Glass drawing process thickness monitoring method and system based on capacitive sensor

Abstract

The invention discloses a glass drawing process thickness monitoring method and system based on a capacitive sensor, and relates to the technical field of glass manufacturing process monitoring, and the method comprises the steps: obtaining an original capacitance signal, a sensor temperature and a drawing speed through a capacitive sensor array; performing temperature compensation on the original capacitance signal; establishing spatial electric field intensity distribution according to the compensated capacitance value difference and converting the spatial electric field intensity distribution into thickness distribution data; calculating a thickness difference value of adjacent positions and determining an abnormal thickness position; calculating the moving time of reaching the heating area according to the abnormal position coordinates; and determining the power adjustment amplitude according to the thickness difference value, and adjusting the heating power at the corresponding moment. According to the invention, high-precision continuous monitoring and automatic regulation and control in the glass drawing process are realized, and the stability of product quality is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of glass manufacturing process monitoring, in particular to a glass drawing process thickness monitoring method and system based on a capacitive sensor. BACKGROUND

[0002] At present, optical measurement, mechanical measurement and other methods are commonly used for thickness monitoring in the glass drawing process, but these methods have problems such as large environmental influence on measurement accuracy and slow response speed.

[0003] The capacitive sensor has the advantages of non-contact, high precision and fast response, and is suitable for thickness monitoring in the glass drawing process. The existing capacitive sensor thickness monitoring method mainly obtains local thickness data through single-point measurement, which is difficult to realize large-area continuous monitoring. Due to the drastic temperature change in the glass drawing process, the measurement accuracy of the sensor is significantly affected by the temperature, resulting in large errors in the measurement results.

[0004] In the prior art, the identification of abnormal thickness mainly relies on manual experience judgment, and lacks automatic identification and timely control means. When abnormal thickness is found, it is impossible to accurately predict the time when the abnormal position reaches the heating area, and it is also impossible to realize accurate heating power adjustment, resulting in product quality fluctuations.

[0005] With the continuous development of glass manufacturing process, the requirement for glass thickness uniformity is becoming higher and higher. The existing technology cannot meet the monitoring needs of high precision, real-time and automation, and it is urgent to develop a thickness monitoring method that can realize large-area continuous monitoring, has temperature compensation function, can automatically identify abnormalities and realize accurate control. SUMMARY

[0006] The purpose of the present application is to provide a glass drawing process thickness monitoring method and system based on a capacitive sensor, which aims to at least solve one of the technical problems existing in the prior art.

[0007] The technical solution of the present application is: a glass drawing process thickness monitoring method based on a capacitive sensor, comprising the following steps: Obtain the original capacitance signal of the glass drawing area through the capacitive sensor array, and synchronously obtain the sensor temperature and the drawing speed; Calculate the capacitance temperature offset according to the sensor temperature, subtract the capacitance temperature offset from the original capacitance signal to obtain the compensated capacitance signal; According to the difference in capacitance value of each sensor electrode in the compensated capacitance signal, establish the spatial electric field intensity distribution, convert the spatial electric field intensity distribution into the glass thickness value of each spatial position, and form the thickness distribution data; Calculate thickness difference values of adjacent spatial positions in the thickness distribution data, determine spatial positions with thickness difference values exceeding a preset fluctuation threshold as abnormal thickness positions, and extract spatial coordinates and corresponding thickness difference values of the abnormal thickness positions; Calculate a moving distance of the abnormal thickness position to the heating area according to the spatial coordinates of the abnormal thickness position and the drawing direction, and calculate a moving time according to the moving distance and the drawing speed; Determine a heating power adjustment amplitude according to the thickness difference value corresponding to the abnormal thickness position, determine a power adjustment time according to the moving time, and apply the heating power adjustment amplitude to the heating area at the power adjustment time.

[0008] Calculate a capacitance temperature offset according to the sensor temperature, subtract the capacitance temperature offset from the original capacitance signal to obtain a compensated capacitance signal, including: Collect a temperature data sequence of the temperature sensor at different time instants, and construct a sensor temperature variation curve in time sequence; Calculate a temperature variation rate of adjacent time instants according to the sensor temperature variation curve, and divide the temperature data sequence into a sensor temperature stable interval and a sensor temperature fluctuation interval based on the temperature variation rate; Calculate a linear relationship between temperature data in the sensor temperature stable interval and the original capacitance signal to obtain a reference temperature correction value; Calculate a linear relationship between sensor temperature and original capacitance signal in a fixed time window in the sensor temperature fluctuation interval to obtain a dynamic capacitance temperature correction value; Weight and combine the reference temperature correction value and the dynamic capacitance temperature correction value to obtain a temperature compensation weight; Calculate a capacitance temperature offset according to the temperature compensation weight, subtract the capacitance temperature offset from the original capacitance signal to obtain a preliminary compensated capacitance signal; Perform time sequence accumulation processing on the preliminary compensated capacitance signal, calculate a change trend of the capacitance value with time, perform nonlinear correction on the preliminary compensated capacitance signal according to the change trend, and obtain a compensated capacitance signal.

[0009] Establish a spatial electric field intensity distribution according to the capacitance value difference of each sensor electrode in the compensated capacitance signal, convert the spatial electric field intensity distribution into glass thickness values of each spatial position, and form thickness distribution data, including: Calculate the capacitance value difference between the sensor electrodes according to the compensated capacitance signal; In the measurement plane formed by the sensor electrodes, scan the capacitance value difference in the horizontal direction to obtain a horizontal direction field strength component, scan the capacitance value difference in the vertical direction to obtain a vertical direction field strength component, and orthogonally synthesize the horizontal direction field strength component and the vertical direction field strength component to obtain an initial field strength distribution; The initial field intensity distribution is divided into a measurement region in a measurement plane according to an initial field intensity distribution, a gradient change of the initial field intensity distribution in the measurement region is extracted, and the initial field intensity distribution is optimized by taking the gradient change as a constraint condition to obtain a spatial electric field intensity distribution; The spatial electric field intensity distribution is converted into an in-medium field intensity distribution through the dielectric properties of the glass medium, and a medium non-uniform correction coefficient is calculated based on the in-medium field intensity distribution; The in-medium field intensity distribution is corrected according to the medium non-uniform correction coefficient to obtain a glass thickness value at each spatial position; Position coordinate data of the sensor electrode in the measurement plane is collected, and the glass thickness value is reconstructed into thickness distribution data according to the position coordinate data.

[0010] The initial field intensity distribution is divided into a measurement region in a measurement plane according to an initial field intensity distribution, a gradient change of the initial field intensity distribution in the measurement region is extracted, and the initial field intensity distribution is optimized by taking the gradient change as a constraint condition to obtain a spatial electric field intensity distribution includes: The field intensity values of the initial field intensity distribution in the measurement plane are obtained, the change law of the field intensity values in the horizontal direction and the vertical direction is calculated respectively, the position of the measurement region boundary is determined according to the change law, and the measurement plane is divided into a plurality of measurement regions; The initial field intensity distribution between adjacent measurement regions is extracted, the gradient change of the initial field intensity distribution is calculated, and the measurement region boundary constraint is established according to the gradient change; The initial field intensity distribution in each measurement region is fitted to obtain a measurement region field intensity distribution, and the gradient difference of the field intensity distribution of adjacent measurement regions is calculated according to the measurement region boundary constraint; The gradient difference and the measurement region boundary constraint are compared to determine the boundary position exceeding the constraint, and the boundary position is marked as a gradient change abnormal point; A correction step is set at the gradient change abnormal point, the gradient difference is calculated in a decreasing manner according to the correction step, and when the gradient difference after the decrease is less than the measurement region boundary constraint, the gradient correction value is recorded; The gradient correction value is applied to the initial field intensity distribution at the gradient change abnormal point to obtain an optimized field intensity distribution; The optimized field intensity distribution and the measurement region field intensity distribution are combined and reconstructed to generate a spatial electric field intensity distribution.

[0011] The thickness difference value of adjacent spatial positions in the thickness distribution data is calculated, and the spatial position whose thickness difference value exceeds a preset fluctuation threshold is determined as an abnormal thickness position includes: The coordinate information of adjacent spatial positions in the thickness distribution data is obtained, the thickness difference value between adjacent spatial positions is measured, and the thickness difference value is constructed into a thickness difference value matrix according to the coordinate information; extracting decomposition coefficients of the thickness difference matrix, identifying fluctuation amplitudes of the thickness difference according to the decomposition coefficients, and determining a spatial position where the fluctuation amplitude exceeds a preset fluctuation threshold as a candidate abnormal position; extracting the fluctuation amplitude and the thickness difference of the candidate abnormal position, sorting the candidate abnormal positions according to the fluctuation amplitudes, and determining a candidate abnormal position with the largest fluctuation amplitude as a central abnormal position; analyzing fluctuation amplitude attenuation rules of the candidate abnormal positions based on the central abnormal position, dividing an abnormal area according to the fluctuation amplitude attenuation rules, and selecting a spatial position with the largest thickness difference in the abnormal area as an abnormal thickness position.

