Glass insulator intelligent production line control method based on digital twinning

Abstract

The application discloses a glass insulator intelligent production line control method based on digital twinning, relates to the technical field of intelligent manufacturing and industrial automation, and comprises the following steps: acquiring real-time temperature field data of the surface of a glass insulator blank to be subjected to toughening treatment placed in a toughening treatment station; and driving a preset thermal-mechanical coupling physical model based on the real-time temperature field data. The application acquires real-time temperature field data of the surface of the glass insulator blank through an infrared thermal imager, drives the thermal-mechanical coupling physical model to solve a virtual measured stress field, compares the stress field with an ideal target stress field of a corresponding type in an ideal stress model database to obtain a stress deviation, generates an optimized adjustment instruction through a model predictive control (MPC) algorithm model, controls an array of tuyeres of a toughening air grid device to perform partitioned differential cooling, forms a closed loop control circuit, makes the residual stress distribution in the glass insulator uniform, effectively eliminates the hidden danger of “zero value self-explosion”, and improves the product qualification rate.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing and industrial automation technology, specifically to a control method for an intelligent production line of glass insulators based on digital twins. Background Technology

[0002] In the glass insulator manufacturing industry, with the advancement of "Industry 4.0," digital twin technology is widely used. By constructing virtual models, it enables functions such as production guidance, real-time monitoring, and predictive maintenance, significantly improving production efficiency and product quality. Currently, some factories are leveraging the Internet of Things (IoT) for real-time management of production equipment, using a combination of digital twins and deep learning to build models and solve the problem of integrating physical and virtual information. As a key component in power systems, the core performance of glass insulators, such as mechanical strength and thermal stability, largely depends on the tempering process. Traditional tempering processes use fixed air pressure and time for production, which is essentially an open-loop control method.

[0003] However, in actual production, the initial temperature of the glass droplets fluctuates slightly, and the ambient temperature and humidity also change constantly. Open-loop control cannot effectively cope with these interfering factors. As a result, the residual stress inside the produced glass insulators is uneven, leading to a low yield. More seriously, this uneven internal stress can also cause the difficult-to-detect "zero-value spontaneous explosion" hazard, posing a significant threat to the safe and stable operation of the power system. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a digital twin-based intelligent production line control method for glass insulators. This method solves the problem in existing technologies where uneven distribution of internal residual stress in glass insulators during the tempering process, caused by initial temperature fluctuations and changes in ambient temperature and humidity, leads to low yield and the potential for "zero-value spontaneous explosion."

[0005] To achieve the above objectives, the present invention provides the following technical solution: a control method for an intelligent production line of glass insulators based on digital twins, the method comprising: S1. Obtain real-time temperature field data of the surface of a glass insulator blank to be tempered, which is placed in the tempering station. S2. Based on the real-time temperature field data, drive the preset thermo-mechanical coupling physical model to calculate the stress distribution inside the glass insulator blank in real time, and thereby obtain the virtual measured stress field. S3. Retrieve the ideal target stress field corresponding to the model and specifications of the glass insulator blank from the preset ideal stress model database; S4. Compare the virtual measured stress field with the ideal target stress field across the entire domain or at key points to calculate the stress deviation characterizing the difference between the two. S5. Based on the stress deviation, generate optimized adjustment instructions for the tempered air grid device under zone control through a preset control decision model; S6. The optimization and adjustment command is sent to the execution controller through the data communication interface, and the execution controller drives the multiple independently adjustable air nozzle arrays of the tempered air grid device to perform zoned differential cooling on the glass insulator blank. The method involves repeatedly executing steps such as acquiring real-time temperature field data, solving the virtual measured stress field, calculating stress deviation, and generating and executing optimization adjustment instructions to form a closed-loop control loop until the tempering process of the glass insulator blank is completed.

[0006] Further, real-time temperature field data of the surface of the glass insulator blank is obtained, including: At least one infrared thermal imager is used to perform continuous non-contact scanning of all or key surfaces of the glass insulator blank; The infrared thermal imager will send the raw thermal imaging data, which includes multiple pixels and their corresponding temperature values, to the data acquisition and synchronization module. The data acquisition and synchronization module preprocesses the raw thermal imaging data, including noise filtering and coordinate registration, to generate structured real-time temperature field data that can be directly used by the thermo-mechanical coupling physical model.

