An exposure machine fault diagnosis method and system based on data processing

Abstract

The present application relates to the field of data processing, and particularly relates to a kind of exposure machine fault diagnosis method and system based on data processing.Exposure machine fault diagnosis method and system based on data processing include: obtaining gray exposure image and vibration data;Hough curve detection is carried out to gray exposure image, the number of curves in each gray exposure image is obtained, and the number of votes corresponding to each curve number value;Visual exposure index of gray exposure image is calculated;Two-dimensional fast Fourier transform is carried out to gray exposure image, and frequency spectrum result is obtained;Frequency ratio of gray exposure image is calculated;Exposure forming property index of gray exposure image is calculated;When exposure forming property index is less than or equal to preset fault threshold, vibration root cause index is calculated according to vibration data, and wear vibration amount is calculated according to vibration root cause index;The fault reason of exposure machine is determined according to wear vibration amount.The present application can improve the identification precision of exposure machine fault type.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing, in particular to an exposure machine fault diagnosis method and system based on data processing. BACKGROUND

[0002] An exposure machine generally refers to a device used for manufacturing semiconductor devices, display screens or pattern transfer in a lithography process. The exposure machine exposes a predetermined pattern on the wafer surface by irradiating light through a mask onto a photoresist-coated wafer. The exposure machine is widely used in microelectronic manufacturing, especially in the lithography process of the semiconductor industry. The main function of the exposure machine is to irradiate light through a mask (usually a template containing a pattern) onto a photosensitive material (such as photoresist) in order to form a specific micro pattern on a substrate (such as a silicon wafer or a glass substrate). Specific functions include lithography, resolution control, and pattern transfer. Exposure machine failure can involve many aspects, including hardware failure, software failure, environmental factor changes, etc., resulting in various parameters not meeting expectations during the exposure process.

[0003] Exposure machine failure specifically affects the visibility of the exposure pattern. This includes problems such as blurring, mispositioning, stretching, and distortion of the formed pattern. Pattern visibility problems can be caused by many reasons, such as vibration failure caused by system wear, or mechanical vibration failure caused by unbalance or eccentricity of the stepping motor during operation. In order to achieve fault diagnosis, the actual pattern formation visibility problem needs to be identified, and the corresponding fault reason is comprehensively judged in combination with the actual device operation data, so as to eliminate the reason diagnosis deviation problem caused by multiple reasons resulting in a single effect. SUMMARY

[0004] The present application provides an exposure machine fault diagnosis method and system based on data processing to solve the existing problems.

[0005] The exposure machine fault diagnosis method based on data processing provided by the present application adopts the following technical scheme:

[0006] An embodiment of the present application provides an exposure machine fault diagnosis method based on data processing, which comprises the following steps:

[0007] Obtain the gray exposure image corresponding to the exposure image after imaging of the exposure machine and the vibration data in the exposure image formation process of each exposure image;

[0008] Perform Hough curve detection on the gray exposure image to obtain the number of curves in each gray exposure image and the voting number corresponding to each curve number value;

[0009] According to the curve number value and the voting number corresponding to each curve number value, calculate the visual exposure index of the gray exposure image;

[0010] performing two-dimensional fast Fourier transform on the gray-scale exposure image to obtain a frequency spectrum result;

[0011] calculating a frequency ratio of the gray-scale exposure image according to the frequency spectrum result;

[0012] determining an exposure forming index of the gray-scale exposure image according to the visual exposure index and the frequency ratio;

[0013] when the exposure forming index is greater than a preset fault threshold, determining that the exposure machine is in normal operation; when the exposure forming index is less than or equal to the preset fault threshold, calculating a vibration root cause index according to the vibration data, and calculating a wear vibration amount according to the vibration root cause index;

[0014] determining a fault cause of the exposure machine according to the wear vibration amount.