[0012] calculating a moving distance of the abnormal thickness position to the heating area according to the spatial coordinates of the abnormal thickness position and the drawing direction, and calculating a moving time according to the moving distance and the drawing speed, including: determining a drawing direction of the drawing process according to the spatial coordinates of the abnormal thickness position, projecting the spatial coordinates onto the drawing direction, and calculating a straight-line distance from the abnormal thickness position to a boundary of the heating area; arranging a stress detection point on a path from the abnormal thickness position to the boundary of the heating area, collecting stress data of the stress detection point, and calculating a path deformation variable; compensating the path deformation variable and the straight-line distance to obtain an actual moving distance of the abnormal thickness position to the heating area; obtaining a real-time rotating speed of the drawing equipment, and converting the real-time rotating speed into a reference speed in the drawing process; collecting stress change data along the path of the actual moving distance, establishing a speed change rule according to the stress change data, and calculating actual motion speeds of each path segment in combination with the reference speed; calculating a cumulative motion time according to the lengths of each path segment in the actual moving distance and the corresponding actual motion speeds, and taking the cumulative motion time as the moving time.

[0013] determining a heating power adjustment amplitude according to the thickness difference corresponding to the abnormal thickness position, determining a power adjustment time according to the moving time, and applying the heating power adjustment amplitude to the heating area at the power adjustment time, including: collecting glass sheet deformation data of the abnormal thickness position, and calculating a temperature compensation value according to the deformation data and the thickness difference corresponding to the abnormal thickness position; collecting heat conduction data of the heating area, calculating a power reference value according to the heat conduction data and the temperature compensation value, and taking the power reference value as an initial heating power; collecting a temperature distribution of the heating area, calculating a power correction value according to the temperature distribution, and combining the power correction value and the initial heating power to generate a heating power adjustment amplitude; Measure the temperature response curve of the heating area, determine the temperature response hysteresis based on the rise time and shift time of the temperature response curve, and use the temperature response hysteresis as the power adjustment time. At the power adjustment moment, the heating power adjustment range is applied to the heating area by the heating controller.

[0014] This invention provides a glass drawing process thickness monitoring system based on a capacitive sensor, the system comprising: The capacitance signal acquisition module is used to acquire the original capacitance signal of the glass drawing region through a capacitive sensor array, and simultaneously acquire the sensor temperature and drawing speed. The temperature compensation module is used to calculate the capacitor temperature offset based on the sensor temperature, subtract the capacitor temperature offset from the original capacitor signal, and obtain the compensated capacitor signal. The thickness calculation module is used to establish the spatial electric field intensity distribution based on the difference in capacitance values ​​of each sensor electrode in the compensation capacitor signal, and convert the spatial electric field intensity distribution into glass thickness values ​​at each spatial location to form thickness distribution data. An abnormal location determination module is used to calculate the thickness difference between adjacent spatial locations in the thickness distribution data, determine the spatial location where the thickness difference exceeds a preset fluctuation threshold as an abnormal thickness location, and extract the spatial coordinates of the abnormal thickness location and the corresponding thickness difference. The movement time calculation module is used to calculate the movement distance from the abnormal thickness position to the heating area based on the spatial coordinates of the abnormal thickness position and the drawing direction, and to calculate the movement time based on the movement distance and the drawing speed. The heating control module is used to determine the heating power adjustment range based on the thickness difference corresponding to the abnormal thickness position, determine the power adjustment time based on the movement time, and apply the heating power adjustment range to the heating area at the power adjustment time.

[0015] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0016] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing computer program instructions, which, when executed by a processor, implement the steps in any of the aforementioned methods.

[0017] This invention achieves large-area continuous monitoring of the glass drawing region using a capacitive sensor array, overcoming the limitations of traditional single-point measurements. A temperature compensation mechanism effectively eliminates the impact of temperature changes on measurement accuracy, improving measurement precision. High-precision thickness distribution data reconstruction is achieved by establishing a spatial electric field intensity distribution. Abnormal thickness locations are automatically identified and relevant parameters are extracted, avoiding the subjectivity of manual judgment. Precise calculation of movement time based on the spatial information of abnormal locations enables predictive control. Closed-loop automatic adjustment of heating power is achieved through precise calculation of power adjustment amplitude and timing control. The overall solution significantly improves the automation level of the glass drawing process and the stability of product quality, reduces the need for manual intervention, and increases production efficiency. Attached Figure Description

[0018] Figure 1 A flowchart of a glass drawing process thickness monitoring method based on a capacitive sensor provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the determination of the boundary line of the measurement area based on the horizontal change pattern in an embodiment of the present invention. Detailed Implementation

[0019] like Figure 1 As shown, Figure 1 A flowchart of a glass drawing process thickness monitoring method based on a capacitive sensor provided in an embodiment of the present invention is shown. The method includes the following steps: The original capacitance signal of the glass drawing region is obtained by a capacitive sensor array, and the sensor temperature and drawing speed are obtained simultaneously. The capacitor temperature offset is calculated based on the sensor temperature, and the capacitor temperature offset is subtracted from the original capacitor signal to obtain the compensation capacitor signal. The spatial electric field intensity distribution is established based on the difference in capacitance values ​​of each sensor electrode in the compensation capacitor signal. The spatial electric field intensity distribution is then converted into glass thickness values ​​at each spatial location to form thickness distribution data. Calculate the thickness difference between adjacent spatial locations in the thickness distribution data, identify spatial locations where the thickness difference exceeds a preset fluctuation threshold as abnormal thickness locations, and extract the spatial coordinates of the abnormal thickness locations and the corresponding thickness difference. The distance the abnormal thickness location moves from the abnormal thickness location to the heating zone is calculated based on the spatial coordinates of the abnormal thickness location and the drawing direction, and the moving time is calculated based on the moving distance and the drawing speed. The heating power adjustment range is determined based on the thickness difference corresponding to the abnormal thickness location, the power adjustment time is determined based on the movement time, and the heating power adjustment range is applied to the heating area at the power adjustment time.

[0020] The capacitor temperature offset is calculated based on the sensor temperature. This offset is then subtracted from the original capacitor signal to obtain the compensated capacitor signal, which includes: Collect temperature data sequences from the temperature sensor at different times and construct temperature change curves of the sensor in chronological order; The temperature change rate between adjacent moments is calculated based on the sensor temperature change curve, and the temperature data sequence is divided into a stable sensor temperature range and a fluctuating sensor temperature range based on the temperature change rate. The reference temperature correction value is obtained by calculating the linear relationship between the temperature data and the original capacitance signal within the stable temperature range of the sensor. The dynamic capacitor temperature correction value is obtained by calculating the linear relationship between the sensor temperature and the original capacitance signal within a fixed time window within the sensor temperature fluctuation range. The temperature compensation weight is obtained by weighting the reference temperature correction value and the dynamic capacitor temperature correction value. The capacitor temperature offset is calculated based on the temperature compensation weight, and the capacitor temperature offset is subtracted from the original capacitor signal to obtain the preliminary compensated capacitor signal. The initial compensation capacitor signal is subjected to time-series accumulation processing to calculate the change trend of the capacitance value over time. Based on the change trend, the initial compensation capacitor signal is nonlinearly corrected to obtain the compensation capacitor signal.

[0021] First, a PT100 temperature sensor can be used to continuously collect temperature data from the production environment at a sampling frequency of once per second. The temperature data is stored in the form of temperature values ​​and corresponding timestamps, forming a temperature data sequence. This sequence is then used to plot a curve showing the temperature change over time, visually reflecting the fluctuations in ambient temperature during the production process.

[0022] The rate of temperature change is determined by comparing the difference between temperature values ​​at two adjacent time points with the time interval. When the rate of temperature change at multiple consecutive time points is lower than a preset threshold (e.g., 0.05℃ / minute), the current time period is determined to be a stable temperature range; when the rate of temperature change exceeds the threshold, it is determined to be a temperature fluctuation range. For example, a time period in which the temperature change is no more than 0.5℃ within approximately 30 minutes can be classified as a stable range, while periods of rapid temperature change can be classified as fluctuation ranges.

