper motor to move in the three directions of XYZ to perform scanning measurement on multiple feature points of the LED ring light source. The measurement data is collected in real time by the data acquisition module and transmitted to the control system, and the spatial position data of the LED ring light source is obtained after coordinate transformation processing. At the same time, multiple reference marking points are arranged on the surface of the standard silicon wafer simulation board, and these marking points are imaged by an industrial camera. The position coordinates of the marking points are extracted through the image processing algorithm to establish the spatial coordinate system of the simulation board surface. The coordinate systems of the LED ring light source and the standard silicon wafer simulation board are unified through spatial coordinate transformation to obtain the relative position relationship parameters between the two, including position parameter data such as relative distance, tilt angle and horizontal offset. Based on the position parameter data, the kinematic inverse solution algorithm is used to calculate the translation velocity, acceleration and lifting parameters of the standard silicon wafer simulation board in the XY plane, thereby setting the motion parameters of the standard silicon wafer simulation board.

[0022] After obtaining the motion parameters, the lighting parameters of the LED ring light source need to be configured. First, based on the relative position information in the motion parameters, the actual irradiation distance from each light-emitting unit of the LED ring light source to the surface of the standard silicon wafer simulation board is calculated in combination with the light intensity attenuation model, and the working distance parameter is determined according to the irradiation distance. Then, the pitch angle and horizontal rotation angle of the LED ring light source are controlled by the angle adjustment mechanism so that the illumination area can accurately cover the detection area on the surface of the simulation board, thereby obtaining the optimal illumination angle. At the same time, considering the illumination uniformity requirements of different areas, the driving current of the inner ring, middle ring and outer ring LED lamp beads is controlled respectively by the current modulation circuit, the luminous intensity of each area is adjusted, and a light intensity grading control strategy is established, and finally a lighting parameter combination including the illumination angle, working distance and light intensity level is obtained. These parameter combinations are uniformly managed through the parameter configuration module of the control system to ensure that the LED ring light source can maintain stable lighting conditions during the detection process.

[0023] When performing timing matching processing on the combination of motion parameters and lighting parameters, a parameter synchronization mechanism based on the time axis is first established to discretize the motion trajectory of the standard silicon wafer simulation board into key points on the time series, and each key point corresponds to a set of position coordinates and speed parameters. Then the lighting parameters of the LED ring light source are divided according to the same time series to ensure that the motion state and the lighting state can accurately correspond at each time point. A unified clock signal is generated by the timing controller to drive the motion control module and the lighting control module to operate in coordination according to the predetermined timing to form a complete motion-lighting linkage control scheme. Based on this control scheme, the change process of the motion parameters and lighting parameters is further converted into an instruction sequence executable by the device, including motion control instructions, light intensity adjustment instructions, and angle adjustment instructions. These instructions are arranged in chronological order to form a benchmark test sequence for guiding the subsequent test process. Through this timing-based parameter matching method, it is ensured that the lighting characteristics of the LED ring light source and the motion characteristics of the standard silicon wafer simulation board can be accurately synchronized.

[0024] 102. According to the benchmark test sequence, control the LED ring light source to illuminate the standard silicon wafer simulation board, and collect the light field of the LED ring light source irradiating the standard silicon wafer simulation board in multiple regions at different speed gears to obtain multi-dimensional test data; In one embodiment of the present invention, the LED ring light source is controlled to illuminate the standard silicon wafer simulation board according to the benchmark test sequence, and the light field of the LED ring light source irradiated on the standard silicon wafer simulation board is collected in multiple areas at different speed gears to obtain multi-dimensional test data, including: according to the benchmark test sequence, the light source lighting control instruction of the LED ring light source is determined, and the speed gear switching control of the standard silicon wafer simulation board is performed; under the light source lighting control instruction and different speed gears of the standard silicon wafer simulation board, the light field of the inner ring, middle ring and outer ring of the LED ring light source is collected by a photodetector array to obtain light intensity distribution data; the standard defects on the surface of the standard silicon wafer simulation board are imaged and processed by an industrial camera to obtain defect detection image data, and the spectrum of the inner ring, middle ring and outer ring is collected and processed by a spectrum analyzer according to the light intensity distribution data to obtain spectrum characteristic data; the light intensity distribution data, defect detection image data and spectrum characteristic data are integrated and processed to obtain multi-dimensional test data.

[0025] Specifically, the benchmark test sequence contains complete timing information of the LED ring light source lighting parameters and the standard silicon wafer simulation board motion parameters. The control system first parses the benchmark test sequence and extracts the lighting control instructions corresponding to each time point. These instructions include the driving current value, lighting angle adjustment value and working distance setting value of the LED lamp beads in the inner ring, middle ring and outer ring. The control system converts these parameters into specific PWM waveforms and digital control signals through the digital signal processor, and accurately controls the luminous intensity and irradiation angle of the LED ring light source through the drive circuit. At the same time, the motion control parameters are extracted from the benchmark test sequence, and the motion speed of the standard silicon wafer simulation board is graded and set through the motion controller, and a speed gear system including low speed, medium speed and high speed is established, and each gear corresponds to a specific linear speed and acceleration parameters. The motor speed is controlled by the servo drive system to achieve smooth movement of the standard silicon wafer simulation board at different speed gears, ensuring that the light field characteristic data under different motion states can be obtained during the test.