[0015] Optionally, the visual exposure index of the gray-scale exposure image is calculated according to the curve quantity value and the number of votes corresponding to each curve quantity value, and specifically includes:

[0016] for each gray-scale exposure image, summing the number of votes corresponding to the curve quantity value to obtain a line clarity degree of each gray-scale exposure image;

[0017] calculating a gray-scale entropy of each gray-scale exposure image using a gray-scale entropy calculation method;

[0018] taking the reciprocal of the gray-scale entropy of each gray-scale exposure image to obtain a chaos degree of each gray-scale exposure image;

[0019] the visual exposure index of each gray-scale exposure image is calculated using the product of the line clarity degree of each gray-scale exposure image, the chaos degree of each gray-scale exposure image and the curve quantity value of each gray-scale exposure image.

[0020] Optionally, the frequency ratio of the gray-scale exposure image is calculated according to the frequency spectrum result, and specifically includes:

[0021] determining a low-frequency energy and a high-frequency energy from the frequency spectrum result;

[0022] determining the ratio of the low-frequency energy and the high-frequency energy as the frequency ratio.

[0023] Optionally, the low-frequency energy and the high-frequency energy are determined from the frequency spectrum result, and specifically includes:

[0024] determining a frequency spectrum region within a preset amplitude range from the frequency spectrum result;

[0025] determining a low-frequency region as a region in which the frequency spectrum region is less than a preset amplitude threshold;

[0026] determining a high-frequency region as a region in which the frequency spectrum region is equal to or greater than the preset amplitude threshold.

[0027] The low-frequency energy is obtained by summing the amplitude values ​​in the low-frequency region.

[0028] The high-frequency energy is obtained by summing the amplitudes in the high-frequency region.

[0029] Optionally, the exposure shaping index of the grayscale image is determined based on the visual exposure index and frequency ratio, specifically including:

[0030] The ratio of visual exposure index to frequency ratio is calculated to obtain the initial exposure shaping index.

[0031] The initial exposure forming index is normalized to obtain the exposure forming index.

[0032] Optionally, vibration root cause indices are calculated based on vibration data, specifically including:

[0033] The vibration data were converted into vibration curves using the least squares method.

[0034] Determine the maximum and peak values ​​of the vibration curve, as well as the time difference sequence between adjacent maximum and peak values, from the vibration curve.

[0035] The root cause index of vibration is determined based on the maximum and peak values ​​and time difference sequence of the vibration curve.

[0036] Optionally, vibration root cause indices are determined based on the maximum and peak values ​​and time difference sequence of the vibration curve, specifically including:

[0037] Summing all elements in the time difference sequence yields the duration.

[0038] The LOF outlier algorithm is used to calculate the outlier value corresponding to the maximum peak value of each vibration curve;

[0039] The mean of the outliers corresponding to all the maximum peaks of each vibration curve is defined as the outlier.

[0040] The product of duration and outlier is calculated to obtain the vibration root cause index.

[0041] Optionally, the amount of wear vibration is calculated based on vibration root cause indices, specifically including:

[0042] Perform a discrete Fourier transform on the vibration data to obtain the discrete spectrum;

[0043] The total energy of the spectrum is obtained by integrating the energy values ​​at different frequencies in the discrete spectrum.

[0044] The ratio of the total spectral energy to the maximum energy component in the discrete spectrum is defined as the periodicity of the vibration.

[0045] The product of the vibration root cause index and the vibration periodicity is calculated and normalized to obtain the wear vibration amount.

[0046] Optionally, the fault cause of the exposure machine is determined according to the wear vibration amount, and specifically includes:

[0047] When the wear vibration amount is greater than or equal to a preset wear vibration amount, the fault cause is determined to be an exposure machine motion system fault.

[0048] When the wear vibration amount is less than the preset wear vibration amount, the fault cause is determined to be a motor fault.

[0049] The present application provides an exposure machine fault diagnosis system based on data processing, including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the computer program is executed by the processor, the steps of the exposure machine fault diagnosis method based on data processing are implemented.