[0023] Within the stable temperature range, the linear relationship between sensor temperature data and the original capacitance signal is calculated to obtain the reference temperature correction value. Temperature data within the stable range and the corresponding original capacitance values ​​are selected, and a linear fitting method is used to establish the relationship function between temperature and capacitance. By analyzing the change in capacitance value corresponding to each unit change in temperature, the reference temperature correction coefficient is obtained. For example, within a certain stable range, when the temperature changes from 25.0℃ to 25.2℃, the capacitance signal changes from 12.56pF to 12.62pF, thus the reference temperature correction coefficient is 0.3pF / ℃.

[0024] For temperature fluctuation ranges, the dynamic capacitance temperature correction value is obtained by calculating the linear relationship between the sensor temperature and the original capacitance signal within a fixed time window. An appropriate time window (e.g., 10 minutes) is selected, and the correspondence between temperature changes and capacitance changes within the window is analyzed. Data within the window is updated using a sliding mechanism to ensure that the latest temperature-related characteristics on the capacitance are captured. In scenarios with rapid temperature changes, the dynamic correction value more accurately reflects the instantaneous impact of temperature on the capacitance. For example, within a 10-minute window of a certain fluctuation range, if the temperature rapidly rises from 26.5℃ to 28.0℃, the capacitance changes from 13.05pF to 13.52pF, thus calculating a dynamic temperature correction coefficient of 0.31pF / ℃.

[0025] The temperature compensation weight is obtained by weighting the reference temperature correction value and the dynamic capacitor temperature correction value. The weight of the two correction values ​​is determined according to the drasticness of the current temperature change. When the temperature is relatively stable, the reference temperature correction value has a higher weight; when the temperature changes drastically, the dynamic capacitor temperature correction value has a higher weight. The weight allocation adopts a smooth transition method to avoid abrupt changes in the compensation effect. The weight function can be designed according to the temperature change rate: when the temperature change rate is 0.02℃ / minute, the reference correction value has a weight of 0.8 and the dynamic correction value has a weight of 0.2; when the temperature change rate is 0.1℃ / minute, the reference correction value has a weight of 0.3 and the dynamic correction value has a weight of 0.7.

[0026] Select an appropriate reference temperature (e.g., 25℃) and calculate the deviation of the current temperature from the reference temperature. Using temperature compensation weights and the temperature deviation value, determine the capacitor temperature offset through a specific calculation relationship. Process the capacitor temperature offset and the original capacitor signal to obtain a pre-compensated capacitor signal. When the ambient temperature is 27.5℃, there is a 2.5℃ temperature deviation relative to the reference temperature of 25℃. If the compensation weight is 0.32pF / ℃, the capacitor temperature offset can be determined to be 0.8pF. Through capacitor signal processing, the original capacitance value of 13.85pF is converted to a pre-compensated capacitance value of 13.05pF.

[0027] The initial compensation capacitance signal is processed by time-series accumulation to calculate the capacitance value's trend over time. Based on this trend, a nonlinear correction is applied to the initial compensation capacitance signal to obtain the final compensation capacitance signal. A sliding window method is used to smooth the initial compensation signal, eliminating the influence of short-term fluctuations. The long-term trend of the capacitance signal is analyzed to identify capacitance value fluctuations caused by changes in glass thickness rather than temperature. Based on the identification results, a nonlinear correction function is applied to perform a secondary correction on the initial compensation signal, further improving measurement accuracy. The nonlinear correction function considers the nonlinear characteristics of capacitance and temperature across different temperature ranges, providing more accurate compensation for temperature effects. Applying a second-order polynomial correction function to the initially compensated capacitance value reduces the temperature-induced measurement error from ±0.2 pF to ±0.05 pF.

[0028] This invention achieves precise temperature compensation for capacitive sensors by analyzing the relationship between sensor temperature and capacitance signal, effectively eliminating the interference of ambient temperature changes on glass thickness measurement. Temperature compensation is handled in two cases: a stable range and a fluctuating range. A smooth transition is achieved through weighted combination, overcoming the shortcomings of traditional single compensation methods in scenarios with rapid temperature changes. Furthermore, time-series accumulation and nonlinear correction further improve the compensation effect, ensuring that the compensated capacitance signal accurately reflects the actual glass thickness.

[0029] The spatial electric field intensity distribution is established based on the capacitance differences of each sensor electrode in the compensation capacitor signal. This spatial electric field intensity distribution is then converted into glass thickness values ​​at each spatial location, forming thickness distribution data, including: The capacitance difference between the sensor electrodes is calculated based on the compensation capacitor signal. Within the measurement plane formed by the sensor electrodes, the difference in capacitance values ​​is scanned along the horizontal direction to obtain the horizontal field strength component, and the difference in capacitance values ​​is scanned along the vertical direction to obtain the vertical field strength component. The horizontal field strength component and the vertical field strength component are orthogonally combined to obtain the initial field strength distribution. The measurement area is divided according to the initial field strength distribution in the measurement plane. The gradient change of the initial field strength distribution in the measurement area is extracted. The gradient change is used as a constraint to optimize the initial field strength distribution and obtain the spatial electric field strength distribution. The spatial electric field intensity distribution is converted into the field intensity distribution inside the medium through the dielectric properties of the glass medium, and the medium non-uniformity correction coefficient is calculated based on the field intensity distribution inside the medium. The glass thickness value at each spatial location is obtained by correcting the field intensity distribution within the medium based on the medium inhomogeneity correction coefficient. The position coordinate data of the sensor electrode in the measurement plane is collected, and the glass thickness value is reconstructed into thickness distribution data based on the position coordinate data.

[0030] When calculating the capacitance difference between sensor electrodes based on the compensation capacitance signal, capacitive sensors typically consist of multiple electrode arrays arranged in a specific geometry to cover the measurement area of ​​the glass sheet. After the aforementioned temperature compensation processing, the compensation capacitance signal can accurately reflect the capacitance characteristics between the electrodes. For adjacent electrode pairs, their capacitance values ​​can be extracted and the capacitance difference between adjacent electrodes can be calculated. For example, a glass drawing production line uses a 4×4 array of electrodes. When measuring a specific area of ​​the glass sheet, the obtained capacitance difference data between adjacent electrodes shows that the capacitance difference between horizontally adjacent electrode pairs ranges from 0.05pF to 0.15pF, while the capacitance difference between vertically adjacent electrode pairs ranges from 0.07pF to 0.18pF. These differences reflect the spatial distribution of the glass thickness.

[0031] Within the measurement plane formed by the sensor electrodes, the horizontal field strength component is obtained by scanning the capacitance difference along the horizontal direction, and the vertical field strength component is obtained by scanning the capacitance difference along the vertical direction. The horizontal and vertical field strength components are orthogonally combined to obtain the initial field strength distribution. A row-by-row scanning method is used to scan the electrode array horizontally, recording the trend of capacitance difference changes between adjacent electrodes. A specific algorithm converts the capacitance difference into a horizontal electric field strength component. Similarly, the vertical electric field strength component is obtained by scanning column-by-column. For example, in glass thickness detection, during horizontal scanning, the average capacitance difference in the left region is found to be 0.08 pF, while that in the right region is 0.12 pF, which are converted to horizontal field strength components of 0.32 V / mm and 0.48 V / mm, respectively. The vertical scanning results show that the average capacitance difference in the upper region is 0.09 pF, and that in the lower region is 0.14 pF, which are converted to vertical field strength components of 0.36 V / mm and 0.56 V / mm, respectively. Through a specific orthogonal synthesis process, the field strength components in these two directions are combined into an initial field strength distribution matrix, which clearly shows the spatial distribution characteristics of the electric field strength.

[0032] Based on the numerical characteristics of the initial electric field distribution, the measurement plane can be divided into multiple regions, such as regions with uniform intensity and regions with significant intensity gradient changes. Gradient constraints are established by calculating the gradient rate of change of the electric field intensity within each small region. These constraints are then used to optimize and adjust the initial electric field distribution, eliminating noise and unreasonable fluctuations to obtain a smoother and more accurate spatial electric field intensity distribution. For example, in a production line application, the measurement plane was divided into 25 small regions, each 100mm × 100mm in size. Gradient analysis revealed that the gradient rate of change of the electric field intensity at the region boundaries was as high as 30%, while it was only 5% at the region center. An iterative optimization method was used to smooth the electric field distribution in the high-gradient regions, resulting in a spatial electric field intensity distribution map that better conforms to physical laws, with a more continuous and natural gradient change.