[0026] Specifically, after determining the light source lighting control instructions and the movement speed gear, a photodetector array is used to collect multi-region light fields generated by the LED ring light source. The photodetector array consists of a plurality of highly sensitive photodiodes, which are evenly distributed along the ring and cover the inner ring, middle ring and outer ring. Each detector converts the optical signal into an electrical signal through a preamplifier circuit, amplifies and filters it through a signal conditioning circuit, and then realizes digital sampling through a high-speed AD converter. At each speed gear, the control system synchronously collects the light intensity signals of each area according to the preset sampling frequency to obtain complete data on the changes of light intensity with time and spatial position. These data are normalized and spatially reconstructed to form light intensity distribution data that characterizes the spatial light intensity distribution characteristics of the LED ring light source, including the time domain fluctuation characteristics and spatial distribution characteristics of the light intensity.

[0027] Specifically, while collecting the light field, an industrial camera is used to collect images of standard defects on the surface of the standard silicon wafer simulation board. The industrial camera uses a high-resolution CMOS sensor and is equipped with a telecentric lens optical system. It is synchronized with the movement of the standard silicon wafer simulation board through trigger signal control. The image collected by the camera undergoes digital image processing, including preprocessing operations such as image enhancement, noise suppression, and geometric correction, and then the contour features of the defect area are extracted using an edge detection algorithm. For different types of standard defects, the morphological analysis method is used to extract characteristic parameters such as the area, perimeter, and grayscale distribution of the defect to form a complete defect feature description. At the same time, the spectral characteristics of the LED ring light source are measured using a spectrum analyzer. The spectrum analyzer collects spectral information of the inner ring, middle ring, and outer ring through a fiber optic probe to obtain spectral parameters such as wavelength distribution, color temperature, and color rendering index of each area. After the spectral data is calibrated for wavelength and intensity, it is combined with the previously obtained light intensity distribution data to form spectral characteristic data that characterizes the luminous characteristics of the LED ring light source.

[0028] Specifically, in order to achieve a comprehensive evaluation of the performance of LED ring light sources, it is necessary to perform systematic data integration processing on the acquired light intensity distribution data, defect detection image data, and spectral characteristic data. First, a unified data format standard is established to convert different types of data into standardized data structures. Through the data synchronization mechanism, ensure that data from different sources are aligned in the time dimension and establish a correlation between data. A data fusion algorithm is used to fuse multi-source data at the feature level to extract key indicators that reflect the performance characteristics of LED ring light sources. For light intensity distribution data, light intensity uniformity and stability indicators are extracted; for defect detection image data, defect detection sensitivity and accuracy indicators are extracted; for spectral characteristic data, spectral consistency and color reproduction indicators are extracted. These indicators are organized in the form of feature vectors to form multi-dimensional test data that describes the overall performance of LED ring light sources, providing comprehensive data support for subsequent performance evaluation.

[0029] 103. Perform characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results, and construct a light field distortion model that characterizes the degree of light intensity distribution distortion based on the light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results; In one embodiment of the present invention, the characteristic analysis processing of the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results includes: performing Fourier transform processing on the light intensity distribution data to obtain a light intensity distribution spectrum characteristic vector, and discretizing the light intensity fluctuation frequency domain distribution of the LED ring light source according to the light intensity distribution spectrum characteristic vector to obtain light intensity fluctuation frequency characteristic data; performing spatial mapping processing on the light intensity distribution of the three regions of the inner ring, the middle ring and the outer ring according to the spectral characteristic data to obtain a light field spatial distribution characteristic matrix, and performing singular value decomposition operation on the light field spatial distribution characteristic matrix to obtain light field spatial characteristic data; performing edge detection and morphological processing on the defect detection image data to obtain a defect characteristic vector, and identifying and classifying the standard defects according to the defect characteristic vector to obtain a defect detection rate evaluation result.

[0030] Specifically, Fourier transform processing of light intensity distribution data is an important means to analyze the light intensity fluctuation characteristics of LED ring light sources. First, the collected time domain light intensity data is reconstructed into a discrete time series according to the sampling time series, and the data is preprocessed by a windowing function to reduce spectrum leakage. The processed time domain data is transformed by a fast Fourier transform algorithm, and the time domain information is converted to the frequency domain space to obtain the spectrum distribution of light intensity fluctuations. The spectrum data is amplitude normalized and phase corrected, and the amplitude and phase information of the main frequency components are extracted to form a vector describing the spectrum characteristics of the light intensity distribution. Based on the spectrum characteristic vector, the sampling interval is determined by the frequency resolution criterion, and the frequency domain distribution is discretized and sampled to obtain the spectrum characteristics at discrete frequency points. By performing statistical analysis on the sampled data, the spectrum energy distribution, the main frequency component and its modulation depth are calculated, and finally the frequency characteristic data characterizing the light intensity fluctuation characteristics of the LED ring light source are obtained.

[0031] Specifically, the spectral characteristic data contains the spectral information of each area of ​​the LED ring light source, and it is necessary to associate the spectral characteristics with the light intensity distribution through spatial mapping. First, a cylindrical coordinate system is established to convert the spatial position parameters of the three areas of the inner ring, the middle ring and the outer ring into standardized coordinates. The wavelength is integrated for the spectral data of each area, the energy distribution of each wavelength range is calculated, and a corresponding relationship is established with the spatial coordinates. The discrete sampling points are spatially reconstructed by an interpolation algorithm to obtain a continuous light intensity distribution function. The reconstructed light intensity distribution data is organized into a matrix form, in which the rows of the matrix represent the spatial position and the columns represent the light intensity values ​​of different wavelengths, forming a light field spatial distribution feature matrix. The feature matrix is ​​subjected to singular value decomposition and decomposed into three components representing the spatial basis vector, singular value and eigenvector. By analyzing the distribution characteristics of the singular values, the main spatial characteristic patterns are extracted, and the light field distribution is reconstructed in combination with the eigenvectors, and finally the light field spatial characteristic data describing the spatial distribution characteristics of the LED ring light source are obtained.

