Optical glass plate thickness detection method and system

Abstract

The present invention belongs to the field of precision detection technology, and discloses an optical glass plate thickness detection method and system, including collecting optical glass plate parameter data, detection equipment parameter data and detection environment data; preprocessing the collected optical glass plate parameter data, detection equipment parameter data and detection environment data to obtain an optical glass plate feature data set, a detection equipment feature data set and a detection environment feature data set; analyzing the influence of the detection environment feature data set on the optical glass plate feature data set and the detection equipment feature data set to obtain an influence feature data set; training and acquiring a thickness detection error prediction model according to the influence feature data set, and predicting the optical glass plate thickness detection error through the thickness detection error prediction model; comparing the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard; and facilitating improving the accuracy of thickness detection.

Description

Technical Field

[0001] The present invention relates to the field of precision detection technology, and more specifically, to a method and system for detecting the thickness of an optical glass plate. Background Art

[0002] The patent application publication number CN102353915A discloses a device and method for measuring magnetostriction coefficient by using Doppler galvanometer sinusoidally modulated multi-beam laser heterodyne, which belongs to the field of micro-displacement detection technology. It solves the problem that the measurement accuracy of the existing magnetostriction coefficient measurement method cannot meet the requirements of ultra-precision measurement. The present invention makes the plane reflector and the thin glass plate of no thickness parallel to each other and equal in height, and makes the distance d between the thin glass plate of no thickness and the reflecting surface of the plane reflector mm, turns on the driving power of the galvanometer to make the galvanometer start to perform simple harmonic vibration; collects the electrical signal output by the photoelectric detector through the signal processing system, and processes the collected signal to obtain the distance change between the plane reflector and the thin glass plate of no thickness, and obtains the magnetostriction coefficient of the iron-nickel alloy sample to be measured according to the distance change. The present invention is suitable for measuring the magnetostriction coefficient of ferromagnetic bodies.

[0003] The existing optical glass plate thickness detection method and system have the following main problems:

[0004] Traditional methods are difficult to accurately simulate the complex behavior of light in different layers of materials, resulting in inaccurate calculation results of transmittance, refractive index and reflectivity, which affects the reliability of the overall detection and analysis results; the restriction on refractive index is not considered, and when simulating the refraction of light in different media, unreasonable calculation results of refractive index may be caused, which cannot truly reflect the actual performance of optical glass plates under complex conditions; it may cause the simulation results to deviate from the physical reality, thus affecting the subsequent calculation of optical properties; traditional analysis methods cannot capture small surface errors in detail and accurately; such error accumulation will affect the overall quality assessment of optical glass plates, and it is impossible to obtain true flatness results; fitting calculations based on mathematical models are not used, and the high-precision fitting capabilities of Zernike polynomials cannot be fully utilized, resulting in low calculation accuracy, which affects the assessment of the surface flatness of optical glass plates;

[0005] If the F statistic and critical value are not used to judge significance, the analysis process may rely on subjective inference, which may easily lead to deviations and inconsistencies in the analysis results; there is a lack of data-driven significance judgment basis, the scientificity and reliability of the analysis results are low, and it is impossible to rigorously verify the impact of different testing environments on the characteristics of optical glass plates and equipment; there is no dynamic adjustment of the critical value based on variables such as the type of feature data set and the number of samples, and the analysis method may be too rigid; the existing technology does not take into account the diversity and complexity of the data, resulting in a mismatch between the critical value and the actual situation, and cannot effectively distinguish between significant and non-significant effects, thereby reducing the accuracy of the analysis.

[0006] In view of this, the present invention provides a method and system for detecting the thickness of an optical glass plate to solve the above-mentioned problem. Summary of the invention

[0007] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for detecting the thickness of an optical glass plate, comprising:

[0008] S1. Collecting optical glass plate parameter data, testing equipment parameter data and testing environment data;

[0009] S2, preprocessing the collected optical glass plate parameter data, testing equipment parameter data and testing environment data to obtain an optical glass plate feature data set, a testing equipment feature data set and a testing environment feature data set; analyzing the influence of the testing environment feature data set on the optical glass plate feature data set and the testing equipment feature data set to obtain an influence feature data set;

[0010] S3, training a thickness detection error prediction model according to the influencing feature data set, and predicting the thickness detection error of the optical glass plate through the thickness detection error prediction model;

[0011] S4, comparing the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard;

[0012] S5. If the thickness of the glass plate meets the standard, the thickness detection intelligent terminal generates a qualified instruction, indicating that the glass plate thickness detection is qualified; if the thickness of the glass plate does not meet the standard, the thickness detection intelligent terminal generates a failed instruction, generates a warning message, and automatically collects the thickness data of the glass plate that does not meet the standard;

[0013] S6. Construct a thickness non-compliance diagnosis model, input the non-compliance glass plate thickness data into the thickness non-compliance diagnosis model, and predict the non-compliance type of the glass plate thickness; take corresponding improvement measures according to the non-compliance type of the glass plate thickness until the glass plate thickness meets the standard.

[0014] Furthermore, the optical glass plate parameter data includes the surface flatness, transmittance, refractive index and reflectivity of the optical glass plate; the detection equipment parameter data includes the model of the detection equipment, measurement accuracy, vibration frequency of the equipment, detection speed and detection time; the detection environment data includes the detection environment temperature, humidity, light intensity and dust concentration in the air.