[0007] Furthermore, the thermo-mechanical coupling physical model is a high-precision finite element model; The thermo-mechanical coupling physical model is driven by the real-time temperature field data, including: The real-time temperature field data is input into the finite element model as a thermal load boundary condition. Transient heat conduction analysis was performed using the finite element model to calculate the temperature gradient distribution inside the glass insulator blank. Transient heat conduction equation: ; in, Density of glass, unit: ), where c is the unit of specific heat capacity: kg·℃, where k is the thermal conductivity, unit: m·℃, The internal heat source is represented by a value of 0 here. T represents temperature in °C, and t represents time. .

[0008] Based on the temperature gradient distribution, thermal stress analysis and calculation are performed to calculate the thermal stress and residual stress distribution at any position inside the glass insulator blank, and this distribution is used as the virtual measured stress field.

[0009] thermal stress constitutive equation: , ;in, Stress tensor, unit: MPa For the elasticity matrix, For the total strain tensor, For thermal strain tensor, Coefficient of thermal expansion, unit: 1 / ℃ This refers to the change in temperature, in degrees Celsius (°C). It is an identity matrix.

[0010] Furthermore, the method for constructing the ideal stress model database includes: For glass insulators of different models and specifications, multiple physical tempering tests were conducted in advance; In the physical tempering experiment, multiple groups of samples with optimal mechanical and electrical properties were obtained by optimizing process parameters. The sample was subjected to physical testing to obtain data on the actual optimal residual stress distribution inside it. The actual optimal residual stress distribution data is digitized and stored together with the corresponding model and specification information in the ideal stress model database to form the ideal target stress field.

[0011] Furthermore, the virtual measured stress field is compared with the ideal target stress field to calculate the stress deviation, including: In the virtual measured stress field and the ideal target stress field, a set of preset stress monitoring points corresponding to spatial locations are determined; Extract the stress tensor or principal stress value of each stress monitoring point in the virtual measured stress field; Extract the target stress tensor or target principal stress value of each stress monitoring point in the ideal target stress field; The difference or vector difference between the virtual measured stress value and the target stress value is calculated to form a stress deviation matrix or deviation vector containing deviation information of each monitoring point.

[0012] Furthermore, the control decision model is a Model Predictive Control (MPC) algorithm model; Based on the stress deviation, an optimization adjustment instruction is generated, including: The stress deviation matrix or deviation vector is provided as input to the Model Predictive Control (MPC) algorithm model. The Model Predictive Control (MPC) algorithm model performs rolling optimization calculations within a preset control time domain based on the input and the prediction of future states. Model Predictive Control (MPC) objective function formula: ; in, The stress deviation predicted at time t after k steps, in MPa. For prediction in the time domain, values ​​range from 5 to 10. The control adjustment value at time t, in kPa or ), To control the time domain Q and R are weight matrices for stress deviation and control quantity, respectively, and are diagonal matrices with positive elements. The objective of the rolling optimization calculation is to minimize the sum of predicted stress deviations over a future period of time. Output a control sequence for each nozzle array of the tempered air grating device that can achieve the stated objective, and use the first element of the control sequence as the optimization adjustment instruction for the current cycle.

[0013] Furthermore, the control decision model is a fuzzy control algorithm model; Based on the stress deviation, an optimization adjustment instruction is generated, including: The stress deviation is fuzzified and converted into one or more fuzzy linguistic variables. A set of preset fuzzy control rules are applied, the form of which is "if the stress deviation is positive, then the air volume is adjusted to negative". The fuzzy linguistic variables and the fuzzy control rules are processed by the fuzzy inference engine to obtain the fuzzy control output; Fuzzy control membership function formula: Large burden Membership function: ; Zhengda Membership function: ; Where e is the stress deviation, and E is the quantification range of the stress deviation, in MPa, which is set according to the insulator model, such as 10MPa; The fuzzy control output is defuzzified to obtain deterministic optimization adjustment instructions for each nozzle array of the tempered air grating device. Fuzzy control defuzzification, centroid method formula: ; in, Let i be the output membership degree of the i-th fuzzy rule. Let m be the output of the i-th rule, and m be the number of rules, ranging from 9 to 25.