[0032] Specifically, the processing of defect detection image data first requires edge detection and morphological analysis. The gradient of the image is calculated by the Sobel operator or the Canny operator, the edge information in the image is extracted, and the contour features of the defect area are obtained. The edge image is binarized using the adaptive threshold segmentation method to separate the defect area from the background. Then, morphological operations, including basic operations such as expansion, corrosion, opening and closing operations, are used to process the binary image to eliminate noise and improve the defect contour. Based on the processed image, the morphological feature parameters such as the area, perimeter, roundness, and rectangularity of the defect area are calculated, and the grayscale statistical features of the defect area are extracted, including texture features such as average grayscale, grayscale variance, and grayscale co-occurrence matrix. These feature parameters constitute a vector that describes the defect characteristics.

[0033] Specifically, based on the acquired defect feature vectors, standard defects need to be identified and classified. First, a standard defect sample library is established, which contains feature templates of different types of defects, such as scratches, stains, bubbles and other typical defect types. Pattern recognition algorithms, such as support vector machines or neural network classifiers, are used to classify and discriminate the feature vectors of the defects to be tested and determine the types of defects. The credibility of defect identification is evaluated by calculating the similarity between the feature vector and the standard template. For each type of defect, the ratio of the number of detected defects to the actual number of defects is counted, and the detection rate index under different working conditions is calculated. The detection rate data is correlated with the working parameters to establish a quantitative evaluation system for defect detection capabilities, and finally the defect detection rate evaluation results that characterize the detection performance of LED ring light sources are obtained. These evaluation results include statistical indicators such as the detection probability, missed detection rate, and false detection rate of different types of defects, providing an important basis for the performance evaluation of LED ring light sources.

[0034] Further, the light intensity distribution of the inner ring, middle ring and outer ring regions is spatially mapped according to the spectral characteristic data to obtain a light field spatial distribution feature matrix, and the light field spatial distribution feature matrix is ​​subjected to singular value decomposition to obtain light field spatial characteristic data, including: polar coordinate transformation of the spectral characteristic data to obtain angular light intensity distribution data, and the inner ring, middle ring and outer ring regions are uniformly divided into circles according to the angular light intensity distribution data to obtain annular region division parameters; interpolation fitting of the angular light intensity distribution data is performed according to the annular region division parameters to obtain a spatial sampling point sequence, and the The spatial sampling point sequence is radially normalized to obtain standardized light intensity distribution data; the standardized light intensity distribution data is subjected to matrix reconstruction operation to obtain a light field spatial distribution characteristic matrix, and the light field spatial distribution characteristic matrix is ​​subjected to covariance calculation operation to obtain a spatial correlation matrix; eigenvalue calculation operation is performed according to the spatial correlation matrix to obtain a singular value sequence, and the singular value sequence is arranged in descending order to obtain main characteristic components; the main characteristic components are subjected to spatial reprojection processing to obtain reconstructed light field distribution data, and characteristic vector orthogonalization processing is performed according to the reconstructed light field distribution data to obtain light field spatial characteristic data.

[0035] Specifically, after the spectral characteristic data is collected through the rectangular coordinate system, it needs to undergo precise coordinate transformation processing to meet the requirements of the geometric feature analysis of the ring light source. First, a set of coordinate transformation equations is established to convert the position point (x, y) in the rectangular coordinate system into (r, θ) in the polar coordinate system through the inverse tangent function and the square and square root operation, where r represents the distance value from the spatial sampling point to the center of the LED ring light source, and θ represents the angle between the sampling point position vector and the horizontal reference axis. In the polar coordinate system, the spectral energy value of each angular position is extracted according to the principle of uniform angle sampling, and the light intensity value at the corresponding position is obtained by integrating it, thereby forming a complete angular light intensity distribution data set. At the same time, according to the physical size parameters of the inner ring, middle ring and outer ring of the LED ring light source, the radius range of each area is calculated to determine the radial stratification boundary. Using the principle of equal angle division, each annular area is divided into multiple angle units along the circumferential direction. The number of unit divisions is determined according to the spatial resolution requirements, and the general value range is 120-360. By accurately calculating the starting angle, ending angle, inner diameter, outer diameter and other geometric parameters of each angle unit, a complete annular area division parameter system is formed.

[0036] Specifically, based on the annular area division parameters, the cubic spline interpolation algorithm is used to reconstruct and optimize the angular light intensity distribution data. First, a cubic spline function is constructed inside each angular unit. The function coefficients are calculated by analyzing the positions and light intensity values ​​of adjacent data points to ensure that the zero-order, first-order and second-order derivatives of the interpolation curve at the nodes are continuous. Interpolation calculations are performed on each angular unit to generate denser sampling points in both the angular and radial directions. The sampling interval is usually 1 / 10 of the original data interval to obtain a high-precision spatial sampling point sequence. Then a radial normalization function is constructed, and the distance attenuation characteristics of LED light are considered. The inverse square relationship is used to compensate for the light intensity values ​​at different radial positions. By dividing the measured light intensity value of each sampling point by its corresponding normalization coefficient, the light intensity difference caused by the distance change is eliminated, and finally the standard light intensity distribution data uniformed in space is obtained. These data truly reflect the luminous characteristics of the LED ring light source in different areas.