[0015] Furthermore, the method of preprocessing the collected optical glass plate parameter data, detection equipment parameter data and detection environment data to obtain the optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set includes:

[0016] S31. For the optical glass plate parameter data: by using the improved multi-layer Monte Carlo ray tracing algorithm, the propagation behavior of light in the multi-layer medium of the optical glass plate is simulated. By introducing the multi-layer medium model, the refraction and reflection of light in each layer of the optical glass plate medium are accurately simulated to obtain the transmittance, refractive index and reflectivity of the optical glass plate;

[0017] S311, calculated by Snell's law formula, the light rays are The layer of medium enters the optical glass plate The refractive index of the layer medium;

[0018] The Snell's law formula is: ;in, For the light Refractive index in the optical glass plate; For light to enter The refractive index of the optical glass plate; For the light The angle of incidence in the optical glass plate; For the light Refraction angle in a layer of optical glass; is the index of the number of layers of the optical glass plate;

[0019] S312, through the refractive index limiting formula, the light The refractive index in the optical glass plate is limited; the refractive index limiting formula is: ;in, is the speed of light in air; is the propagation speed of light in the optical glass plate; is the total number of layers of optical glass plates; is a constant that adjusts the restriction strength;

[0020] S313, calculating the transmittance and reflectance of light in the optical glass plate by using the Fresnel equation; the Fresnel equation is: ;in, For the light Reflectivity in the optical glass sheet;

[0021] Subtract the reflectivity of light in the optical glass plate from 1 to get the transmittance of light in the optical glass plate. ;

[0022] S314, fitting the surface flatness of the optical glass plate by using Zernike polynomials;

[0023] The Zernike polynomial is: ;in, is the standard form of the polynomial; is the order of the polynomial; is the frequency of the polynomial; is the normalized radius, that is, the relative distance from a point on the surface of the optical glass plate to the geometric center of the glass plate; is the Zernike radial polynomial; is the periodic changing complex exponent of the Zernike polynomial in the azimuth direction; is the phase angle change of the azimuth; is the azimuth of a point on the optical surface; is an imaginary unit;

[0024] S315, normalize the radius by using the radius adjustment formula Dynamic adjustment is performed; the radius adjustment formula is: ;in, It is the actual distance from a point on the surface of the optical glass plate to the geometric center of the glass plate; is the radius of the optical glass plate; To control the azimuth angle to the normalized radius Adjustment factor affecting strength;

[0025] S316, constructing a surface morphology error model, and fitting the error of the optical glass plate surface by using Zernike polynomials; the surface morphology error model is: ;in, is the morphological error of the optical glass plate surface; is the fitting coefficient of Zernike polynomial;

[0026] Fitting coefficients of Zernike polynomials The morphological error of the optical glass surface The fitting coefficients obtained by least squares fitting are , calculate the surface flatness of the optical glass plate through the flatness calculation formula;

[0027] The flatness calculation formula is: ;in, is the surface flatness of the optical glass plate;

[0028] S32, for the detection equipment parameter data and the detection environment data: identifying and extracting features in the detection equipment parameter data and the detection environment data by principal component analysis to obtain a detection environment feature data set and a detection equipment feature data set;

[0029] S33, detecting and eliminating abnormal data in the optical glass plate parameter data, the detection equipment parameter data, and the detection environment data by a box plot method, to obtain a processed optical glass plate feature data set, a detection equipment feature data set, and a detection environment feature data set;

[0030] The processed optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set are normalized and converted into a standard normal distribution to obtain a normalized processed optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set.

[0031] Furthermore, the method of analyzing the influence of the detection environment characteristic data set on the optical glass plate characteristic data set and the detection equipment characteristic data set to obtain the influenced characteristic data set includes:

[0032] S41. Preset two significant judgment hypotheses, including a null hypothesis and an alternative hypothesis; the null hypothesis indicates that the detection environment feature data set has no significant effect on the optical glass plate feature data set and the detection equipment feature data set; the alternative hypothesis indicates that the detection environment feature data set has a significant effect on the optical glass plate feature data set and the detection equipment feature data set;

[0033] S42, collecting an optical glass plate feature data set, a detection equipment feature data set, and a detection environment feature data set to form a sample set, grouping the optical glass plate features and the detection equipment features according to different detection environment features, and calculating the intra-group variance and inter-group variance of each group;

[0034] The within-group variance calculation formula is: ;in, is the within-group variance; is the total sample size; is the number of groups; For the The sample size of the group; For the Group Sample values; For the The mean of the group; is the index of the group; is the index of the sample;

[0035] The inter-group variance calculation formula is: ;in, is the between-group variance; is the total mean of all groups;

[0036] S43, dividing the inter-group variance by the intra-group variance to obtain the calculated F statistic, and comparing the F statistic with the critical value to determine whether the detection environment characteristics have a significant impact on the optical glass plate characteristics and the detection equipment characteristics; the critical value is calculated as follows: ;in, is the critical value; is the feature type of the feature data set;

[0037] S44. If the F statistic is greater than the critical value , then the preset null hypothesis is not established, and the detection environment feature data set has a significant impact on the optical glass plate feature data set and the detection equipment feature data set;

[0038] If the F statistic is less than or equal to the critical value , then the pre-equipment selection hypothesis does not hold, and the detection environment feature data set has no significant impact on the optical glass plate feature data set and the detection equipment feature data set;

[0039] The data collection of the detection environment characteristic data set that has a significant impact on the optical glass plate characteristic data set and the detection equipment characteristic data set is statistically analyzed and integrated to obtain the impact characteristic data set.