[0014] Furthermore, the optimization adjustment command specifically includes command parameters for adjusting at least one of the air pressure, air volume, or air nozzle angle of a specific air nozzle array of the tempered air grid device; The execution controller is a programmable logic controller (PLC) or an industrial personal computer. After receiving the optimization and adjustment command, the execution controller converts it into a drive signal and sends it to the solenoid valve or variable frequency fan driver corresponding to the specific nozzle array.

[0015] Furthermore, the method also includes: Before the control method begins, input the model and specification information of the glass insulator blank to be processed; The method automatically loads the corresponding ideal target stress field from the ideal stress model database based on the model specification information, and selects the matching thermo-mechanical coupling physical model from the model library.

[0016] Furthermore, the data communication interface adopts the high-speed industrial Ethernet protocol; The single-cycle time of the closed-loop control loop is less than a preset time threshold to ensure effective tracking and real-time response to the rapid dynamic changes in the stress field during the tempering process.

[0017] Beneficial effects This invention acquires real-time temperature field data of the glass insulator blank surface using an infrared thermal imager, drives a thermo-mechanical coupling physical model to solve the virtual measured stress field, compares it with the ideal target stress field of the corresponding model in the ideal stress model database to obtain the stress deviation, and then generates optimization and adjustment instructions through the model predictive control MPC algorithm model to control the nozzle array of the tempering air grid device to perform zoned differentiated cooling, forming a closed-loop control loop. This process can respond in real time to the initial temperature fluctuation of the glass droplets and changes in ambient temperature and humidity, improve the control accuracy of the tempering process, make the residual stress distribution inside the glass insulator uniform, effectively eliminate the hidden danger of "zero-value spontaneous explosion", improve the product qualification rate, and meet the higher requirements of modern intelligent manufacturing for product quality and production efficiency. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0019] 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.

[0020] Please see Figure 1 This invention proposes a control method for an intelligent production line of glass insulators based on digital twins, the method comprising: S1. Obtain real-time temperature field data of the surface of a glass insulator blank to be tempered, which is placed in the tempering station. S2. Based on the real-time temperature field data, drive the preset thermo-mechanical coupling physical model to calculate the stress distribution inside the glass insulator blank in real time, and thereby obtain the virtual measured stress field. S3. Retrieve the ideal target stress field corresponding to the model and specifications of the glass insulator blank from the preset ideal stress model database; S4. Compare the virtual measured stress field with the ideal target stress field across the entire domain or at key points to calculate the stress deviation characterizing the difference between the two. S5. Based on the stress deviation, generate optimized adjustment instructions for the tempered air grid device under zone control through a preset control decision model; S6. The optimization and adjustment command is sent to the execution controller through the data communication interface, and the execution controller drives the multiple independently adjustable air nozzle arrays of the tempered air grid device to perform zoned differential cooling on the glass insulator blank. The method involves repeatedly executing steps such as acquiring real-time temperature field data, solving the virtual measured stress field, calculating stress deviation, and generating and executing optimization adjustment instructions to form a closed-loop control loop until the tempering process of the glass insulator blank is completed.

[0021] Digital twins refer to the technology of constructing a virtual model that maps the operating state of a physical entity through real-time data synchronization. Specifically, it can be achieved by combining thermal imaging scanning, finite element simulation and Internet of Things communication technology. It is used to simulate the changes in internal stress of glass during the tempering process in real time, providing a basis for dynamically adjusting process parameters.

[0022] The thermo-mechanical coupling physical model refers to the mathematical equations describing the interaction between the temperature field and the stress field. Specifically, it can be implemented by constructing a finite element model using the transient heat conduction equation and the thermal stress constitutive equation. This model is used to invert the internal stress distribution of the glass based on surface temperature data and identify the deviation between the actual stress and the ideal state. The virtual measured stress field refers to the dynamic stress distribution data inside the glass obtained through simulation calculations. Specifically, it can be implemented by solving the thermo-mechanical coupling equation in real time using finite element analysis software. This model replaces the traditional method of directly measuring internal stress using physical sensors, overcoming the technical limitation of not being able to implant sensors during the high-temperature tempering process.

