Glass ceramic thickness detection method and related device

Abstract

The invention discloses a glass ceramic thickness detection method and a related device, and the method comprises the steps: obtaining a vibration signal generated in a glass ceramic thinning and grinding process, and extracting a vibration dominant frequency, a frequency spectrum centroid frequency and a frequency band energy distribution ratio as thickness characteristic parameters through fast Fourier transform processing; three-dimensional position coordinates of the grinding head are obtained, space mapping and weighted average processing are conducted on the thickness characteristic parameters, and space distribution of the thickness characteristic parameters is obtained; performing process interference compensation and sliding window time sequence smoothing processing according to the grinding process parameters to obtain stable thickness characteristic distribution; based on the stable thickness feature distribution and the current grinding process, final thickness distribution is predicted, and a thickness qualification pre-judgment result is obtained; and extracting a time sequence change characteristic of the thickness characteristic parameter as a vibration evolution index, and performing comprehensive quality evaluation in combination with a thickness eligibility pre-judgment result. According to the scheme, the thickness can be detected in real time in the thinning and grinding process of the microcrystalline glass, and the production efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of glass-ceramic detection, and in particular to a method for detecting the thickness of glass-ceramic and a related device. Background Art

[0002] Glass-ceramics is an advanced composite material formed by partially or completely crystallizing ordinary glass under specific temperature conditions through a controlled crystallization process. Its unique microstructure exhibits a coexistence of glass and microcrystalline phases. This material combines the optical transparency and chemical stability of glass with the excellent mechanical strength and thermal stability of crystalline materials, achieving a hardness of 6-7 on the Mohs scale, a low coefficient of thermal expansion, and excellent impact resistance. Due to these exceptional properties, glass-ceramics has been widely used in high-tech fields such as electronic substrates, optical devices, and architectural decorative materials. In particular, it demonstrates enormous market potential and technological advantages in screen protectors for portable electronic devices such as smartphones and tablets.

[0003] In the existing technology, the thickness detection of microcrystalline glass screen protectors mainly adopts the post-finished product detection mode, that is, after completing all processing steps such as thinning grinding, precision polishing, and cleaning, the product is transferred to a special detection station, and the thickness of the final product is measured and the quality is evaluated using detection equipment such as laser interference, eddy current induction or ultrasonic waves. This detection method requires the establishment of a special detection station and the provision of detection equipment on the production line, and each product must go through additional detection time. On modern high-speed production lines, this additional detection time directly affects the production rhythm and the production capacity of the entire line. Especially in large-scale continuous production modes, the detection link has become a key bottleneck restricting the improvement of production efficiency. With the rapid growth of market demand for microcrystalline glass screen protectors, how to eliminate the extra time required for post-finished product detection has become a core technical issue for improving production efficiency and reducing manufacturing costs. Summary of the Invention

[0004] The main purpose of the present invention is to solve the technical problem that the thickness detection of the existing microcrystalline glass screen protective film requires extra time after the finished product is completed, which affects the production efficiency.

[0005] A first aspect of the present invention provides a method for detecting the thickness of a glass-ceramic. The method comprises: Acquire a vibration signal generated during the thinning and grinding process of the glass-ceramic, perform fast Fourier transform processing on the vibration signal, and extract the vibration main frequency, spectrum centroid frequency, and frequency band energy distribution ratio as thickness characteristic parameters; Obtaining the three-dimensional position coordinates of the grinding head, mapping the thickness characteristic parameters to the three-dimensional position coordinates, and processing the thickness characteristic parameters of adjacent measuring points using a weighted average algorithm to obtain a spatial distribution of the thickness characteristic parameters; Performing process interference compensation on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and performing time series smoothing processing on the compensated thickness characteristic parameters using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution; Calculate the thickness variation of the remaining processing area based on the stable thickness characteristic distribution and the current grinding process, predict the final thickness distribution after the grinding is completed, and obtain the thickness qualification prediction result; The time series variation characteristics of the thickness characteristic parameters during the grinding process are extracted as vibration evolution indicators, combined with the thickness qualification prediction results, to judge the comprehensive quality status of the product and obtain real-time quality assessment results.

[0006] Preferably, the step of obtaining a vibration signal generated during the glass-ceramic thinning and grinding process, performing fast Fourier transform processing on the vibration signal, and extracting the vibration main frequency, spectrum centroid frequency, and frequency band energy distribution ratio as thickness characteristic parameters includes: Obtaining raw vibration data during the glass-ceramic thinning and grinding process, and performing a two-phase response separation process on the raw vibration data based on the difference between the elastic modulus of the glass phase and the elastic modulus of the microcrystalline phase of the glass-ceramic to obtain a glass phase vibration component and a microcrystalline phase vibration component; According to the parameters of the grinding spindle speed and feed speed, the glass phase vibration component and the microcrystalline phase vibration component are synchronously divided and processed during the grinding cycle to eliminate the mechanical vibration interference of the grinding equipment and obtain a synchronous purified vibration signal; Performing fast Fourier transform processing on the synchronous purification vibration signal in the frequency range of 50-1500 Hz, dividing it into three thickness-sensitive bands: low frequency band, mid frequency band, and high frequency band, and obtaining frequency band spectrum data; According to the frequency spectrum data of the frequency bands, the peak frequency in each frequency band is calculated as the main vibration frequency, the weighted average value of the spectrum amplitude is calculated as the spectrum centroid frequency, and the ratio of the energy of each frequency band to the total energy is calculated as the frequency band energy distribution ratio, and the combination is used to form the basic thickness characteristic parameter; The vibration main frequency, spectrum centroid frequency and frequency band energy distribution ratio in the basic thickness characteristic parameters are subjected to inter-band differential operation to generate low-medium frequency differential coefficients, medium-high frequency differential coefficients and full-band comprehensive coefficients, which are used together with the basic thickness characteristic parameters as thickness characteristic parameters.

[0007] Preferably, performing a two-phase response separation process on the original vibration data based on the difference between the glass phase elastic modulus and the microcrystalline phase elastic modulus of the microcrystalline glass to obtain the glass phase vibration component and the microcrystalline phase vibration component includes: According to the crystallinity distribution characteristics in the glass-ceramic preparation process, the volume fraction of the glass phase and the volume fraction of the microcrystalline phase are calculated. Combined with the density difference of each phase, the dual-phase density distribution parameters are obtained. Based on the density distribution parameters and elastic modulus differences of the two phases, the acoustic impedance difference between the glass phase and the microcrystalline phase is calculated, the relationship between the reflection coefficient and the transmission coefficient of the vibration wave at the two phase interface is established, and the response characteristics of the two phase interface are obtained; According to the dual-phase interface response characteristics, the original vibration data is decomposed in the frequency domain, low-frequency vibration components are attributed to glass phase response, and high-frequency vibration components are attributed to microcrystalline phase response, thereby obtaining preliminary phase separation vibration data; Based on the grain size distribution and orientation distribution characteristics of the microcrystalline phase, the microcrystalline phase response part in the preliminary phase separation vibration data is corrected by grain scattering to eliminate the influence of grain boundary scattering on vibration propagation, and the glass phase vibration component and the microcrystalline phase vibration component are obtained.

[0008] Preferably, the obtaining of the three-dimensional position coordinates of the grinding head, mapping the thickness characteristic parameters to the three-dimensional position coordinates, and processing the thickness characteristic parameters of adjacent measuring points using a weighted average algorithm to obtain the spatial distribution of the thickness characteristic parameters include: Obtain the position coordinate data of the grinding head in the X-axis, Y-axis, and Z-axis directions, and divide the workpiece surface into regular grids based on the path spacing and overlap rate of the grinding tracks to obtain spatial grid units; The low-medium frequency differential coefficient, the medium-high frequency differential coefficient and the full-band comprehensive coefficient of the thickness characteristic parameter are temporally and spatially matched with the three-dimensional position coordinates of the grinding head at the corresponding moment, and assigned to the corresponding spatial grid units to obtain gridded thickness characteristic data; According to the effective radius range of vibration propagation in the glass-ceramics, the adjacent influence area of ​​each grid unit is determined, and the spatial distance weight coefficient between adjacent grid units is calculated to obtain the spatial weight matrix; According to the spatial weight matrix, a weighted average operation is performed on the gridded thickness characteristic data, and a boundary constraint correction process is applied to the grid cells in the edge area of ​​the workpiece to obtain the spatial distribution of the thickness characteristic parameters.

[0009] Preferably, the process interference compensation is performed on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and the compensated thickness characteristic parameters are smoothed in time series using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution, including: Acquiring real-time change data of the grinding force, calculating a grinding force reference offset according to a workpiece stiffness decreasing law, and performing grinding force offset compensation on the glass phase response component and the microcrystalline phase response component in the spatial distribution of the thickness characteristic parameter to obtain dual-phase compensation data; According to the spatial gradient change of the grinding feed direction, the dual-phase compensation data is subjected to directional weighted correction to eliminate the anisotropic effect of the grinding track direction on the thickness feature detection, thereby obtaining an isotropic thickness feature distribution; Based on the time scale difference between the grinding cycle and the thickness change cycle, a double-layer nested sliding window is set. The inner window tracks the short-term changes in the grinding cycle, and the outer window tracks the long-term trend of the thickness change. The isotropic thickness characteristic distribution is smoothed by a double layer time series to obtain multi-scale smoothed data. The continuity index of the multi-scale smoothed data in the spatially adjacent regions and the stability index of the temporally continuous windows are calculated. When the continuity index and the stability index simultaneously meet the convergence conditions, the corresponding data are output as a stable thickness characteristic distribution.

[0010] Preferably, based on the time scale difference between the grinding cycle and the thickness change cycle, a double-layer nested sliding window is set, the inner window tracks the short-term changes in the grinding cycle, and the outer window tracks the long-term trend of the thickness change. The isotropic thickness characteristic distribution is subjected to double-layer time series smoothing to obtain multi-scale smoothed data, including: According to the grinding spindle speed and the grinding head feed speed, the time length of a single grinding cycle is calculated, and the time span of the inner sliding window is set to 2-5 times the length of the grinding cycle to obtain the short-time window parameters; Based on the physical continuity constraint of the workpiece thickness change, the time scale of significant thickness feature changes is calculated, the time span of the outer sliding window is set to 1.5-3 times the thickness change time scale, the nesting ratio relationship between the outer window and the inner window is determined, and the long-term window parameters are obtained; Within the short-time window parameter range, high-frequency noise filtering and grinding vibration synchronization processing are performed on the isotropic thickness characteristic distribution to retain effective signal components related to thickness and obtain short-time smoothed data; Within the long-term window parameter range, trend extraction and outlier detection processing are performed on the short-term smoothed data, and segmented weighted averaging is performed in combination with the stage characteristics of the grinding process to obtain multi-scale smoothed data; The smoothness evaluation index of the multi-scale smoothed data in a continuous time period is calculated. When the smoothness evaluation index reaches a convergence threshold, the data processing is confirmed to be completed and the result is output.

[0011] Preferably, the calculating of the thickness variation of the remaining processing area based on the stable thickness characteristic distribution and the current grinding process, predicting the final thickness distribution after the grinding is completed, and obtaining the thickness qualification prediction result includes: According to the current grinding completion ratio and grinding trajectory path data, the spatial distribution of the remaining unprocessed area is identified, the boundary gradient change rate between the processed area and the remaining area is calculated, and the thickness transition characteristics between the areas are obtained; Based on the thickness decreasing law of the processed area in the stable thickness characteristic distribution, combined with the current workpiece stiffness decreasing coefficient and the accumulated wear of the grinding tool, the material removal efficiency variation coefficient of the remaining processed area in the subsequent grinding process is calculated to obtain the nonlinear thickness variation; Based on the thickness transition characteristics and nonlinear thickness variation between the regions, a thickness prediction operation is performed on the remaining unprocessed region, and a predicted thickness distribution map after grinding is generated in combination with the boundary constraints of the edge region; The thickness deviation value and thickness uniformity index of each area in the predicted thickness distribution diagram are calculated, the thickness deviation value is compared with the allowable tolerance range of the technical specification, and the thickness uniformity index is matched and evaluated with the optical performance requirements to obtain a thickness qualification prediction result.

