A PID-based fireproof tempered glass heating control system

Abstract

This invention relates to the field of fire-resistant tempered glass processing control technology, specifically a fire-resistant tempered glass control system based on PID regulation, comprising: a data acquisition module to acquire surface temperature distribution data and physical dimension parameters of the target processed part; a preprocessing module to extract the average surface temperature and calculate the surface temperature rise rate; a feature calculation module to determine the core temperature state based on the physical dimension parameters, average surface temperature, and a preset thermal conduction state space model, and to generate thermal gradient decoupling feature quantities by combining the surface temperature rise rate and storing them as historical thermal gradient curves in a time series; a variable parameter control module to compare the feature quantities with preset thresholds to generate dynamic proportional-integral-derivative parameters; and an execution output module to calculate the target output value and convert it into a PWM signal to control the output power of the heating actuator; thereby achieving coordinated control of heating rate, heat homogenization quality, and heating power output.

Description

Technical Field

[0001] This invention relates to the field of fireproof glass tempering processing control technology, specifically a fireproof glass tempering control system based on PID regulation. Background Technology

[0002] As the tempering process of fireproof glass develops towards production beyond the preset area specifications, multiple specifications, and continuous production, temperature control during the heating process has become a key factor affecting the quality of finished products and production stability. In order to achieve uniform heating and stable tempering of fireproof glass during the heating stage, higher requirements are placed on the real-time sensing and dynamic adjustment capabilities of the heating control system. In the existing heating control process for tempered fireproof glass, the glass surface temperature or furnace temperature signal is usually used as the main feedback basis, and the heating power is adjusted by a proportional-integral-derivative control method with fixed parameters. However, fireproof glass exhibits significant thickness-to-thickness heat conduction lag during the heating process, especially for products with a thickness greater than the preset thickness threshold. The surface temperature change and the core temperature response are often asynchronous. Relying solely on surface temperature for control makes it difficult to accurately reflect the true thermal state inside the glass. This can easily lead to a high heating intensity even when the surface temperature rises too quickly and the core temperature does not keep up, resulting in problems such as surface overheating, excessive thickness-to-thickness temperature difference, insufficient heat uniformity, and unstable subsequent stress distribution. At the same time, the infrared temperature acquisition process is also susceptible to factors such as reflection, stains, missing frames, and batch specification changes, which amplify the fluctuations in the input signal and further affect the accuracy of the control parameter judgment. Therefore, effectively processing the surface temperature data during the heating stage of fireproof glass, calculating the core thermal state in conjunction with parameters such as glass thickness, and then dynamically adjusting the heating control parameters based on the difference in heat conduction between the core and the surface to form a stable control mechanism suitable for continuous tempering production lines is crucial for improving the heating uniformity of fireproof glass, suppressing temperature overshoot, and ensuring the consistency of finished product quality. Summary of the Invention

[0003] The purpose of this invention is to provide a PID-based fireproof glass tempering heating control system to solve the following technical problems: The aim is to transform the traditional static feedback that relies solely on surface temperature into a dynamic adjustment feedback that considers the hysteresis of heat conduction between the core and the surface, thereby achieving coordinated control of heating rate, heat homogenization quality, and heating power output.

[0004] The objective of this invention can be achieved through the following technical solutions: A fireproof tempered glass heating control system based on PID regulation, the system comprising: The data acquisition module is used to acquire surface temperature distribution data and physical dimensional parameters of fireproof glass; The preprocessing module is used to filter the surface temperature distribution data to extract the average surface temperature and calculate the surface temperature rise rate based on the average surface temperature. The feature calculation module is used to determine the core temperature state based on physical size parameters, average surface temperature and a preset thermal conduction state space model for characterizing the thickness thermal conduction characteristics. It combines the surface temperature rise rate to generate thermal gradient decoupling feature quantities and stores the thermal gradient decoupling feature quantities as historical thermal gradient curves in time series. The variable parameter control module is used to compare the thermal gradient decoupling feature with a preset threshold and generate dynamic proportional-integral-differential parameters based on the comparison result. The output module is used to calculate the target output value based on the dynamic proportional-integral-derivative parameters and convert the target output value into a pulse width modulation signal to control the output power of the heating actuator that is connected to the system.

[0005] Preferably, the preprocessing module includes: The filtering unit is used to perform sliding window mean filtering on the surface temperature distribution data and remove outlier data points to extract the average surface temperature. The differential unit is used to calculate the difference between the average surface temperature of the current sampling period and the average surface temperature of the previous sampling period according to a preset sampling period, so as to obtain the surface temperature rise rate.

[0006] Preferably, the feature calculation module includes: The state iteration unit is used to input physical size parameters and average surface temperature into a preset heat conduction state space model for iterative calculation to output the core temperature state. The feature composite unit is used to weight the surface temperature rise rate based on a preset first engineering calibration coefficient and to weight the core temperature state based on a preset second engineering calibration coefficient. The difference between the weighted surface temperature rise rate and the weighted core temperature state is calculated to generate a thermal gradient decoupling feature quantity.

[0007] Preferably, the preset threshold includes a positive threshold, and the variable parameter control module includes: The first adjustment unit is used to obtain the system's preset basic proportional coefficient and basic differential coefficient. When the thermal gradient decoupling feature is greater than the positive threshold, the difference between the thermal gradient decoupling feature and the positive threshold is used as the independent variable to exponentially decay the basic proportional coefficient to obtain the target proportional coefficient, and to linearly increase the basic differential coefficient to obtain the target differential coefficient. The target proportional coefficient and the target differential coefficient are output as dynamic proportional integral differential parameters. The second adjustment unit is used to output the basic proportional coefficient and basic differential coefficient as dynamic proportional-integral-differential parameters when the thermal gradient decoupling characteristic quantity is less than or equal to the positive threshold, and to introduce an integral coefficient to eliminate steady-state error.

[0008] Preferably, the system also includes: The quality inspection module is used to obtain the surface compressive stress value of the tempered fireproof glass. The adaptive correction module is used to extract the historical thermal gradient curve of the fireproof glass during the heating stage when the absolute value of the difference between a consecutive preset number of surface compressive stress values ​​and the target median value is greater than a preset tolerance threshold, and to fine-tune the thermal conductivity in the thermal conduction state space model based on the historical thermal gradient curve; when the absolute value of the difference is less than or equal to the preset tolerance threshold, the current thermal conductivity of the thermal conduction state space model remains unchanged.