[0023] Among them, the ideal target stress field refers to the optimal residual stress distribution data that has been pre-calibrated through experiments. Specifically, it can be achieved by obtaining the actual optimal data through multiple physical tempering experiments combined with destructive testing, and then storing it digitally to provide a benchmark reference value for stress control.

[0024] Among them, the closed-loop control loop refers to the dynamic adjustment mechanism formed by real-time data acquisition, simulation calculation and execution feedback. Specifically, it can be realized by a cyclic process of periodic temperature scanning, stress deviation calculation and nozzle parameter adjustment, which is used to continuously eliminate the impact of process interference factors on tempering quality.

[0025] The core innovation of this invention lies in constructing a thermo-mechanical coupling model using digital twin technology to calculate the internal stress of the glass in real time. Combined with closed-loop control to dynamically adjust the cooling parameters of each zone, the traditional open-loop control is upgraded to real-time regulation based on virtual stress feedback. This solves the technical defect of uneven tempering stress caused by environmental fluctuations, thereby improving the finished product qualification rate and eliminating the risk of zero-value spontaneous explosion.

[0026] The working process and principle of this invention are as follows: A digital twin-based intelligent production line control method for glass insulators achieves real-time optimization of the tempering process by constructing a closed-loop control circuit. First, an infrared thermal imager acquires real-time temperature field data of the surface of the glass insulator blank to be tempered. Then, a preset thermo-mechanical coupling physical model calculates the stress distribution inside the glass insulator blank based on this temperature field data, forming a virtual measured stress field. Next, an ideal target stress field corresponding to the model and specifications of the glass insulator blank is retrieved from an ideal stress model database. By comparing the virtual measured stress field with the ideal target stress field, the stress deviation characterizing the difference between the two is calculated. Based on this stress deviation, a preset control decision model generates optimization adjustment instructions for the tempering air vent device under zoned control. This instruction is sent to the execution controller via a data communication interface, driving multiple independently adjustable air nozzle arrays of the tempering air vent device to perform zoned differentiated cooling of the glass insulator blank. The entire process is executed cyclically until the tempering process ends, thereby achieving dynamic optimization control of the tempering process.

[0027] As a preferred embodiment, the present invention is implemented as follows: Two high-resolution infrared thermal imagers are installed in the tempering station of the glass insulator production line to perform omnidirectional scanning of the glass insulator blanks to be tempered. The temperature field data collected by the infrared thermal imagers is preprocessed by a data acquisition and synchronization module to generate structured real-time temperature field data. A pre-established thermo-mechanical coupling physical model is a high-precision finite element model. This model receives the real-time temperature field data as thermal load boundary conditions and calculates the stress distribution inside the glass insulator blank through transient heat conduction analysis and thermal stress analysis, forming a virtual measured stress field. An ideal stress model database stores ideal target stress fields for different models and specifications of glass insulators. The control system retrieves the ideal target stress field corresponding to the model and specification of the currently processed glass insulator blank from the database. The virtual measured stress field and the ideal target stress field are compared at multiple preset stress monitoring points to calculate the stress deviation matrix. The control decision model employs a Model Predictive Control (MPC) algorithm, using the stress deviation matrix as input. Within a preset control time domain, it performs rolling optimization calculations to generate optimized adjustment instructions for each nozzle array of the tempered glass insulator. These instructions are transmitted to a Programmable Logic Controller (PLC) via a high-speed industrial Ethernet protocol. The PLC converts the instructions into drive signals and sends them to the solenoid valves or variable frequency fan drivers of the tempered glass insulator, enabling precise control of the air pressure, airflow, or nozzle angle of each independent nozzle array. This allows for differentiated cooling of the glass insulator blank in different zones. The entire closed-loop control circuit executes cyclically at a high frequency, with each cycle time controlled within a preset time threshold, ensuring effective tracking and real-time response to the rapid dynamic changes in the stress field during tempering.

[0028] The present invention further proposes: At least one infrared thermal imager is used to perform continuous non-contact scanning of all or key surfaces of the glass insulator blank; The infrared thermal imager will send the raw thermal imaging data, which includes multiple pixels and their corresponding temperature values, to the data acquisition and synchronization module. The data acquisition and synchronization module preprocesses the raw thermal imaging data, including noise filtering and coordinate registration, to generate structured real-time temperature field data that can be directly used by the thermo-mechanical coupling physical model.