[0012] Preferably, the extraction of the time series variation characteristics of the thickness characteristic parameters during the grinding process as a vibration evolution index, combined with the thickness qualification prediction result, judges the comprehensive quality status of the product, and obtains a real-time quality assessment result, including: Gradient analysis and inflection point detection are performed on the variation trajectory of the low-medium frequency differential coefficient, medium-high frequency differential coefficient, and full-band comprehensive coefficient of the thickness characteristic parameter on the grinding time axis, and the vibration amplitude attenuation rate, frequency drift rate, and phase jump number are extracted as vibration evolution indicators; Based on the vibration response differences of the dual-phase structure of the glass-ceramic, the vibration evolution index is subjected to glass phase abnormal pattern recognition and microcrystalline phase abnormal pattern recognition, and signs of microcrack extension and stress concentration are detected to obtain a risk assessment of internal defects in the material; Cross-validate and analyze the thickness qualification prediction results with the material internal defect risk assessment, calculate the comprehensive weight coefficient of thickness geometry qualification and structural integrity qualification, and obtain a multi-dimensional quality comprehensive score; Based on the multi-dimensional quality comprehensive score, the product quality level is divided into three levels: excellent, qualified and risky. The defect type and risk degree of risky products are marked, and a real-time quality assessment result including quality grade and risk warning information is generated.

[0013] A second aspect of the present invention provides a device for detecting the thickness of a glass-ceramic. The device comprises: A vibration signal processing module is used to obtain the vibration signal generated during the thinning and grinding process of the micro-ceramic glass, perform fast Fourier transform processing on the vibration signal, and extract the vibration main frequency, spectrum centroid frequency and frequency band energy distribution ratio as thickness characteristic parameters; A spatial mapping module is used to obtain the three-dimensional position coordinates of the grinding head, map the thickness characteristic parameters to the three-dimensional position coordinates, and use a weighted average algorithm to process the thickness characteristic parameters of adjacent measuring points to obtain the spatial distribution of the thickness characteristic parameters; A data calibration module is used to perform process interference compensation on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and to perform time series smoothing on the compensated thickness characteristic parameters using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution; A thickness prediction module is used to calculate the thickness variation of the remaining processing area based on the stable thickness characteristic distribution and the current grinding process, predict the final thickness distribution after the grinding is completed, and obtain a thickness qualification prediction result; The quality assessment module is used to extract the time-series variation characteristics of the thickness characteristic parameters during the grinding process as vibration evolution indicators, combine the thickness qualification prediction results, judge the comprehensive quality status of the product, and obtain real-time quality assessment results.

[0014] The third aspect of the present invention provides a device for detecting the thickness of microcrystalline glass, comprising: a memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected through lines; the at least one processor calls the instructions in the memory so that the device for detecting the thickness of microcrystalline glass performs the steps of the above-mentioned method for detecting the thickness of microcrystalline glass.

[0015] A fourth aspect of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions that, when executed on a computer, enable the computer to execute the steps of the above-mentioned method for detecting the thickness of microcrystalline glass.

[0016] The existing microcrystalline glass screen protector thickness detection technology adopts a post-finished product detection mode, which requires a special detection station to be set up on the production line. Each product must go through additional detection time, which directly affects production efficiency. The present invention obtains the vibration signal generated during the thinning and grinding process of the microcrystalline glass, and converts the grinding process itself into a detection signal source. When the grinding tool contacts the workpiece, the propagation characteristics of the vibration wave excited inside it are directly physically related to the thickness of the workpiece. The micron-level change in thickness will produce a measurable frequency shift in the vibration spectrum. The main frequency of the vibration, the frequency spectrum centroid frequency and the energy distribution ratio of the frequency band are extracted in real time through fast Fourier transform as thickness characteristic parameters, thus realizing the simultaneous processing and detection. At the same time, the three-dimensional position coordinates of the grinding head are obtained and mapped in real time with the thickness characteristic parameters. The existing position feedback function of the grinding equipment is used to realize the precise spatial positioning of the thickness information. The weighted average algorithm is used to process the data of adjacent measuring points. The thickness distribution map is synchronously constructed in the process of the grinding track covering the surface of the workpiece, avoiding the extra spatial scanning time of the traditional method. The thickness characteristics are further compensated in real time based on the grinding process parameters and time-series smoothing is performed using a sliding window algorithm. After obtaining a stable thickness characteristic distribution, the thickness variation of the remaining area is predicted based on the current grinding process, and the final thickness qualification is judged. Finally, a comprehensive quality assessment is performed based on the time-series variation characteristics of the thickness characteristic parameters. This technical approach transforms the traditional "processing time + inspection time" serial model into a "processing and inspection" parallel model. The inspection function is fully integrated into the grinding process and has prediction and real-time feedback capabilities. This completely eliminates the additional time required for post-finished product inspection and maximizes production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying any creative work.

[0018] Figure 1 Schematic diagram of an embodiment of a method for detecting thickness of glass-ceramics according to an embodiment of the present invention; Figure 2 Schematic diagram of an embodiment of a device for detecting thickness of glass-ceramics according to an embodiment of the present invention; Figure 3 Schematic diagram of an embodiment of a glass-ceramic thickness detection device in an embodiment of the present invention.

[0019] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings and in conjunction with the embodiments. DETAILED DESCRIPTION

[0020] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0021] It should be noted that if the embodiments of the present invention involve directional indications (such as up, down, left, right, front, back, etc.), the directional indications are only used to explain the relative position relationship, movement status, etc. between the various components under a certain specific posture. If the specific posture changes, the directional indications will also change accordingly.

[0022] In addition, the descriptions of "first", "second", etc. in the present invention are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, "and / or" in the full text includes three solutions. Taking A and / or B as an example, it includes technical solution A, technical solution B, and technical solution that satisfies both A and B. In addition, the technical solutions between the various embodiments can be combined with each other, and must be based on the ability of ordinary technicians in this field to implement. When the combination of technical solutions is mutually contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0023] An embodiment of the present application provides a method for detecting the thickness of glass-ceramics. Figure 1 A flow chart of a method for measuring the thickness of glass-ceramics provided in one embodiment of the present application. In this embodiment, the method includes: See also Figure 1 , obtaining a vibration signal generated during the thinning and grinding process of the micro-ceramic glass, performing fast Fourier transform processing on the vibration signal, and extracting the vibration main frequency, spectrum centroid frequency and frequency band energy distribution ratio as thickness characteristic parameters; In one embodiment of the present invention, the method of obtaining a vibration signal generated during the thinning and grinding process of the glass-ceramics, performing fast Fourier transform processing on the vibration signal, and extracting the vibration main frequency, spectrum centroid frequency, and frequency band energy distribution ratio as thickness characteristic parameters includes: Obtaining raw vibration data during the glass-ceramic thinning and grinding process, and performing a two-phase response separation process on the raw vibration data based on the difference between the elastic modulus of the glass phase and the elastic modulus of the microcrystalline phase of the glass-ceramic to obtain a glass phase vibration component and a microcrystalline phase vibration component; According to the parameters of the grinding spindle speed and feed speed, the glass phase vibration component and the microcrystalline phase vibration component are synchronously divided and processed during the grinding cycle to eliminate the mechanical vibration interference of the grinding equipment and obtain a synchronous purified vibration signal; Performing fast Fourier transform processing on the synchronous purification vibration signal in the frequency range of 50-1500 Hz, dividing it into three thickness-sensitive bands: low frequency band, mid frequency band, and high frequency band, and obtaining frequency band spectrum data; According to the frequency spectrum data of the frequency bands, the peak frequency in each frequency band is calculated as the main vibration frequency, the weighted average value of the spectrum amplitude is calculated as the spectrum centroid frequency, and the ratio of the energy of each frequency band to the total energy is calculated as the frequency band energy distribution ratio, and the combination is used to form the basic thickness characteristic parameter; The vibration main frequency, spectrum centroid frequency and frequency band energy distribution ratio in the basic thickness characteristic parameters are subjected to inter-band differential operation to generate low-medium frequency differential coefficients, medium-high frequency differential coefficients and full-band comprehensive coefficients, which are used together with the basic thickness characteristic parameters as thickness characteristic parameters.

[0024] The following is a detailed description of the steps involved in the above embodiment: During the glass-ceramic thinning and grinding process, raw vibration data is acquired using a piezoelectric vibration sensor mounted on the workpiece clamping device. The sampling frequency is set to 5000 Hz to ensure coverage of the required frequency range. Glass-ceramic consists of two distinct phases: a glass phase and a microcrystalline phase. The glass phase has an elastic modulus of approximately 70-80 GPa, while the microcrystalline phase has an elastic modulus of approximately 110-130 GPa. This significant difference in the elastic moduli of the two phases results in different response characteristics during vibration wave propagation. The specific implementation process of the dual-phase response separation processing is as follows: first, the original vibration data is preprocessed to remove the DC component and 50Hz power frequency interference, and then the frequency domain separation is performed based on the difference in the physical response characteristics of the glass phase and the microcrystalline phase. Due to the low elastic modulus of the glass phase, it mainly produces a resonant response in the low-frequency range, while the high elastic modulus of the microcrystalline phase makes it mainly produce a resonant response in the high-frequency range. Therefore, 300Hz is used as the dividing frequency, and a Butterworth digital filter is used to design a low-pass filter (cut-off frequency 300Hz, order 4) to extract vibration components below 300Hz. These components mainly come from the elastic response of the glass phase and are defined as glass phase vibration components; at the same time, a high-pass filter (cut-off frequency 300Hz, order 4) is used to extract vibration components above 300Hz. These components mainly come from the elastic response of the microcrystalline phase and are defined as microcrystalline phase vibration components. For example, when the original vibration signal contains 200Hz with an amplitude of 0.5m / s 2 and 800Hz amplitude 0.3m / s 2 When the components are present, the glass phase vibration components are 200Hz and 0.5m / s after filtering and separation. 2, the microcrystalline phase vibration component is 800Hz, 0.3m / s 2 This separation process utilizes the inherent physical properties of the dual-phase structure of glass-ceramics and can decompose the vibration response of the composite material into a single-phase response with a clear origin, providing a more detailed data basis for subsequent precise thickness analysis.

[0025] The duration of a single grinding cycle is calculated based on the real-time spindle speed and feed rate parameters output by the grinding equipment's CNC system. The specific implementation process for synchronous grinding cycle segmentation is as follows: Taking a spindle speed of 3000 rpm as an example, the grinding cycle time is calculated as 60 / 3000 = 0.02 seconds. The time series data of the glass phase vibration component and the microcrystalline phase vibration component are segmented into fixed time windows of 0.02 seconds. Each window contains 100 sampling points (5000 Hz × 0.02 seconds). This ensures that each data segment corresponds to a complete grinding cycle. The mechanical vibration interference of grinding equipment mainly includes the 50Hz fundamental frequency vibration generated by the rotation of the spindle, the 100Hz frequency harmonic vibration of the transmission system, and other periodic vibrations of the cooling system. The frequency characteristics of these interferences are integer multiples of the spindle speed. In each 0.02-second time window, a notch filter is used to eliminate specific frequency components related to the spindle speed, such as 50Hz, 100Hz, and 150Hz. The center frequency of the notch filter is precisely set to the spindle frequency and its harmonics, and the bandwidth is set to ±2Hz to avoid over-filtering. After this periodic segmentation and synchronous filtering processing, the vibration data in each time window retains only the components related to the material response, eliminating the regular interference generated by the operation of the equipment. The resulting synchronous purified vibration signal can truly reflect the vibration characteristics of the microcrystalline glass material during the grinding process. The key to this processing method is to use the periodic characteristics of the grinding process to effectively separate the random material response signal from the regular equipment interference signal.