[0009] Preferably, the data acquisition module includes: The infrared scanning unit is used to acquire a two-dimensional temperature matrix of the fireproof glass surface at a preset acquisition frequency as surface temperature distribution data. The manufacturing execution acquisition unit is used to read the thickness parameters of the current batch of fireproof glass from the manufacturing execution system as physical dimension parameters.

[0010] Preferably, the system further includes: a mode switching module, used to disable the variable parameter control module and switch the control mode to a fixed parameter control mode when the physical size parameter is less than or equal to a preset thickness threshold, so as to control the output power of the heating actuator through fixed proportional integral derivative parameters; and to keep the variable parameter control module in an active state when the physical size parameter is greater than the preset thickness threshold.

[0011] Preferably, the preset thickness threshold is 3mm, and the heating actuator includes a solid-state relay and a heating element.

[0012] The beneficial effects of this invention are: 1. This invention calculates the temperature of the glass core by introducing a thermal conduction state-space model and generates a thermal gradient decoupling characteristic quantity by combining the surface temperature rise rate to dynamically adjust the proportional, integral, and differential parameters. This effectively solves the problem of surface overheating caused by the hysteresis of thermal conduction in the thickness direction and realizes coordinated control of the heating process and improved heat homogenization quality. 2. This invention uses sliding window mean filtering to remove outlier data points and extract the average surface temperature. At the same time, it performs differential calculation to obtain the surface temperature rise rate, effectively isolating the interference of measurement noise on the front-end data and providing a stable and reliable input for subsequent thermal state estimation and variable parameter control. 3. This invention, by calibrating and weighting the surface temperature rise rate and the core temperature state and then subtracting them, explicitly introduces the core thermal state, which cannot be directly measured, into the control link, thereby realizing online observation and quantitative expression of the thick-section heat conduction hysteresis of fireproof glass and avoiding misjudging abnormal conditions as normal temperature rise. 4. This invention uses threshold segmentation for proportional-integral-derivative gain scheduling. When thermal imbalance is severe, the proportional coefficient is exponentially decayed and the derivative coefficient is linearly increased. When the temperature approaches uniformity, the basic parameters are restored and integral action is introduced, thus achieving a precise temperature control effect where the more dangerous the thermal gradient, the stronger the suppression. 5. This invention introduces a cross-process closed-loop correction mechanism. When a systematic deviation occurs in the compressive stress on the surface of multiple finished glass pieces, the historical thermal gradient curve is extracted to fine-tune the thermal conductivity, thereby realizing the long-term self-correction capability of using the finished product quality results to correct the front-end heating model. 6. This invention utilizes infrared scanning to acquire a two-dimensional temperature matrix and automatically reads batch thickness parameters from the manufacturing execution system, avoiding the limitations of single-point temperature measurement and manual input errors, ensuring the accuracy and synchronization of the information required for thermal state calculation, and realizing the automation of control input; 7. This invention sets up a mode switching mechanism based on a thickness threshold. When the temperature difference of the ultra-thin glass core is extremely small, it automatically degrades to a fixed proportional integral differential parameter control mode, which clarifies the applicable boundary of the system, avoids power jitter caused by variable parameter control, and ensures the reliability when switching between products of different specifications. Attached Figure Description

[0013] Other features, objects, and advantages of the invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 This is a schematic diagram of a fireproof glass tempering heating control system based on PID regulation, provided as an embodiment of this application. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] Please see Figure 1 A fireproof glass tempering heating control system based on PID regulation, the system includes: a data acquisition module, used to acquire surface temperature distribution data and physical dimensional parameters of fireproof glass; The preprocessing module is used to filter the surface temperature distribution data to extract the average surface temperature and calculate the surface temperature rise rate based on the average surface temperature. The feature calculation module is used to determine the core temperature state based on physical size parameters, average surface temperature and a preset thermal conduction state space model for characterizing the thickness thermal conduction characteristics. It combines the surface temperature rise rate to generate thermal gradient decoupling feature quantities and stores the thermal gradient decoupling feature quantities as historical thermal gradient curves in time series. The variable parameter control module is used to compare the thermal gradient decoupling feature with a preset threshold and generate dynamic proportional-integral-differential parameters based on the comparison result. The output module is used to calculate the target output value based on the dynamic proportional-integral-derivative parameters and convert the target output value into a pulse width modulation signal to control the output power of the heating actuator that is connected to the system.