[0037] Specifically, the standardized light intensity distribution data needs to be represented by a two-dimensional matrix after system reconstruction. First, determine the number of rows and columns of the matrix. The number of rows corresponds to the number of angular sampling points, which is generally 360 to ensure that the angular resolution reaches 1 degree. The number of columns corresponds to the number of radial layers, which varies from 12 to 24 layers according to the stratification requirements of the inner ring, middle ring, and outer ring. The sampling data is located and mapped according to the two dimensions of angle and radius, and the standardized light intensity value corresponding to each spatial position is filled into the corresponding position of the matrix to form a complete light field spatial distribution feature matrix. The spatial correlation of the feature matrix is ​​calculated using statistical methods. The correlation coefficient of the light intensity values ​​of any two positions in the matrix is ​​calculated to obtain the correlation coefficient value that measures the degree of association of the spatial point pairs. The correlation coefficients of all position pairs are rearranged according to the spatial correspondence, and an N×N-dimensional spatial correlation matrix is ​​constructed, where N is the total number of sampling points of the feature matrix, and the matrix element value represents the degree of correlation of the light intensity distribution between the corresponding spatial positions.

[0038] Specifically, eigenvalue analysis is performed based on the spatial correlation matrix, and the characteristic equation is solved by numerical calculation method. All eigenvalues ​​of the N×N matrix are calculated by an iterative algorithm, and each eigenvalue corresponds to a basic mode that reflects the spatial distribution characteristics of the light field. The calculated eigenvalues ​​are sorted in descending order according to the numerical value to obtain a singular value sequence that characterizes the importance of different spatial modes. By analyzing the attenuation law of the singular values, the number of main characteristic components is determined. Generally, the eigenvectors corresponding to the eigenvalues ​​with a cumulative contribution rate of more than 95% are selected as the main characteristic components. These main characteristic components contain the key characteristic information of the spatial distribution of the light field, and their numerical values ​​directly reflect the weight of the characteristic mode in the overall light field distribution.

[0039] Specifically, the main characteristic components are reprojected in the spatial domain, and the information of the characteristic space is mapped back to the physical space. First, the eigenvectors corresponding to each characteristic component are calculated, and these eigenvectors constitute an orthogonal basis for describing the spatial distribution of the light field. By linearly combining the main characteristic components with the corresponding eigenvectors, the spatial distribution of the light field is reconstructed to obtain the reconstructed light field distribution data reflecting the main spatial characteristics of the LED ring light source. Then, the eigenvectors in the reconstructed data are Schmidt orthogonalized to eliminate the possible linear correlation between the vectors and obtain mutually orthogonal characteristic basis vectors. Based on these orthogonalized eigenvectors, the spatial structural characteristics of the light field distribution are extracted, including the light intensity uniformity index, the spatial symmetry index, the edge sharpness index, etc. These indicators constitute the complete light field spatial characteristic data, which comprehensively characterizes the spatial luminescence characteristics of the LED ring light source.

[0040] Furthermore, the light field distortion model that characterizes the degree of light intensity distribution distortion based on the light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results includes: standardizing the light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results to obtain a normalized feature vector; constructing a three-dimensional feature space based on the normalized feature vector, and performing a principal component analysis operation on the three-dimensional feature space to obtain a main feature component; performing feature weight calculation processing on the main feature component to obtain a contribution coefficient of each dimensional feature, and performing a linear combination operation on the main feature component based on the contribution coefficient to obtain light field distortion characteristic parameters; establishing a nonlinear regression equation based on the light field distortion characteristic parameters, and iteratively solving the coefficients of the nonlinear regression equation by the least squares method to obtain a light field distortion model that characterizes the degree of light intensity distribution distortion.

[0041] Specifically, when standardizing the light field spatial characteristic data, light intensity fluctuation frequency characteristic data, and defect detection rate evaluation results, it is first necessary to solve the problem of inconsistent dimensions of different data types. By calculating the mean and standard deviation of each type of data, the original data is centered and the standard deviation is normalized. Specifically, for the light field spatial characteristic data, it contains the light intensity distribution values ​​of 360×24 spatial sampling points. Each data point is subtracted from the average value of all sampling points and divided by the standard deviation; for the light intensity fluctuation frequency characteristic data, it contains the amplitude of 128 discrete sampling points in the frequency domain, and the mean elimination and variance normalization are also performed; for the defect detection rate evaluation results, it contains the detection probability values ​​of different types of defects. These probability values ​​are logarithmically transformed and then standardized. After standardization, the three types of data are mapped to a distribution interval with a mean of 0 and a variance of 1, forming a standardized feature space with equal weights. These standardized data are combined into a high-dimensional feature vector in the order of spatial characteristics, frequency characteristics, and detection rate characteristics. The vector dimension is the sum of all feature dimensions, that is, 8640+128+10 (assuming there are 10 defect types), thereby obtaining a normalized feature vector.

[0042] Specifically, when constructing a three-dimensional feature space based on normalized eigenvectors, it is necessary to first reduce the dimensionality of high-dimensional feature data. The principal component analysis (PCA) method is used to first calculate the covariance matrix of the normalized eigenvector, and the matrix dimension is 8778×8778 (i.e., the square of 8640+128+10). The covariance matrix is ​​subjected to eigenvalue decomposition, and the eigenvalues ​​and eigenvectors are solved by the Jacobi iteration method. The convergence threshold in the iterative process is set to 1e-6, and the maximum number of iterations is 1000. The obtained eigenvalues ​​are sorted from large to small, and the eigenvectors corresponding to the first three largest eigenvalues ​​are selected as the principal component basis vectors. These three eigenvectors constitute the basis of the new three-dimensional feature space, where the first eigenvector mainly reflects the spatial distribution characteristics, the second eigenvector mainly reflects the frequency characteristics, and the third eigenvector mainly reflects the detection rate characteristics. The original normalized eigenvector is projected onto these three basis vectors to obtain the coordinate representation in the new three-dimensional feature space, namely, the principal eigencomponent. Each principal eigencomponent is the projection value of the original feature in the corresponding direction, which contains the main change information in that direction.