[0050] The technical scheme of the present application has the following advantages:

[0051] In the embodiment of the present application, the visual analysis of the formed pattern is performed, the relevant discrimination index is established, and the real-time data of other dimensions are combined according to the index results to achieve the purpose of exposure machine fault anomaly monitoring. On the basis of ensuring exposure anomaly monitoring, the present application further analyzes the pattern forming visual anomaly type, combines the real-time sensor dimension data, establishes a relevant diagnosis decision tree, and comprehensively judges the exposure machine fault type to realize fault anomaly monitoring and improve the identification accuracy of the exposure machine fault type. In the process of comprehensively judging the fault type combined with the sensor dimension data, the vibration root cause index is different, and the identified fault type is different. In order to achieve accurate identification effect, the present application performs data analysis and detection on the vibration type of the exposure machine, and judges the real-time wear vibration amount through the monitoring result, and accurately obtains the corresponding red line discrimination value according to the vibration amount. BRIEF DESCRIPTION OF DRAWINGS

[0052] In order to more clearly illustrate the technical scheme in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiment or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0053] Figure 1 A flowchart of an exposure machine fault diagnosis method based on data processing provided by an embodiment of the present application;

[0054] Figure 2 A structural diagram of an exposure machine fault diagnosis system based on data processing provided by an embodiment of the present application. Detailed Implementation

[0055] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a data processing-based exposure machine fault diagnosis method proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0056] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0057] The following description, in conjunction with the accompanying drawings, details a specific scheme for a data processing-based exposure machine fault diagnosis method provided by the present invention.

[0058] This invention provides a data processing-based method and system for diagnosing exposure machine faults. Please refer to [link / reference]. Figure 1 The diagram illustrates a flowchart of a data processing-based exposure machine fault diagnosis method according to an embodiment of the present invention, the method comprising the following steps:

[0059] S101. Acquire the grayscale exposure image corresponding to the exposure image after imaging by the exposure machine, as well as the vibration data during the forming process of each exposure image.

[0060] For example, during the operation of the exposure machine, data acquisition devices and sensors are configured, mainly including two types: image data acquisition of the formed pattern and vibration data acquisition during the forming process. Specifically, after the film exit / development process of the exposure machine, exposure images are acquired using a high-resolution industrial camera or a microscopic imaging system; the acquisition parameters can be: resolution greater than or equal to 2... The sampling frequency is 5-10 images randomly sampled per batch, with consistent lighting conditions and magnification. Different patterns are tested once after molding. Output data includes the original grayscale image and the standardized image. For vibration data acquisition during the molding process, a triaxial accelerometer is installed on the motor mount, guide rail, or platform to collect vibration signals in real time during equipment operation. Acquisition parameters: sampling frequency greater than or equal to 10kHz, measurement range: 50g, sampling time covering one complete molding cycle; output data: original vibration waveform.

[0061] The quality of the exposure machine is generally monitored and identified to a certain extent by the direct production result, i.e., the quality of the exposure pattern. The main reason is that, in the exposure machine fault type, the abnormal vibration in the equipment (the cause is the unbalanced motion system), the result caused by the fault is that the abnormal vibration affects the stability of the exposure machine, and the fuzzy exposure pattern is generated; the uneven photoresist, due to the precision problem of the coating machine, the coating thickness of the photoresist is uneven, which causes the image quality difference in different areas during pattern transfer, forming a fuzzy effect; the mechanical positioning error, the fault of the stepping motor, the motor or part of the guide rail, the transmission shaft, will cause the inaccurate alignment of the exposure machine, thereby causing the pattern blur.

[0062] Therefore, the fault results of multiple causes are all directed to the visual effect of the final exposure pattern, including possible blur, stretching, deformation, etc.

[0063] Therefore, first, the pattern result is analyzed, and then, according to the analysis result, the related fault type diagnosis analysis is performed in combination with real-time multi-dimensional data. First, the exposure pattern result needs to be preliminarily analyzed and identified to obtain general indicators, i.e., exposure forming indicators, which are lower than a certain standard, and it is determined that the exposure fault needs to be diagnosed and analyzed. Secondly, in combination with the change characteristics of other dimensions of the real-time equipment, the corresponding diagnostic decision tree is established, and then the fault type of the exposure machine is determined.

[0064] Specifically, the image data of the formed pattern is obtained, and each forming result has a batch of imaging data results. The pattern image is a gray scale image. It should be noted that the blur, deformation or stretching property of the pattern is the main performance of the forming result after the fault of the exposure machine. In the equipment motion system, wear, aging and loosening occur during use, which will cause noise in the motion system, and then cause random irregular vibration of the equipment, resulting in a blurred pattern.