[0040] Furthermore, the training method of the thickness detection error prediction model includes:

[0041] The data set is divided into a training set, a validation set and a test set, and a thickness detection error prediction model is constructed; the sample set is a subset of the data set, and each sample set includes a historical impact feature data set and the corresponding optical glass plate thickness detection error;

[0042] The thickness detection error prediction model includes input data and output labels; the input data of the model is a historical impact feature data set, and the output label of the model is an optical glass plate thickness detection error; RBF is selected as the kernel function of the model, and the hyperparameters of the model are initialized. The hyperparameters of the model include a regularization parameter C and a γ value of the RBF kernel function; the thickness detection error prediction model is a support vector machine model;

[0043] Use mean absolute error as the loss function to measure the error between the model's predicted value and the actual value; use the training set data to train the model and use the Adam optimizer to minimize the loss function; use the validation set to evaluate the performance of the model and tune the model's hyperparameters until the model performance no longer improves significantly or reaches the preset stopping condition;

[0044] The test set is used to evaluate the performance of the model in the prediction task, and the current influencing feature data set is input into the trained optical glass plate thickness detection error to obtain the optical glass plate thickness detection error.

[0045] Furthermore, the method of comparing the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard includes:

[0046] If the predicted optical glass plate thickness detection error is less than the preset optical glass plate thickness detection error threshold, it is determined that the glass plate thickness meets the standard;

[0047] If the predicted optical glass plate thickness detection error is greater than or equal to a preset optical glass plate thickness detection error threshold, it is determined that the glass plate thickness does not meet the standard.

[0048] Furthermore, the substandard glass plate thickness data includes an actual thickness value of the glass plate, a location where the thickness does not meet the standard, a glass plate detection time, a surface flatness of the glass plate, an optical transmittance, an optical refractive index, and a reflectivity of the glass plate.

[0049] Furthermore, the method for constructing the thickness non-standard diagnosis model includes:

[0050] The training data set is divided into a training set, a validation set, and a test set to construct a thickness non-compliance diagnosis model; the sample set is a subset of the data set, and each sample set includes historical non-compliance glass plate thickness data and the corresponding non-compliance type of glass plate thickness;

[0051] The thickness non-compliance diagnosis model includes an input layer, a self-attention layer, a feedforward network layer and an output layer; the input layer of the model is used to input historical non-compliance glass plate thickness data, and the output layer of the model is used to output the non-compliance type of glass plate thickness; a softmax function is used as an activation function; the thickness non-compliance diagnosis model is a self-attention mechanism model;

[0052] Use multi-classification cross entropy as the model's loss function to measure the difference between the model's predicted value and the actual value; use the training set to train the thickness substandard diagnosis model, and update the model parameters through the back propagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the thickness substandard diagnosis model by calculating the accuracy index;

[0053] The SGD optimization algorithm was selected as the optimizer, and the model was tuned according to the performance feedback of the validation set. The model parameters were adjusted until the performance no longer improved significantly or the preset number of iterations was reached. The test set was used to evaluate the performance of the model in the prediction task, and the trained thickness non-compliance diagnosis model was used to predict the current non-compliant glass plate thickness data to obtain the non-compliant glass plate thickness type.

[0054] Furthermore, the method of taking corresponding improvement measures according to the type of glass sheet thickness that does not meet the standard until the glass sheet thickness meets the standard includes:

[0055] The types of glass plate thickness that do not meet the standards include glass plate thickness thinner than the standard glass plate thickness, glass plate thickness thicker than the standard glass plate thickness and thickness unevenness rate; for glass plate thickness thinner than the standard glass plate thickness, improvement measures include adjusting the raw material formula and optimizing production process parameters; for glass plate thickness thicker than the standard glass plate thickness, improvement measures include reducing the amount of raw materials used and adjusting the pressure and time parameters of the pressing process; for thickness unevenness rate, improvement measures include improving mold design and optimizing production process parameters.

[0056] An optical glass plate thickness detection system, comprising:

[0057] A data collection module, used to collect optical glass plate parameter data, testing equipment parameter data and testing environment data;

[0058] The data processing module is used to pre-process the collected optical glass plate parameter data, the detection equipment parameter data and the detection environment data to obtain the optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set; analyze the influence of the detection environment feature data set on the optical glass plate feature data set and the detection equipment feature data set to obtain the influence feature data set;

[0059] A detection error prediction module is used to obtain a thickness detection error prediction model based on the influencing feature data set training, and to predict the thickness detection error of the optical glass plate through the thickness detection error prediction model;

[0060] A glass plate thickness assessment module is used to compare the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard;

[0061] Thickness detection warning module: if the thickness of the glass plate meets the standard, the thickness detection intelligent terminal generates a qualified instruction, prompting that the glass plate thickness detection is qualified; if the thickness of the glass plate does not meet the standard, the thickness detection intelligent terminal generates a failed instruction, generates a warning message, and automatically collects the thickness data of the glass plate that does not meet the standard;

[0062] The thickness control module is used to construct a thickness non-compliance diagnosis model, input the non-compliance glass plate thickness data into the thickness non-compliance diagnosis model, and predict the non-compliance type of glass plate thickness; take corresponding improvement measures according to the non-compliance type of glass plate thickness until the glass plate thickness meets the standard.

[0063] The technical effects and advantages of the optical glass plate thickness detection method and system of the present invention are as follows:

[0064] By using the improved multi-layer Monte Carlo ray tracing algorithm, the refraction and reflection behavior of light in the multi-layer medium of the optical glass plate can be accurately simulated. Through this simulation, the transmittance, refractive index and reflectivity of the optical glass plate can be calculated more accurately to ensure the high accuracy and reliability of the measurement data; through Snell's law and the refractive index limit formula, the refractive index of light in different media can be dynamically adjusted; the refractive index limit can better reflect the actual performance of the optical glass plate under different conditions, thereby avoiding unreasonable physical phenomena and improving the physical authenticity of the simulation; Zernike polynomials are an effective tool for analyzing and fitting optical surface morphological errors; through this method, the errors on the surface of the optical glass plate can be fitted in detail, and then the surface flatness can be calculated through the fitting coefficient; this flatness calculation method based on mathematical models is more accurate than traditional measurement methods;