[0033] The conversion of the spatial electric field intensity distribution to the field intensity distribution within a medium requires consideration of parameters such as the dielectric constant and dielectric loss of the glass material, establishing the propagation law of the electric field in the medium. For typical flat glass, the dielectric constant is usually between 4.0 and 8.0. Based on this characteristic, the electric field intensity distribution in air can be converted to the field intensity distribution inside the medium through a specific field intensity conversion relationship. Considering the non-uniformity of the glass material, it is necessary to calculate the medium non-uniformity correction coefficient. For example, the dielectric constant of the glass sheet from a float glass production line is 5.8. Through the dielectric property conversion relationship, the field intensity of 0.5 V / mm in air corresponds to a field intensity of 0.086 V / mm inside the medium. Through analysis of multiple test points, it was found that there is an approximately 8% difference in dielectric properties between the edge region and the center region of the glass. Based on this, a medium non-uniformity correction coefficient matrix was calculated for subsequent correction.

[0034] Utilizing the inverse relationship between glass thickness and electric field strength, a specific conversion function is used to transform the corrected electric field distribution within the medium into a thickness distribution. This conversion function must consider the electrical properties and geometric shape of the glass. The conversion parameters are obtained by measuring and fitting standard glass samples of known thickness. For example, in a glass thickness measurement, the corrected electric field distribution showed a field strength of 0.084 V / mm in the central region and 0.092 V / mm in the edge region. The thickness conversion function calculated the thickness to be 4.02 mm in the central region and 3.68 mm in the edge region. This result shows a 96% agreement with the verification results obtained using mechanical measurement methods.

[0035] The physical coordinates of each electrode within the measurement plane are precisely measured to establish a correspondence between the electrode positions and their actual spatial locations. Using this correspondence, the calculated thickness values ​​at each point are mapped onto the actual spatial coordinates, forming a complete thickness distribution dataset. For example, a 4×4 electrode array covers a 500mm×500mm measurement area, and the position coordinates of each electrode are precisely measured and recorded using a laser positioning device. The calculated thickness values ​​at the 16 measurement points are then expanded into 100×100 high-density data points using an interpolation algorithm, generating a thickness distribution map with a resolution of 5mm. This map visually displays the trend of glass thickness variation throughout the entire area.

[0036] In practice, multiple capacitive sensor electrodes are arranged to form an electrode array covering the entire measurement area of ​​the glass plate. The compensated capacitance signal is acquired, and the capacitance difference between adjacent electrodes is calculated. Capacitance difference scanning is performed along the horizontal and vertical directions, and the scanning results are converted into field strength components in the corresponding directions. An initial field strength distribution is obtained through orthogonal synthesis. Gradient analysis is performed on the initial field strength distribution by region, and optimization is performed based on gradient constraints to obtain an accurate spatial electric field strength distribution. Considering the dielectric properties of the glass material, the spatial electric field strength is converted into a field strength distribution within the medium, and a medium inhomogeneity correction coefficient is calculated. The field strength within the medium is corrected based on the correction coefficient, and the glass thickness value at each point is obtained through a thickness conversion function. The actual position coordinates of the electrodes are collected, the thickness values ​​are mapped to the actual spatial positions, and a high-resolution thickness distribution map is generated through an interpolation algorithm.

[0037] This invention achieves accurate conversion from capacitance signal to glass thickness distribution through multi-dimensional analysis and processing of capacitance sensor signals. It overcomes measurement errors caused by the non-uniformity of the dielectric properties of glass materials, significantly improving the spatial resolution and accuracy of thickness monitoring. Employing field strength gradient constraint optimization and dielectric inhomogeneity correction compensation techniques, it effectively solves edge effects and electric field distortion problems. It can acquire real-time thickness distribution data across the entire glass sheet, accurately capturing thickness fluctuation trends and providing timely and effective data support for production process adjustments.

[0038] The measurement area is divided within the measurement plane based on the initial electric field distribution. The gradient change of the initial electric field distribution within the measurement area is extracted, and the gradient change is used as a constraint to optimize the initial electric field distribution, resulting in a spatial electric field intensity distribution including: Obtain the initial field strength distribution field strength values ​​within the measurement plane, calculate the variation patterns of the field strength values ​​in the horizontal and vertical directions respectively, determine the boundary line positions of the measurement area based on the variation patterns, and divide the measurement plane into multiple measurement areas; Extract the initial field strength distribution between adjacent measurement areas, calculate the gradient change of the initial field strength distribution, and establish boundary constraints of the measurement area based on the gradient change; The initial field strength distribution within each measurement region is fitted to obtain the field strength distribution of the measurement region. The gradient difference between the field strength distributions of adjacent measurement regions is calculated based on the boundary constraints of the measurement region. The gradient difference is compared with the boundary constraints of the measurement area to determine the boundary location that exceeds the constraints, and the boundary location is marked as anomaly point of gradient change. Set a correction step size at the abnormal gradient change point, and calculate the gradient difference by decreasing the correction step size. When the reduced gradient difference is less than the boundary constraint of the measurement area, record it as the gradient correction value. The gradient correction value is applied to the initial field strength distribution at the gradient change anomaly point to obtain the optimized field strength distribution; The optimized field strength distribution is combined with the field strength distribution of the measurement area to reconstruct the spatial electric field strength distribution.

[0039] The initial electric field strength distribution is obtained within the measurement plane. The variation patterns of the electric field strength values ​​in the horizontal and vertical directions are calculated. Based on these variation patterns, the boundary lines of the measurement areas are determined, and the measurement plane is divided into multiple measurement regions. The initial electric field strength distribution data obtained through the sensor electrode array is presented as a two-dimensional matrix, where each element represents the electric field strength value at the corresponding location. For thickness detection in glass drawing processes, the difference in electric field strength values ​​between adjacent points in the horizontal direction can be calculated first, and a difference curve can be plotted. For example, in an application case on a glass production line, an electric field strength value was sampled every 50 mm along the horizontal direction. It was found that the electric field strength value difference abruptly occurred at x=250 mm and x=550 mm, reaching 0.15 V / mm and 0.18 V / mm respectively, significantly higher than the average difference of 0.06 V / mm in other areas. Similarly, analysis along the vertical direction revealed abrupt changes in electric field strength values ​​at y=200 mm and y=650 mm, with differences of 0.17 V / mm and 0.20 V / mm respectively. These abrupt change points become the natural boundaries of the measurement area, which can be used to divide the entire measurement plane into 3×3 measurement areas, totaling 9 areas.

[0040] Extract the electric field strength (EV) value sequence from the common boundary of adjacent measurement areas and calculate the spatial gradient of these EV values. Taking the common boundary between area 1 and area 2 as an example, among the previously divided nine areas, extract 10 equally spaced EV sampling points on the boundary, and calculate the EV gradient value sequence as follows: 0.042, 0.046, 0.051, 0.048, 0.072, 0.085, 0.047, 0.045, 0.043, 0.044 V / mm / mm. Through statistical analysis, the normal gradient range of this boundary is determined to be 0.040 to 0.055 V / mm / mm, and this range is set as the constraint condition for this boundary. Repeat the above process for all area boundaries to establish a complete set of boundary constraints.

[0041] For each defined measurement region, a polynomial fitting method was used to fit the field intensity distribution within the region. In region 1, 25 uniformly distributed field intensity sampling points were selected, and the field intensity distribution function of this region was obtained through two-dimensional polynomial fitting. This operation was repeated for all nine regions to obtain their respective field intensity distribution functions. The gradient values ​​of the field intensity distribution at the boundary between adjacent regions were calculated, and the gradient difference was obtained. Taking the common boundary between region 1 and region 2 as an example, the fitted gradient difference sequence is: 0.002, 0.003, 0.004, 0.003, 0.032, 0.045, 0.002, 0.003, 0.002, 0.003 V / mm / mm. Compared with the previously established boundary constraints, the gradient difference at points 5 and 6 significantly exceeds the constraint range.

[0042] By comparing the gradient differences point by point with the boundary constraint values, it is possible to determine which locations exhibit abnormal gradient changes. In the aforementioned case, the gradient differences at points 5 and 6 on the common boundary between regions 1 and 2 are 0.032 and 0.045 V / mm / mm, respectively, exceeding the constraint upper limit of 0.015 V / mm / mm. Therefore, these two points are marked as gradient change anomalies. Similarly, by examining the boundaries of all adjacent regions, a total of 15 gradient change anomalies were identified. These points are mainly distributed at the region boundaries, reflecting unreasonable abrupt changes in the field strength distribution at these locations.