[0065] In addition, when the stepping motor is unbalanced or eccentric during work, periodic mechanical vibration (when the eccentricity occurs, the stepping motor generates a displacement eccentric vibration once per revolution), which affects the stability of the exposure machine, and the displacement deviation in the guide rail powered by the stepping motor, which causes the final pattern imaging result to have certain stretching and deformation characteristics.

[0066] That is, although the visual effect of the pattern exposure result is caused by vibration, the all-around irregular noise vibration caused by the aging of the components in the equipment and the periodic vibration caused by the eccentricity (abnormal positioning) of the stepping motor are different, and the fault reasons are also different, so they need to be distinguished.

[0067] Therefore, in the imaging gray pattern, if the blur of its visual effect is high, the corresponding exposure index is biased to the aging of the exposure machine motion system; on the contrary, if its blur is general, but the pattern produces obvious stretching or deformation, the exposure index is biased to the exposure machine stepping motor eccentricity (positioning abnormality) caused, and the diagnosis results are different.

[0068] S102, Hough curve detection is performed on the gray exposure image to obtain a curve number value in each gray exposure image and a voting number corresponding to each curve number value.

[0069] In this embodiment, the pattern gray imaging result is obtained, Hough curve (including Hough circle, straight line, etc.) detection is performed thereon, and the curve number value in the pattern is obtained The detection result of the first pattern forming and the corresponding Hough voting number (prior art) The voting number of the first straight line is represented), the more the number of straight lines, the higher the voting number, the higher the pattern definition, and the worse the stretching effect; in addition, the higher the definition, the lower the pixel gray entropy (presently calculated, which will not be described here) in the image, and the more disordered.

[0070] S103, calculating the visual exposure index of the gray exposure image according to the curve number value and the voting number corresponding to each curve number value.

[0071] In this embodiment, the visual exposure index of the gray exposure image is calculated according to the curve number value and the voting number corresponding to each curve number value, specifically including:

[0072] For each gray exposure image, the voting numbers corresponding to the curve number values are summed to obtain the line definition degree of each gray exposure image;

[0073] The gray entropy of each gray exposure image is calculated using the gray entropy calculation method;

[0074] The gray entropy of each gray exposure image is taken as the reciprocal to obtain the disorder degree of each gray exposure image;

[0075] The product of the line definition degree of each gray exposure image, the disorder degree of each gray exposure image, and the curve number value of each gray exposure image is used to calculate the visual exposure index of each gray exposure image.

[0076] According to the curve number value and the voting number corresponding to each curve number value, the visual exposure index of the gray exposure image is calculated, and the calculation method can be: ​​

[0077]

[0078] wherein, represents the visual exposure index of the th gray-scale exposure image, represents the curve number value of the th gray-scale exposure image, represents the gray-scale entropy of the th gray-scale exposure image, represents the voting number of each curve in the th gray-scale exposure image; represents the line clarity of the th gray-scale exposure image, represents the confusion degree of the th gray-scale exposure image.

[0079] In the formula, the higher the voting number, the higher the regularity and clarity of the detected line in the image, and the better the pattern effect generated by the corresponding exposure machine device; otherwise, the line clarity effect is weaker, the blur and irregularity are higher, and the possibility of exposure machine failure is higher. Further, the visual exposure index is obtained in combination with the frequency characteristics. The frequency characteristics can more quantitatively analyze the blur effect in the time domain in addition to the visual effect. When there are more blurred edges or blurred organizations in the time domain, i.e., in the image, the low-frequency energy proportion in the frequency domain is high because the high-frequency region is usually composed of regions with large gray-scale difference such as edges.

[0080] S104, performing two-dimensional fast Fourier transform on the gray-scale exposure image to obtain a frequency spectrum result.

[0081] S105, calculating a frequency ratio of the gray-scale exposure image according to the frequency spectrum result.

[0082] In this embodiment, the frequency ratio of the gray-scale exposure image is calculated according to the frequency spectrum result, specifically including: determining a low-frequency energy and a high-frequency energy from the frequency spectrum result; and determining the ratio of the low-frequency energy and the high-frequency energy as the frequency ratio.