[0016] This embodiment provides a PID-based heating control mechanism for fireproof tempered glass. Specifically, this mechanism is deployed in a continuous fireproof tempered glass production line, with the front section being a loading platform, the middle section being a multi-temperature zone heating furnace, and the rear section being a cooling zone and a cold end testing platform. The target workpiece enters the heating furnace in a single piece along the ceramic roller conveyor. During the entire heating process, the system no longer uses the glass surface temperature as the only feedback quantity, but further calculates the thermal state of the glass core and uses the change in thermal gradient between the core and the surface to adjust the PID parameters online. Specifically, when a piece of fireproof glass enters the heating furnace, the temperature acquisition device installed on the furnace top continuously outputs two-dimensional temperature distribution data of the glass surface, and the manufacturing execution system simultaneously sends out the corresponding thickness parameters of the glass; the preprocessing module first extracts and filters the effective area of ​​the two-dimensional temperature distribution data to obtain the average surface temperature at the current sampling time; The average surface temperature is compared with the average surface temperature of the previous sampling period to calculate the surface temperature rise rate. The feature calculation module then uses the thickness parameter, the current average surface temperature and the preset heat conduction state space model together to estimate the core temperature state at the current moment in an iterative manner. Furthermore, the surface temperature rise rate and the core temperature state are combined to generate a thermal gradient decoupling characteristic quantity to characterize the current degree of thermal imbalance. To illustrate the data flow process, an example is given below: Suppose that the average surface temperature of an 8mm fireproof glass in three adjacent sampling periods is 640℃, 646℃, and 651℃ respectively, and the sampling period is 0.1s. Then, the surface temperature rise rate in the second period can be considered as 60℃ / s, and the surface temperature rise rate in the third period can be considered as 50℃ / s. If the state-space model estimates the core temperature to be 620°C in the third cycle, the system can construct a characteristic quantity that simultaneously reflects whether the surface heating rate is too fast and whether the core heat transfer response is sluggish. This characteristic quantity is not a simple temperature difference, but rather incorporates both the surface heating trend and the core heat conduction sluggishness. Assuming that the thermal gradient decoupling characteristic value for this cycle is 18 after engineering calibration, it indicates that there is a thermal imbalance state where the surface temperature rise rate is higher than the preset rate threshold and the core temperature has not reached the target following temperature. If this value gradually decreases to 5 or 2 in subsequent cycles, it indicates that the glass is entering the homogenization stage. The variable parameter control module compares this characteristic with a preset threshold to determine which PID parameters to use in the current control cycle. When the thermal gradient decoupling characteristic is too large, it indicates that the glass surface absorbs heat too quickly and the heat transfer to the core is relatively lagging. If the proportional adjustment is maintained at a value greater than the preset gain coefficient, it is easy to cause the surface temperature to overshoot. Therefore, the system reduces the proportional action and enhances the differential suppression. When the characteristic value falls back to near the threshold, it indicates that the glass is approaching a uniform temperature. The system restores the basic PID parameters and gradually introduces integral action to reduce steady-state error. The system presets or the manufacturing execution system issues the target process temperature. The execution output module calculates the target output value based on the real-time updated PID parameters and the current temperature deviation. Then, it converts the target output value into a pulse width modulation signal to drive the solid-state relay to modulate the duty cycle of the heating wire, thereby changing the output power of the corresponding temperature zone. In addition, as an anomaly handling mechanism, when surface temperature distribution data experiences short-term frame loss, local areas become unavailable, or thickness parameters are not synchronized in a timely manner, the system prioritizes calling the average surface temperature of the previous valid sampling period and the most recent valid thickness parameter, and marks the current period as a degraded calculation period. If valid temperature data cannot be obtained for several consecutive cycles, the dynamic PID parameters will be temporarily frozen to maintain the most recent stable output until the data is recovered. If the core temperature estimated by the model exceeds the allowable range of the process, such as being higher than the surface temperature or experiencing a non-physical jump, then the state solution for that cycle is considered abnormal, the feature quantity is directly bypassed for updating, and only the basic PID control is retained to prevent control quantity oscillation. In the tempering process of 8mm high borosilicate fireproof glass with a specification of 2440×3660mm, when the glass enters the third heating zone, the infrared scanning results show that the surface temperature rises rapidly, while the estimated core temperature follows more slowly. The system continuously obtains a relatively large thermal gradient decoupling characteristic value between 3.2s and 4.5s, so it actively reduces the heating power in this region to prevent the upper surface of the glass from softening first. After 5 seconds, the characteristic value drops back down, and the system gradually resumes normal proportional and integral regulation to make the glass achieve a more uniform thickness temperature distribution before it leaves the furnace. The purpose of this step is to transform the traditional static feedback that relies solely on surface temperature into a dynamic adjustment feedback that considers the hysteresis of heat conduction between the core and the surface, thereby achieving coordinated control of heating rate, heat homogenization quality, and heating power output.

[0017] In a preferred embodiment of the present invention, the preprocessing module includes: a filtering unit, used to perform sliding window mean filtering on the surface temperature distribution data and remove abnormal data points to extract the average surface temperature; and a difference unit, used to perform difference calculation between the average surface temperature of the current period and the average surface temperature of the previous sampling period according to a preset sampling period to obtain the surface temperature rise rate.

[0018] This embodiment provides a preprocessing mechanism for temperature front-end data purification; specifically, in a continuous tempering furnace, the infrared scanning device is easily affected by glass edge reflection, furnace internal heat radiation interference, and local stains. If the original two-dimensional temperature matrix is ​​directly sent into the subsequent control link, the surface temperature rise rate will fluctuate frequently, resulting in control deviation of the dynamic PID parameters. Therefore, this embodiment further incorporates a filtering unit and a differential unit in the aforementioned overall control process to improve the stability of the temperature input; Specifically, the filtering unit performs sliding window mean filtering on the two-dimensional temperature matrix collected in each sampling period; the glass surface can be divided into several adjacent small regions, for example, each small region covers 3×3 temperature sampling points; For each small region, first remove outliers that deviate significantly from the surrounding data, then calculate the local mean within the window, and finally average the local mean of the effective region to obtain the average surface temperature of the glass in the current period. For ease of understanding, specific examples are given below; assume that within a certain sampling period, the 3×3 temperature submatrices extracted from the effective region are: 641, 642, 643; 640, 700, 644; 639, 641, 642; The 700℃ value is significantly higher than the surrounding data and can be considered an outlier caused by a bright spot or a bad pixel. After removing this value, the average of the remaining 8 values ​​is approximately 641.5℃. Therefore, 641.5℃ is taken as the filtering result for this sub-region. If the glass has four effective sub-regions with filtering results of 641.5℃, 642.0℃, 640.8℃ and 641.7℃ respectively, then the overall average surface temperature can be taken as approximately 641.5℃. The differential unit compares this value with the average surface temperature of the previous sampling period; assuming the previous period was 636.5℃ and the sampling period is still 0.1s, the surface temperature rise rate of this period can be converted to 50℃ / s. Furthermore, if the average surface temperatures of two adjacent cycles are 641.5℃ and 641.6℃ respectively, the calculated surface temperature rise rate is only 1℃ / s. When this rate is fed into subsequent feature calculations, it will be identified as a gradual heating phase that tends towards homogenization, rather than a rapid heating phase. This shows that the combined effect of the filtering unit and the differential unit can reduce the amplification effect of the original temperature noise on the control parameters. In addition, in the anomaly handling mechanism, if the number of valid sampling points in a certain window is lower than the preset ratio, for example, if there are only 2 valid points left in a 3×3 window, the result of that window will not be included in the overall average calculation of the current period; if the number of valid windows in the whole glass is insufficient, for example, if the number of valid windows is less than 60% of the total number of windows, the average surface temperature of the current period will directly use the data of the previous valid period and set a low confidence mark in the internal system. For differential calculation, if there are multiple missing frames between the previous valid period and the current period, the differential unit calculates the temperature rise rate according to the actual time interval, instead of calculating it according to a single sampling period, in order to avoid artificially high slopes. On the same high borosilicate fireproof glass production line, when a piece of glass passes through the middle of the heating furnace, there is local dust accumulation in the furnace top viewing window, which causes a momentary abnormal increase in the infrared reading in a certain area. If no pretreatment is performed, the system will mistakenly identify it as a sudden overheating of the glass surface and trigger strong suppression control. After adopting this embodiment, the abnormal points were eliminated, the average surface temperature maintained a smooth increase, and the surface temperature rise rate curve did not show a peak, so the subsequent control module could continue to adjust the power according to the actual thermal state. The purpose of this step is to provide a stable and reliable input for subsequent thermal state estimation and PID parameter variation, thereby achieving effective isolation between temperature measurement noise and control jitter.