[0043] For identified gradient anomalies, an initial correction step size of 0.001 V / mm / mm is set. Taking point 5 as an example, its gradient difference is 0.032 V / mm / mm, while the upper constraint limit is 0.015 V / mm / mm, requiring incremental correction. The gradient difference is gradually decreased: 0.032 - 0.001 = 0.031, still exceeding the constraint; 0.031 - 0.001 = 0.030, still exceeding the constraint; this continues until the 18th decrease: 0.032 - 0.018 = 0.014 V / mm / mm. At this point, the gradient difference is less than the upper constraint limit of 0.015 V / mm / mm, therefore the gradient correction value is recorded as 0.018 V / mm / mm. This process is repeated for all gradient anomalies to obtain the corresponding gradient correction value set.

[0044] For each gradient anomaly point, the electric field strength values ​​of that point and its surrounding points are adjusted based on the calculated gradient correction value. The adjustment range is typically a 3×3 grid area centered on the anomaly point, with the adjustment magnitude decreasing proportionally to the distance from the anomaly point. For point 5, its gradient correction value of 0.018 V / mm / mm is applied to the initial electric field distribution. After adjustment, the electric field strength value at this point changes from 0.58 V / mm to 0.562 V / mm, and the surrounding 8 points are also fine-tuned proportionally. In this way, unreasonable abrupt changes in the electric field distribution are eliminated, making the spatial variation of the electric field strength smoother and more natural.

[0045] The optimized electric field distribution is combined with the electric field distribution of the measurement area to reconstruct a spatial electric field intensity distribution. After correcting all gradient change anomalies, the correction results are applied to the electric field distribution of each measurement area. The original electric field distribution in non-anomaly areas is retained, while the electric field distribution in anomaly areas is replaced with the corrected electric field distribution. Boundary smoothing techniques are used to ensure a natural and continuous transition of electric field intensity between different areas.

[0046] like Figure 2 The diagram illustrates the determination of the measurement region boundary based on the horizontal variation law in this embodiment. The horizontal axis represents the horizontal position, and the vertical axis represents the rate of change of field strength. It visually compares the gradient abrupt change detection results of the existing global smoothing algorithm with those of this scheme. The curve of this scheme shows significant peaks at x=250mm and x=550mm, with the rate of change surging to approximately 0.15V / mm and 0.18V / mm, respectively, significantly exceeding the average background difference of approximately 0.06V / mm. These two gradient abrupt change points are accurately captured by the system and marked by dashed lines as boundary line 1 and boundary line 2. This visually verifies that by detecting drastic changes in the horizontal field strength gradient, the region boundary can be effectively located, thereby dividing the measurement plane into independent measurement regions for subsequent fine-tuning and correction.

[0047] This invention effectively solves the problems of abrupt changes in field strength and boundary discontinuities in glass thickness monitoring caused by traditional capacitive sensors by dividing the initial field strength distribution into regions and optimizing the gradient. By establishing reasonable regional boundary constraints and identifying and correcting anomalies in gradient changes, the optimized electric field strength distribution better conforms to physical laws, eliminating field strength anomalies caused by measurement errors and environmental interference. This improves the spatial continuity and accuracy of the electric field strength distribution, providing more reliable basic data for subsequent glass thickness calculations, thereby enhancing the accuracy and stability of thickness monitoring in glass drawing processes.

[0048] Calculating the thickness difference between adjacent spatial locations in the thickness distribution data, and identifying spatial locations where the thickness difference exceeds a preset fluctuation threshold as abnormal thickness locations includes: Obtain the coordinate information of adjacent spatial positions in the thickness distribution data, measure the thickness difference between adjacent spatial positions, and construct a thickness difference matrix according to the coordinate information; Extract the decomposition coefficients of the thickness difference matrix, identify the fluctuation amplitude of the thickness difference based on the decomposition coefficients, and determine the spatial location where the fluctuation amplitude exceeds the preset fluctuation threshold as a candidate anomaly location. Extract the fluctuation amplitude and thickness difference of candidate anomaly locations, sort the candidate anomaly locations according to the fluctuation amplitude, and determine the candidate anomaly location with the largest fluctuation amplitude as the central anomaly location; The fluctuation amplitude attenuation law of candidate anomaly locations is analyzed based on the central anomaly location. Anomaly regions are divided according to the fluctuation amplitude attenuation law. The spatial location with the largest thickness difference within the anomaly region is selected as the anomaly thickness location.

[0049] In the glass drawing process, the thickness distribution data collected by capacitive sensors is typically represented as a set of thickness values ​​at different locations in a two-dimensional space. Each location has specific spatial coordinates and is adjacent to its surrounding locations. For a glass sheet with dimensions of 1000mm × 800mm, thickness data can be collected at a sampling interval of 10mm, forming a 100×80 thickness data grid. For each location in the grid, the thickness difference between it and its four adjacent locations is calculated. For example, if a location has coordinates (320mm, 450mm) and a thickness of 4.02mm, and its right-side adjacent location has coordinates (330mm, 450mm) and a thickness of 4.05mm, then the calculated thickness difference to the right in the horizontal direction is 0.03mm. Similarly, the thickness differences between this location and its adjacent locations in the top, bottom, and left directions are calculated to be -0.01mm, 0.02mm, and -0.04mm, respectively. The four-dimensional thickness differences of all locations are organized into a thickness difference matrix, which contains information on thickness changes in the horizontal and vertical directions and can reflect the gradient characteristics of the thickness distribution on the glass sheet surface.

[0050] The constructed thickness difference matrix is ​​subjected to singular value decomposition (SVD) to obtain its principal decomposition coefficients. In glass drawing thickness monitoring, the top 10 principal decomposition coefficients are typically selected for analysis based on the characteristics of the production process. These coefficients reflect the main variation patterns in the thickness difference matrix. Using a reconstruction algorithm, the thickness difference fluctuation amplitude at each spatial location can be calculated using these principal decomposition coefficients. For example, in a glass production line case, the top 5 principal coefficients obtained after decomposition are 1.85, 1.42, 0.98, 0.76, and 0.53, respectively. These coefficients are used to calculate the distribution of thickness difference fluctuation amplitude across the entire glass sheet surface. Based on the quality requirements of the glass production process, a preset fluctuation threshold of 0.08 mm is set. That is, when the thickness difference fluctuation amplitude at a certain spatial location exceeds 0.08 mm, it is considered that there may be a thickness anomaly at that location. By comparing the fluctuation amplitude with the preset threshold, 27 candidate anomaly locations are identified on the entire glass sheet. These locations are mainly distributed in the edge area and certain specific areas in the middle of the glass sheet.

[0051] For the 27 identified candidate anomaly locations, their fluctuation amplitude and thickness difference data in four directions were extracted. The fluctuation amplitudes, from highest to lowest, were: 0.152mm, 0.143mm, 0.137mm, 0.128mm, 0.125mm, etc., with corresponding spatial coordinates of (680mm, 320mm), (690mm, 320mm), (670mm, 330mm), (670mm, 310mm), (700mm, 320mm), etc. These candidate anomaly locations were sorted from largest to smallest fluctuation amplitude, and the location with the largest fluctuation amplitude (680mm, 320mm) was determined as the central anomaly location. The thickness differences in the four directions at this location were: -0.125mm above, -0.138mm below, -0.092mm to the left, and -0.145mm to the right, indicating that the thickness of the surrounding area was significantly higher than that at this location, forming a local "concave" feature.

[0052] Using the established central anomaly location (680mm, 320mm) as a reference point, the spatial distribution and fluctuation amplitude trends of surrounding candidate anomaly locations are analyzed. By calculating the distance from each candidate anomaly location to the central location and the corresponding fluctuation amplitude, the attenuation law of fluctuation amplitude with distance is obtained. In this case, the average fluctuation amplitude of candidate anomaly locations within 10mm of the central location is 0.145mm, within 20mm is 0.132mm, within 30mm is 0.118mm, within 40mm is 0.095mm, and within 50mm is 0.085mm. Based on this attenuation law, the area with fluctuation amplitude above 0.085mm can be identified as an anomaly region. This region is approximately elliptical, covering an area with a radius of about 50mm around the central location, and contains 15 candidate anomaly locations. Among these 15 locations, by comparing the absolute values ​​of the thickness differences in the four directions, it was found that the right-hand thickness difference at location (690mm, 320mm) reached -0.158mm, which is the largest thickness difference within the abnormal area. Therefore, this location was determined as the final abnormal thickness location, representing the point on the glass sheet where the thickness change is most drastic, requiring close monitoring and handling.

[0053] The abnormal thickness locations identified by the above methods reflect potential localized thickness fluctuations during glass drawing. These anomalies are often related to improper adjustment of drawing machinery parameters, melting temperature fluctuations, or uneven localized cooling. By accurately locating these anomalies, production parameters can be adjusted in a targeted manner, such as adjusting the drawing speed, optimizing the cooling airflow distribution, or correcting the melting temperature, thereby improving the thickness uniformity of the glass sheet.