[0083] The low-frequency energy and the high-frequency energy are determined from the frequency spectrum result, specifically including: determining a frequency spectrum region within a preset amplitude range from the frequency spectrum result; determining a region with a spectrum region less than a preset amplitude threshold as a low-frequency region; determining a region with a spectrum region equal to or greater than the preset amplitude threshold as a high-frequency region; summing the amplitudes in the low-frequency region to obtain the low-frequency energy; and summing the amplitudes in the high-frequency region to obtain the high-frequency energy.

[0084] For example, a two-dimensional FFT (Fast Fourier Transform) is performed on the pattern image to obtain a frequency spectrum result. Frequencies within the frequency spectrum are divided into three parts , and the frequencies in the first part are regarded as a low-frequency region, and the frequencies in the other two parts are regarded as high-frequency regions. Herein, the first part is 1 / 3 of the frequency range , and the preset amplitude threshold is set empirically and can be modified according to actual needs, and is not specifically limited herein. The sum of amplitudes corresponding to the low-frequency region is obtained as a low-frequency energy , and the other two parts are regarded as high-frequency energies .

[0085] The calculation method of the frequency ratio of the gray exposure image can be as follows:

[0086]

[0087] , wherein the frequency ratio of the i-th gray exposure image is represented by , the low-frequency energy is represented by , and the high-frequency energy is represented by . The greater the frequency ratio is, the higher the proportion of the low-frequency region is, and the higher the image blur effect is.

[0088] S106, determining an exposure forming property index of the gray exposure image according to the visual exposure index and the frequency ratio.

[0089] In this embodiment, the exposure forming property index of the gray exposure image is determined according to the visual exposure index and the frequency ratio, and specifically includes:

[0090] calculating the ratio of the visual exposure index and the frequency ratio to obtain an initial exposure forming property index;

[0091] normalizing the initial exposure forming property index to obtain the exposure forming property index.

[0092] The calculation method of the exposure forming property index can be as follows:

[0093]

[0094] , wherein the exposure forming property index of the i-th gray exposure image is represented by

[0095] . The higher the index is, the clearer the image is, and the higher the frequency ratio is, the more blurred the image is.

[0096] ​​​​​S107, when the exposure formability index is greater than the preset failure threshold, determining that the exposure machine is in normal operation; when the exposure formability index is less than or equal to the preset failure threshold, calculating a vibration root cause index according to the vibration data, and calculating a wear vibration amount according to the vibration root cause index.

[0097] In this embodiment, the vibration root cause index is calculated according to the vibration data, specifically including: using the least square method to convert the vibration data into a vibration curve; determining the maximum peak value of the vibration curve and the time difference sequence between adjacent maximum peak values from the vibration curve; determining the vibration root cause index according to the maximum peak value of the vibration curve and the time difference sequence.

[0098] According to the maximum peak value of the vibration curve and the time difference sequence, the vibration root cause index is determined, specifically including: summing all elements in the time difference sequence to obtain a duration; using the LOF outlier algorithm to calculate the outlier value corresponding to each maximum peak value of each vibration curve; the mean of the outlier values corresponding to all maximum peak values of each vibration curve is determined as the outlier; the product of the duration and the outlier is calculated to obtain the vibration root cause index.

[0099] According to the vibration root cause index, the wear vibration amount is calculated, specifically including: performing discrete Fourier transform on the vibration data to obtain a discrete frequency spectrum; integrating the energy values at different frequencies in the discrete frequency spectrum to obtain a total spectrum energy; the ratio of the total spectrum energy to the maximum energy component in the discrete frequency spectrum is determined as the vibration periodicity; the product of the vibration root cause index and the vibration periodicity is calculated, and normalized to obtain the wear vibration amount.

[0100] Exemplarily, the preset failure threshold is set to 0.5 here, the preset failure threshold here is set according to experience, which can be modified according to actual needs, and is not specifically limited here. When is in the range , it is considered that the exposure formability is blurred, at this time the exposure machine device may have a fault, which needs to be further analyzed; otherwise, when it is in the range , it is considered that the exposure formability is clear, and there is no fault diagnosis result.

[0101] Further, for the case that the exposure formability is blurred, it is necessary to further analyze the specific cause in combination with the running parameters in the actual device environment.