[0029] Specifically, the infrared thermal imager scans at a rate of 10-30 frames per second, with each pixel corresponding to a spatial resolution of 0.5mm × 0.5mm. The data acquisition and synchronization module performs noise filtering on each frame of thermal imaging data, for example, using median filtering with a 3×3 pixel window to eliminate single-point temperature anomalies. Coordinate registration uses a pre-stored blank geometric model to convert the two-dimensional thermal imaging pixel coordinates into three-dimensional spatial coordinates, generating a four-dimensional array containing X, Y, and Z coordinates and temperature values. This array is stored in matrix form, with row numbers corresponding to spatial grid numbers and column elements containing coordinate and temperature data, which can be directly input into the thermal load boundary condition interface of the finite element model. This process transforms unstructured image data into temperature field data with clear physical meaning, ensuring that the input data for subsequent stress field calculations has temporal synchronization and spatial consistency.

[0030] As a preferred embodiment, the solution of the present invention is implemented as follows: To acquire real-time temperature field data of the glass insulator blank surface, two high-resolution infrared thermal imagers were used to continuously scan the entire surface of the glass insulator blank using a non-contact method. The two infrared thermal imagers were installed on opposite sides of the tempering station to ensure complete coverage of the glass insulator blank surface. The scanning frequency of the infrared thermal imagers was set to 30Hz, and the spatial resolution was 640×480 pixels.

[0031] The infrared thermal imager acquires raw thermal imaging data containing multiple pixels and their corresponding temperature values, and transmits it in real time to the data acquisition and synchronization module via industrial Ethernet. The data acquisition and synchronization module uses a high-performance industrial computer equipped with dedicated image processing software.

[0032] The data acquisition and synchronization module preprocesses the raw thermal imaging data, including noise filtering and coordinate registration. Noise filtering uses a median filtering algorithm to remove salt-and-pepper noise from the thermal images. Coordinate registration uses pre-calibrated reference points to spatially align and stitch the images acquired by the two thermal imagers, forming a complete surface temperature distribution map of the glass insulator blank.

[0033] After preprocessing, structured, real-time temperature field data is generated that can be directly used by the thermo-mechanical coupling physical model. This data is stored in matrix form, with each element corresponding to a discrete point on the surface of the glass insulator blank, containing the three-dimensional spatial coordinates and temperature value of that point.

[0034] This invention further proposes a high-precision finite element model for the thermo-mechanical coupling physical model; The thermo-mechanical coupling physical model is driven by the real-time temperature field data, including: The real-time temperature field data is input into the finite element model as a thermal load boundary condition. Transient heat conduction analysis was performed using the finite element model to calculate the temperature gradient distribution inside the glass insulator blank. Transient heat conduction equation: ; in, Density of glass, unit: ), where c is the unit of specific heat capacity: kg·℃, where k is the thermal conductivity, unit: m·℃, The internal heat source is represented by a value of 0 here. T represents temperature in °C, and t represents time. .

[0035] Based on the temperature gradient distribution, thermal stress analysis and calculation are performed to calculate the thermal stress and residual stress distribution at any position inside the glass insulator blank, and this distribution is used as the virtual measured stress field.

[0036] thermal stress constitutive equation: , ;in, Stress tensor, unit: MPa For the elasticity matrix, For the total strain tensor, For thermal strain tensor, Coefficient of thermal expansion, unit: 1 / ℃ This refers to the change in temperature, in degrees Celsius (°C). It is an identity matrix.

[0037] Specifically, real-time temperature field data, acquired by an infrared thermal imager, is mapped onto the surface nodes of the finite element model as thermal boundary conditions. The transient heat conduction equation, discretized using the finite element method, forms stiffness and mass matrices, and the temperature distribution at each time step is iteratively solved using the conjugate gradient method. The temperature gradient calculation results are transmitted to the thermal stress analysis module, where thermal strain and mechanical strain are decoupled using constitutive equations, and stress components at each integration point are calculated separately. For example, in a 10mm thick glass blank, when the surface temperature decreases at a rate of 50℃ / s, the finite element model can calculate that the tensile stress at 3mm from the surface reaches 12MPa, while the compressive stress in the core is -8MPa. This stress distribution data is output as nodal stress or element-averaged stress, forming benchmark data for comparison with the ideal target stress field. By accurately solving the coupling relationship between the transient temperature field and the stress field, the spatial resolution and temporal continuity of the virtual measured stress field are effectively improved, providing high-precision input for subsequent control decisions.