[0019] In a preferred embodiment of the present invention, the feature calculation module includes: a state iteration unit, used to input physical size parameters and average surface temperature into a preset heat conduction state space model for iterative calculation, so as to output the core temperature state; The feature composite unit is used to weight the surface temperature rise rate based on a preset first engineering calibration coefficient and to weight the core temperature state based on a preset second engineering calibration coefficient. The difference between the weighted surface temperature rise rate and the weighted core temperature state is calculated to generate a thermal gradient decoupling feature quantity.

[0020] This embodiment provides a thermal state calculation mechanism based on thermal conduction iteration and feature composite; specifically, based on the above-mentioned average surface temperature and surface temperature rise rate, there is still a defect, that is, although the system can obtain the specific temperature rise rate of the glass surface, it cannot evaluate the relative lag of the heat transfer response inside the glass. For thick fireproof glass, judging the heating state solely by surface temperature and its rate of change may still lead to the misidentification of abnormal conditions such as surface overheating and core heat transfer lag as normal temperature rise. Therefore, this embodiment further introduces state iteration units and feature composite units to construct feature quantities that reflect the core-surface coupling relationship. Specifically, the state iteration unit can adopt a one-dimensional thickness-direction heat conduction discrete state space model, simplifying the glass thickness direction into several layers of nodes, such as surface nodes, intermediate layer nodes, and core nodes. In each sampling period, the system uses the current average surface temperature as the boundary input and combines it with the thickness parameter of the glass sheet to recursively update the temperature of each layer node. Specifically, the discrete state-space model of one-dimensional thickness heat conduction is based on the principle of energy conservation. It multiplies the temperature difference between the current layer node temperature and the adjacent layer node temperature, the thermal conductivity, and the time step as the temperature increment of the current layer. Taking heat transfer from the surface to the intermediate layer as an example, the new temperature of the intermediate layer is equal to the temperature of the intermediate layer in the previous cycle plus the temperature difference between the surface and the intermediate layer. Similarly, the update of the core temperature depends on the heat transfer from the intermediate layer to the core. In this way, even without directly embedding sensors inside the glass, the current core temperature can be obtained based on the laws of heat conduction; specific numerical examples are given below. Assume an 8mm glass sheet is simplified into three layers, corresponding to the surface layer, transition layer, and core; the temperatures of the three layers in the previous cycle were 640℃, 626℃, and 615℃, respectively. The average surface temperature obtained after preprocessing in the current cycle is 646℃; after iteration of the state space model according to the preset thermal conductivity and time step, the new three-layer temperatures are approximately 646℃, 632℃ and 620℃, of which 620℃ can be taken as the current core temperature state. The system pre-calibrates the ambient reference temperature inside the heating furnace; during feature composite, the specific rules are as follows: obtain the surface temperature rise rate and multiply it by the first engineering calibration coefficient as the temperature rise driving term, obtain the difference between the core temperature state and the ambient reference temperature and multiply it by the second engineering calibration coefficient as the core response term, and subtract the core response term from the temperature rise driving term to obtain the thermal gradient decoupling feature quantity that can quantify the degree of surface heat excess. At this point, if the current surface temperature rise rate is 60℃ / s, and the first engineering calibration coefficient is given as 0.2 and the second engineering calibration coefficient is given as 0.03 during engineering calibration, then the weighted surface temperature rise contribution is 12 and the weighted core temperature contribution is 18.6. Considering that this feature is used to characterize the degree of imbalance between the surface heating trend and the core following ability, the two can be subtracted or mapped to an equivalent difference in a preset direction; For example, after zero-point translation, the thermal gradient decoupling characteristic is 12. If the surface temperature rise rate drops to 35℃ / s and the core temperature rises to 628℃ in the next cycle, the corresponding characteristic will decrease significantly, reflecting that the thermal state of the core and surface tends to be coordinated. In practical implementation, the first engineering calibration factor can be used to amplify or compress the influence of the surface temperature rise rate on the judgment of thermal imbalance, while the second engineering calibration factor is used to correct the sensitivity of the core temperature state under different thickness specifications. For thicker fireproof glass, the core heat transfer lag is more pronounced. The second engineering calibration coefficient can be appropriately increased to make the characteristic quantity reflect the problem of insufficient core following earlier. For products with thickness close to the lower limit, this weight can be appropriately reduced to avoid excessive sensitivity to small temperature differences. In addition, in the anomaly handling mechanism, if the core temperature obtained by the state space model after a certain cycle is lower than the ambient reference temperature, or the positive jump amplitude compared to the previous cycle exceeds the error allowable range, such as exceeding the preset jump threshold, it is considered that the model input is inconsistent with the physical state. The result of this cycle will not participate in feature composition, but will be replaced by the previous effective core temperature state for smoothing. If the glass thickness parameter changes, for example, if the current batch changes from 6mm to 8mm, the system first loads the corresponding model parameter template before allowing the formal iteration calculation; before the change is completed, the feature quantity update is temporarily suspended to prevent model mismatch. When continuously producing 8mm fireproof glass on the same tempering production line, the operator kept the heating zone temperature set constant. However, due to the low initial temperature of the raw material in a certain batch of glass, the core heating was significantly delayed. The average surface temperature appeared to be rising as planned. If only the surface temperature is relied upon, the stable state of the process will be misidentified. In this embodiment, the state iteration unit continuously provides a low core temperature state, and the feature composite unit generates a large thermal gradient decoupling feature quantity accordingly, which promptly triggers subsequent control suppression to prevent the surface from continuing to heat up prematurely. The purpose of this step is to explicitly introduce the core thermal state, which cannot be directly measured, into the control link, thereby enabling online observation and quantitative expression of the thickness-direction thermal conduction hysteresis.