[0026] The synchronously purified vibration signals were analyzed in the frequency domain using a fast Fourier transform algorithm. A Hanning window function was used to reduce spectral leakage. The analysis window length was set to 1024 sampling points, with an overlap of 50%, resulting in a spectral resolution of 4.88 Hz. The frequency range of 50-1500 Hz was divided into three thickness-sensitive bands based on the following criteria: the low-frequency band of 50-300 Hz corresponds to the first- and second-order bending vibration modes of the glass-ceramic workpiece. Changes in thickness cause changes in the overall stiffness of the workpiece, resulting in frequency shifts in these modes. The mid-frequency band of 300-800 Hz corresponds to the workpiece's higher-order bending and torsional modes, which are more sensitive to local thickness variations. The high-frequency band of 800-1500 Hz corresponds to surface wave propagation and stress wave propagation within the material, primarily reflecting changes in surface roughness and internal stress states. The frequency-amplitude distribution array is calculated by the FFT algorithm in each frequency band. The low frequency band contains 51 frequency points (from 50Hz to 300Hz, with an interval of 4.88Hz), the mid-frequency band contains 101 frequency points (from 300Hz to 800Hz), and the high frequency band contains 144 frequency points (from 800Hz to 1500Hz). The acquisition process of the frequency band spectrum data is as follows: the amplitude of the FFT result within each frequency band is extracted to form the amplitude spectrum array of the frequency band. For example, the amplitude of 2.1m / s is extracted at the 150Hz frequency point in the low frequency band. 2 , the amplitude of 1.8m / s was extracted at the 550Hz frequency point in the mid-frequency band 2 This frequency band processing can quantify the vibration responses generated by different physical mechanisms separately, providing a complete data basis for multi-dimensional thickness feature analysis.

[0027] The specific process for calculating the three basic thickness characteristic parameters based on the amplitude spectrum array of each frequency band is as follows: the main vibration frequency is determined by searching for the frequency point corresponding to the maximum amplitude within each frequency band. The search algorithm traverses the amplitudes of all frequency points within the frequency band and records the maximum amplitude and its corresponding frequency value as the main vibration frequency of the frequency band. The spectrum centroid frequency is calculated by multiplying the frequency value of each frequency point within the frequency band by its amplitude, summing all the products, and dividing by the sum of all amplitudes within the frequency band. The mathematical expression is that the spectrum centroid frequency equals the sum of the products of frequency and amplitude divided by the sum of amplitudes. The frequency band energy distribution ratio is calculated by summing the squared amplitudes of all frequency points within the frequency band to obtain the energy value of the frequency band, and then dividing it by the sum of the squared amplitudes of all frequency points in the full frequency band of 50-1500Hz to obtain the proportion of the energy of the frequency band to the total energy. The basic thickness characteristic parameters refer to the numerical combination of these three parameters within the three frequency bands, totaling nine parameter values. For example, when monitoring a vibration signal at a specific moment, the dominant frequency of the low-frequency band is 150Hz, the frequency spectrum centroid is 180Hz, and the frequency band energy distribution ratio is 0.45. The corresponding parameters for the mid-frequency band are 550Hz, 520Hz, and 0.35, and the corresponding parameters for the high-frequency band are 1200Hz, 1100Hz, and 0.20. These parameters can describe the characteristics of the vibration signal from multiple dimensions, such as frequency distribution, energy distribution, and dominant frequency. When the thickness of the workpiece changes, these parameters will change accordingly, thus enabling indirect thickness measurement.

[0028] The specific implementation process of the inter-band differential operation of the basic thickness characteristic parameters is as follows: the low-medium frequency differential coefficient is calculated by subtracting the characteristic parameters corresponding to the medium frequency band from each characteristic parameter of the low frequency band, that is, the main vibration frequency of the low frequency band minus the main vibration frequency of the medium frequency band, the centroid frequency of the low frequency band spectrum minus the centroid frequency of the medium frequency band spectrum, and the energy distribution ratio of the low frequency band band minus the energy distribution ratio of the medium frequency band band, to obtain three low-medium frequency differential coefficients; the medium-high frequency differential coefficient adopts the same subtraction operation, and subtracts the corresponding parameters of the high frequency band from each characteristic parameter of the medium frequency band to obtain three medium-high frequency differential coefficients; the calculation process of the full-band comprehensive coefficient is to perform weighted averaging on the similar characteristic parameters of the three frequency bands according to the energy distribution ratio of each frequency band, that is, multiplying a certain characteristic parameter of the low frequency band by the energy ratio of the low frequency band, multiplying the corresponding parameter of the medium frequency band by the energy ratio of the medium frequency band, and multiplying the corresponding parameter of the high frequency band by the energy ratio of the high frequency band, and adding the three items to obtain the full-band comprehensive coefficient of the characteristic parameter. For example, the low-mid frequency differential coefficient of the main vibration frequency is 150Hz - 550Hz = -400Hz, the mid-high frequency differential coefficient is 550Hz - 1200Hz = -650Hz, and the comprehensive coefficient for the entire frequency band is 150 × 0.45 + 550 × 0.35 + 1200 × 0.20 = 500Hz. The thickness characteristic parameter is a complete set of 15 values, including 9 basic parameters and 6 differential coefficients. The physical significance of the differential operation lies in its ability to eliminate interference components common to all frequency bands, highlight the response differences between different frequency bands caused by thickness changes, and improve the sensitivity and anti-interference ability of thickness detection.

[0029] In one embodiment of the present invention, the dual-phase response separation processing is performed on the original vibration data based on the difference between the glass phase elastic modulus and the microcrystalline phase elastic modulus of the microcrystalline glass to obtain the glass phase vibration component and the microcrystalline phase vibration component, including: According to the crystallinity distribution characteristics in the glass-ceramic preparation process, the volume fraction of the glass phase and the volume fraction of the microcrystalline phase are calculated. Combined with the density difference of each phase, the dual-phase density distribution parameters are obtained. Based on the density distribution parameters and elastic modulus differences of the two phases, the acoustic impedance difference between the glass phase and the microcrystalline phase is calculated, the relationship between the reflection coefficient and the transmission coefficient of the vibration wave at the two phase interface is established, and the response characteristics of the two phase interface are obtained; According to the dual-phase interface response characteristics, the original vibration data is decomposed in the frequency domain, low-frequency vibration components are attributed to glass phase response, and high-frequency vibration components are attributed to microcrystalline phase response, thereby obtaining preliminary phase separation vibration data; Based on the grain size distribution and orientation distribution characteristics of the microcrystalline phase, the microcrystalline phase response part in the preliminary phase separation vibration data is corrected by grain scattering to eliminate the influence of grain boundary scattering on vibration propagation, and the glass phase vibration component and the microcrystalline phase vibration component are obtained.

[0030] The following is a detailed description of the steps involved in the above embodiment: The specific implementation process of calculating the volume fraction of the two phases based on the crystallinity distribution characteristics in the microcrystalline glass preparation process is as follows: the crystallinity distribution characteristics are obtained by recording the preparation process parameters. The crystallinity distribution characteristics refer to the difference in the degree of crystallization in different regions of the microcrystalline glass during the crystallization heat treatment process, which is manifested as the spatial distribution of the percentage of the crystalline phase in the total volume. The preparation process parameters include nucleation temperature, crystallization temperature, holding time and cooling rate. These parameters have a certain corresponding relationship with crystallinity. The method for calculating the volume fraction of the glass phase and the volume fraction of the microcrystalline phase is: based on the set value of the nucleation temperature in the range of 480-520℃, combined with the set value of the crystallization temperature in the range of 780-820℃ and the holding time, the corresponding crystallinity value is obtained by querying the preparation process database. For example, when the nucleation temperature is 500℃, the crystallization temperature is 800℃, and the holding time is 2 hours, the corresponding crystallinity is 65%, that is, the volume fraction of the microcrystalline phase is 0.65, and the volume fraction of the glass phase is 0.35. The calculation process combined with the density difference of each phase is: the density of the glass phase is approximately 2.4g / cm 3 The density of the microcrystalline phase is about 2.7g / cm 3 The two-phase density distribution parameter is calculated by the weighted average of volume fraction and density, that is, the volume fraction of the glass phase multiplied by the density of the glass phase plus the volume fraction of the microcrystalline phase multiplied by the density of the microcrystalline phase, for example, 0.35×2.4+0.65×2.7=2.595g / cm 3 The dual-phase density distribution parameter reflects the spatial distribution characteristics of the internal density of glass-ceramics. This parameter can quantify the physical properties of the dual-phase structure while ensuring the real-time and economical acquisition of the parameter.

[0031] The specific implementation process of calculating the acoustic impedance difference based on the difference in dual-phase density distribution parameters and elastic modulus is as follows: Acoustic impedance is defined as the product of material density and sound wave propagation velocity, and sound wave propagation velocity is equal to the square root of the elastic modulus divided by the density. The elastic modulus of the glass phase is 75GPa, corresponding to a sound wave velocity of 5590m / s, and the acoustic impedance of the glass phase is 2.4×5590=13416kg / (m 2 ·s); the elastic modulus of the microcrystalline phase is 120GPa, the corresponding sound wave velocity is 6667m / s, and the acoustic impedance of the microcrystalline phase is 2.7×6667=18001kg / (m 2 ·s), the acoustic impedance difference is 18001-13416=4585kg / (m 2·s). The two-phase interface response characteristic quantitatively describes the physical phenomenon of reflection and transmission of vibration waves at the interface between the glass phase and the microcrystalline phase, including two key parameters: reflection coefficient and transmission coefficient. The reflection coefficient is calculated as the difference in acoustic impedance divided by the sum of the acoustic impedances: 4585÷(13416+18001)=0.146; the transmission coefficient is calculated as twice the acoustic impedance of the incident medium divided by the sum of the acoustic impedances: 2×13416÷(13416+18001)=0.854. The two-phase interface response characteristic indicates that 14.6% of the vibration wave energy is reflected at the phase interface, while 85.4% of the energy is transmitted into the other phase. This energy distribution pattern determines the propagation characteristics and amplitude distribution of vibration waves of different frequencies in the two-phase structure, providing a physical basis for phase separation of vibration signals.

[0032] The specific implementation process for frequency-domain decomposition of raw vibration data based on the response characteristics of the two-phase interface is to utilize the physical properties of reflection and transmission coefficients to classify vibration signals by frequency range. The physical principle of frequency-domain decomposition is that low-frequency vibrations have longer wavelengths and can propagate across multiple phase interfaces. During propagation, they are primarily influenced by the low acoustic impedance of the glass phase, so the low-frequency vibration components primarily reflect the response characteristics of the glass phase. High-frequency vibrations have shorter wavelengths and primarily propagate within a single phase domain. The high acoustic impedance of the microcrystalline phase makes them more sensitive to high-frequency vibrations, so the high-frequency vibration components primarily reflect the response characteristics of the microcrystalline phase. The specific operation process is to perform a fast Fourier transform on the raw vibration data to obtain a frequency-domain spectrum. Based on the 400Hz cutoff frequency determined by the acoustic impedance difference, the components below 400Hz in the spectrum are extracted as low-frequency vibration components and attributed to the glass phase response, while the components above 400Hz are extracted as high-frequency vibration components and attributed to the microcrystalline phase response. The two spectral data are then subjected to an inverse fast Fourier transform to obtain the corresponding time-domain vibration signals. The initial phase separation vibration data refers to the vibration data obtained by frequency domain decomposition and is classified by phase, including low-frequency time domain signals belonging to the glass phase and high-frequency time domain signals belonging to the microcrystalline phase. For example, when the original vibration signal contains 200Hz amplitude 0.8m / s 2 and 600Hz amplitude 0.5m / s 2 When the components are present, the glass phase response signal is obtained after frequency domain decomposition as 200Hz, 0.8m / s 2 , the microcrystalline phase response signal is 600Hz, 0.5m / s 2 This decomposition method can distinguish the vibration response of the composite material according to its physical source, providing a data basis for subsequent precise phase separation analysis.