[0102] Specifically, when it is determined that the pattern forming result has a visual problem, through the analysis of the vibration logic in the foregoing, it may be caused by random vibration of the motion system inside the device, or it may be caused by periodic regular vibration of the eccentricity (abnormal positioning) of the stepping motor.

[0103] In addition, for the identification of photoresist unevenness, this part can be directly diagnosed by detecting the photoresist thickness, rather than requiring data analysis to distinguish and diagnose due to the complexity of vibration.

[0104] Therefore, the next step is to analyze real-time vibration data to obtain the corresponding root cause indicators and wear vibration amount.

[0105] Specifically, real-time vibration data collected by sensors is obtained to get the corresponding vibration set: ,in, Indicates the first A collection of vibration data of a grayscale exposure image during the forming process; This indicates the first [number]th ... Vibration value.

[0106] The moving platform, stepper motor, drive shaft, and guide rails of an exposure machine need to maintain high-precision movement. If the moving parts of these components are unbalanced, or if wear, aging, or loosening occurs during use, vibration will occur. This vibration will be irregular and random, resulting in blurred pattern images rather than stretching or deformation. Furthermore, if the stepper motor is eccentric during operation (positioning abnormality), it will cause mechanical vibration (when eccentricity occurs, the stepper motor generates a directional vibration with each revolution), thus affecting the stability of the exposure machine and producing periodic vibrations. The direct result of this is stretching and deformation of the pattern image. Therefore, the distinction between these two types of vibration lies in their randomness, periodicity, and continuity. We will first analyze the random effects of vibration, i.e., the root cause indicators.

[0107] Specifically, the sequence is transformed into an vibration curve using the least squares method, and the vibration data peaks (maximum peaks) of the vibration curve are obtained. (in, The more frequently the peak index changes, and the greater the time difference between peaks... The larger the value, the higher the root cause index of vibration, that is, the higher the randomness of vibration.

[0108] Based on the maximum and peak values ​​and time difference sequence of the vibration curve, the calculation method for determining the root cause index of vibration can be as follows:

[0109]

[0110] in, Indicates the first Vibration root cause index of grayscale exposure images Indicated by Peak As a sample, the mean of the outlier values ​​(LOF outlier algorithm) for each peak is calculated. The larger this value is, the greater the difference in variation between peaks and the higher the randomness. Represents the first in the time difference sequence One element, Indicates the duration.

[0111] When vibration data originates from the motion system within the exposure machine, including issues such as aging, the vibration changes are highly random. Vibration might be intense one second and slow down the next, with a significant difference in peak values ​​and a relatively long interval between peaks. Therefore, the root cause index value of vibration is obtained by quantifying the aforementioned peak value differences.

[0112] Furthermore, the higher the vibration root cause index, the higher the randomness, and the higher the probability of vibration caused by aging and wear of the motion system, and the higher the corresponding amount of wear vibration.

[0113] The calculation method for wear vibration can be as follows:

[0114]

[0115] in, Indicates the first The amount of wear and vibration in grayscale exposed images. Indicates the first Vibration root cause index of grayscale exposure images Represents the frequency values ​​in the discrete spectrum The energy value below, This represents the integral over the spectral energy, i.e., the total spectral energy. This represents the maximum energy component in the discrete spectrum. This represents the normalization formula.

[0116] The amount of wear vibration is directly proportional to the magnitude of the vibration root cause index; that is, the higher the randomness of vibration, the higher the amount of wear vibration. Simultaneously, the latter indicates that in the process of analyzing the randomness of vibration, the periodicity of vibration is obtained through spectrum analysis, and when the ratio... The smaller the value, the higher the proportion of the largest energy component in the total energy, and the higher the periodicity of the original sequence, because the frequency domain of the vibration data sequence is concentrated around a single data point. Therefore, the higher the wear vibration amount, the higher the probability of pattern exposure anomalies caused by motion system malfunctions; conversely, the smaller the wear vibration amount, the higher the probability of anomalies caused by motor eccentricity (positioning anomalies), because a smaller value means lower randomness and higher periodicity, as the energy is more concentrated around a single frequency.