[0043] Specifically, when calculating the feature weights of the main feature components, it is necessary to comprehensively consider the physical meaning and statistical significance of the features of each dimension. First, the contribution rate of the eigenvalues ​​in the directions of the three main feature components is calculated, and each eigenvalue is divided by the sum of all eigenvalues ​​to obtain the relative contribution ratio. Then, a weight modulation factor based on the Fisher discriminant criterion is introduced. By calculating the ratio of the inter-class discreteness to the intra-class discreteness of each dimensional feature in samples with different light field distortion degrees, the Fisher score reflecting the feature discrimination ability is obtained. The Fisher score is converted into a weight coefficient in the interval [0,1] by exponential mapping. At the same time, considering the physical meaning of different features, a priori weight matrix is ​​introduced, and the element values ​​in the matrix are determined according to expert knowledge and historical data statistics. The contribution rate, Fisher weight and prior weight are weighted averaged to obtain the final contribution coefficient of each dimensional feature. Based on these contribution coefficients, the three main characteristic components are linearly combined. The combination process adopts a weighted summation method. Each characteristic component is multiplied by the corresponding contribution coefficient and then added. Finally, a one-dimensional light field distortion characteristic parameter is obtained. This parameter comprehensively reflects the degree of distortion of the light field in three aspects: spatial distribution, frequency characteristics and defect detection.

[0044] Specifically, based on the calculated characteristic parameters of light field distortion, an accurate mathematical model needs to be established to describe the degree of light field distortion. First, a nonlinear regression equation is constructed, in which the independent variable is the characteristic parameter of light field distortion and the dependent variable is the measured value of light field distortion. Considering the nonlinear relationship between light field distortion and characteristic parameters, a polynomial function is selected as the basic model, including quadratic terms, cubic terms and cross terms, with a total of 10 undetermined coefficients. Then, these coefficients are solved by the least squares method, the error square sum objective function is established, and the Levenberg-Marquardt algorithm is used for iterative optimization. In each iteration, the error between the model prediction value and the measured value under the current coefficient is first calculated, and then the Jacobian matrix is ​​constructed to solve the normal equation to obtain the correction amount of the coefficient. To prevent overfitting, the Tikhonov regularization term is introduced, and the regularization parameter is determined by the L curve method. When the coefficient change of two consecutive iterations is less than the preset threshold (usually 1e-8) or the maximum number of iterations (set to 1000) is reached, the iteration is stopped. The final nonlinear regression equation is the light field distortion model that characterizes the degree of light intensity distribution distortion. This model can accurately predict the degree of distortion under given light field characteristic parameters.

[0045] 104. Calculate the light intensity uniformity deviation value and defect detection capability score of the LED ring light source based on the light field distortion model, and generate a test evaluation report of the LED ring light source based on the light intensity uniformity deviation value and defect detection capability score.

[0046] In one embodiment of the present invention, the calculation of the light intensity uniformity deviation value and the defect detection capability score of the LED ring light source according to the light field distortion model, and the generation of a test evaluation report of the LED ring light source according to the light intensity uniformity deviation value and the defect detection capability score include: performing numerical calculation processing on the light field distortion model to obtain the light intensity distortion coefficient of each region, and performing statistical distribution operation on the light intensity distortion coefficient to obtain a light intensity uniformity deviation curve; performing weighted accumulation processing on the light intensity distortion coefficient of each region according to the light intensity uniformity deviation curve to obtain a comprehensive distortion index, and performing normalization operation on the comprehensive distortion index , and obtain the light intensity uniformity deviation value; classify and statistically process the defect detection rate evaluation result to obtain the detection probability distribution of each type of defects, and perform error analysis calculation based on the detection probability distribution to obtain detection confidence data; perform weight allocation processing on the detection probability distribution of each type of defects based on the detection confidence data to obtain a detection capability weight matrix, and perform feature fusion calculation on the detection capability weight matrix to obtain a defect detection capability score; perform multi-dimensional indicator integration processing on the light intensity uniformity deviation value and the defect detection capability score to obtain a comprehensive performance evaluation index, and generate a test evaluation report based on the comprehensive performance evaluation index.

[0047] Specifically, when numerically calculating the light field distortion model, the model first needs to be applied to the inner ring, middle ring and outer ring of the LED ring light source. By substituting the light field characteristic parameters of each region into the nonlinear regression equation, the light intensity distortion coefficient of each region is calculated. During the calculation process, the Runge-Kutta method is used to numerically solve the nonlinear equation, the solution step is set to 0.01, and the calculation accuracy is controlled within the range of 1e-6. For each region, 360 uniformly distributed sampling points are taken along the circumferential direction, and the local light intensity distortion coefficient is calculated at each sampling point. These coefficients reflect the degree to which the light intensity distribution of the LED ring light source at different positions deviates from the ideal uniform distribution. Then, the light intensity distortion coefficients of the three regions are subjected to probability statistical analysis, and the probability density distribution function of the distortion coefficient of each region is calculated. The kernel density estimation method is adopted, and the discrete distortion coefficient data is smoothed using the Gaussian kernel function, and the bandwidth of the kernel function is determined by the Silverman criterion. The probability density curves of the three regions are plotted in the same coordinate system to form a complete light intensity uniformity deviation curve, which reflects the distribution law and severity of light intensity distortion in different regions.