[0065] Using F statistics and critical values ​​to determine significance avoids subjective inference and provides a basis for judgment based on data; through variance analysis, the specific impact of the test environment characteristics on the characteristics of the optical glass plate and the test equipment characteristics can be determined more accurately, improving the scientific nature of the analysis results; the calculation formula of the critical value is dynamically adjusted according to the feature type of the feature data set, which provides flexibility for different types of data. The critical value can reflect the changes in variables such as feature type and sample size, making the analysis process more accurate and practical; through significance analysis, it can be found which test environment characteristics have a significant impact on the characteristics of the optical glass plate and equipment, helping to optimize the test process and equipment settings, thereby improving product quality, reducing errors and improving test efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 It is a schematic diagram of a process of detecting the thickness of an optical glass plate of the present invention;

[0067] Figure 2 The figure is a schematic structural diagram of an optical glass plate thickness detection system of the present invention. DETAILED DESCRIPTION

[0068] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only 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.

[0069] Example 1: Please refer to Figure 1 As shown, this embodiment provides a method for detecting the thickness of an optical glass plate, comprising:

[0070] S1. Collecting optical glass plate parameter data, testing equipment parameter data and testing environment data;

[0071] S2, preprocessing the collected optical glass plate parameter data, testing equipment parameter data and testing environment data to obtain an optical glass plate feature data set, a testing equipment feature data set and a testing environment feature data set; analyzing the influence of the testing environment feature data set on the optical glass plate feature data set and the testing equipment feature data set to obtain an influence feature data set;

[0072] S3, training a thickness detection error prediction model according to the influencing feature data set, and predicting the thickness detection error of the optical glass plate through the thickness detection error prediction model;

[0073] S4, comparing the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard;

[0074] S5. If the thickness of the glass plate meets the standard, the thickness detection intelligent terminal generates a qualified instruction, indicating that the glass plate thickness detection is qualified; if the thickness of the glass plate does not meet the standard, the thickness detection intelligent terminal generates a failed instruction, generates a warning message, and automatically collects the thickness data of the glass plate that does not meet the standard;

[0075] S6. Construct a thickness non-compliance diagnosis model, input the non-compliance glass plate thickness data into the thickness non-compliance diagnosis model, and predict the non-compliance type of the glass plate thickness; take corresponding improvement measures according to the non-compliance type of the glass plate thickness until the glass plate thickness meets the standard.

[0076] The parameter data of optical glass plates include the surface flatness, transmittance, refractive index and reflectivity of the optical glass plates; the parameter data of testing equipment include the model of the testing equipment, measurement accuracy, vibration frequency of the equipment, testing speed and testing time; the testing environment data include the testing environment temperature, humidity, light intensity and dust concentration in the air.

[0077] The method of preprocessing the collected optical glass plate parameter data, the detection equipment parameter data and the detection environment data to obtain the optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set includes:

[0078] S31. For the optical glass plate parameter data: by using the improved multi-layer Monte Carlo ray tracing algorithm, the propagation behavior of light in the multi-layer medium of the optical glass plate is simulated. By introducing the multi-layer medium model, the refraction and reflection of light in each layer of the optical glass plate medium are accurately simulated to obtain the transmittance, refractive index and reflectivity of the optical glass plate;

[0079] S311, calculated by Snell's law formula, the light rays are The layer of medium enters the optical glass plate The refractive index of the layer medium;

[0080] The formula for Snell's law is: ;in, For the light Refractive index in the optical glass plate; For light to enter The refractive index of the optical glass plate; For the light The angle of incidence in the optical glass plate; For the light Refraction angle in a layer of optical glass; is the index of the number of layers of the optical glass plate;

[0081] S312, through the refractive index limiting formula, the light The refractive index in the optical glass plate is limited; the refractive index limitation formula is: ;in, is the speed of light in air; is the propagation speed of light in the optical glass plate; is the total number of layers of optical glass plates; is a constant that adjusts the restriction strength;

[0082] For example, the total number of layers of optical glass plates is 4, and the speed of light propagation in the air is m / s, assuming that the propagation speed of light in the first layer of the optical glass plate is m / s, and the propagation speed of the second layer is m / s, and the propagation speed of the third layer is m / s, and the propagation speed of the fourth layer is m / s, constant for adjusting the limiting strength is 2;

[0083] The refractive index of each layer is calculated using the refractive index limiting formula:

[0084] The refractive index of light in the first layer of optical glass plate is: ;

[0085] The refractive index of light in the second layer of the optical glass plate is: ;

[0086] The refractive index of light in the third layer of the optical glass plate is: ;

[0087] The refractive index of light in the fourth layer of the optical glass plate is: ;

[0088] S313, calculating the transmittance and reflectance of light in the optical glass plate by using the Fresnel equation; the Fresnel equation is: ;in, For the light Reflectivity in the optical glass layer;

[0089] Subtract the reflectivity of light in the optical glass plate from 1 to get the transmittance of light in the optical glass plate. ;

[0090] S314, fitting the surface flatness of the optical glass plate by using Zernike polynomials;

[0091] The Zernike polynomials are: ;in, is the standard form of the polynomial; is the order of the polynomial; is the frequency of the polynomial; is the normalized radius, that is, the relative distance from a point on the surface of the optical glass plate to the geometric center of the glass plate; is the Zernike radial polynomial; is the periodic changing complex exponent of the Zernike polynomial in the azimuth direction; is the phase angle change of the azimuth; is the azimuth of a point on the optical surface; is an imaginary unit;

[0092] S315, normalize the radius by using the radius adjustment formula Perform dynamic adjustment; the radius adjustment formula is: ;in, It is the actual distance from a point on the surface of the optical glass plate to the geometric center of the glass plate; is the radius of the optical glass plate; To control the azimuth angle to the normalized radius Adjustment factor affecting strength;