[0054] This invention achieves precise identification of abnormal thickness locations during glass drawing through meticulous analysis of thickness distribution data. Employing thickness difference matrix decomposition technology, it effectively extracts the main variation patterns of thickness distribution, eliminates random noise interference, and improves the accuracy of anomaly detection. The anomaly region division method, determined through fluctuation amplitude analysis and spatial attenuation laws, makes the identification of abnormal thickness locations more objective and reasonable, avoiding misjudgments of single-point outliers. This invention can monitor thickness distribution changes in real time during glass drawing, promptly detect local thickness anomalies, improve the thickness uniformity and quality stability of glass products, and reduce material waste and energy consumption.

[0055] The distance traveled from the location of the abnormal thickness to the heating zone is calculated based on the spatial coordinates of the abnormal thickness location and the drawing direction. The travel time is then calculated based on the travel distance and the drawing speed, including: The drawing direction of the drawing process is determined based on the spatial coordinates of the abnormal thickness location. The spatial coordinates are then projected onto the drawing direction, and the straight-line distance from the abnormal thickness location to the boundary of the heating area is calculated. Stress detection points are set up along the path from the abnormal thickness location to the boundary of the heating area, stress data of the stress detection points are collected, and path deformation is calculated. The path deformation and straight-line distance are compensated for to obtain the actual moving distance from the abnormal thickness location to the heating area; Obtain the real-time rotation speed of the drawing equipment and convert the real-time rotation speed into a reference speed during the drawing process; Stress change data is collected along the actual moving distance path, and a speed change law is established based on the stress change data. The actual movement speed of each path segment is calculated in combination with the reference speed. The cumulative movement time is calculated based on the length of each path segment within the actual movement distance and the corresponding actual movement speed, and the cumulative movement time is used as the movement time.

[0056] In the glass drawing process, the location of abnormal thickness detected by capacitive sensors has clear spatial coordinates, typically representing the distance from the starting point of the drawing process. For example, on a glass production line, the coordinates of the detected abnormal thickness location are (580mm, 320mm), which represent a position 580mm laterally and 320mm longitudinally from the starting point of the drawing process. According to the process layout, the glass drawing direction is the longitudinal movement from the heating zone to the cooling zone. To calculate the distance from the abnormal thickness location to the heating zone, the coordinates of this location need to be projected onto the drawing direction. Since the drawing direction is parallel to the longitudinal axis, the projected position is 320mm. The boundary of the heating zone is 80mm on the longitudinal axis, so the straight-line distance from the abnormal thickness location to the boundary of the heating zone is 240mm. This straight-line distance is the basis for subsequent calculations, but due to the deformation of the glass during the drawing process, the actual movement distance needs further correction.

[0057] During the glass drawing process, thermal and mechanical stresses are generated, causing a deviation between the actual movement path and the theoretical straight path. To accurately calculate the actual movement distance, five stress detection points were evenly distributed along the path from the abnormal thickness location to the boundary of the heating zone, at distances of 80mm, 120mm, 160mm, 200mm, and 240mm from the boundary of the heating zone. Stress data at these detection points were collected using a photoelastic stress detection device, yielding values ​​of 12.5MPa, 11.8MPa, 10.2MPa, 8.7MPa, and 7.3MPa, respectively. These data indicate that the closer to the heating zone, the greater the stress experienced by the glass. Based on the relationship between stress and deformation, the local deformation at each detection point was calculated. In this case, the deformations at the five detection points were 1.6mm, 1.2mm, 0.9mm, 0.7mm, and 0.5mm, respectively. Summing these local deformations yielded a total path deformation of 4.9mm.

[0058] Considering the deformation characteristics during glass drawing, the actual distance traveled from the abnormal thickness location to the heating zone should be a combination of the straight-line distance and the path deformation. Based on the properties of the glass material and the drawing process parameters, a specific compensation function is used for calculation. Substituting the 4.9 mm path deformation into the compensation function and combining it with the 240 mm straight-line distance, the actual travel distance is calculated to be 244.9 mm. This result indicates that due to the deformation during the drawing process, the abnormal thickness location needs to travel 4.9 mm more than the theoretical straight-line distance to reach the boundary of the heating zone.

[0059] Glass drawing equipment typically consists of multiple sets of drive rollers, and the drawing speed of the glass is adjusted by controlling the rotational speed of these rollers. To accurately calculate the travel time at locations of abnormal thickness, real-time operating parameters of the drawing equipment are required. The main drive roller of the drawing equipment has a diameter of 200 mm, and its real-time monitored rotational speed is 15 rpm. Based on the circumference and rotational speed of the drive roller, a reference speed for drawing can be obtained. The drive roller has a circumference of 628 mm and rotates 15 times per minute; therefore, the reference speed is 9420 mm / min, or 157 mm / s. This reference speed represents the ideal glass movement speed, but in actual drawing processes, due to temperature changes and uneven stress distribution, the actual movement speed of the glass at different locations may vary.

[0060] To calculate the movement time more accurately, the velocity change of the glass during the drawing process needs to be considered. Stress change rate data were collected at the aforementioned five detection points along the actual movement path, with results of 0.15 MPa / s, 0.12 MPa / s, 0.09 MPa / s, 0.07 MPa / s, and 0.05 MPa / s, respectively. These data reflect the stress change trend over time and can be used to deduce the velocity change law. According to the rheological properties of glass, there is a corresponding relationship between the stress change rate and the velocity change. Using a specific conversion function and a reference velocity of 157 mm / s, the actual movement velocities at the five detection points were calculated to be 160.5 mm / s, 158.8 mm / s, 156.9 mm / s, 155.1 mm / s, and 153.2 mm / s, respectively. This indicates that as the glass gradually moves away from the heating area, its movement velocity decreases slightly, which is consistent with the physical property of glass shrinking and hardening after cooling.

[0061] The actual moving distance of 244.9 mm was divided into five path segments, each with lengths of 40 mm, 40 mm, 40 mm, 40 mm, and 84.9 mm. Based on the actual movement speed of each segment, the movement time for each path segment can be calculated. Specifically, the calculated movement times for the five path segments are 0.249 s, 0.252 s, 0.255 s, 0.258 s, and 0.554 s, respectively. Summing these times yields a total movement time of 1.568 s, meaning it takes 1.568 s for the abnormal thickness location to move from its current position to the boundary of the heating area. This movement time ensures that the heating parameters are adjusted in place by the time the abnormal thickness reaches the heating area, effectively correcting the thickness anomaly. This time value is transmitted to the heating control unit for precise control of the timing of heating power adjustments, ensuring the accuracy of thickness correction.

[0062] This invention provides crucial technical support for real-time thickness control in glass drawing processes by accurately calculating the travel distance and time from abnormal thickness locations to the heating zone. It fully considers deformation factors and speed variation patterns during glass drawing, achieving precise measurement of the travel distance through stress detection and compensation calculations; simultaneously, it combines real-time equipment rotation speed and path segment speed variations to accurately calculate the travel time. Compared to traditional methods that simplify calculations by only considering theoretical straight-line distance and average speed, this invention significantly improves calculation accuracy, allowing for more precise timing of heating parameter adjustments. It effectively addresses dynamic changes during glass drawing, promptly correcting thickness anomalies, improving the thickness uniformity and quality stability of glass products, reducing scrap rates, and increasing production efficiency.

[0063] The heating power adjustment range is determined based on the thickness difference corresponding to the abnormal thickness location, and the power adjustment time is determined based on the movement time. Applying the heating power adjustment range to the heating area at the power adjustment time includes: Collect deformation data of glass sheet at locations with abnormal thickness, and calculate temperature compensation values ​​based on the deformation data and the thickness difference corresponding to the abnormal thickness locations; Collect heat conduction data of the heating area, calculate a power reference value based on the heat conduction data and temperature compensation value, and use the power reference value as the initial heating power; Collect the temperature distribution of the heating area, calculate the power correction value based on the temperature distribution, and combine the power correction value with the initial heating power to generate the heating power adjustment range. Measure the temperature response curve of the heating area, determine the temperature response hysteresis based on the rise time and shift time of the temperature response curve, and use the temperature response hysteresis as the power adjustment time. At the power adjustment moment, the heating power adjustment range is applied to the heating area by the heating controller.