[0048] Specifically, based on the obtained light intensity uniformity deviation curve, it is necessary to perform quantitative weighted analysis on the light intensity distortion coefficient of each region. First, the statistical moments of the deviation curve of each region are calculated, including mean, variance, skewness and kurtosis. These statistics reflect the different statistical characteristics of light intensity distortion in each region. Then, according to the importance of each region in practical applications, the regional weight coefficient is set, and the weight ratio of the inner ring, middle ring and outer ring is allocated according to 4:3:3. The light intensity distortion coefficient of each region is multiplied by the corresponding weight coefficient, and the weighted results of all regions are accumulated to obtain a comprehensive distortion index reflecting the overall distortion degree. Considering the comparability requirements between different batches of LED ring light sources, the comprehensive distortion index is normalized. The normalization process uses the minimum-maximum normalization method to map the original comprehensive distortion index to the [0,1] interval. When mapping, the reference range determined by the historical test data is referred to, and finally the standardized light intensity uniformity deviation value is obtained. This deviation value can quantitatively characterize the light intensity distribution uniformity level of the LED ring light source.

[0049] Specifically, when classifying and statistically processing the defect detection rate evaluation results, it is necessary to establish a special statistical analysis system for each type of defect. First, the defects are divided into different categories such as scratches, pitting, bubbles, stains, etc. according to their morphological characteristics, and each category is divided into three levels: large, medium, and small according to the defect size. For each category and level of defect samples, calculate its detection probability under different working conditions (including different light intensity levels and detection speeds). Using the stratified sampling method, repeat the detection experiment 100 times under each working condition, and record the number of defects successfully detected in each experiment. According to the principle of the law of large numbers, calculate the average detection rate and standard deviation of each defect type under each working condition. Using the Bootstrap resampling technique, estimate the confidence interval of the detection probability through 1000 resamplings, and construct a complete detection probability distribution curve. Perform variance analysis on these distribution data, calculate the ratio of the between-group variance to the within-group variance, and evaluate the significance of the impact of different working conditions on the detection rate. The detection confidence data obtained through these statistical analyses include the expected value, fluctuation range and confidence level of the detection rate of each defect type.

[0050] Specifically, based on the detection confidence data, a reasonable weight distribution mechanism needs to be established. First, a hierarchical analysis model is constructed, and defect type, size grade and working conditions are used as evaluation factors at different levels. The pairwise comparison matrix between factors is established through the expert scoring method, and the matrix size is n×n, where n is the total number of all evaluation factors. The comparison matrix is ​​tested for consistency, and the consistency index CI and consistency ratio CR are calculated. When CR is less than 0.1, the judgment matrix is ​​considered to have satisfactory consistency. Then the eigenvalue method is used to calculate the weight vector, and the eigenvector corresponding to the maximum eigenvalue is standardized to obtain the weight coefficient of each factor. These weight coefficients are multiplied by the corresponding detection probability to construct the detection capability weight matrix. The weight matrix is ​​subjected to eigenvalue decomposition, and the eigenvector corresponding to the maximum eigenvalue is extracted as the principal component. Through the principal component score calculation method, all detection indicators are integrated into a unified defect detection capability score, which comprehensively reflects the detection capability of the LED ring light source for various defects.

[0051] Specifically, when integrating the multi-dimensional indicators of light intensity uniformity deviation value and defect detection ability score, it is necessary to establish a scientific comprehensive evaluation system. First, a standardized model of evaluation indicators is constructed, and the light intensity uniformity deviation value is converted into a fuzzy evaluation value through a semi-trapezoidal membership function by a fuzzy mathematical method. The parameters of the membership function are determined by genetic algorithm optimization based on a large amount of historical test data. At the same time, a five-level evaluation standard is established for defect detection ability scoring, and the closeness of the score value to each level of standards is calculated by the distance discrimination method. Then a fuzzy comprehensive evaluation matrix is ​​established, and the row vectors of the matrix represent different evaluation indicators, and the column vectors represent different evaluation levels. The weighted average model is used to calculate the comprehensive membership, and the weight coefficient is determined by the entropy weight method. The discrete degree of the indicator is considered in the calculation. The eccentricity of the comprehensive membership is calculated to determine the final evaluation level. Based on the evaluation results, a test evaluation report in a standard format is generated. The report content includes the test environment parameter record, the original test data of each indicator, the intermediate results of the data processing process, the final quantitative evaluation results, the equipment operation status record and the test conclusion. Through this multi-dimensional indicator integration method, the overall performance of the LED ring light source can be comprehensively and objectively evaluated.

[0052] In this embodiment, the LED ring light source installed on the angle-adjustable bracket and the standard silicon wafer simulation board are calibrated and initialized to obtain a benchmark test sequence; the LED ring light source is controlled to illuminate the standard silicon wafer simulation board according to the benchmark test sequence, and multi-region light field acquisition is performed at different speed gears to obtain multi-dimensional test data; the multi-dimensional test data is analyzed for characteristics to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results, and then a light field distortion model is constructed; the light intensity uniformity deviation value and defect detection capability score are calculated according to the light field distortion model, and a test evaluation report is generated. This method constructs a light field distortion model through multi-region acquisition and data analysis, and can accurately evaluate the uniformity of light intensity distribution and defect detection capability. Improve the effect of LED ring light sources in high-precision detection of silicon wafers.

[0053] The above describes the LED ring light source testing method in the embodiment of the present invention. The following describes the LED ring light source testing device in the embodiment of the present invention. Figure 2 , an embodiment of the LED ring light source testing device in the embodiment of the present invention includes: A pre-processing module 201 is used to perform calibration initialization processing on the LED ring light source mounted on the angle-adjustable bracket and the standard silicon wafer simulation board, and obtain a benchmark test sequence of the LED ring light source; The test module 202 is used to control the LED ring light source to irradiate the standard silicon wafer simulation board according to the benchmark test sequence, and to collect multi-area light fields irradiated by the LED ring light source on the standard silicon wafer simulation board at different speed gears to obtain multi-dimensional test data; The analysis module 203 is used to perform characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results, and to construct a light field distortion model that characterizes the degree of light intensity distribution distortion based on the light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results; The evaluation module 204 is used to calculate the light intensity uniformity deviation value and the defect detection capability score of the LED ring light source according to the light field distortion model, and generate a test evaluation report of the LED ring light source according to the light intensity uniformity deviation value and the defect detection capability score.