[0093] S316, constructing a surface morphology error model, and fitting the error of the optical glass plate surface by using Zernike polynomials; the surface morphology error model is: in, is the morphological error of the optical glass plate surface; is the fitting coefficient of Zernike polynomial;

[0094] Fitting coefficients of Zernike polynomials The morphological error of the optical glass surface The fitting coefficients obtained by least squares fitting are , calculate the surface flatness of the optical glass plate through the flatness calculation formula;

[0095] The flatness calculation formula is: ;in, is the surface flatness of the optical glass plate;

[0096] S32, for the detection equipment parameter data and the detection environment data: identifying and extracting features in the detection equipment parameter data and the detection environment data by principal component analysis to obtain a detection environment feature data set and a detection equipment feature data set;

[0097] S33, detecting and eliminating abnormal data in the optical glass plate parameter data, the detection equipment parameter data, and the detection environment data by a box plot method, to obtain a processed optical glass plate feature data set, a detection equipment feature data set, and a detection environment feature data set;

[0098] The processed optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set are normalized and converted into a standard normal distribution to obtain a normalized processed optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set.

[0099] The method of analyzing the influence of the detection environment characteristic data set on the optical glass plate characteristic data set and the detection equipment characteristic data set to obtain the influenced characteristic data set includes:

[0100] S41. Preset two significant judgment hypotheses, including a null hypothesis and an alternative hypothesis; the null hypothesis indicates that the detection environment feature data set has no significant effect on the optical glass plate feature data set and the detection equipment feature data set; the alternative hypothesis indicates that the detection environment feature data set has a significant effect on the optical glass plate feature data set and the detection equipment feature data set;

[0101] S42, collecting an optical glass plate feature data set, a detection equipment feature data set, and a detection environment feature data set to form a sample set, grouping the optical glass plate features and the detection equipment features according to different detection environment features, and calculating the intra-group variance and inter-group variance of each group;

[0102] The within-group variance calculation formula is: ;in, is the within-group variance; is the total sample size; is the number of groups; For the The sample size of the group; For the Group Sample values; For the The mean of the group; is the index of the group; is the index of the sample;

[0103] The inter-group variance calculation formula is: ;in, is the between-group variance; is the total mean of all groups;

[0104] S43, dividing the inter-group variance by the intra-group variance to obtain the calculated F statistic, and comparing the F statistic with the critical value to determine whether the detection environment characteristics have a significant impact on the optical glass plate characteristics and the detection equipment characteristics; the critical value is calculated as follows: ;in, is the critical value; is the feature type of the feature data set;

[0105] S44. If the F statistic is greater than the critical value , then the preset null hypothesis is not established, and the detection environment feature data set has a significant impact on the optical glass plate feature data set and the detection equipment feature data set;

[0106] If the F statistic is less than or equal to the critical value , then the pre-equipment selection hypothesis does not hold, and the detection environment feature data set has no significant impact on the optical glass plate feature data set and the detection equipment feature data set;

[0107] The data collection of the detection environment characteristic data set that has a significant impact on the optical glass plate characteristic data set and the detection equipment characteristic data set is statistically analyzed and integrated to obtain the impact characteristic data set.

[0108] The training method of the thickness detection error prediction model includes:

[0109] The data set is divided into a training set, a validation set and a test set, and a thickness detection error prediction model is constructed; the sample set is a subset of the data set, and each sample set includes a historical impact feature data set and the corresponding optical glass plate thickness detection error;

[0110] The thickness detection error prediction model includes input data and output labels; the input data of the model is a historical impact feature data set, and the output label of the model is an optical glass plate thickness detection error; RBF is selected as the kernel function of the model, and the hyperparameters of the model are initialized. The hyperparameters of the model include a regularization parameter C and a γ value of the RBF kernel function; the thickness detection error prediction model is a support vector machine model;

[0111] The mean absolute error is used as the loss function to measure the error between the model's predicted value and the actual value; the mean absolute error loss function is: ;in, is the mean absolute error; is the number of data sets; For the The true value of the optical glass plate thickness detection error; For the The predicted value of the optical glass plate thickness detection error;

[0112] Use the training set data to train the model and minimize the loss function using the Adam optimizer. Use the validation set to evaluate the performance of the model and tune the model's hyperparameters until the model performance no longer improves significantly or reaches the preset stopping condition.

[0113] The test set is used to evaluate the performance of the model in the prediction task, and the current influencing feature data set is input into the trained optical glass plate thickness detection error to obtain the optical glass plate thickness detection error.

[0114] The method of comparing the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard includes:

[0115] If the predicted optical glass plate thickness detection error is less than the preset optical glass plate thickness detection error threshold, it is determined that the glass plate thickness meets the standard;

[0116] If the predicted optical glass plate thickness detection error is greater than or equal to a preset optical glass plate thickness detection error threshold, it is determined that the glass plate thickness does not meet the standard.

[0117] The substandard glass plate thickness data includes the actual thickness value of the glass plate, the location where the thickness does not meet the standard, the glass plate detection time, the surface flatness of the glass plate, the optical transmittance, the optical refractive index and the reflectivity of the glass plate.