[0021] In a preferred embodiment of the present invention, the preset threshold includes a positive threshold, and the variable parameter control module includes: a first adjustment unit, used to obtain the system preset basic proportional coefficient and basic differential coefficient, and when the thermal gradient decoupling feature is greater than the positive threshold, use the difference between the thermal gradient decoupling feature and the positive threshold as the independent variable to perform exponential decay processing on the basic proportional coefficient to obtain the target proportional coefficient, and perform linear increase processing on the basic differential coefficient to obtain the target differential coefficient, and output the target proportional coefficient and the target differential coefficient as dynamic proportional integral differential parameters; The second adjustment unit is used to output the basic proportional coefficient and basic differential coefficient as dynamic proportional-integral-differential parameters when the thermal gradient decoupling characteristic quantity is less than or equal to the positive threshold, and to introduce an integral coefficient to eliminate steady-state error.

[0022] This embodiment provides a PID gain scheduling mechanism based on threshold segmentation; specifically, on the basis of being able to generate thermal gradient decoupling feature, another problem still needs to be solved, namely, how exactly this feature is applied to the controller parameters. If it is only used as an alarm indicator and not included in the PID parameter calculation, true active control cannot be formed. Therefore, this embodiment further stipulates that when the characteristic quantity crosses the positive threshold, the proportional coefficient and the derivative coefficient are dynamically adjusted in different directions; when the characteristic quantity falls back to within the threshold, the normal PID working mode is restored and integral action is introduced. Specifically, the system pre-stores a set of basic proportional coefficients and basic differential coefficients; for example, the basic proportional coefficient for a certain temperature zone can be set to 2.0, the basic differential coefficient can be set to 0.4, and the positive threshold can be set to 10. When the thermal gradient decoupling characteristic value is no greater than 10, it indicates that although there is a certain difference between the core and the surface of the glass, it is still within an acceptable range. At this time, the controller still mainly uses the basic proportional coefficient and the basic differential coefficient, and introduces an integral coefficient, such as 0.1, to gradually eliminate the steady-state error. When the thermal gradient decoupling characteristic value is greater than 10, it indicates that the surface temperature rises too quickly and the core does not follow up enough, and the system enters the first adjustment path. A simple deduction can be made; assuming the current feature quantity is 16, then the difference between it and the positive threshold is 6; specifically, the structured logic of the exponential decay processing is as follows: preset a decay base less than 1, such as 0.9, use the difference between the thermal gradient decoupled feature quantity and the positive threshold as the exponent, multiply the basic proportional coefficient by the corresponding exponent of the decay base, so as to achieve the weakening effect that the larger the error, the faster the proportional effect decays; The logic of linear increase is as follows: a fixed linear gain rate is preset, such as 0.05. The difference is multiplied by the linear gain rate to obtain the differential increment, and the differential increment is added to the basic differential coefficient. The first adjustment unit exponentially decreases the base proportional coefficient based on the above rules, while linearly increasing the base differential coefficient; for example, combining the aforementioned assumptions, namely, a decay base of 0.9, a linear gain ratio of 0.05, and a difference of 6, the target proportional coefficient is calculated as follows: ; The target differential coefficient is equal to the basic differential coefficient 0.4 plus the product of 0.05 and 6, which is 0.7. In this way, the proportional action is weakened, which prevents the controller from continuing to heat up significantly because the surface temperature is lower than the set value; the differential action is enhanced, which is used to suppress the tendency of rapid temperature rise in advance. If the characteristic value increases to 20 in the next cycle, the difference becomes 10, the target proportional coefficient will decrease further, and the target differential coefficient will increase further, thus forming a stronger temperature overshoot suppression effect. Conversely, when the characteristic value of the subsequent heat dissipation stage drops to 8, the system will no longer continue to execute exponential decay and linear gain, but will restore the basic proportional coefficient of 2.0 and the basic differential coefficient of 0.4, and re-enable the integral coefficient of 0.1; This is because during the homogenization stage, the main task has shifted from preventing temperature overshoot to eliminating steady-state error. At this time, an appropriate amount of integral action is beneficial to make the overall temperature of the glass closer to the target process window. Furthermore, to avoid abrupt parameter switching, a buffer zone can be set between the first and second adjustment units; for example, when the characteristic value fluctuates between 9.5 and 10.5, a hard switch is not performed directly, but an interpolation transition is used to make the PID parameters change smoothly within the continuous cycle; this can avoid power jitter caused by crossing back and forth near the threshold. As a protective mechanism, if the target proportional coefficient after exponential decay is lower than a preset lower limit, such as lower than 0.5, it will be clamped to 0.5 to prevent the proportional effect from being too weak and causing the system to adjust lag. If the target differential coefficient after linear increase is higher than the preset upper limit, for example, higher than 1.2, it will be forcibly limited to within 1.2 to prevent excessive amplification of measurement noise; if the integral term accumulates too much in a long-term low-error state, integral separation or integral freezing can be performed simultaneously when switching to the first adjustment path to avoid obvious backlash after returning to the homogenization stage. When heating 8mm fireproof glass on the same production line, when the glass enters the critical temperature range close to the softening zone, the thermal gradient decoupling characteristic value rises rapidly from 7 to 15. The system immediately reduces the proportional coefficient and increases the differential coefficient, resulting in the solid-state relay output duty cycle being reduced and the furnace wire power peak being reduced. Once the internal temperature of the glass gradually catches up and the characteristic value drops back below 9, the controller restores the basic parameters and restarts the integral action, making the temperature more stable before exiting the furnace. The purpose of this mechanism is to directly map the thermal state judgment result into executable PID parameter adjustment rules, thereby achieving the control effect of stronger suppression when the thermal gradient is more dangerous and more sufficient steady-state correction when the thermal state is more balanced.