[0054] In an embodiment of the present invention, the LED ring light source test device runs the above-mentioned LED ring light source test method, and the LED ring light source test device obtains a benchmark test sequence by calibrating and initializing the LED ring light source and the standard silicon wafer simulation board installed on the angle-adjustable bracket; controls the LED ring light source to illuminate the standard silicon wafer simulation board according to the benchmark test sequence, and performs multi-region light field acquisition at different speed gears to obtain multi-dimensional test data; performs characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results, and then constructs a light field distortion model; calculates the light intensity uniformity deviation value and defect detection capability score according to the light field distortion model, and generates a test evaluation report. This method constructs a light field distortion model through multi-region acquisition and data analysis, and can accurately evaluate the uniformity of light intensity distribution and defect detection capability. Improve the effect of LED ring light sources in high-precision detection of silicon wafers.

[0055] above Figure 2 The LED ring light source testing device in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The LED ring light source testing device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0056] Figure 33 is a schematic diagram of the structure of an LED ring light source test device provided by an embodiment of the present invention. The LED ring light source test device 300 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 310 (for example, one or more processors) and a memory 320, and one or more storage media 330 (for example, one or more mass storage device terminals) storing application programs 333 or data 332. Among them, the memory 320 and the storage medium 330 can be short-term storage or permanent storage. The program stored in the storage medium 330 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the LED ring light source test device 300. Furthermore, the processor 310 may be configured to communicate with the storage medium 330, and execute a series of instruction operations in the storage medium 330 on the LED ring light source test device 300 to implement the steps of the above-mentioned LED ring light source test method.

[0057] The LED ring light source testing device 300 may further include one or more power supplies 340, one or more wired or wireless network interfaces 350, one or more input and output interfaces 360, and / or one or more operating systems 331, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Figure 3 The structure of the LED ring light source testing device shown does not constitute a limitation on the LED ring light source testing device provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0058] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the LED ring light source testing method.

[0059] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0060] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0061] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for testing an LED ring light source, characterized in that: The LED ring light source testing method comprises: Performing calibration initialization processing on an LED ring light source mounted on an angle-adjustable bracket and a standard silicon wafer simulation board, and obtaining a benchmark test sequence of the LED ring light source; Controlling the LED ring light source to illuminate the standard silicon wafer simulation board according to the benchmark test sequence, and collecting the light field illuminated by the LED ring light source on the standard silicon wafer simulation board in multiple regions at different speed gears to obtain multi-dimensional test data; Performing characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data, and defect detection rate evaluation results, and constructing a light field distortion model that characterizes the degree of light intensity distribution distortion based on the light field spatial characteristic data, light intensity fluctuation frequency characteristic data, and defect detection rate evaluation results; The light intensity uniformity deviation value and the defect detection capability score of the LED ring light source are calculated according to the light field distortion model, and a test evaluation report of the LED ring light source is generated according to the light intensity uniformity deviation value and the defect detection capability score.

2. The LED ring light source testing method according to claim 1, characterized in that: The step of performing calibration initialization processing on the LED ring light source mounted on the angle-adjustable bracket and the standard silicon wafer simulation board and obtaining a benchmark test sequence of the LED ring light source comprises: Measuring the positional relationship between the LED ring light source and the standard silicon wafer simulation board to obtain position parameter data, and setting the motion parameters of the standard silicon wafer simulation board according to the position parameter data; According to the motion parameters, the lighting parameters of the LED ring light source are configured to obtain a parameter combination of lighting angle, working distance and light intensity level; The motion parameters and the parameter combinations are subjected to timing matching processing to obtain a motion-lighting linkage control scheme, and a benchmark test sequence is generated according to the motion-lighting linkage control scheme.

3. The LED ring light source testing method according to claim 1, characterized in that: The LED ring light source is controlled to illuminate the standard silicon wafer simulation board according to the benchmark test sequence, and the light field illuminated by the LED ring light source on the standard silicon wafer simulation board is collected in multiple regions at different speed gears to obtain multi-dimensional test data including: According to the benchmark test sequence, determining the light source illumination control instruction of the LED ring light source, and performing speed gear switching control on the standard silicon wafer simulation board; Under the light source illumination control instruction and different speed gears of the standard silicon wafer simulation board, light field collection is performed on the inner ring, middle ring and outer ring of the LED ring light source through a photodetector array to obtain light intensity distribution data; The standard defects on the surface of the standard silicon wafer simulation board are imaged and processed by an industrial camera to obtain defect detection image data, and the spectrum of the inner ring, middle ring and outer ring regions is collected and processed by a spectrum analyzer according to the light intensity distribution data to obtain spectrum characteristic data; The light intensity distribution data, defect detection image data and spectral characteristic data are subjected to data integration processing to obtain multi-dimensional test data.