[0033] The specific implementation process for grain scattering correction based on the grain size and orientation distribution characteristics of the microcrystalline phase is as follows: The grain size and orientation distribution characteristics are indirectly determined through the preparation process parameters. Grain size is primarily controlled by the crystallization temperature and holding time. Higher crystallization temperatures and longer holding times result in larger grain sizes, with grain diameters ranging from 5 to 15 μm within the temperature range of 780-820°C. Orientation distribution is primarily influenced by the cooling rate: rapid cooling results in random orientation, while slow cooling leads to preferred orientation. Grain boundary scattering refers to the energy attenuation and direction change that occurs when vibration waves propagate within the microcrystalline phase and encounter grain boundaries. This phenomenon reduces the amplitude of high-frequency vibration components and deflects the propagation path. The grain scattering correction calculation process involves calculating a scattering attenuation factor based on the average grain diameter. The scattering attenuation factor is inversely proportional to the product of the grain diameter and the vibration frequency. A scattering directionality correction factor is calculated based on the degree of randomness in the orientation distribution. The correction factor for random orientation is 1.0, while that for preferred orientation is 0.8-1.2. For the microcrystalline phase response part in the preliminary phase separation vibration data, its amplitude is divided by the scattering attenuation factor for amplitude correction, and its phase is subtracted from the delay caused by scattering for phase correction to obtain the corrected microcrystalline phase vibration component. For example, when the average grain diameter is 10μm and the orientation is randomly distributed, the scattering attenuation factor of the 600Hz vibration component is 0.85, and the corrected amplitude is 0.5÷0.85=0.588m / s 2 The glass phase vibration components remain initially separated because the glass phase is amorphous and has no grain boundary scattering. This correction eliminates the interference of the microcrystalline structure on vibration propagation, obtaining a signal that truly reflects the intrinsic vibration characteristics of each phase, significantly improving the accuracy and stability of thickness detection.

[0034] Please continue reading Figure 1 , obtaining the three-dimensional position coordinates of the grinding head, mapping the thickness characteristic parameters to the three-dimensional position coordinates, processing the thickness characteristic parameters of adjacent measuring points using a weighted average algorithm, and obtaining the spatial distribution of the thickness characteristic parameters; In one embodiment of the present invention, the three-dimensional position coordinates of the grinding head are obtained, the thickness characteristic parameters are mapped to the three-dimensional position coordinates, and the thickness characteristic parameters of adjacent measuring points are processed using a weighted average algorithm to obtain the spatial distribution of the thickness characteristic parameters, including: Obtain the position coordinate data of the grinding head in the X-axis, Y-axis, and Z-axis directions, and divide the workpiece surface into regular grids based on the path spacing and overlap rate of the grinding tracks to obtain spatial grid units; The low-medium frequency differential coefficient, the medium-high frequency differential coefficient and the full-band comprehensive coefficient of the thickness characteristic parameter are temporally and spatially matched with the three-dimensional position coordinates of the grinding head at the corresponding moment, and assigned to the corresponding spatial grid units to obtain gridded thickness characteristic data; According to the effective radius range of vibration propagation in the glass-ceramics, the adjacent influence area of ​​each grid unit is determined, and the spatial distance weight coefficient between adjacent grid units is calculated to obtain the spatial weight matrix; According to the spatial weight matrix, a weighted average operation is performed on the gridded thickness characteristic data, and a boundary constraint correction process is applied to the grid cells in the edge area of ​​the workpiece to obtain the spatial distribution of the thickness characteristic parameters.

[0035] The following is a detailed description of the steps involved in the above embodiment: The specific implementation process for obtaining the grinding head's position coordinate data in the X, Y, and Z axes is as follows: Real-time 3D position coordinates of the grinding head are obtained through encoder feedback from the grinding equipment's numerical control system. The numerical control system outputs position data every 1 millisecond with an accuracy of 1 micron. The X-axis represents the grinding head's position along the workpiece's length, the Y-axis represents its position along the workpiece's width, and the Z-axis represents its position along the workpiece's thickness. A grinding track is the path the grinding head moves along the workpiece surface, including linear scanning, spiral scanning, or raster scanning. Path spacing refers to the distance between two adjacent grinding paths, and the overlap ratio refers to the percentage of the path width that overlaps adjacent grinding paths. The regular meshing process involves determining the grid cell size based on the path spacing and overlap ratio. For example, when the path spacing is 0.5 mm and the overlap ratio is 20%, the grid cell size is set to 0.4 mm × 0.4 mm. The workpiece surface is then divided into evenly spaced sections according to this size, forming a two-dimensional grid array. Spatial grid cells are each independent region of the workpiece surface, each with a unique grid coordinate identifier. For example, for a 50mm x 30mm workpiece surface, a 0.4mm pitch division yields 125 x 75 spatial grid cells, for a total of 9,375 cells. This regular gridding discretizes the continuous workpiece surface into a finite number of manageable spatial cells, providing a unified coordinate framework for subsequent spatial data processing while ensuring that the grid density matches the grinding accuracy.

[0036] The specific implementation process for temporally and spatially synchronizing the low-mid-frequency differential coefficients, mid-high-frequency differential coefficients, and full-band comprehensive coefficients with the three-dimensional position coordinates of the grinding head is as follows: timestamp synchronization technology is used to ensure the temporal correspondence between the thickness characteristic parameters and the position coordinates. The vibration signal acquisition system and the CNC positioning system use a unified clock reference, achieving a time synchronization accuracy of 0.1 millisecond. Temporally and spatially synchronizing the thickness characteristic parameters obtained at the same moment is paired with the position coordinates of the grinding head. The specific operation process is as follows: the low-mid-frequency differential coefficients, mid-high-frequency differential coefficients, and full-band comprehensive coefficients at a specific moment are read, along with the X, Y, and Z coordinates of the grinding head at that moment; the corresponding spatial grid cell is determined based on the X and Y coordinates of the grinding head, and the three differential coefficient values ​​at that moment are assigned to the corresponding grid cell. For example, at t = 100ms, the low-medium frequency differential coefficient is -400Hz, the medium-high frequency differential coefficient is -650Hz, the full-band comprehensive coefficient is 500Hz, and the grinding head position is X = 10.2mm, Y = 5.8mm. This set of data is assigned to the spatial grid unit with grid coordinates of (26, 15). Gridded thickness feature data refers to the set of thickness feature parameter values ​​contained in each spatial grid unit after spatial allocation. This spatiotemporal matching process can convert the thickness feature parameters of the time series into spatially distributed thickness feature data, realizing the mapping conversion from time domain information to spatial domain information, and ensuring that the thickness characteristics of each spatial position have clear data support.

[0037] The specific implementation process for determining adjacent influence regions based on the effective radius of vibration propagation in glass-ceramics is as follows: The effective radius refers to the spatial distance range within which vibration waves propagating in glass-ceramics can produce a significant response. This range is primarily influenced by the material's elastic modulus, density, and vibration frequency. Acoustic theory calculates that the effective radius of vibration waves in glass-ceramics is approximately 2-3 mm. Beyond this distance, the vibration response amplitude decays to a negligible level. The adjacent influence region is defined as the area encompassing all grid cells within the effective radius, centered on a particular grid cell. The process for determining the adjacent influence region is as follows: A circular region with an effective radius of 2.5 mm is drawn around the center point of each spatial grid cell. All grid cells within this circular region constitute the adjacent influence region of that central cell. The spatial distance weight coefficient is a weight calculated based on the Euclidean distance between grid cells. Closer distances result in a greater weight, while greater distances result in a smaller weight. The calculation method involves measuring the straight-line distance between the center point of the central grid cell and the center point of each grid cell within the adjacent influence region. The weight coefficient is calculated using a Gaussian function, which is equal to the exponential value of the negative square of the distance divided by twice the variance. The spatial weight matrix is ​​a two-dimensional array containing weight coefficients between all grid cells, with each row and column corresponding to a different grid cell. For example, when the distance between two grid cells is 1 mm, the weight coefficient is 0.8; when the distance is 2 mm, the weight coefficient is 0.3; and when the distance exceeds 3 mm, the weight coefficient approaches 0. This weight calculation method quantifies the influence of spatial position on vibration propagation and provides a physical basis for spatial data fusion.

[0038] The specific implementation process for weighted averaging gridded thickness characteristic data based on the spatial weight matrix is ​​as follows: For each spatial grid cell, the thickness characteristic parameter values ​​of each grid cell within its adjacent influence area are multiplied by the corresponding weight coefficient. The sum is then divided by the sum of the weight coefficients to obtain the weighted average thickness characteristic parameter for that grid cell. Weighted averaging can improve the accuracy and stability of single-point measurements by leveraging information from adjacent areas. The specific calculation process is as follows: For grid cell (i, j), the low-mid-frequency differential coefficient, mid-high-frequency differential coefficient, and full-band comprehensive coefficient of all grid cells within its adjacent influence area are read. The weighted average of these three parameters is then calculated. The weighted average of each parameter is equal to the sum of the product of the parameter value and the weight coefficient for each adjacent cell divided by the sum of the weight coefficients. Boundary constraint correction refers to the process of applying special processing to grid cells at the edge of the workpiece. Because the adjacent influence area of ​​edge grid cells is incomplete, direct weighted averaging will result in deviations. The boundary constraint correction method is: for edge grid cells, only the adjacent cells located within the workpiece are considered, and the weight coefficients are normalized to ensure that the sum of the weight coefficients is 1. The spatial distribution of thickness characteristic parameters refers to the distribution of thickness characteristic parameters across the workpiece surface for all spatial grid cells after weighted averaging and boundary correction. For example, the original low-to-mid-frequency differential coefficient for a grid cell is -400Hz, but after weighted averaging across 3×3 adjacent regions, it is corrected to -385Hz. The standard deviation of the spatial distribution is reduced from 50Hz to 20Hz. This spatial processing eliminates random errors associated with single-point measurements, improving the continuity and reliability of the spatial distribution of thickness characteristic parameters while maintaining the accuracy of data in edge areas.

[0039] Please continue reading Figure 1 , performing process interference compensation on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and performing time series smoothing processing on the compensated thickness characteristic parameters using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution; In one embodiment of the present invention, the process interference compensation is performed on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and the compensated thickness characteristic parameters are smoothed in time series using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution, including: Acquiring real-time change data of the grinding force, calculating a grinding force reference offset according to a workpiece stiffness decreasing law, and performing grinding force offset compensation on the glass phase response component and the microcrystalline phase response component in the spatial distribution of the thickness characteristic parameter to obtain dual-phase compensation data; According to the spatial gradient change of the grinding feed direction, the dual-phase compensation data is subjected to directional weighted correction to eliminate the anisotropic effect of the grinding track direction on the thickness feature detection, thereby obtaining an isotropic thickness feature distribution; Based on the time scale difference between the grinding cycle and the thickness change cycle, a double-layer nested sliding window is set. The inner window tracks the short-term changes in the grinding cycle, and the outer window tracks the long-term trend of the thickness change. The isotropic thickness characteristic distribution is smoothed by a double layer time series to obtain multi-scale smoothed data. The continuity index of the multi-scale smoothed data in the spatially adjacent regions and the stability index of the temporally continuous windows are calculated. When the continuity index and the stability index simultaneously meet the convergence conditions, the corresponding data are output as a stable thickness characteristic distribution.

[0040] The following is a detailed description of the steps involved in the above embodiment: The specific implementation process for obtaining real-time grinding force data is as follows: tangential, normal, and axial force data are obtained during the grinding process using a three-axis dynamometer mounted on the grinding spindle. The sampling frequency is set to 1000 Hz to capture the dynamic changes in grinding force. The workpiece stiffness reduction law refers to the physical law that as the thickness of the microcrystalline glass decreases, the bending stiffness of the workpiece decreases according to the cube of the thickness. When the thickness decreases from 0.5mm to 0.3mm, the stiffness decreases by approximately 65%. The grinding force baseline offset refers to the deviation of the grinding force from the initial baseline value due to the change in workpiece stiffness. The calculation method is: if the baseline grinding force at the start of grinding is 0.5mm thick and is recorded as 100N, and the actual grinding force at the current thickness of 0.3mm is 75N, then the grinding force baseline offset is 25N. Grinding force offset compensation refers to the data processing process that corrects thickness characteristic parameters based on the grinding force baseline offset. The glassy and microcrystalline phase response components have different sensitivities to grinding force variations. The correction factor for the glassy phase response component is 0.8 times the grinding force offset, and the correction factor for the microcrystalline phase response component is 1.2 times the grinding force offset. The specific compensation operation is: add the grinding force offset multiplied by 0.8 to the glassy phase response component, and add the grinding force offset multiplied by 1.2 to the microcrystalline phase response component. For example, when the grinding force baseline offset is 25N, the correction value for the glassy phase response component is 20Hz, and the correction value for the microcrystalline phase response component is 30Hz. Dual-phase compensation data refers to the data set of the glassy and microcrystalline phase response components after grinding force offset compensation. This compensation process eliminates the interference of workpiece stiffness changes on vibration characteristic detection, ensuring that thickness measurement accuracy is not affected by the grinding process.