[0064] In the glass drawing process, glass sheets at locations with abnormal thickness will deform due to factors such as temperature and stress. This deformation is correlated with the thickness difference. The abnormal thickness location detected by a capacitive sensor has a thickness difference of -0.158mm, indicating that it is 0.158mm thinner than the standard thickness. To accurately calculate the required temperature compensation value, deformation data at this location needs to be collected. A high-precision laser displacement sensor was used to measure the deformation at the abnormal location, finding a vertical deformation of 0.072mm and a lateral deformation of 0.035mm. Combining the relationship between deformation and thickness difference, the required temperature compensation value was calculated. In this case, considering the -0.158mm thickness difference, the 0.072mm vertical deformation, and the 0.035mm lateral deformation, the temperature compensation value was calculated to be 11.5℃ using thermal expansion coefficient conversion. This means that the temperature in this area needs to be increased by 11.5℃ to correct the thickness difference.

[0065] Heat conduction data of the heating area was collected, and a power reference value was calculated based on the heat conduction data and temperature compensation value. This power reference value was then used as the initial heating power. The thermal conductivity of the glass material affects the heating effect; therefore, heat conduction data of the heating area needs to be collected to accurately control the heating power. Temperature changes were measured at different locations within the heating area using a temperature sensor array, and the thermal conductivity coefficient was calculated. The measured thermal conductivity coefficient of the glass sheet in the heating area was 1.05 W / (m·K), and the thermal diffusivity was 0.58 mm. 2 / s. Based on these heat conduction data, combined with the aforementioned temperature compensation value of 11.5℃, the required power baseline value is calculated. Considering the heating area size of 200mm × 120mm and the glass thickness of 4mm, the power baseline value is calculated to be 382W using the heat conduction equation. This means that in order to raise the temperature at the location of abnormal thickness by 11.5℃, an initial heating power of 382W is required.

[0066] Uneven temperature distribution in the heating area leads to variations in heating effectiveness, affecting the accuracy of thickness correction. Temperature distribution in the heating area was acquired using an infrared thermal imager, measuring a center temperature of 535℃ and an average edge temperature of 512℃, with a temperature gradient of 0.23℃ / mm. Based on the uneven temperature distribution, a power correction value was calculated to ensure more concentrated heat application to areas of abnormal thickness. In this case, considering that the abnormal thickness location would shift to the right side of the heating area, where the temperature is 7℃ lower than the center, additional power was required for compensation. Using the relationship between temperature distribution and power, a power correction value of 27W was calculated. Combining this 27W power correction value with the initial heating power of 382W, the final heating power adjustment range was determined to be 409W. This adjustment range takes into account the uneven temperature distribution, enabling more accurate correction of thickness differences at abnormal thickness locations.

[0067] There is a time delay between the heating device receiving the power adjustment command and the actual temperature reaching the target value, which affects the accuracy of thickness correction. To determine the appropriate power adjustment timing, the temperature response curve of the heating area needs to be measured. A fast-response temperature sensor is installed in the heating area to record the temperature change process after a power change. It was measured that when the power increases from 350W to 450W, the heating area temperature rises from 520℃ to 540℃ in 0.85s, with a delay of 0.12s for the temperature to begin rising significantly, and a slope of 23.5℃ / s during the temperature rise. Based on this temperature response curve, combined with the previously calculated travel time of 1.568s from the abnormal thickness location to the heating area, the temperature response lag is determined. Since the heating temperature needs to reach the target value precisely when the abnormal thickness location reaches the heating area, the heating power adjustment needs to be initiated earlier. Based on the 0.85s required for temperature rise, the power adjustment should be initiated 0.85s earlier, i.e., the temperature response lag is 0.85s. Based on the movement time of 1.568s, the power adjustment time was determined to be 0.718s after the abnormal thickness position was detected.

[0068] The heating controller receives a power adjustment command and adjusts the power to the target value at a specified time. Upon detecting an abnormal thickness location, a timer starts counting. When the timer reaches 0.718 seconds, the heating controller receives a command to adjust the heating power from the current value to 409W. The heating controller employs a PID control algorithm with parameters set to P=0.85, I=0.12, and D=0.03 to ensure the smoothness and accuracy of power adjustment. During power adjustment, the controller monitors the current and voltage of the heating element in real time to ensure that the actual output power matches the target value. Simultaneously, it monitors the real-time temperature of the heating area and fine-tunes the power output through a feedback mechanism, making temperature changes more precise. After power adjustment, the temperature of the heating area gradually rises and stabilizes at 546.5℃, perfectly matching the target temperature of 546.5℃. With precise temperature adjustment, the local thermal expansion effect of the glass sheet in this area corrects the thickness, reducing the abnormal thickness difference from -0.158mm to -0.025mm, meeting the process requirements for thickness uniformity. The entire power adjustment process lasted 2.5 seconds, after which the heating power gradually returned to normal, completing the correction process for the abnormal thickness location.

[0069] This invention achieves dynamic correction of abnormal thickness during glass drawing by precisely calculating the adjustment range and timing of heating power. It enables accurate temperature compensation for specific thickness differences, achieving directional thickness control. By precisely calculating the power adjustment timing, the timing problem between heating response and glass movement is solved, ensuring that the heating effect acts on the target location. This invention significantly improves the accuracy of thickness control in glass drawing processes, reduces thickness fluctuations, and enhances product quality stability.

[0070] The glass drawing process thickness monitoring system based on a capacitive sensor provided in this embodiment of the invention includes: The capacitance signal acquisition module is used to acquire the original capacitance signal of the glass drawing region through a capacitive sensor array, and simultaneously acquire the sensor temperature and drawing speed. The temperature compensation module is used to calculate the capacitor temperature offset based on the sensor temperature, subtract the capacitor temperature offset from the original capacitor signal, and obtain the compensated capacitor signal. The thickness calculation module is used to establish the spatial electric field intensity distribution based on the difference in capacitance values ​​of each sensor electrode in the compensation capacitor signal, and convert the spatial electric field intensity distribution into glass thickness values ​​at each spatial location to form thickness distribution data. An abnormal location determination module is used to calculate the thickness difference between adjacent spatial locations in the thickness distribution data, determine the spatial location where the thickness difference exceeds a preset fluctuation threshold as an abnormal thickness location, and extract the spatial coordinates of the abnormal thickness location and the corresponding thickness difference. The movement time calculation module is used to calculate the movement distance from the abnormal thickness position to the heating area based on the spatial coordinates of the abnormal thickness position and the drawing direction, and to calculate the movement time based on the movement distance and the drawing speed. The heating control module is used to determine the heating power adjustment range based on the thickness difference corresponding to the abnormal thickness position, determine the power adjustment time based on the movement time, and apply the heating power adjustment range to the heating area at the power adjustment time.

[0071] One technical solution provided in this embodiment of the invention is an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0072] One technical solution provided in this embodiment of the invention is a computer-readable storage medium storing a computer program, wherein the processor executes the computer program to implement the steps in any of the aforementioned methods.

[0073] The specific embodiments described above are preferred embodiments of the present invention and are not intended to limit the specific scope of the present invention. The scope of the present invention includes, but is not limited to, these specific embodiments. All equivalent changes made in accordance with the shape and structure of the present invention are within the protection scope of the present invention.

Claims

1. A method for monitoring the thickness of glass drawing process based on a capacitive sensor, characterized in that, Includes the following steps: The original capacitance signal of the glass drawing region is obtained by a capacitive sensor array, and the sensor temperature and drawing speed are obtained simultaneously. The capacitor temperature offset is calculated based on the sensor temperature, and the capacitor temperature offset is subtracted from the original capacitor signal to obtain the compensation capacitor signal. The spatial electric field intensity distribution is established based on the difference in capacitance values ​​of each sensor electrode in the compensation capacitor signal. The spatial electric field intensity distribution is then converted into glass thickness values ​​at each spatial location to form thickness distribution data. Calculate the thickness difference between adjacent spatial locations in the thickness distribution data, identify spatial locations where the thickness difference exceeds a preset fluctuation threshold as abnormal thickness locations, and extract the spatial coordinates of the abnormal thickness locations and the corresponding thickness difference. The distance the abnormal thickness location moves from the abnormal thickness location to the heating zone is calculated based on the spatial coordinates of the abnormal thickness location and the drawing direction, and the moving time is calculated based on the moving distance and the drawing speed. The heating power adjustment range is determined based on the thickness difference corresponding to the abnormal thickness location, the power adjustment time is determined based on the movement time, and the heating power adjustment range is applied to the heating area at the power adjustment time.