[0038] This invention further proposes a method for constructing an ideal stress model database, including: For glass insulators of different models and specifications, multiple physical tempering tests were conducted in advance; In the physical tempering experiment, multiple groups of samples with optimal mechanical and electrical properties were obtained by optimizing process parameters. The sample was subjected to physical testing to obtain data on the actual optimal residual stress distribution inside it. The actual optimal residual stress distribution data is digitized and stored together with the corresponding model and specification information in the ideal stress model database to form the ideal target stress field.

[0039] The physical tempering experiment needs to cover glass insulators of different models and specifications to ensure the comprehensiveness of the database; the optimization of process parameters involves iterative adjustment of the combination of tempering temperature, cooling time and wind pressure parameters to screen out samples that meet the standards for mechanical strength and electrical performance; physical testing uses destructive or non-destructive stress testing equipment, such as X-ray diffractometers or photoelastic testing devices, to directly measure the residual stress distribution inside the sample; digital processing includes converting the test data into a three-dimensional stress field matrix and establishing a mapping relationship with the model and specifications to form a structured database.

[0040] Specifically, for a certain type of glass insulator, the initial tempering temperature was set to 680℃ and the cooling air pressure to 8kPa in the physical tempering experiment. By adjusting the opening sequence of the air nozzle array, 10 samples with different combinations of process parameters were obtained. Four-point bending strength tests were performed on each sample, and qualified samples with a bending strength of 120MPa or higher were selected. The internal residual stress distribution was then measured using an X-ray diffractometer. The stress distribution data of the qualified samples was converted into a three-dimensional matrix containing X, Y, and Z-axis stress components and stored in association with the model label. For example, for an insulator of model XZ-120, the compressive stress value in the key area of ​​the ideal target stress field is 35±2MPa, and the tensile stress value does not exceed 5MPa. Through this database, the closed-loop control system can directly call the ideal stress field data matching the current production model, ensuring the accuracy of the comparison benchmark, thereby improving the precision of tempering process control and the consistency of finished products.

[0041] As a preferred embodiment, the solution of the present invention is implemented as follows: For glass insulators of different models and specifications, multiple physical tempering experiments were conducted in advance. During these experiments, process parameters were optimized to obtain multiple sets of samples with optimal mechanical and electrical properties. Physical testing was performed on the samples to obtain data on the actual optimal residual stress distribution within them. This data was then digitized and stored, along with the corresponding model and specification information, in an ideal stress model database to form an ideal target stress field.

[0042] Specifically, ten physical tempering experiments were conducted on a certain type of suspension glass insulator. In each experiment, process parameters such as tempering temperature, holding time, and cooling air pressure were adjusted. For example, the tempering temperature was adjusted in 10-degree intervals from 650℃ to 750℃; the holding time was adjusted in 1-minute intervals from 3 minutes to 8 minutes; and the cooling air pressure was adjusted in 0.05-MPa intervals from 0.3MPa to 0.6MPa. The optimal combination of process parameters was determined using orthogonal experimental design.

[0043] Further, the obtained samples underwent mechanical property testing, including flexural strength testing and thermal shock testing. The flexural strength test employed the three-point bending method, and the testing equipment was a universal testing machine. The thermal shock test was conducted according to GB / T2611 standard, and the testing equipment was a programmable temperature controlled thermal shock chamber. Five groups of samples with the best mechanical properties were selected for electrical performance testing, including power frequency dry flashover testing and impact withstand voltage testing. Based on this, the three groups of samples with the best overall performance were determined.

[0044] Physical testing was performed on these three groups of samples, and the internal residual stress distribution was measured using the photoelastic method. Specifically, a stress meter was used for measurement, with a precision of 0.1 MPa. During measurement, the samples were cut into 3 mm thick slices, and the distribution of isochromatic lines was observed and recorded under polarized light. Digital image processing technology was then used to convert the isochromatic line diagrams into stress distribution data.