[0023] In a preferred embodiment of the present invention, the system further includes: a quality detection module for acquiring the surface compressive stress value of the tempered fireproof glass; and an adaptive correction module for extracting the historical thermal gradient curve of the fireproof glass during the heating stage when the absolute value of the difference between a consecutive preset number of surface compressive stress values ​​and the target median value is greater than a preset tolerance threshold, and fine-tuning the thermal conductivity in the thermal conduction state space model based on the historical thermal gradient curve; and maintaining the current thermal conductivity of the thermal conduction state space model unchanged when the absolute value of the difference is less than or equal to the preset tolerance threshold.

[0024] This embodiment provides a cross-process closed-loop correction mechanism; specifically, the aforementioned solution can complete online control during the heating stage of a single piece of glass, but in long-term operation, it will still encounter situations such as furnace aging, seasonal temperature differences, and fluctuations in the original glass formula. At this point, even if the control logic of the heating stage is stable, the thermal conductivity in the preset heat conduction state space model may still gradually deviate from the actual material state; if no a posteriori correction is made, the core temperature estimation will become increasingly distorted. Therefore, this embodiment introduces a quality inspection module and an adaptive correction module in the cold end inspection stage, so that the stress results of the finished product can be used to reversely correct the heating model; Specifically, after the glass has been tempered and cooled to a detectable state, the quality inspection module acquires its surface compressive stress value; the system establishes a correspondence between the heating record and the finished product stress record for each piece of glass; when the absolute value of the difference between the surface compressive stress value and the target median value of a consecutive preset number of finished glass pieces, such as 3 consecutive pieces, is greater than the preset tolerance threshold, the adaptive correction module extracts the historical thermal gradient curves corresponding to these 3 pieces of glass during the heating stage, and fine-tunes the thermal conductivity in the thermal conduction state space model accordingly; Microscopic extrapolation can be performed; assuming the target median is 95 MPa and the tolerance threshold is 5 MPa; the measured surface compressive stresses of three consecutive glass pieces are 86 MPa, 87 MPa and 85 MPa, respectively. The absolute values ​​of their differences from the target median are 9, 8 and 10, respectively, all exceeding the tolerance threshold. This indicates that this deviation is not random noise, but may reflect a systematic deviation in the estimation of the core surface thermal state during the heating stage. Further examination of the historical thermal gradient curves of these three glass pieces reveals that they are generally higher than the normal reference curve for a long period during the middle and later stages of heating. This indicates that the model may have overestimated the core heating rate, and the actual thermal conductivity is weaker than the current model setting. In this case, the thermal conductivity can be slightly adjusted downwards, for example, from 1.00 to 0.97. Specifically, when fine-tuning the thermal conductivity, the adaptive correction module follows the rule based on integral deviation ratio mapping; it extracts the actual historical thermal gradient curves of the three target workpieces during the heating stage and calculates the time integral value of the deviation of the characteristic quantity from the normal reference curve in each sampling period; the average of this integral value is taken as the overall deviation. Set a base adjustment factor, such as 0.001, and multiply the overall deviation by this base adjustment factor to obtain the correction step size; if the overall deviation is positive, it indicates that there has been an overestimation in the long term, and the thermal conductivity is subtracted from the correction step size; If the overall deviation is negative, it indicates an underestimation, so the thermal conductivity is increased by the correction step size. Through this two-layer adaptive logic that uses the continuous compressive stress deviation as a trigger and the integral value of the thermal gradient curve as a quantitative correction dimension, the aforementioned adjustment from 1.00 to 0.97 has a definite mathematical and physical basis. After fine-tuning, the next batch of glass will be estimated to have a lower core temperature under the same surface temperature input, thus enabling subsequent control to suppress the surface from heating up too quickly earlier. Conversely, if the surface compressive stress values ​​of three consecutive glass panes are 103MPa, 104MPa and 102MPa respectively, all of which are higher than the target median and exceed the tolerance, and the historical thermal gradient curve shows that the characteristic values ​​are generally low, it may indicate that the model underestimates the internal heat transfer capacity and the system is too conservative. In this case, the thermal conductivity can be slightly increased, for example, from 1.00 to 1.02. Furthermore, adaptive correction does not involve drastically rewriting the model, but rather iterates in small steps; each correction can be limited to a preset fine-tuning ratio, for example, between ±0.01 and ±0.03. After correction, the system can set an observation window, for example, to continuously observe the stress results of the next 5 or 10 pieces of glass to confirm whether the model converges to the target; this can avoid excessive model oscillation caused by abnormalities in individual batches of raw materials. As an anomaly handling mechanism, if the stress deviation directions of three consecutive glass pieces are inconsistent, such as one being too high, one being too low, and one being normal, the model correction will not be triggered, because this is more likely to be random process fluctuations rather than model systematic errors. If the cold end stress testing equipment itself malfunctions during calibration, the test results for this cycle will not be used for correction. If historical thermal gradient curves are missing or cannot be matched one-to-one with finished product records, the existing model parameters will be retained, and correction will be performed after subsequent complete data is available. After the fireproof glass production line had been running continuously for several weeks, the surface compressive stress of the finished product was continuously low due to changes in the radiation environment inside the furnace and the aging of the furnace wires. After the system detected that the stress values ​​of three consecutive pieces were lower than the target median and exceeded the tolerance, it automatically traced the heating trajectory of these three pieces of glass and found that their historical thermal gradient curves were generally too high. Therefore, it lowered the thermal conductivity. After running for a subsequent shift, the stress of the finished product gradually returned to the target window, without the need for manual readjustment of the entire controller. The purpose of this mechanism is to use the finished product quality results to correct the front-end thermal state model in reverse, thereby achieving long-term self-correction capability across processes.

[0025] In a preferred embodiment of the present invention, the data acquisition module includes: an infrared scanning unit, used to acquire a two-dimensional temperature matrix of the fireproof glass surface at a preset acquisition frequency as surface temperature distribution data; and a manufacturing execution acquisition unit, used to read the thickness parameters of the current batch of fireproof glass from the manufacturing execution system as physical dimension parameters.