[0117] S108. Determine the cause of the exposure machine malfunction based on the amount of wear and vibration.

[0118] In the embodiment, the failure cause of the exposure machine is determined according to the wear vibration amount, and specifically includes: when the wear vibration amount is greater than or equal to a preset wear vibration amount, the failure cause is determined as a failure of a motion system of the exposure machine; and when the wear vibration amount is less than the preset wear vibration amount, the failure cause is determined as a motor failure.

[0119] Exemplarily, the vibration root cause index can be solved by establishing a decision tree. The decision tree is established, that is, when the wear vibration amount is greater than or equal to a preset wear vibration amount (0.7, which can also be referred to as a red line discrimination value), the decision tree considers that the failure cause is in the motion system of the exposure machine; otherwise, when the wear vibration amount is less than the preset wear vibration amount, the decision tree considers that the failure cause is in the eccentricity of the stepping motor (positioning abnormality). The preset wear vibration amount is an actual value calculated according to historical experience, and can be changed according to actual requirements, and is not limited in specific numerical values. The above failure identification and determination is based on the existence of the failure, that is, based on the satisfaction of the exposure forming index discrimination threshold.

[0120] Thus, the present application is completed.

[0121] In summary, in the embodiment of the present application, the exposure machine failure abnormality monitoring purpose is achieved by visually analyzing the formed pattern, establishing related discrimination indexes, and combining other dimension real-time data according to the index results. On the basis of ensuring exposure abnormality monitoring, the present application further analyzes the pattern forming visual abnormality type, combines the real-time sensor dimension data, establishes a related diagnosis decision tree, comprehensively judges the exposure machine failure type, realizes the failure abnormality monitoring, and improves the identification precision of the exposure machine failure type. In the process of comprehensively judging the failure type in combination with the sensor dimension data, the vibration root cause index is different, and the identified failure type is different. In order to achieve accurate identification effect, the present application analyzes and detects the vibration type of the exposure machine, and judges the real-time wear vibration amount through the monitoring result, and accurately obtains the corresponding red line discrimination value according to the vibration amount.

[0122] The present application also provides an exposure machine failure diagnosis system based on data processing, please refer to Figure 2 which shows a structure diagram of an exposure machine failure diagnosis system based on data processing provided by an embodiment of the present application, and the system includes a data acquisition module 101, a data processing module 102 and a failure detection module 103.

[0123] The data acquisition module 101 is used for acquiring the gray exposure image corresponding to the exposure image after the exposure machine imaging and the vibration data in the forming process of each image;

[0124] ​The data processing module 102 is configured to perform Hough curve detection on the gray exposure image to obtain a curve quantity value of each gray exposure image and a voting number corresponding to each curve quantity value; calculate a visual exposure index of the gray exposure image according to the curve quantity value and the voting number corresponding to each curve quantity value; perform two-dimensional fast Fourier transform on the gray exposure image to obtain a frequency spectrum result; calculate a frequency ratio of the gray exposure image according to the frequency spectrum result; determine an exposure forming property index of the gray exposure image according to the visual exposure index and the frequency ratio; when the exposure forming property index is greater than a preset fault threshold, determine that the exposure machine is in normal operation; when the exposure forming property index is less than or equal to the preset fault threshold, calculate a vibration root cause index according to the vibration data, and calculate a wear vibration amount according to the vibration root cause index.

[0125] The fault detection module 103 is configured to determine a fault cause of the exposure machine according to the wear vibration amount.

[0126] It should be noted that the system provided in the above embodiments is only exemplified by the division of the above functional modules, and in actual application, the above functions can be completed by different functional modules according to needs, that is, the internal structure of the computer device is divided into different functional modules to complete all or part of the functions described above. In addition, the exposure machine fault diagnosis system and the exposure machine fault diagnosis method provided in the above embodiments belong to the same concept, and the specific implementation process is described in the method embodiments, which will not be repeated here.

[0127] It should be noted that the above-mentioned embodiments of the present application are only for description, and do not represent the advantages and disadvantages of the embodiments. The processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are also possible or may be advantageous.