[0041] The specific implementation process for directional weighted correction based on spatial gradient variations in the grinding feed direction is as follows: The grinding feed direction refers to the direction of movement of the grinding head relative to the workpiece surface, including the positive X-axis, the positive Y-axis, and their combination. Spatial gradient variation refers to the difference in the rate of change of the thickness characteristic parameter in different directions. Due to the directional characteristics of the grinding track, the sensitivity of thickness variation at the same location in different directions varies. Anisotropy refers to the phenomenon in which the thickness characteristic detection exhibits different response characteristics in different spatial directions due to the grinding track direction, manifesting as higher detection accuracy along the grinding direction than in the perpendicular grinding direction. The operational process of directional weighted correction is as follows: the gradient values ​​of the dual-phase compensation data in the X-axis and Y-axis directions are calculated. The gradient value is equal to the difference in the thickness characteristic parameter between adjacent grid cells divided by the grid spacing. The directional weight coefficient is determined based on the grinding feed direction: the weight coefficient along the grinding direction is 1.0, the weight coefficient perpendicular to the grinding direction is 0.7, and the weight coefficient at a 45-degree angle is 0.85. The gradient value in each direction is multiplied by the corresponding weight coefficient and then vectorized to obtain the corrected thickness characteristic parameter. For example, when grinding along the positive X-axis, the X-direction gradient is 50 Hz / mm, the Y-direction gradient is 30 Hz / mm, and the resulting gradient after directional weighting correction is 45 Hz / mm. An isotropic thickness characteristic distribution refers to a data distribution state in which, after directional weighting correction, the thickness characteristic parameters exhibit consistent response characteristics in all spatial directions. This correction eliminates the systematic impact of grinding process directionality on inspection results, ensuring that the thickness characteristic distribution truly reflects the workpiece's geometric characteristics rather than process characteristics.

[0042] The specific implementation process of setting up a double-layer nested sliding window based on the time scale difference between the grinding cycle and the thickness variation cycle is as follows: the grinding cycle is the time it takes for the grinding spindle to complete one rotation. For example, at a spindle speed of 3000 rpm, the grinding cycle is 0.02 seconds. The thickness variation cycle is the time required for a significant change in workpiece thickness, which is mainly determined by the grinding feed rate. For example, at a feed rate of 50 mm / min, the thickness variation cycle is approximately 2-5 seconds. The double-layer nested sliding window is a time data processing framework consisting of an inner window and an outer window. The inner window time length is set to 3-5 times the grinding cycle, that is, 0.06-0.1 seconds, and is used to track short-term changes within the grinding cycle. The outer window time length is set to 2-3 times the thickness variation cycle, that is, 4-15 seconds, and is used to track long-term thickness trends. The operational process of double-layer time series smoothing is as follows: within the inner window range, a moving average filter is applied to the isotropic thickness characteristic distribution to eliminate high-frequency noise and random disturbances during the grinding process; within the outer window range, a trend filter is applied to the data processed by the inner window to extract the long-term variation trend of the thickness characteristic and suppress medium-frequency interference. For example, inner window processing can reduce the standard deviation of the thickness characteristic parameter from ±15Hz to ±8Hz, and outer window processing can further reduce it to ±3Hz. Multi-scale smoothed data refers to time series data obtained after double smoothing with the inner and outer windows. This data not only retains the effective information of thickness variation but also filters out noise interference at various time scales. This double-layer processing method can simultaneously handle the short-term randomness of the grinding process and the long-term trend of thickness variation, significantly improving the stability and reliability of the data.

[0043] The specific implementation process for calculating the continuity index and stability index of multi-scale smoothed data is as follows: The continuity index is a quantitative indicator of the smoothness of the thickness characteristic parameter variation within a spatially adjacent region. It is calculated as the inverse of the standard deviation of the thickness characteristic parameter difference between adjacent grid cells. A larger continuity index indicates better spatial continuity. The spatially adjacent region refers to a 3×3 grid area centered on a grid cell and containing 9 grid cells. The stability index is a quantitative indicator of the degree of fluctuation of the thickness characteristic parameter within a temporally continuous window. It is calculated as the inverse of the coefficient of variation of the thickness characteristic parameter at 10 consecutive time points. A larger stability index indicates better temporal stability. The temporally continuous window refers to 10 consecutive data acquisition time points, corresponding to a time span of 0.1 seconds. The convergence condition refers to the judgment criteria that both the continuity index and the stability index reach preset thresholds. The continuity index threshold is set at 50, and the stability index threshold is set at 30. These two thresholds are set based on the accuracy requirements of microcrystalline glass thickness detection and statistical analysis of actual test data. The specific judgment process is as follows: the continuity index and stability index are calculated every 0.01 seconds. When the results of five consecutive calculations meet the threshold requirements, the convergence condition is determined to be met, and the multi-scale smoothed data at the current moment is output as a stable thickness characteristic distribution. For example, when the continuity index reaches 55 and the stability index reaches 35, the system determines that the data is stable and outputs the results. A stable thickness characteristic distribution refers to thickness characteristic parameter distribution data that has been confirmed by convergence judgment and has good spatial continuity and temporal stability. This convergence judgment mechanism ensures the quality and reliability of the output data and avoids the accumulation of errors caused by subsequent processing when the data is not stable.

[0044] In one embodiment of the present invention, based on the time scale difference between the grinding cycle and the thickness change cycle, a double-layer nested sliding window is set, the inner window tracks the short-term changes in the grinding cycle, and the outer window tracks the long-term trend of the thickness change. The isotropic thickness characteristic distribution is subjected to double-layer time series smoothing to obtain multi-scale smoothed data, including: According to the grinding spindle speed and the grinding head feed speed, the time length of a single grinding cycle is calculated, and the time span of the inner sliding window is set to 2-5 times the length of the grinding cycle to obtain the short-time window parameters; Based on the physical continuity constraint of the workpiece thickness change, the time scale of significant thickness feature changes is calculated, the time span of the outer sliding window is set to 1.5-3 times the thickness change time scale, the nesting ratio relationship between the outer window and the inner window is determined, and the long-term window parameters are obtained; Within the short-time window parameter range, high-frequency noise filtering and grinding vibration synchronization processing are performed on the isotropic thickness characteristic distribution to retain effective signal components related to thickness and obtain short-time smoothed data; Within the long-term window parameter range, trend extraction and outlier detection processing are performed on the short-term smoothed data, and segmented weighted averaging is performed in combination with the stage characteristics of the grinding process to obtain multi-scale smoothed data; The smoothness evaluation index of the multi-scale smoothed data in a continuous time period is calculated. When the smoothness evaluation index reaches a convergence threshold, the data processing is confirmed to be completed and the result is output.

[0045] The following is a detailed description of the steps involved in the above embodiment: The specific implementation process for calculating the duration of a single grinding cycle based on the grinding spindle speed and grinding head feed rate is as follows: the grinding spindle speed is obtained from the grinding equipment's encoder feedback in revolutions per minute (rpm), and the grinding head feed rate is obtained from the CNC system's servo drive feedback in millimeters per minute (mm / min). The duration of a single grinding cycle is equal to 60 seconds divided by the grinding spindle speed. For example, at a spindle speed of 3000 rpm, the duration of a single grinding cycle is 60 ÷ 3000 = 0.02 seconds. The inner sliding window is a time window used to process short-term data changes. Its time span is set to 2-5 times the grinding cycle duration. This range is based on the periodic characteristics of the vibration signal during the grinding process and the need for noise suppression. For a grinding cycle of 0.02 seconds, the time span of the inner sliding window is set to 0.04-0.1 seconds. The specific value is determined by the signal-to-noise ratio requirement. A larger time span is selected for lower signal-to-noise ratios to achieve better smoothing. The short-term window parameters refer to the combination of the inner sliding window's time span, window overlap, and sliding step size. The time span determines the degree of smoothing, the overlap is set to 50% to ensure data continuity, and the sliding step size is set to one-tenth of the grinding cycle length to ensure sufficient temporal resolution. For example, when selecting 3 times the grinding cycle length, the short-term window parameters are a time span of 0.06 seconds, a 50% overlap, and a sliding step size of 0.002 seconds. A range of 2-5 times the time span effectively suppresses high-frequency random noise while maintaining sensitivity to short-term variations within the grinding cycle, ensuring the accuracy and stability of short-term data processing.

[0046] The specific implementation process for calculating the timescale of significant thickness changes based on the physical continuity constraint of workpiece thickness variation is as follows: The physical continuity constraint states that the thickness change during the grinding process of a glass-ceramic workpiece must follow the physical laws of material removal. Specifically, the rate of thickness change is limited by grinding parameters and material properties, and no instantaneous jumps occur. The timescale of significant thickness changes is calculated by analyzing the grinding feed rate and material removal rate. At a feed rate of 50 mm / min and a material removal rate of 5 μm / pass, a 10 μm thickness reduction takes approximately 2-4 seconds, representing the timescale of significant thickness changes. The time span of the outer sliding window is set to 1.5-3 times the thickness change timescale, or 3-12 seconds. This range is determined by the need for long-term trend analysis and anomaly detection. The nested proportional relationship, defined as the ratio of the outer window time span to the inner window time span, is controlled between 30 and 200 times to ensure that the two windows process signal features at different timescales. For example, when the inner window span is 0.06 seconds and the outer window span is 6 seconds, the nesting ratio is 100x. The long-term window parameters refer to the combination of the outer sliding window's time span, the trend extraction method, and the anomaly detection threshold. The time span is 3-12 seconds, with linear regression used for trend extraction and a threshold of 3 standard deviations for anomaly detection. A 1.5-3x multiplier range captures long-term trends in thickness changes while avoiding response delays caused by oversmoothing, ensuring accurate and timely long-term data processing.

[0047] The specific implementation process for high-frequency noise filtering and grinding vibration synchronization processing for isotropic thickness characteristic distribution within a short-time window parameter range is as follows: High-frequency noise filtering is performed using a Butterworth low-pass digital filter with a cutoff frequency set to 5 times the grinding fundamental frequency and an order set to 4 to achieve good frequency response characteristics. Grinding vibration synchronization processing involves aligning the thickness characteristic signal with the rotational phase of the grinding spindle. By detecting the zero-phase signal of the grinding spindle as a synchronization reference, the thickness characteristic data is rearranged and averaged according to the grinding cycle. The effective signal component, which is the vibration signal component directly related to the workpiece thickness variation, is primarily distributed within the grinding fundamental frequency and its 2-3 harmonics. Signals within this frequency range are extracted using a bandpass filter. The specific processing flow is as follows: The isotropic thickness characteristic distribution data is low-pass filtered to remove high-frequency noise, then band-pass filtered to extract the effective signal component, and finally, phase synchronization processing is performed to eliminate the random effects of grinding vibration. For example, the original signal's signal-to-noise ratio (SNR) is 15dB, which increases to 25dB after high-frequency noise filtering and further to 35dB after synchronization. Short-term smoothed data, obtained after high-frequency noise filtering and grinding vibration synchronization, retains the rapid response to thickness changes while eliminating short-term random disturbances. This processing method significantly improves the SNR and stability of the data while maintaining high sensitivity to thickness changes.