[0026] This embodiment provides a front-end data acquisition mechanism for tempering production lines. Specifically, in the aforementioned control logic, if the input relies solely on the temperature at a single point in the furnace or the thickness manually entered by the operator, there are two obvious drawbacks: firstly, the temperature at a single point cannot reflect the actual temperature distribution on the glass surface. Secondly, manual entry of thickness is prone to batch errors and specification errors, leading to incorrect calls to the heat conduction model. Therefore, this embodiment explicitly adopts an infrared scanning unit and a manufacturing execution acquisition unit to form an automatic, synchronous, and traceable data entry point. Specifically, the infrared scanning unit is installed at the observation position on the top of the heating furnace and scans the glass passing through the roller conveyor below at a preset acquisition frequency, outputting a surface temperature matrix; the so-called two-dimensional temperature matrix can be understood as a temperature grid arranged along the length and width of the glass. For example, in a simplified scenario, a single sampling can yield a 4×3 temperature matrix, where each element corresponds to the instantaneous temperature value of a region on the glass surface; the manufacturing execution acquisition unit then reads the thickness parameters of the corresponding glass from the manufacturing execution system based on the current batch work order, barcode, or equipment cycle time identifier, and binds them to the same glass record along the time axis with the temperature matrix. A simple deduction can be made; assuming the infrared scan results of a certain piece of glass are: 638, 640, 639; 641, 643, 642; 644, 645, 643; 646, 647, 645; This matrix indicates that the glass exhibits a gradual heating trend along the transport direction; simultaneously, the manufacturing execution system returns the thickness parameter of the glass sheet as 8 mm. The subsequent preprocessing module and feature calculation module can then work together based on this temperature matrix and the thickness parameter; if the next piece of glass is a 6mm product, the manufacturing execution acquisition unit will synchronously return 6mm, so that the state space model will automatically switch to the corresponding thickness template without the need for manual file modification by the operator; Furthermore, the acquisition frequency can be matched according to the furnace speed and control cycle; for example, when the roller conveyor linear speed is high, increasing the infrared scanning frequency helps to track temperature changes more closely; when the furnace speed is low, a stable frequency can be maintained appropriately to reduce the computational burden. For the manufacturing execution acquisition unit, the thickness parameter can be not only a single value, but can also include the batch number, product type or specification group number, for subsequent control modules to perform process mapping. As an anomaly handling mechanism, if the infrared scanning unit fails to form a complete temperature matrix in a certain cycle, the missing area can be marked as null and handed over to the preprocessing module for point filling or downgrading; if the manufacturing execution system communication is interrupted, the system can read the local task parameters that have been cached before the glass is mounted. If the local cache is also missing, dynamic thermal gradient control is not allowed. Instead, the basic process mode is maintained until the correct specifications are restored. If the infrared scan detects that the glass has not fully entered the observation area, such as only scanning the glass leading edge, the current matrix is ​​only used for preheating tracking and not for complete thermal state calculation. On the same 2440×3660mm fireproof glass tempering line, a shift continuously produces two types of products: 8mm and 6mm. The infrared scanning unit continuously outputs a two-dimensional temperature matrix of each piece of glass at different positions in the furnace, while the manufacturing execution acquisition unit synchronously delivers the corresponding thickness according to the batch sequence. Because of the automatic matching at the input end, the system can call different heat conduction models in a timely manner when switching products, avoiding control deviations caused by estimating 6mm glass as an 8mm model; The purpose of this step is to ensure that the temperature and thickness information required for thermal state calculation are accurate and synchronized, thereby achieving automated control input and batch consistency.

[0027] In a preferred embodiment of the present invention, the system further includes: a mode switching module, used to disable the variable parameter control module and switch the control mode to a fixed parameter control mode when the physical size parameter is less than or equal to a preset thickness threshold, so as to control the output power of the heating actuator through fixed proportional integral differential parameters; and to keep the variable parameter control module in an active state when the physical size parameter is greater than the preset thickness threshold. The preset thickness threshold is 3mm, and the heating actuator includes a solid-state relay and heating wire.

[0028] This embodiment provides a control mode switching mechanism under applicable boundaries; specifically, the aforementioned thermal gradient decoupling and dynamic parameter variation logic mainly rely on the premise that there is an observable hysteresis in the thickness-direction thermal conduction of the glass; this premise holds true in fireproof glass of conventional thickness; However, when the glass thickness is too thin, the core-surface temperature difference is very small, the thermal inertia difference is significantly compressed, and the calculated thermal gradient decoupling characteristic quantity is easy to approach the measurement noise level. If variable parameter control is still forcibly used at this time, unnecessary power jitter may be caused by noise driving. Therefore, this embodiment sets a mode switching module in the system, which automatically degenerates to fixed parameter control mode when a specific thickness condition is met. Specifically, before each piece of glass enters the heating furnace, the mode switching module determines whether the threshold condition is met based on the read thickness parameters. When the thickness parameter is less than or equal to the preset thickness threshold, the system disables the variable parameter control module and no longer adjusts the PID parameters online based on the thermal gradient decoupling characteristic. Instead, it directly uses pre-calibrated fixed proportional, integral, and derivative parameters to control the output power of the heating actuator. In this embodiment, the target workpiece is fireproof glass with a preset thickness threshold of 3mm, and the heating actuator includes a solid-state relay and a heating element. In other words, for ultra-thin fireproof glass of 3mm and below, the system uses fixed PID parameters and pulse width modulation signal (PWM signal) to drive solid-state relay, which in turn controls the duty cycle of the heating element. A simple deduction can be made; assuming that the thickness of a certain piece of glass is 2.8mm, after the mode switching module receives this thickness, it directly sets the fixed parameter mode flag; At this point, even if the preprocessing module can still output the average surface temperature and surface temperature rise rate, the feature calculation module can continue to record in the background, but these data will no longer be included in the online gain scheduling. The controller calls fixed parameters, such as a proportional coefficient of 1.6, an integral coefficient of 0.08, and a derivative coefficient of 0.3, to perform conventional closed-loop temperature control for the current temperature zone; conversely, if the thickness of the next glass piece is 4mm, the marker is removed and dynamic variable parameter control driven by thermal gradient is re-enabled. Furthermore, mode switching can be linked with batch management; when the entire batch of glass consists of products with a thickness of 3mm or less, the system can switch the control mode once before the entire batch begins. When there is mixed production on the production line, the judgment and switching are performed independently at the unit level; in order to prevent the controller state from being discontinuous due to frequent mode switching, the switching module can reset the integral accumulation value or perform soft transition processing when the mode changes. As an exception handling mechanism, if the thickness parameter is missing, the system will not default to dynamic mode, but will instead prioritize the more secure fixed parameter control mode and prompt the user to switch back after the thickness data is recovered. If the thickness parameter is exactly 3mm, then the fixed parameter control is directly adopted according to the rule of less than or equal to the threshold; if the equipment is in the moment of mode switching and the previous piece of glass has not been completely discharged from the furnace, then the control status of each piece is maintained by the piece number isolation method to avoid the mode of the later piece affecting the tail section heating of the previous piece. On the same tempered glass production line, when producing 8mm high borosilicate fireproof glass in the morning, the system adopts a dynamic PID strategy driven by thermal gradient throughout the entire process. When switching to 3mm ultra-thin fireproof glass in the afternoon, the mode switching module automatically shields the variable parameter control after reading the thickness parameter, and switches to control the solid-state relay and heating wire with fixed proportional integral and derivative parameters. Since the core-surface temperature difference of ultra-thin glass is already very small, this degradation mode avoids noise interference with parameter scheduling and maintains the stability of the heating process. The purpose of this mechanism is to clarify the applicable boundaries of the system and provide a stable degradation path outside the boundaries, thereby achieving reliable switching of control strategies and anomaly fallback for products of different thicknesses.