[0118] The construction method of the thickness non-standard diagnosis model includes:

[0119] The training data set is divided into a training set, a validation set, and a test set to construct a thickness non-compliance diagnosis model; the sample set is a subset of the data set, and each sample set includes historical non-compliance glass plate thickness data and the corresponding non-compliance type of glass plate thickness;

[0120] The thickness non-compliance diagnosis model includes an input layer, a self-attention layer, a feedforward network layer and an output layer; the input layer of the model is used to input historical non-compliance glass plate thickness data, and the output layer of the model is used to output the non-compliance type of glass plate thickness; a softmax function is used as an activation function; the thickness non-compliance diagnosis model is a self-attention mechanism model;

[0121] Multi-classification cross entropy is used as the loss function of the model to measure the difference between the predicted value and the actual value of the model; the multi-classification cross entropy loss function is: ;in, is the average loss of the dataset; is the total number of samples in the data set; The number of glass sheets that do not meet the thickness standards; For the The samples for The true labels of each type; The model predicts The samples belong to The probability of each type; is the index of the sample;

[0122] The training set is used to train the thickness substandard diagnosis model, and the model parameters are updated through the back propagation algorithm to minimize the loss function. The validation set is used to evaluate the performance of the thickness substandard diagnosis model by calculating the accuracy index.

[0123] The SGD optimization algorithm is selected as the optimizer. The model is tuned according to the performance feedback of the validation set. The model parameters are adjusted until the performance is no longer significantly improved or the preset number of iterations is reached.

[0124] The test set is used to evaluate the performance of the model in the prediction task, and the trained thickness non-compliance diagnosis model is used to predict the current non-compliance glass plate thickness data to obtain the non-compliance type of glass plate thickness.

[0125] According to the type of glass plate thickness that does not meet the standard, corresponding improvement measures are taken until the glass plate thickness meets the standard. The methods include:

[0126] The types of glass plate thickness that do not meet the standards include glass plate thickness thinner than the standard glass plate thickness, glass plate thickness thicker than the standard glass plate thickness and thickness unevenness rate; for glass plate thickness thinner than the standard glass plate thickness, improvement measures include adjusting the raw material formula and optimizing the production process parameters; for glass plate thickness thicker than the standard glass plate thickness, improvement measures include reducing the amount of raw materials and adjusting the pressure and time parameters of the pressing process; for thickness unevenness rate, improvement measures include improving mold design and optimizing production process parameters.

[0127] The preset optical glass plate thickness detection error threshold is set by the staff, and the thickness detection errors of different optical glass plates are collected by the thickness detection intelligent terminal, and the average value of the plurality of optical glass plate thickness detection errors is taken as the preset optical glass plate thickness detection error threshold;

[0128] This embodiment, by using an improved multi-layer Monte Carlo ray tracing algorithm, can accurately simulate the refraction and reflection behavior of light in the multi-layer medium of an optical glass plate. Through this simulation, the transmittance, refractive index and reflectivity of the optical glass plate can be calculated more accurately to ensure the high accuracy and reliability of the measurement data; through Snell's law and the refractive index limit formula, the refractive index of light in different media can be dynamically adjusted; the refractive index limit can better reflect the actual performance of the optical glass plate under different conditions, thereby avoiding unreasonable physical phenomena and improving the physical authenticity of the simulation; Zernike polynomials are an effective tool for analyzing and fitting optical surface morphological errors; through this method, the errors on the surface of the optical glass plate can be fitted in detail, and then the surface flatness can be calculated through the fitting coefficient; this flatness calculation method based on a mathematical model is more accurate than traditional measurement methods;

[0129] Using F statistics and critical values ​​to determine significance avoids subjective inference and provides a basis for judgment based on data; through variance analysis, the specific impact of the test environment characteristics on the characteristics of the optical glass plate and the test equipment characteristics can be determined more accurately, improving the scientific nature of the analysis results; the calculation formula of the critical value is dynamically adjusted according to the feature type of the feature data set, which provides flexibility for different types of data. The critical value can reflect the changes in variables such as feature type and sample size, making the analysis process more accurate and practical; through significance analysis, it can be found which test environment characteristics have a significant impact on the characteristics of the optical glass plate and equipment, helping to optimize the test process and equipment settings, thereby improving product quality, reducing errors and improving test efficiency.

[0130] Example 2: Please refer to Figure 2 As shown, this embodiment provides an optical glass plate thickness detection system, including:

[0131] A data collection module, used to collect optical glass plate parameter data, testing equipment parameter data and testing environment data;

[0132] The data processing module is used to pre-process the collected optical glass plate parameter data, the detection equipment parameter data and the detection environment data to obtain the optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set; analyze the influence of the detection environment feature data set on the optical glass plate feature data set and the detection equipment feature data set to obtain the influence feature data set;

[0133] A detection error prediction module is used to obtain a thickness detection error prediction model based on the influencing feature data set training, and to predict the thickness detection error of the optical glass plate through the thickness detection error prediction model;

[0134] A glass plate thickness assessment module is used to compare the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard;

[0135] Thickness detection warning module: if the thickness of the glass plate meets the standard, the thickness detection intelligent terminal generates a qualified instruction, prompting that the glass plate thickness detection is qualified; if the thickness of the glass plate does not meet the standard, the thickness detection intelligent terminal generates a failed instruction, generates a warning message, and automatically collects the thickness data of the glass plate that does not meet the standard;

[0136] The thickness control module is used to construct a thickness non-compliance diagnosis model, input the non-compliance glass plate thickness data into the thickness non-compliance diagnosis model, and predict the non-compliance type of glass plate thickness; take corresponding improvement measures according to the non-compliance type of glass plate thickness until the glass plate thickness meets the standard.

[0137] Since the electronic device introduced in this embodiment is an electronic device used to implement the optical glass plate thickness detection method and system in the embodiment of this application, based on the optical glass plate thickness detection method and system introduced in the embodiment of this application, the technical personnel in this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not introduced in detail here. As long as the technical personnel in this field implement the electronic device used in the optical glass plate thickness detection method and system in the embodiment of this application, it belongs to the scope of protection of this application.