[0048] The specific implementation process for trend extraction and outlier detection of short-term smoothed data within the long-term window parameter range is as follows: Trend extraction uses a least-squares linear regression algorithm, and a linear fit is performed on the short-term smoothed data within the long-term window range to extract the long-term trend of the thickness characteristics. Outlier detection refers to the process of identifying and removing data points that deviate from the normal trend. The detection criterion is that the deviation of the data point from the trend line exceeds three standard deviations. Data points identified as outliers are replaced with the trend value. The stage characteristics of the grinding process refer to the different characteristics exhibited at different stages of the grinding process, including high grinding forces in the early stage of grinding, stable conditions in the middle stage, and reduced grinding forces in the late stage. These stage characteristics can affect the variation pattern of the thickness characteristics. Segmented weighted averaging is a processing method that assigns different weights to data from different time periods based on the stage characteristics of the grinding process. The weight for the stable grinding period is set to 1.0, and the weights for the early and late stages of grinding are set to 0.8 to reduce the impact of the transition stage on the overall trend. The specific operation process is as follows: the long-term window is divided into 3-5 time periods according to the grinding process. Trend extraction and outlier detection are performed on the short-term smoothed data within each time period. Then, a weighted average is calculated according to the stage weights. For example, when the long-term window is 6 seconds, it is divided into three stages: initial 1 second, stable 4 seconds, and late 1 second, with weights of 0.8, 1.0, and 0.8, respectively. Multi-scale smoothed data refers to data obtained after trend extraction, outlier detection, and segmented weighted averaging. This data reflects the long-term trend of thickness variation while maintaining good data quality. This multi-scale processing method can simultaneously handle short-term fluctuations and long-term trends, obtaining high-quality thickness characteristic data.

[0049] The specific implementation process for calculating the smoothness evaluation index of multi-scale smoothed data over a continuous time period is as follows: The smoothness evaluation index is a numerical indicator used to quantify the degree of data smoothness. It is represented by the ratio of the standard deviation to the mean of the data over a continuous time period. A smaller ratio indicates smoother data. The continuous time period is set to 10 data collection cycles, corresponding to a time span of approximately 0.1 seconds. The standard deviation and mean of the multi-scale smoothed data are calculated over this time period. The convergence threshold is a standard value used to determine whether data processing is complete. It is set to a smoothness evaluation index less than 0.05. This threshold is selected based on the accuracy requirements of microcrystalline glass thickness testing and statistical analysis of actual test data. The specific calculation process is as follows: The smoothness evaluation index is calculated every data collection cycle. The smoothness evaluation index is obtained by dividing the standard deviation of the 10 data points in the continuous time period by the mean. This index is then compared with the convergence threshold of 0.05. Data processing is confirmed to be complete when the results of three consecutive calculations are less than the convergence threshold. For example, when the standard deviation of the data over a continuous time period is 2 Hz and the mean is 500 Hz, the smoothness evaluation index is 0.004, which is less than the convergence threshold of 0.05 and meets the convergence condition. The determination of data processing completion ensures that the output data quality meets the expected requirements, and the output is multi-scale smoothed data that meets the smoothness requirements. A convergence threshold of 0.05 ensures that the output data smoothness meets the accuracy requirements of subsequent thickness prediction and quality assessment, while also avoiding response delays caused by excessive processing.

[0050] Please continue reading Figure 1 , according to the stable thickness characteristic distribution and the current grinding process, the thickness variation of the remaining processing area is calculated, the final thickness distribution after the grinding is completed is predicted, and the thickness qualification prediction result is obtained; In one embodiment of the present invention, the thickness variation of the remaining processing area is calculated based on the stable thickness characteristic distribution and the current grinding process, and the final thickness distribution after the grinding is completed is predicted to obtain the thickness qualification prediction result, including: According to the current grinding completion ratio and grinding trajectory path data, the spatial distribution of the remaining unprocessed area is identified, the boundary gradient change rate between the processed area and the remaining area is calculated, and the thickness transition characteristics between the areas are obtained; Based on the thickness decreasing law of the processed area in the stable thickness characteristic distribution, combined with the current workpiece stiffness decreasing coefficient and the accumulated wear of the grinding tool, the material removal efficiency variation coefficient of the remaining processed area in the subsequent grinding process is calculated to obtain the nonlinear thickness variation; Based on the thickness transition characteristics and nonlinear thickness variation between the regions, a thickness prediction operation is performed on the remaining unprocessed region, and a predicted thickness distribution map after grinding is generated in combination with the boundary constraints of the edge region; The thickness deviation value and thickness uniformity index of each area in the predicted thickness distribution diagram are calculated, the thickness deviation value is compared with the allowable tolerance range of the technical specification, and the thickness uniformity index is matched and evaluated with the optical performance requirements to obtain a thickness qualification prediction result.

[0051] The following is a detailed description of the steps involved in the above embodiment: The specific implementation process for identifying the spatial distribution of the remaining unmachined area based on the current grinding completion ratio and grinding path data is as follows: The current grinding completion ratio is obtained through the progress monitoring function of the grinding equipment's CNC system and represents the percentage of the workpiece surface area that has been ground to the total surface area. The grinding path data includes the coordinates of all path points traversed by the grinding head and is updated in real time by the CNC system's path recording function. The remaining unmachined area refers to the area on the workpiece surface not yet covered by the grinding head and is calculated by subtracting the area of ​​the machined area from the total workpiece surface area. The identification process involves dividing the workpiece surface into a regular grid, marking the grid cells that the grinding head has passed through as the machined area, and unmarked grid cells as the remaining unmachined area. The boundary gradient change rate refers to the spatial change rate of the thickness characteristic parameter at the junction of the machined area and the remaining area. It is calculated by dividing the difference in thickness characteristic parameters between adjacent grid cells by the grid spacing. The specific calculation process is as follows: At the boundary between the machined area and the remaining area, three grid cells on each side of the boundary are selected. The gradient of the thickness characteristic parameter between these six cells is calculated, and then the gradient values ​​are statistically analyzed to obtain the boundary gradient change rate. For example, if the thickness characteristic parameter at the boundary of the machined region is 500 Hz and at the boundary of the remaining region is 520 Hz, and the grid spacing is 0.4 mm, then the boundary gradient change rate is 50 Hz / mm. The inter-region thickness transition characteristic refers to the change pattern of the thickness characteristic parameter when transitioning from the machined region to the remaining region, including the value, direction, and spatial distribution of the gradient change rate. This boundary analysis can reveal the continuity of thickness variation during the grinding process and provide important spatial constraints for predicting the thickness distribution in the remaining region.

[0052] The specific implementation process for calculating the material removal efficiency variation coefficient based on the thickness decrease law of the machined area within a stable thickness characteristic distribution is as follows: The thickness decrease law refers to the mathematical law that determines the thickness change of the machined area as the grinding process progresses. It is extracted through time-series analysis of the thickness characteristic distribution of the machined area. The workpiece stiffness decrease coefficient is the proportional coefficient of the decrease in overall workpiece stiffness according to the cubic relationship of thickness as the workpiece thickness decreases. When the thickness decreases by 10% from the initial value, the stiffness decrease coefficient is approximately 0.27. The accumulated wear of the grinding tool is calculated based on grinding time, grinding parameters, and machined area, indicating the impact of the degree of grinding tool dullness on machining efficiency. The material removal efficiency variation coefficient is the ratio of the change in material removal rate of the remaining machining area relative to the machined area, comprehensively considering the effects of workpiece stiffness changes and tool wear. The calculation process is as follows: Analyze the thickness decrease data of the machined area and extract the functional relationship between material removal efficiency and thickness; adjust this functional relationship based on the current workpiece stiffness decrease coefficient; a decrease in stiffness will result in an increase in material removal efficiency of approximately 15-25%; and further correct it based on the accumulated wear of the grinding tool; tool wear will result in a decrease in material removal efficiency of approximately 5-15%. The nonlinear thickness variation refers to the deviation of the thickness variation of the remaining machined area from the linear expected value after taking into account the stiffness change and tool wear. For example, if the linear expected thickness variation is 50μm, it will increase by 12μm after taking into account the stiffness decrease and decrease by 5μm after taking into account tool wear, resulting in a final nonlinear thickness variation of 57μm. This nonlinear analysis can accurately predict the processing effect in the later stages of grinding, avoiding prediction errors caused by changes in the workpiece and tool state.

[0053] The specific implementation process for thickness prediction based on inter-regional thickness transition characteristics and nonlinear thickness variation is as follows: Thickness prediction refers to the numerical calculation process of estimating the final thickness distribution of the remaining unmachined area based on known information. The prediction method is: starting from the boundary conditions provided by the inter-regional thickness transition characteristics, spatial interpolation is performed into the remaining unmachined area. The interpolation results are corrected by combining the nonlinear thickness variation to ensure that the predicted value conforms to the physical laws of material removal. Boundary constraints refer to the geometric and physical constraints of the workpiece edge region, including support conditions, boundary effects, and stress distribution characteristics at the edge. The specific operation process is: the remaining unmachined area is divided into several sub-regions, and the cubic spline interpolation method is used to predict the thickness of each sub-region. The boundary constraints are applied to the edge regions, and the interpolation results are corrected to account for the influence of edge effects. The predicted results are superimposed with the nonlinear thickness variation to obtain the final thickness prediction value. The predicted thickness distribution map is a visual representation of the thickness distribution of the remaining unmachined area after grinding, presented as a 2D contour map or a 3D surface map. For example, the predicted thickness at the center of the remaining area is 0.32mm, while the edge is predicted to be 0.34mm due to boundary effects, resulting in a thickness variation of 0.31-0.35mm across the entire area. This predictive calculation determines the final thickness distribution before grinding is complete, enabling early quality assessment and process optimization.

[0054] The specific implementation process for calculating the thickness deviation value and thickness uniformity index in the predicted thickness distribution map is as follows: The thickness deviation value refers to the difference between the thickness value of each area in the predicted thickness distribution map and the target thickness value. It is calculated by subtracting the absolute value of the target thickness of 0.30mm from the predicted thickness of each grid cell. The thickness uniformity index is a quantitative indicator of the uniformity of the thickness distribution in the predicted thickness distribution map. It is calculated as the inverse of the standard deviation of the predicted thickness distribution. A larger value indicates a more uniform thickness distribution. The allowable tolerance range of the technical specification refers to the thickness tolerance requirement for microcrystalline glass screen protectors, set at ±5% of the target thickness, i.e., 0.285-0.315mm. The optical performance requirements refer to the optical transparency and optical uniformity requirements of the screen protector. The thickness uniformity index must be greater than 200 to meet the optical performance requirements. The comparison and judgment process is as follows: the thickness deviation value of all grid cells in the predicted thickness distribution map is calculated, and the number and proportion of grid cells with deviation values ​​exceeding the allowable tolerance range are counted. The thickness uniformity index of the entire predicted thickness distribution map is calculated and compared with the optical performance threshold of 200. The matching evaluation process is as follows: when the proportion of grid cells with thickness deviations exceeding the tolerance range is less than 5% and the thickness uniformity index is greater than 200, it is judged as qualified; when any of the conditions is not met, it is judged as unqualified. The thickness qualification prediction result refers to the final quality status judgment of the product based on the analysis of the predicted thickness distribution map, including the classification of qualified, unqualified and risk level. For example, when the thickness deviation values ​​of 95% of the regions are within the range of ±0.010mm and the thickness uniformity index is 250, the thickness qualification prediction result is qualified. This prediction mechanism can determine the final quality status of the product before the grinding process is completed, realizing the early identification of unqualified products and the effective allocation of production resources.

[0055] Please continue reading Figure 1 , extract the time series variation characteristics of the thickness characteristic parameters during the grinding process as the vibration evolution index, combine the thickness qualification prediction results, judge the comprehensive quality status of the product, and obtain the real-time quality assessment results.

[0056] In one embodiment of the present invention, the time-series variation characteristics of the thickness characteristic parameters during the grinding process are extracted as vibration evolution indicators, combined with the thickness qualification prediction results to judge the comprehensive quality status of the product and obtain real-time quality assessment results, including: Gradient analysis and inflection point detection are performed on the variation trajectory of the low-medium frequency differential coefficient, medium-high frequency differential coefficient, and full-band comprehensive coefficient of the thickness characteristic parameter on the grinding time axis, and the vibration amplitude attenuation rate, frequency drift rate, and phase jump number are extracted as vibration evolution indicators; Based on the vibration response differences of the dual-phase structure of the glass-ceramic, the vibration evolution index is subjected to glass phase abnormal pattern recognition and microcrystalline phase abnormal pattern recognition, and signs of microcrack extension and stress concentration are detected to obtain a risk assessment of internal defects in the material; Cross-validate and analyze the thickness qualification prediction results with the material internal defect risk assessment, calculate the comprehensive weight coefficient of thickness geometry qualification and structural integrity qualification, and obtain a multi-dimensional quality comprehensive score; Based on the multi-dimensional quality comprehensive score, the product quality level is divided into three levels: excellent, qualified and risky. The defect type and risk degree of risky products are marked, and a real-time quality assessment result including quality grade and risk warning information is generated.