[0029] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A fireproof tempered glass heating control system based on PID regulation, characterized in that, The system includes: The data acquisition module is used to acquire surface temperature distribution data and physical dimensional parameters of fireproof glass; The preprocessing module is used to filter the surface temperature distribution data to extract the average surface temperature, and calculate the surface temperature rise rate based on the average surface temperature. The feature calculation module is used to determine the core temperature state based on the physical size parameters, the average surface temperature and the preset thermal conduction state space model for characterizing the thickness thermal conduction characteristics, generate thermal gradient decoupling feature quantities in combination with the surface temperature rise rate, and store the thermal gradient decoupling feature quantities as historical thermal gradient curves in time series. The variable parameter control module is used to compare the thermal gradient decoupling feature with a preset threshold and generate dynamic proportional-integral-differential parameters based on the comparison result. The execution output module is used to calculate the target output value based on the dynamic proportional-integral-derivative parameters, and convert the target output value into a pulse width modulation signal to control the output power of the heating actuator that is communicatively connected to the system.

2. The fireproof tempered glass heating control system based on PID regulation according to claim 1, characterized in that, The preprocessing module includes: The filtering unit is used to perform sliding window mean filtering on the surface temperature distribution data and remove outlier data points to extract the average surface temperature. The differential unit is used to perform differential calculation between the average surface temperature of the current period and the average surface temperature of the previous sampling period according to a preset sampling period, so as to obtain the surface temperature rise rate.

3. The fireproof tempered glass heating control system based on PID regulation according to claim 1, characterized in that, The feature calculation module includes: The state iteration unit is used to input the physical size parameters and the average surface temperature into the preset heat conduction state space model for iterative calculation, so as to output the core temperature state. The feature composite unit is used to weight the surface temperature rise rate based on a preset first engineering calibration coefficient, and to weight the core temperature state based on a preset second engineering calibration coefficient, and to calculate the difference between the weighted surface temperature rise rate and the weighted core temperature state to generate the thermal gradient decoupling feature quantity.

4. The fireproof tempered glass heating control system based on PID regulation according to claim 1, characterized in that, The preset threshold includes a positive threshold, and the variable parameter control module includes: The first adjustment unit is used to obtain the system's preset basic proportional coefficient and basic differential coefficient. When the thermal gradient decoupling feature is greater than the positive threshold, the basic proportional coefficient is exponentially decayed to obtain the target proportional coefficient, and the basic differential coefficient is linearly increased to obtain the target differential coefficient. The target proportional coefficient and the target differential coefficient are output as the dynamic proportional-integral-differential parameter. The second adjustment unit is used to output the basic proportional coefficient and the basic differential coefficient as the dynamic proportional-integral-differential parameter when the thermal gradient decoupling feature is less than or equal to the positive threshold, and to introduce an integral coefficient to eliminate steady-state error.

5. The fireproof tempered glass heating control system based on PID regulation according to claim 1, characterized in that, The system also includes: The quality inspection module is used to obtain the surface compressive stress value of the tempered fireproof glass. An adaptive correction module is used to extract the historical thermal gradient curve of the fireproof glass during the heating stage when the absolute value of the difference between the surface compressive stress value and the target median value for a consecutive preset number of times is greater than a preset tolerance threshold, and to fine-tune the thermal conductivity in the thermal conduction state space model based on the historical thermal gradient curve; when the absolute value of the difference is less than or equal to the preset tolerance threshold, the current thermal conductivity of the thermal conduction state space model is kept unchanged.

6. The fireproof tempered glass heating control system based on PID regulation according to claim 1, characterized in that, The data acquisition module includes: An infrared scanning unit is used to acquire a two-dimensional temperature matrix of the fireproof glass surface at a preset acquisition frequency as the surface temperature distribution data. The manufacturing execution acquisition unit is used to read the thickness parameters of the current batch of fireproof glass from the manufacturing execution system as the physical dimension parameters.

7. The fireproof tempered glass heating control system based on PID regulation according to claim 1, characterized in that, The system further includes a mode switching module, used to disable the variable parameter control module and switch the control mode to a fixed parameter control mode when the physical size parameter is less than or equal to a preset thickness threshold, so as to control the output power of the heating actuator through fixed proportional integral derivative parameters; and to keep the variable parameter control module in an active state when the physical size parameter is greater than the preset thickness threshold.

8. The fireproof tempered glass heating control system based on PID regulation according to claim 7, characterized in that, The preset thickness threshold is 3mm, and the heating actuator includes a solid-state relay and a heating element.

Images