2. The method according to claim 1, characterized in that, The capacitor temperature offset is calculated based on the sensor temperature. This offset is then subtracted from the original capacitor signal to obtain the compensated capacitor signal, which includes: Collect temperature data sequences from the temperature sensor at different times and construct temperature change curves of the sensor in chronological order; The temperature change rate between adjacent moments is calculated based on the sensor temperature change curve, and the temperature data sequence is divided into a stable sensor temperature range and a fluctuating sensor temperature range based on the temperature change rate. The reference temperature correction value is obtained by calculating the linear relationship between the temperature data and the original capacitance signal within the stable temperature range of the sensor. The dynamic capacitor temperature correction value is obtained by calculating the linear relationship between the sensor temperature and the original capacitance signal within a fixed time window within the sensor temperature fluctuation range. The temperature compensation weight is obtained by weighting the reference temperature correction value and the dynamic capacitor temperature correction value. The capacitor temperature offset is calculated based on the temperature compensation weight, and the capacitor temperature offset is subtracted from the original capacitor signal to obtain the preliminary compensated capacitor signal. The initial compensation capacitor signal is subjected to time-series accumulation processing to calculate the change trend of the capacitance value over time. Based on the change trend, the initial compensation capacitor signal is nonlinearly corrected to obtain the compensation capacitor signal.

3. The method according to claim 1, characterized in that, The spatial electric field intensity distribution is established based on the capacitance differences of each sensor electrode in the compensation capacitor signal. This spatial electric field intensity distribution is then converted into glass thickness values ​​at each spatial location, forming thickness distribution data, including: The capacitance difference between the sensor electrodes is calculated based on the compensation capacitor signal. Within the measurement plane formed by the sensor electrodes, the difference in capacitance values ​​is scanned along the horizontal direction to obtain the horizontal field strength component, and the difference in capacitance values ​​is scanned along the vertical direction to obtain the vertical field strength component. The horizontal field strength component and the vertical field strength component are orthogonally combined to obtain the initial field strength distribution. The measurement area is divided according to the initial field strength distribution in the measurement plane. The gradient change of the initial field strength distribution in the measurement area is extracted. The gradient change is used as a constraint to optimize the initial field strength distribution and obtain the spatial electric field strength distribution. The spatial electric field intensity distribution is converted into the field intensity distribution inside the medium through the dielectric properties of the glass medium, and the medium non-uniformity correction coefficient is calculated based on the field intensity distribution inside the medium. The glass thickness value at each spatial location is obtained by correcting the field intensity distribution within the medium based on the medium inhomogeneity correction coefficient. The position coordinate data of the sensor electrode in the measurement plane is collected, and the glass thickness value is reconstructed into thickness distribution data based on the position coordinate data.

4. The method according to claim 3, characterized in that, The measurement area is divided within the measurement plane based on the initial electric field distribution. The gradient change of the initial electric field distribution within the measurement area is extracted, and the gradient change is used as a constraint to optimize the initial electric field distribution, resulting in a spatial electric field intensity distribution including: Obtain the initial field strength distribution field strength values ​​within the measurement plane, calculate the variation patterns of the field strength values ​​in the horizontal and vertical directions respectively, determine the boundary line positions of the measurement area based on the variation patterns, and divide the measurement plane into multiple measurement areas; Extract the initial field strength distribution between adjacent measurement areas, calculate the gradient change of the initial field strength distribution, and establish boundary constraints of the measurement area based on the gradient change; The initial field strength distribution within each measurement region is fitted to obtain the field strength distribution of the measurement region. The gradient difference between the field strength distributions of adjacent measurement regions is calculated based on the boundary constraints of the measurement region. The gradient difference is compared with the boundary constraints of the measurement area to determine the boundary location that exceeds the constraints, and the boundary location is marked as anomaly point of gradient change. Set a correction step size at the abnormal gradient change point, and calculate the gradient difference by decreasing the correction step size. When the reduced gradient difference is less than the boundary constraint of the measurement area, record it as the gradient correction value. The gradient correction value is applied to the initial field strength distribution at the gradient change anomaly point to obtain the optimized field strength distribution; The optimized field strength distribution is combined with the field strength distribution of the measurement area to reconstruct the spatial electric field strength distribution.

5. The method according to claim 1, characterized in that, Calculating the thickness difference between adjacent spatial locations in the thickness distribution data, and identifying spatial locations where the thickness difference exceeds a preset fluctuation threshold as abnormal thickness locations includes: Obtain the coordinate information of adjacent spatial positions in the thickness distribution data, measure the thickness difference between adjacent spatial positions, and construct a thickness difference matrix according to the coordinate information; Extract the decomposition coefficients of the thickness difference matrix, identify the fluctuation amplitude of the thickness difference based on the decomposition coefficients, and determine the spatial location where the fluctuation amplitude exceeds the preset fluctuation threshold as a candidate anomaly location. Extract the fluctuation amplitude and thickness difference of candidate anomaly locations, sort the candidate anomaly locations according to the fluctuation amplitude, and determine the candidate anomaly location with the largest fluctuation amplitude as the central anomaly location; The fluctuation amplitude attenuation law of candidate anomaly locations is analyzed based on the central anomaly location. Anomaly regions are divided according to the fluctuation amplitude attenuation law. The spatial location with the largest thickness difference within the anomaly region is selected as the anomaly thickness location.

6. The method according to claim 1, characterized in that, The distance traveled from the location of the abnormal thickness to the heating zone is calculated based on the spatial coordinates of the abnormal thickness location and the drawing direction. The travel time is then calculated based on the travel distance and the drawing speed, including: The drawing direction of the drawing process is determined based on the spatial coordinates of the abnormal thickness location. The spatial coordinates are then projected onto the drawing direction, and the straight-line distance from the abnormal thickness location to the boundary of the heating area is calculated. Stress detection points are set up along the path from the abnormal thickness location to the boundary of the heating area, stress data of the stress detection points are collected, and path deformation is calculated. The path deformation and straight-line distance are compensated for to obtain the actual moving distance from the abnormal thickness location to the heating area; Obtain the real-time rotation speed of the drawing equipment and convert the real-time rotation speed into a reference speed during the drawing process; Stress change data is collected along the actual moving distance path, and a speed change law is established based on the stress change data. The actual movement speed of each path segment is calculated in combination with the reference speed. The cumulative movement time is calculated based on the length of each path segment within the actual movement distance and the corresponding actual movement speed, and the cumulative movement time is used as the movement time.

7. The method according to claim 1, characterized in that, The heating power adjustment range is determined based on the thickness difference corresponding to the abnormal thickness location, and the power adjustment time is determined based on the movement time. Applying the heating power adjustment range to the heating area at the power adjustment time includes: Collect deformation data of glass sheet at locations with abnormal thickness, and calculate temperature compensation values ​​based on the deformation data and the thickness difference corresponding to the abnormal thickness locations; Collect heat conduction data of the heating area, calculate a power reference value based on the heat conduction data and temperature compensation value, and use the power reference value as the initial heating power; Collect the temperature distribution of the heating area, calculate the power correction value based on the temperature distribution, and combine the power correction value with the initial heating power to generate the heating power adjustment range. Measure the temperature response curve of the heating area, determine the temperature response hysteresis based on the rise time and shift time of the temperature response curve, and use the temperature response hysteresis as the power adjustment time. At the power adjustment moment, the heating power adjustment range is applied to the heating area by the heating controller.

8. A glass drawing process thickness monitoring system based on a capacitive sensor, used to implement the method described in any one of claims 1-7, characterized in that, The system includes: The capacitance signal acquisition module is used to acquire the original capacitance signal of the glass drawing region through a capacitive sensor array, and simultaneously acquire the sensor temperature and drawing speed. The temperature compensation module is used to calculate the capacitor temperature offset based on the sensor temperature, subtract the capacitor temperature offset from the original capacitor signal, and obtain the compensated capacitor signal. The thickness calculation module is used to establish the spatial electric field intensity distribution based on the difference in capacitance values ​​of each sensor electrode in the compensation capacitor signal, and convert the spatial electric field intensity distribution into glass thickness values ​​at each spatial location to form thickness distribution data. An abnormal location determination module is used to calculate the thickness difference between adjacent spatial locations in the thickness distribution data, determine the spatial location where the thickness difference exceeds a preset fluctuation threshold as an abnormal thickness location, and extract the spatial coordinates of the abnormal thickness location and the corresponding thickness difference. The movement time calculation module is used to calculate the movement distance from the abnormal thickness position to the heating area based on the spatial coordinates of the abnormal thickness position and the drawing direction, and to calculate the movement time based on the movement distance and the drawing speed. The heating control module is used to determine the heating power adjustment range based on the thickness difference corresponding to the abnormal thickness position, determine the power adjustment time based on the movement time, and apply the heating power adjustment range to the heating area at the power adjustment time.

9. An electronic device, characterized in that, include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer program instructions that, when executed by a processor, implement the steps of the method as described in any one of claims 1 to 7.

Images