[0138] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0139] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for detecting the thickness of an optical glass plate, characterized in that: include: S1. Collecting optical glass plate parameter data, testing equipment parameter data and testing environment data; S2, preprocessing the collected optical glass plate parameter data, testing equipment parameter data and testing environment data to obtain an optical glass plate feature data set, a testing equipment feature data set and a testing environment feature data set; analyzing the influence of the testing environment feature data set on the optical glass plate feature data set and the testing equipment feature data set to obtain an influence feature data set; S3, training a thickness detection error prediction model according to the influencing feature data set, and predicting the thickness detection error of the optical glass plate through the thickness detection error prediction model; S4, comparing the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard; S5. If the thickness of the glass plate meets the standard, the thickness detection intelligent terminal generates a qualified instruction, indicating that the glass plate thickness detection is qualified; if the thickness of the glass plate does not meet the standard, the thickness detection intelligent terminal generates a failed instruction, generates a warning message, and automatically collects the thickness data of the glass plate that does not meet the standard; S6. Construct a thickness non-compliance diagnosis model, input the non-compliance glass plate thickness data into the thickness non-compliance diagnosis model, and predict the non-compliance type of the glass plate thickness; take corresponding improvement measures according to the non-compliance type of the glass plate thickness until the glass plate thickness meets the standard; The optical glass plate parameter data includes the surface flatness, transmittance, refractive index and reflectivity of the optical glass plate; the detection equipment parameter data includes the model of the detection equipment, the measurement accuracy, the vibration frequency of the equipment, the detection speed and the detection time; the detection environment data includes the detection environment temperature, humidity, light intensity and dust concentration in the air; The method of preprocessing the collected optical glass plate parameter data, detection equipment parameter data and detection environment data to obtain the optical glass plate feature data set, the detection equipment feature data set and the detection environment feature data set comprises: S31. For the optical glass plate parameter data: by using the improved multi-layer Monte Carlo ray tracing algorithm, the propagation behavior of light in the multi-layer medium of the optical glass plate is simulated. By introducing the multi-layer medium model, the refraction and reflection of light in each layer of the optical glass plate medium are accurately simulated to obtain the transmittance, refractive index and reflectivity of the optical glass plate; S311. Calculate the refractive index of the light when it enters the a+1th layer of the optical glass plate from the ath layer of the optical glass plate by using Snell's law formula; The Snell's law formula is: m a sinθ a =m a+1 sinθ a+1 ; Among them, m a is the refractive index of the light in the ath optical glass plate; m a+1 is the refractive index of the light when it enters the a+1th optical glass plate; θ a is the incident angle of the light in the a-th optical glass plate; θ a+1 is the refraction angle of the light in the a+1th optical glass plate; a is the index of the number of optical glass plates; S312, limiting the refractive index of the light in the a-th layer of the optical glass plate by using a refractive index limiting formula; the refractive index limiting formula is: Where c is the speed of light in air; v a is the propagation speed of light in the optical glass plate; L is the total number of layers of the optical glass plate; k is a constant for adjusting the limiting intensity; S313, calculating the transmittance and reflectance of light in the optical glass plate by using the Fresnel equation; the Fresnel equation is: Among them, R(θ a ) is the reflectivity of light in the a-th optical glass plate; Subtract the reflectivity of the light in the optical glass plate from 1 to get the transmittance T (θ a ); S314, fitting the surface flatness of the optical glass plate by using Zernike polynomials; The Zernike polynomial is: in, is the standard form of the polynomial; q is the order of the polynomial; p is the frequency of the polynomial; ρ is the normalized radius, that is, the relative distance from a point on the surface of the optical glass plate to the geometric center of the glass plate; is the Zernike radial polynomial; is the complex exponent of the periodic variation of the Zernike polynomial in the azimuth direction; pφ is the phase angle variation of the azimuth; φ is the azimuth of a point on the optical surface; i* is the imaginary unit; S315. Dynamically adjust the normalized radius ρ using a radius adjustment formula; the radius adjustment formula is: Wherein, d is the actual distance from a point on the surface of the optical glass plate to the geometric center of the glass plate; r is the radius of the optical glass plate; α is the adjustment coefficient for controlling the intensity of the influence of the azimuth angle on the normalized radius ρ; S316, constructing a surface morphology error model, and fitting the error of the optical glass plate surface by using Zernike polynomials; the surface morphology error model is: Where S(ρ,φ) is the morphological error of the optical glass plate surface; bqp is the fitting coefficient of the Zernike polynomial; The fitting coefficient bqp of the Zernike polynomial is obtained by fitting the morphological error S(ρ,φ) of the optical glass plate surface through the least square method. Based on the fitting coefficient bqp obtained by fitting, the surface flatness of the optical glass plate is calculated through the flatness calculation formula. The flatness calculation formula is: Wherein, u is the surface flatness of the optical glass plate; S32, for the detection equipment parameter data and the detection environment data: identifying and extracting features in the detection equipment parameter data and the detection environment data by principal component analysis to obtain a detection environment feature data set and a detection equipment feature data set; S33, detecting and eliminating abnormal data in the optical glass plate parameter data, the detection equipment parameter data, and the detection environment data by a box plot method, to obtain a processed optical glass plate feature data set, a detection equipment feature data set, and a detection environment feature data set; Normalizing the processed optical glass plate feature data set, the detection equipment feature data set, and the detection environment feature data set, and converting them into a standard normal distribution, to obtain a normalized processed optical glass plate feature data set, the detection equipment feature data set, and the detection environment feature data set; The method of analyzing the influence of the detection environment characteristic data set on the optical glass plate characteristic data set and the detection equipment characteristic data set to obtain the influenced characteristic data set comprises: S41. Preset two significant judgment hypotheses, including a null hypothesis and an alternative hypothesis; the null hypothesis indicates that the detection environment feature data set has no significant effect on the optical glass plate feature data set and the detection equipment feature data set; the alternative hypothesis indicates that the detection environment feature data set has a significant effect on the optical glass plate feature data set and the detection equipment feature data set; S42, collecting an optical glass plate feature data set, a detection equipment feature data set, and a detection environment feature data set to form a sample set, grouping the optical glass plate features and the detection equipment features according to different detection environment features, and calculating the intra-group variance and inter-group variance of each group; The within-group variance calculation formula is: in, is the within-group variance; V is the total sample size; j is the number of groups; v g is the number of samples in the g-th group; ygh is the h-th sample value in the g-th group; is the average value of the gth group; g is the index of the group; h is the index of the sample; The inter-group variance calculation formula is: in, is the between-group variance; is the total mean of all groups; S43, dividing the inter-group variance by the intra-group variance to obtain the calculated F statistic, and comparing the F statistic with the critical value to determine whether the detection environment characteristics have a significant impact on the optical glass plate characteristics and the detection equipment characteristics; the critical value is calculated as follows: Among them, F cl is the critical value; o is the feature type of the feature data set; S44. If the F statistic is greater than the critical value F cl , then the preset null hypothesis is not established, and the detection environment feature data set has a significant impact on the optical glass plate feature data set and the detection equipment feature data set; If the F statistic is less than or equal to the critical value F cl , then the pre-equipment selection hypothesis does not hold, and the detection environment feature data set has no significant impact on the optical glass plate feature data set and the detection equipment feature data set; The data collection of the detection environment characteristic data set that has a significant impact on the optical glass plate characteristic data set and the detection equipment characteristic data set is statistically analyzed and integrated to obtain the impact characteristic data set.