[0128] Each embodiment in the specification is described in a progressive manner, and the same or similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

[0129] The above only describes the preferred embodiments of the present application and does not limit the present application. Any modification, equivalent replacement, improvement, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A data processing-based fault diagnosis method for an exposure machine, characterized by, The method comprises the following steps: acquiring a gray exposure image corresponding to an exposure image after imaging of an exposure machine and vibration data in the exposure process of each exposure image; performing Hough curve detection on the gray exposure image to obtain a curve number value in each gray exposure image and a voting number corresponding to each curve number value; for each gray exposure image, summing the voting numbers corresponding to the curve number values to obtain a line clarity degree of each gray exposure image; calculating a gray entropy of each gray exposure image using a gray entropy calculation method; taking the reciprocal of the gray entropy of each gray exposure image to obtain a confusion degree of each gray exposure image; calculating a visual exposure index of each gray exposure image using the product of the line clarity degree of each gray exposure image, the confusion degree of each gray exposure image and the curve number value of each gray exposure image; performing two-dimensional fast Fourier transform on the gray exposure image to obtain a frequency spectrum result; calculating a frequency ratio of the gray exposure image according to the frequency spectrum result; calculating a ratio of the visual exposure index and the frequency ratio to obtain an initial exposure forming index; normalizing the initial exposure forming index to obtain an exposure forming index; when the exposure forming index is greater than a preset fault threshold, determining that the exposure machine is in normal operation; when the exposure forming index is less than or equal to the preset fault threshold, using a least square method to convert the vibration data into a vibration curve; determining a maximum peak value of the vibration curve and a time difference sequence between adjacent maximum peak values from the vibration curve; determining a vibration root cause index according to the maximum peak value of the vibration curve and the time difference sequence; performing discrete Fourier transform on the vibration data to obtain a discrete frequency spectrum; integrating energy values at different frequencies in the discrete frequency spectrum to obtain a total spectrum energy; determining a vibration periodicity as a ratio of the total spectrum energy to a maximum energy component in the discrete frequency spectrum; calculating a product of the vibration root cause index and the vibration periodicity and normalizing to obtain a wear vibration amount; determining a fault cause of the exposure machine according to the wear vibration amount.

2. The exposure apparatus failure diagnosing method based on data processing according to claim 1, characterized by, The calculation of the frequency ratio of the gray exposure image according to the frequency spectrum result specifically comprises: determining a low-frequency energy and a high-frequency energy from the frequency spectrum result; determining the frequency ratio as a ratio of the low-frequency energy and the high-frequency energy.

3. The exposure apparatus failure diagnosing method based on data processing according to claim 2, characterized by, The determination of the low-frequency energy and the high-frequency energy from the frequency spectrum result specifically comprises: determining a frequency spectrum region within a preset amplitude range from the frequency spectrum result; determining a low-frequency region as a region in which the frequency spectrum region is less than a preset amplitude threshold; determining a high-frequency region as a region in which the frequency spectrum region is equal to or greater than the preset amplitude threshold; summing amplitudes in the low-frequency region to obtain the low-frequency energy; summing amplitudes in the high-frequency region to obtain the high-frequency energy.

4. The exposure apparatus failure diagnosing method based on data processing according to one of claims 1, wherein The determination of the vibration root cause index according to the maximum peak value of the vibration curve and the time difference sequence specifically comprises: summing all elements in the time difference sequence to obtain a duration. The LOF outlier algorithm is used to calculate the outlier value corresponding to the maximum peak value of each vibration curve; The mean of the outlier values corresponding to all the maximum peak values of each vibration curve is determined as the outlier; The product of the duration and the outlier is calculated to obtain a vibration root cause index.

5. The exposure apparatus failure diagnosing method based on data processing according to one of claims 1, wherein The determination of the failure cause of the exposure machine according to the wear vibration quantity specifically comprises: When the wear vibration quantity is greater than or equal to a preset wear vibration quantity, it is determined that the failure cause is an exposure machine motion system failure; When the wear vibration quantity is less than the preset wear vibration quantity, it is determined that the failure cause is a motor failure.

6. A data processing-based fault diagnosis system for an exposure machine, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized by The computer program, when executed by a processor, implements the steps of the data processing-based exposure machine failure diagnosis method according to any one of claims 1-5.

Images