4. The LED ring light source testing method according to claim 3, characterized in that: The characteristic analysis and processing of the multi-dimensional test data to obtain light field space characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results includes: Performing Fourier transform processing on the light intensity distribution data to obtain a light intensity distribution spectrum feature vector, and discretizing and sampling the light intensity fluctuation frequency domain distribution of the LED ring light source according to the light intensity distribution spectrum feature vector to obtain light intensity fluctuation frequency characteristic data; Performing spatial mapping processing on the light intensity distribution of the inner ring, the middle ring and the outer ring according to the spectral characteristic data to obtain a light field spatial distribution characteristic matrix, and performing singular value decomposition operation on the light field spatial distribution characteristic matrix to obtain light field spatial characteristic data; The defect detection image data is subjected to edge detection and morphological processing to obtain a defect feature vector, and the standard defects are identified and classified according to the defect feature vector to obtain a defect detection rate evaluation result.

5. The LED ring light source testing method according to claim 4, characterized in that: The light intensity distribution of the inner ring, the middle ring and the outer ring is spatially mapped according to the spectral characteristic data to obtain a light field spatial distribution characteristic matrix, and a singular value decomposition operation is performed on the light field spatial distribution characteristic matrix to obtain light field spatial characteristic data, including: Performing polar coordinate transformation processing on the spectral characteristic data to obtain angular light intensity distribution data, and evenly dividing the inner ring, middle ring and outer ring regions into three regions according to the angular light intensity distribution data to obtain annular region division parameters; Performing interpolation fitting processing on the angular light intensity distribution data according to the annular area division parameters to obtain a spatial sampling point sequence, and performing radial normalization processing on the spatial sampling point sequence to obtain standardized light intensity distribution data; Performing a matrix reconstruction operation on the standardized light intensity distribution data to obtain a light field spatial distribution characteristic matrix, and performing a covariance calculation process on the light field spatial distribution characteristic matrix to obtain a spatial correlation matrix; Performing eigenvalue calculation processing according to the spatial correlation matrix to obtain a singular value sequence, and performing a descending order operation on the singular value sequence to obtain a main eigencomponent; The main characteristic components are spatially reprojected to obtain reconstructed light field distribution data, and characteristic vector orthogonalization is performed on the reconstructed light field distribution data to obtain light field spatial characteristic data.

6. The LED ring light source testing method according to claim 1, characterized in that: The light field distortion model characterizing the degree of light intensity distribution distortion is constructed based on the light field spatial characteristic data, the light intensity fluctuation frequency characteristic data and the defect detection rate evaluation result, including: Standardizing the light field spatial characteristic data, the light intensity fluctuation frequency characteristic data and the defect detection rate evaluation result to obtain a normalized feature vector; Constructing a three-dimensional feature space according to the normalized feature vector, and performing a principal component analysis operation on the three-dimensional feature space to obtain a principal feature component; Performing feature weight calculation processing on the main feature components to obtain contribution coefficients of features in each dimension, and performing linear combination operations on the main feature components according to the contribution coefficients to obtain light field distortion feature parameters; A nonlinear regression equation is established according to the light field distortion characteristic parameters, and the coefficients of the nonlinear regression equation are iteratively solved by the least square method to obtain a light field distortion model that characterizes the degree of light intensity distribution distortion.

7. The LED ring light source testing method according to claim 1, characterized in that: The step of calculating the light intensity uniformity deviation value and the defect detection capability score of the LED ring light source according to the light field distortion model, and generating a test evaluation report of the LED ring light source according to the light intensity uniformity deviation value and the defect detection capability score includes: Performing numerical calculation processing on the light field distortion model to obtain light intensity distortion coefficients of each region, and performing statistical distribution calculation based on the light intensity distortion coefficients to obtain a light intensity uniformity deviation curve; According to the light intensity uniformity deviation curve, weighted accumulation processing is performed on the light intensity distortion coefficients of each region to obtain a comprehensive distortion index, and normalization operation is performed on the comprehensive distortion index to obtain a light intensity uniformity deviation value; Performing classified statistical processing on the defect detection rate evaluation results to obtain the detection probability distribution of each type of defect, and performing error analysis calculation based on the detection probability distribution to obtain detection confidence data; Performing weight distribution processing on the detection probability distribution of the various defects according to the detection confidence data to obtain a detection capability weight matrix, and performing feature fusion operation on the detection capability weight matrix to obtain a defect detection capability score; The light intensity uniformity deviation value and the defect detection capability score are subjected to multi-dimensional indicator integration processing to obtain a comprehensive performance evaluation indicator, and a test evaluation report is generated based on the comprehensive performance evaluation indicator.

8. A LED ring light source testing device, characterized in that: The LED annular light source testing device comprises: A preprocessing module, used for performing calibration initialization processing on the LED ring light source mounted on the angle-adjustable bracket and the standard silicon wafer simulation board, and obtaining a benchmark test sequence of the LED ring light source; A test module, used to control the LED ring light source to irradiate the standard silicon wafer simulation board according to the benchmark test sequence, and to collect multi-area light fields irradiated by the LED ring light source on the standard silicon wafer simulation board at different speed gears to obtain multi-dimensional test data; An analysis module is used to perform characteristic analysis on the multi-dimensional test data to obtain light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results, and to construct a light field distortion model that characterizes the degree of light intensity distribution distortion based on the light field spatial characteristic data, light intensity fluctuation frequency characteristic data and defect detection rate evaluation results; An evaluation module is used to calculate the light intensity uniformity deviation value and defect detection capability score of the LED ring light source according to the light field distortion model, and generate a test evaluation report of the LED ring light source according to the light intensity uniformity deviation value and defect detection capability score.

9. An LED ring light source testing device, characterized in that: The LED ring light source testing device comprises: a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to enable the LED ring light source testing device to perform the steps of the LED ring light source testing method according to any one of claims 1 to 7.

10. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed by the processor, the steps of the LED ring light source testing method as described in any one of claims 1 to 7 are implemented.

Images