2. The method for detecting the thickness of an optical glass plate according to claim 1, characterized in that: The training method of the thickness detection error prediction model includes: The data set is divided into a training set, a validation set and a test set, and a thickness detection error prediction model is constructed; the sample set is a subset of the data set, and each sample set includes a historical impact feature data set and the corresponding optical glass plate thickness detection error; The thickness detection error prediction model includes input data and output labels; the input data of the model is a historical impact feature data set, and the output label of the model is an optical glass plate thickness detection error; RBF is selected as the kernel function of the model, and the hyperparameters of the model are initialized. The hyperparameters of the model include a regularization parameter C and a γ value of the RBF kernel function; the thickness detection error prediction model is a support vector machine model; Use mean absolute error as the loss function to measure the error between the model's predicted value and the actual value; use the training set data to train the model and use the Adam optimizer to minimize the loss function; use the validation set to evaluate the performance of the model and tune the model's hyperparameters until the model performance no longer improves significantly or reaches the preset stopping condition; The test set is used to evaluate the performance of the model in the prediction task, and the current influencing feature data set is input into the trained optical glass plate thickness detection error to obtain the optical glass plate thickness detection error.

3. The method for detecting the thickness of an optical glass plate according to claim 2, characterized in that: The method of comparing the predicted optical glass plate thickness detection error with a preset optical glass plate thickness detection error threshold to determine whether the glass plate thickness meets the standard includes: If the predicted optical glass plate thickness detection error is less than the preset optical glass plate thickness detection error threshold, it is determined that the glass plate thickness meets the standard; If the predicted optical glass plate thickness detection error is greater than or equal to a preset optical glass plate thickness detection error threshold, it is determined that the glass plate thickness does not meet the standard.

4. The method for detecting the thickness of an optical glass plate according to claim 3, characterized in that: The substandard glass plate thickness data includes the actual thickness value of the glass plate, the location where the thickness does not meet the standard, the glass plate detection time, the surface flatness of the glass plate, the optical transmittance, the optical refractive index and the reflectivity of the glass plate.

5. The method for detecting the thickness of an optical glass plate according to claim 4, characterized in that: The method for constructing the thickness non-standard diagnosis model includes: The training data set is divided into a training set, a validation set, and a test set to construct a thickness non-compliance diagnosis model; the sample set is a subset of the data set, and each sample set includes historical non-compliance glass plate thickness data and the corresponding non-compliance type of glass plate thickness; The thickness non-compliance diagnosis model includes an input layer, a self-attention layer, a feedforward network layer and an output layer; the input layer of the model is used to input historical non-compliance glass plate thickness data, and the output layer of the model is used to output the non-compliance type of glass plate thickness; a softmax function is used as an activation function; the thickness non-compliance diagnosis model is a self-attention mechanism model; Use multi-classification cross entropy as the model's loss function to measure the difference between the model's predicted value and the actual value; use the training set to train the thickness substandard diagnosis model, and update the model parameters through the back propagation algorithm to minimize the loss function; use the validation set to evaluate the performance of the thickness substandard diagnosis model by calculating the accuracy index; The SGD optimization algorithm was selected as the optimizer, and the model was tuned according to the performance feedback of the validation set. The model parameters were adjusted until the performance no longer improved significantly or the preset number of iterations was reached. The test set was used to evaluate the performance of the model in the prediction task, and the trained thickness non-compliance diagnosis model was used to predict the current non-compliant glass plate thickness data to obtain the non-compliant glass plate thickness type.

6. The method for detecting the thickness of an optical glass plate according to claim 5, characterized in that: The method of taking corresponding improvement measures according to the type of glass sheet thickness not meeting the standard until the glass sheet thickness meets the standard includes: The types of glass plate thickness that do not meet the standards include glass plate thickness thinner than the standard glass plate thickness, glass plate thickness thicker than the standard glass plate thickness and thickness unevenness rate; for glass plate thickness thinner than the standard glass plate thickness, improvement measures include adjusting the raw material formula and optimizing production process parameters; for glass plate thickness thicker than the standard glass plate thickness, improvement measures include reducing the amount of raw materials used and adjusting the pressure and time parameters of the pressing process; for thickness unevenness rate, improvement measures include improving mold design and optimizing production process parameters.

Images