[0057] The following is a detailed description of the steps involved in the above embodiment: The specific implementation process for gradient analysis and inflection point detection of the variation trajectory of the low-mid-frequency differential coefficient, mid-high-frequency differential coefficient, and full-band comprehensive coefficient along the grinding time axis is as follows: The variation trajectory refers to the numerical sequence of thickness characteristic parameters changing with grinding time. A time series curve is formed by arranging the continuously collected differential coefficient data in chronological order. Gradient analysis is the process of calculating the numerical difference between adjacent time points in the variation trajectory. The gradient value is equal to the numerical difference between two adjacent points divided by the time interval. A positive gradient indicates an increase in the parameter, while a negative gradient indicates a decrease in the parameter. Inflection point detection is the process of identifying the location of a significant change in curvature in the variation trajectory. This is achieved using a second-order difference algorithm. When the sign of the first-order difference values ​​of three consecutive time points changes, the midpoint is identified as an inflection point. The vibration amplitude decay rate refers to the rate at which the vibration signal amplitude decreases over time. It is calculated by analyzing the time variation of the differential coefficient amplitude and is calculated as the rate of change of the ratio of the current amplitude to the initial amplitude over time. The frequency drift rate refers to the rate at which the main frequency of the vibration signal shifts over time. It is calculated by analyzing the time variation of the differential coefficient frequency characteristics. The number of phase jumps refers to the frequency of sudden changes in the phase of the vibration signal, and is calculated by detecting the discontinuities in the phase of the differential coefficient. For example, when the low-medium frequency differential coefficient changes from -400 Hz to -380 Hz within 2 seconds, the gradient value is 10 Hz / s; when five inflection points are detected, this corresponds to five phase jumps. The vibration evolution index refers to a set of parameters used to describe the time evolution characteristics of the vibration signal, including three key indicators: vibration amplitude attenuation rate, frequency drift rate, and number of phase jumps. This time series analysis can capture the dynamic changes in the vibration characteristics of microcrystalline glass during the grinding process, providing a sensitive indicator signal for identifying changes in the internal state of the material.

[0058] The specific implementation process for identifying abnormal patterns in vibration evolution indicators based on the vibration response differences of the two-phase structure of microcrystalline glass is as follows: Abnormal pattern identification of the glass phase refers to the data analysis process of identifying abnormal conditions based on the vibration response characteristics of the glass phase. Normal vibration evolution indicators for the glass phase range from an amplitude decay rate of 0.02-0.05 / s, a frequency drift rate of 1-3 Hz / s, and 2-5 phase jumps per minute. Abnormal pattern identification of the microcrystalline phase refers to the data analysis process of identifying abnormal conditions based on the vibration response characteristics of the microcrystalline phase. Normal vibration evolution indicators for the microcrystalline phase range from an amplitude decay rate of 0.03-0.08 / s, a frequency drift rate of 2-6 Hz / s, and 3-8 phase jumps per minute. Microcrack propagation signs refer to the characteristic change patterns in the vibration evolution indicators exhibited when microcracks grow within the material, mainly manifested as a sudden increase in the amplitude decay rate, an accelerated frequency drift rate, and a significant increase in the number of phase jumps. Stress concentration signs refer to characteristic patterns of changes in vibration evolution indicators that occur when stress distribution within a material is uneven. These are primarily manifested by periodic variations in the frequency drift rate and clustered distributions of phase jumps. The specific identification process involves comparing measured vibration evolution indicators with normal ranges. Indicators outside these ranges are labeled as abnormal. The combined patterns of abnormal indicators are analyzed. When the amplitude decay rate increases by more than 50% and the number of phase jumps increases by more than 100%, these indicators are identified as signs of microcrack growth. When the frequency drift rate exhibits periodic fluctuations with an amplitude exceeding 30% of the average value, these indicators are identified as signs of stress concentration. Material internal defect risk assessment quantitatively evaluates the severity and development trend of internal defects based on the results of abnormal pattern recognition. The assessment results are categorized into three levels: low risk, medium risk, and high risk. For example, when microcrack growth signs are detected and the growth rate exceeds a threshold, the risk is assessed as high; when stress concentration signs are detected but are mild, the risk is assessed as medium. This dual-phase differential analysis accurately identifies different types of defects within glass-ceramics, significantly improving the accuracy and reliability of defect detection.

[0059] The specific implementation process of cross-validation analysis of thickness conformity prediction results and material internal defect risk assessment is as follows: Cross-validation analysis refers to a data processing method that compares and analyzes two different assessment results to improve the reliability of judgment. Thickness geometric conformity refers to the degree of conformity of the product's geometric dimensions based on the thickness measurement results. The numerical range is 0-100 points, with 90 points and above being excellent, 70-90 points being qualified, and below 70 points being unqualified. Structural integrity conformity refers to the degree of conformity of the product's structural integrity based on the material internal defect risk assessment. The numerical range is 0-100 points, with low risk corresponding to 85-100 points, medium risk corresponding to 60-85 points, and high risk corresponding to 0-60 points. The comprehensive weight coefficient refers to the relative importance coefficient of each assessment dimension in the multidimensional quality assessment. The weight coefficient of thickness geometric conformity is 0.6, and the weight coefficient of structural integrity conformity is 0.4. The weight distribution is based on the application requirements and failure mode analysis of the microcrystalline glass screen protector. The calculation process is as follows: When the thickness pre-qualification result is qualified and the material internal defect risk assessment is low, the thickness geometric qualification score is 90 points and the structural integrity qualification score is 95 points. If the two assessment results disagree, the score corresponding to the lower assessment result is used conservatively. The multidimensional quality comprehensive score is equal to the thickness geometric qualification score multiplied by 0.6 plus the structural integrity qualification score multiplied by 0.4. For example, if the thickness geometric qualification score is 90 points and the structural integrity qualification score is 95 points, the multidimensional quality comprehensive score is 90 × 0.6 + 95 × 0.4 = 92 points. This cross-validation mechanism comprehensively considers the product's geometric accuracy and structural integrity, avoiding the limitations of a single assessment dimension and improving the comprehensiveness and accuracy of quality judgments.

[0060] The specific implementation process for categorizing product quality based on a multi-dimensional quality comprehensive score is as follows: Product quality grading refers to the process of categorizing products into different quality grades based on the comprehensive score. Excellent quality corresponds to a multi-dimensional quality comprehensive score of 85-100, indicating that the product achieves excellent levels of thickness accuracy and structural integrity, suitable for high-end applications. Acceptable quality corresponds to a multi-dimensional quality comprehensive score of 70-85, indicating that the product meets basic quality requirements but has room for improvement, suitable for general applications. Risk quality corresponds to a multi-dimensional quality comprehensive score of 0-70, indicating that the product has obvious quality issues or potential risks and is not recommended for direct use. Defect type marking is the process of identifying specific defect categories for risk-rated products, including thickness tolerance, microcracks, and stress concentration. Risk severity marking is the process of quantifying the risk level of risk-rated products into three levels: minor risk, moderate risk, and severe risk. The specific marking process is as follows: a multi-dimensional quality comprehensive score of 50-70 is marked as minor risk; a score of 30-50 is marked as moderate risk; and a score below 30 is marked as severe risk. Quality grade and risk warning information includes the product's quality grade, specific defect type, risk level, and recommended treatment measures. Real-time quality assessment results are dynamically generated product quality status reports during the grinding process, with the assessment results updated every 5 seconds. For example, the real-time quality assessment result for a product is "Qualified Grade, Overall Score 78 Points, No Obvious Defects, Recommended for Normal Use." The excellent grade threshold of 85-100 points ensures the screening criteria for high-quality products. The qualified grade range of 70-85 points meets the quality requirements of most application scenarios, while the risk level classification below 70 points effectively identifies problematic products and prevents them from entering the market.

[0061] The above describes the thickness detection method of the micro-ceramic glass in the embodiment of the present invention. The following describes the thickness detection device of the micro-ceramic glass in the embodiment of the present invention. Figure 2 An embodiment of a device for detecting the thickness of glass-ceramics according to an embodiment of the present invention includes: The vibration signal processing module 101 is used to obtain the vibration signal generated during the thinning and grinding process of the micro-ceramic glass, perform fast Fourier transform processing on the vibration signal, and extract the vibration main frequency, spectrum centroid frequency and frequency band energy distribution ratio as thickness characteristic parameters; A spatial mapping module 102 is configured to obtain the three-dimensional position coordinates of the grinding head, map the thickness characteristic parameters to the three-dimensional position coordinates, and process the thickness characteristic parameters of adjacent measuring points using a weighted average algorithm to obtain a spatial distribution of the thickness characteristic parameters; The data calibration module 103 is used to perform process interference compensation on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and perform time series smoothing processing on the compensated thickness characteristic parameters using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution; The thickness prediction module 104 is used to calculate the thickness variation of the remaining processing area based on the stable thickness characteristic distribution and the current grinding process, predict the final thickness distribution after the grinding is completed, and obtain the thickness qualification prediction result; The quality assessment module 105 is used to extract the time-series variation characteristics of the thickness characteristic parameters during the grinding process as vibration evolution indicators, and combine the thickness qualification prediction results to judge the comprehensive quality status of the product and obtain real-time quality assessment results.

[0062] above Figure 2 The thickness detection device of the microcrystalline glass in the embodiment of the present invention is described in detail from the perspective of modular functional entities. The thickness detection equipment of the microcrystalline glass in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0063] Figure 3 : This is a schematic structural diagram of a glass-ceramic thickness detection device provided by an embodiment of the present invention. The glass-ceramic thickness detection device 200 may have relatively large differences due to different configurations or performances, and may include one or more processors 210 (for example, one or more processors) and a memory 220, and one or more storage media 230 (for example, one or more mass storage device terminals) storing application programs 233 or data 232. Among them, the memory 220 and the storage medium 230 can be temporary storage or permanent storage. The program stored in the storage medium 230 may include one or more modules (not shown in the figure), each module of which may include a series of instruction operations in the glass-ceramic thickness detection device 200. Furthermore, the processor 210 may be configured to communicate with the storage medium 230, and execute a series of instruction operations in the storage medium 230 on the glass-ceramic thickness detection device 200 to implement the steps of the above-mentioned glass-ceramic thickness detection method.

[0064] The glass-ceramic thickness detection device 200 may further include one or more power supplies 240, one or more wired or wireless network interfaces 250, one or more input and output interfaces 260, and / or one or more operating systems 231, such as Windows Server, Mac OS X, Unix, Linux, FreeBSD, etc. It will be understood by those skilled in the art that Figure 3The structure of the glass-ceramic thickness detection device shown does not constitute a limitation on the glass-ceramic thickness detection device provided by the present invention, and may include more or fewer components than shown in the figure, or a combination of certain components, or a different arrangement of components.

[0065] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. The computer-readable storage medium stores instructions, which, when executed on a computer, cause the computer to execute the steps of the method for detecting the thickness of the microcrystalline glass.

[0066] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, devices, and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0067] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0068] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made by using the contents of the present invention description and drawings under the inventive concept of the present invention, or direct / indirect application in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A method for detecting the thickness of glass-ceramics, characterized in that: include: Acquire a vibration signal generated during the thinning and grinding process of the glass-ceramic, perform fast Fourier transform processing on the vibration signal, and extract the vibration main frequency, spectrum centroid frequency, and frequency band energy distribution ratio as thickness characteristic parameters; Obtaining the three-dimensional position coordinates of the grinding head, mapping the thickness characteristic parameters to the three-dimensional position coordinates, and processing the thickness characteristic parameters of adjacent measuring points using a weighted average algorithm to obtain a spatial distribution of the thickness characteristic parameters; Performing process interference compensation on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and performing time series smoothing processing on the compensated thickness characteristic parameters using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution; Calculate the thickness variation of the remaining processing area based on the stable thickness characteristic distribution and the current grinding process, predict the final thickness distribution after the grinding is completed, and obtain the thickness qualification prediction result; The time series variation characteristics of the thickness characteristic parameters during the grinding process are extracted as vibration evolution indicators, combined with the thickness qualification prediction results, to judge the comprehensive quality status of the product and obtain real-time quality assessment results.

2. The method for detecting the thickness of glass-ceramics according to claim 1, wherein: The method of obtaining a vibration signal generated during the thinning and grinding process of the glass-ceramics, performing fast Fourier transform processing on the vibration signal, and extracting the vibration main frequency, spectrum centroid frequency, and frequency band energy distribution ratio as thickness characteristic parameters includes: Obtaining raw vibration data during the glass-ceramic thinning and grinding process, and performing a two-phase response separation process on the raw vibration data based on the difference between the elastic modulus of the glass phase and the elastic modulus of the microcrystalline phase of the glass-ceramic to obtain a glass phase vibration component and a microcrystalline phase vibration component; According to the parameters of the grinding spindle speed and feed speed, the glass phase vibration component and the microcrystalline phase vibration component are synchronously divided and processed during the grinding cycle to eliminate the mechanical vibration interference of the grinding equipment and obtain a synchronous purified vibration signal; Performing fast Fourier transform processing on the synchronous purification vibration signal in the frequency range of 50-1500 Hz, dividing it into three thickness-sensitive bands: low frequency band, mid frequency band, and high frequency band, and obtaining frequency band spectrum data; According to the frequency spectrum data of the frequency bands, the peak frequency in each frequency band is calculated as the main vibration frequency, the weighted average value of the spectrum amplitude is calculated as the spectrum centroid frequency, and the ratio of the energy of each frequency band to the total energy is calculated as the frequency band energy distribution ratio, and the combination is used to form the basic thickness characteristic parameter; The vibration main frequency, spectrum centroid frequency and frequency band energy distribution ratio in the basic thickness characteristic parameters are subjected to inter-band differential operation to generate low-medium frequency differential coefficients, medium-high frequency differential coefficients and full-band comprehensive coefficients, which are used together with the basic thickness characteristic parameters as thickness characteristic parameters.

3. The method for detecting the thickness of glass-ceramics according to claim 2, wherein: The method of performing a two-phase response separation process on the original vibration data according to the difference between the elastic modulus of the glass phase and the elastic modulus of the microcrystalline phase of the microcrystalline glass to obtain the glass phase vibration component and the microcrystalline phase vibration component includes: According to the crystallinity distribution characteristics in the glass-ceramic preparation process, the volume fraction of the glass phase and the volume fraction of the microcrystalline phase are calculated. Combined with the density difference of each phase, the dual-phase density distribution parameters are obtained. Based on the density distribution parameters and elastic modulus differences of the two phases, the acoustic impedance difference between the glass phase and the microcrystalline phase is calculated, the relationship between the reflection coefficient and the transmission coefficient of the vibration wave at the two phase interface is established, and the response characteristics of the two phase interface are obtained; According to the dual-phase interface response characteristics, the original vibration data is decomposed in the frequency domain, low-frequency vibration components are attributed to glass phase response, and high-frequency vibration components are attributed to microcrystalline phase response, thereby obtaining preliminary phase separation vibration data; Based on the grain size distribution and orientation distribution characteristics of the microcrystalline phase, the microcrystalline phase response part in the preliminary phase separation vibration data is corrected by grain scattering to eliminate the influence of grain boundary scattering on vibration propagation, and the glass phase vibration component and the microcrystalline phase vibration component are obtained.

4. The method for detecting the thickness of glass-ceramics according to claim 1, wherein: The method of obtaining the three-dimensional position coordinates of the grinding head, mapping the thickness characteristic parameters to the three-dimensional position coordinates, and processing the thickness characteristic parameters of adjacent measuring points using a weighted average algorithm to obtain a spatial distribution of the thickness characteristic parameters includes: Obtain the position coordinate data of the grinding head in the X-axis, Y-axis, and Z-axis directions, and divide the workpiece surface into regular grids based on the path spacing and overlap rate of the grinding tracks to obtain spatial grid units; The low-medium frequency differential coefficient, the medium-high frequency differential coefficient and the full-band comprehensive coefficient of the thickness characteristic parameter are temporally and spatially matched with the three-dimensional position coordinates of the grinding head at the corresponding moment, and assigned to the corresponding spatial grid units to obtain gridded thickness characteristic data; According to the effective radius range of vibration propagation in the glass-ceramics, the adjacent influence area of ​​each grid unit is determined, and the spatial distance weight coefficient between adjacent grid units is calculated to obtain the spatial weight matrix; According to the spatial weight matrix, a weighted average operation is performed on the gridded thickness characteristic data, and a boundary constraint correction process is applied to the grid cells in the edge area of ​​the workpiece to obtain the spatial distribution of the thickness characteristic parameters.

5. The method for detecting the thickness of glass-ceramics according to claim 1, wherein: The method of performing process interference compensation on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and performing time series smoothing processing on the compensated thickness characteristic parameters using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution includes: Acquiring real-time change data of the grinding force, calculating a grinding force reference offset according to a workpiece stiffness decreasing law, and performing grinding force offset compensation on the glass phase response component and the microcrystalline phase response component in the spatial distribution of the thickness characteristic parameter to obtain dual-phase compensation data; According to the spatial gradient change of the grinding feed direction, the dual-phase compensation data is subjected to directional weighted correction to eliminate the anisotropic effect of the grinding track direction on the thickness feature detection, thereby obtaining an isotropic thickness feature distribution; Based on the time scale difference between the grinding cycle and the thickness change cycle, a double-layer nested sliding window is set. The inner window tracks the short-term changes in the grinding cycle, and the outer window tracks the long-term trend of the thickness change. The isotropic thickness characteristic distribution is smoothed by a double layer time series to obtain multi-scale smoothed data. The continuity index of the multi-scale smoothed data in the spatially adjacent regions and the stability index of the temporally continuous windows are calculated. When the continuity index and the stability index simultaneously meet the convergence conditions, the corresponding data are output as a stable thickness characteristic distribution.

6. The method for detecting the thickness of glass-ceramics according to claim 5, wherein: Based on the time scale difference between the grinding cycle and the thickness change cycle, a double-layer nested sliding window is set, the inner window tracks the short-term changes in the grinding cycle, and the outer window tracks the long-term trend of the thickness change. The isotropic thickness characteristic distribution is subjected to double-layer time series smoothing to obtain multi-scale smoothed data, including: According to the grinding spindle speed and the grinding head feed speed, the time length of a single grinding cycle is calculated, and the time span of the inner sliding window is set to 2-5 times the length of the grinding cycle to obtain the short-time window parameters; Based on the physical continuity constraint of the workpiece thickness change, the time scale of significant thickness feature changes is calculated, the time span of the outer sliding window is set to 1.5-3 times the thickness change time scale, the nesting ratio relationship between the outer window and the inner window is determined, and the long-term window parameters are obtained; Within the short-time window parameter range, high-frequency noise filtering and grinding vibration synchronization processing are performed on the isotropic thickness characteristic distribution to retain effective signal components related to thickness and obtain short-time smoothed data; Within the long-term window parameter range, trend extraction and outlier detection processing are performed on the short-term smoothed data, and segmented weighted averaging is performed in combination with the stage characteristics of the grinding process to obtain multi-scale smoothed data; The smoothness evaluation index of the multi-scale smoothed data in a continuous time period is calculated. When the smoothness evaluation index reaches a convergence threshold, the data processing is confirmed to be completed and the result is output.

7. The method for detecting the thickness of glass-ceramics according to claim 1, wherein: The method of calculating the thickness variation of the remaining processing area based on the stable thickness characteristic distribution and the current grinding process, predicting the final thickness distribution after the grinding is completed, and obtaining the thickness qualification prediction result includes: According to the current grinding completion ratio and grinding trajectory path data, the spatial distribution of the remaining unprocessed area is identified, the boundary gradient change rate between the processed area and the remaining area is calculated, and the thickness transition characteristics between the areas are obtained; Based on the thickness decreasing law of the processed area in the stable thickness characteristic distribution, combined with the current workpiece stiffness decreasing coefficient and the accumulated wear of the grinding tool, the material removal efficiency variation coefficient of the remaining processed area in the subsequent grinding process is calculated to obtain the nonlinear thickness variation; Based on the thickness transition characteristics and nonlinear thickness variation between the regions, a thickness prediction operation is performed on the remaining unprocessed region, and a predicted thickness distribution map after grinding is generated in combination with the boundary constraints of the edge region; The thickness deviation value and thickness uniformity index of each area in the predicted thickness distribution diagram are calculated, the thickness deviation value is compared with the allowable tolerance range of the technical specification, and the thickness uniformity index is matched and evaluated with the optical performance requirements to obtain a thickness qualification prediction result.

8. The method for detecting thickness of glass-ceramics according to claim 1, wherein: The time series variation characteristics of the thickness characteristic parameters during the grinding process are extracted as vibration evolution indicators, combined with the thickness qualification prediction results to judge the comprehensive quality status of the product and obtain real-time quality assessment results, including: Gradient analysis and inflection point detection are performed on the variation trajectory of the low-medium frequency differential coefficient, medium-high frequency differential coefficient, and full-band comprehensive coefficient of the thickness characteristic parameter on the grinding time axis, and the vibration amplitude attenuation rate, frequency drift rate, and phase jump number are extracted as vibration evolution indicators; Based on the vibration response differences of the dual-phase structure of the glass-ceramic, the vibration evolution index is subjected to glass phase abnormal pattern recognition and microcrystalline phase abnormal pattern recognition, and signs of microcrack extension and stress concentration are detected to obtain a risk assessment of internal defects in the material; Cross-validate and analyze the thickness qualification prediction results with the material internal defect risk assessment, calculate the comprehensive weight coefficient of thickness geometry qualification and structural integrity qualification, and obtain a multi-dimensional quality comprehensive score; Based on the multi-dimensional quality comprehensive score, the product quality level is divided into three levels: excellent, qualified and risky. The defect type and risk degree of risky products are marked, and a real-time quality assessment result including quality grade and risk warning information is generated.

9. A device for detecting the thickness of glass-ceramics, characterized in that: The glass-ceramics thickness detection device adopts the glass-ceramics thickness detection method according to any one of claims 1 to 8, and the glass-ceramics thickness detection device includes: A vibration signal processing module is used to obtain the vibration signal generated during the thinning and grinding process of the micro-ceramic glass, perform fast Fourier transform processing on the vibration signal, and extract the vibration main frequency, spectrum centroid frequency and frequency band energy distribution ratio as thickness characteristic parameters; A spatial mapping module is used to obtain the three-dimensional position coordinates of the grinding head, map the thickness characteristic parameters to the three-dimensional position coordinates, and use a weighted average algorithm to process the thickness characteristic parameters of adjacent measuring points to obtain the spatial distribution of the thickness characteristic parameters; A data calibration module is used to perform process interference compensation on the spatial distribution of the thickness characteristic parameters according to the grinding process parameters, and to perform time series smoothing on the compensated thickness characteristic parameters using a sliding window averaging algorithm to obtain a stable thickness characteristic distribution; A thickness prediction module is used to calculate the thickness variation of the remaining processing area based on the stable thickness characteristic distribution and the current grinding process, predict the final thickness distribution after the grinding is completed, and obtain a thickness qualification prediction result; The quality assessment module is used to extract the time-series variation characteristics of the thickness characteristic parameters during the grinding process as vibration evolution indicators, combine the thickness qualification prediction results, judge the comprehensive quality status of the product, and obtain real-time quality assessment results.

10. A glass-ceramic thickness detection device, characterized in that: The device for detecting the thickness of glass-ceramics includes: a memory and at least one processor, wherein the memory stores instructions; The at least one processor calls the instructions in the memory to enable the glass-ceramic thickness detection device to perform the steps of the glass-ceramic thickness detection method according to any one of claims 1 to 8.

Images