[0045] Finally, the obtained stress distribution data is digitally processed. A complete stress field model is constructed using a three-dimensional interpolation algorithm. This model, along with the corresponding glass insulator model and specification information, is stored in an ideal stress model database. The database uses a relational database management system, such as MySQL, to establish a model-stress field correspondence table.

[0046] This invention further proposes a method for calculating stress deviation by comparing a virtual measured stress field with an ideal target stress field, including: In the virtual measured stress field and the ideal target stress field, a set of preset stress monitoring points corresponding to spatial locations are determined; Extract the stress tensor or principal stress value of each stress monitoring point in the virtual measured stress field; extract the target stress tensor or target principal stress value of each stress monitoring point in the ideal target stress field; Calculate the difference or vector difference between the virtual measured stress value and the target stress value to form a stress deviation matrix or deviation vector containing deviation information of each monitoring point.

[0047] The spatial locations of stress monitoring points are pre-defined based on the geometry of the glass insulator blank and stress-sensitive areas, covering key locations on both the surface and interior. The extraction of stress tensors or principal stress values ​​is achieved through the post-processing module of the finite element model, using numerical interpolation to obtain the stress components corresponding to the monitoring point coordinates. The calculation of differences or vector differences employs either the subtraction of tensor components one by one or the Euclidean distance calculation of principal stress vectors, with the row and column indices of the deviation matrix mapped one-to-one with the spatial coordinates of the monitoring points.

[0048] As a preferred embodiment, the solution of the present invention is implemented as follows: In both the virtual measured stress field and the ideal target stress field, a set of preset stress monitoring points corresponding to spatial locations are determined. These monitoring points can be selected at key locations on the glass insulator blank, such as edges, centers, and stress concentration areas. Specifically, 100 monitoring points can be evenly distributed on the surface of the insulator, and 50 monitoring points can be selected at key locations inside.

[0049] For each stress monitoring point, the stress tensor or principal stress value is extracted from the virtual measured stress field. The stress tensor contains six components: σxx, σyy, σzz, τxy, τyz, and τzx. The principal stress values ​​include the maximum principal stress σ1, the intermediate principal stress σ2, and the minimum principal stress σ3.

[0050] Furthermore, the target stress tensor or target principal stress value corresponding to each stress monitoring point is extracted from the ideal target stress field. These target values ​​represent the ideal stress distribution state.

[0051] Therefore, the difference or vector difference between the virtual measured stress value and the target stress value is calculated. For the stress tensor, the difference between the six components can be calculated; for the principal stress values, the difference between the three principal stresses is calculated.

[0052] Finally, the calculated difference information is organized into a stress deviation matrix or deviation vector. For example, for n monitoring points, a 6n×1 stress deviation vector can be obtained (if a stress tensor is used) or a 3n×1 stress deviation vector (if principal stress values ​​are used).

[0053] The present invention further proposes: The stress deviation matrix or deviation vector is provided as input to the Model Predictive Control (MPC) algorithm model. The Model Predictive Control (MPC) algorithm model performs rolling optimization calculations within a preset control time domain based on the input and the prediction of future states. Model Predictive Control (MPC) objective function formula: ; in, The stress deviation predicted at time t after k steps, in MPa. For prediction in the time domain, values ​​range from 5 to 10. The control adjustment value at time t, in kPa. To control the time domain Q and R are weight matrices for stress deviation and control quantity, respectively, and are diagonal matrices with positive elements. The objective of the rolling optimization calculation is to minimize the sum of predicted stress deviations over a future period of time. Output a control sequence for each nozzle array of the tempered air grating device that can achieve the stated objective, and use the first element of the control sequence as the optimization adjustment instruction for the current cycle.

[0054] Specifically, in the process of controlling the tempered glass grating, after the stress deviation matrix is ​​input into the MPC algorithm model in real time, the future... The stress field state is predicted step by step, based on the dynamic characteristics of the current nozzle control parameters and the thermo-mechanical coupling model. The stress deviation and control adjustment are weighted and summed using an objective function, while satisfying... Under time-domain constraints, the optimal control sequence is obtained by solving the problem. Only the first adjustment instruction of the control sequence is executed each time, and data is re-acquired and continuously optimized in the next control cycle. This method, through the combination of feedforward prediction and feedback correction, effectively compensates for the time-lag effect of temperature field changes during tempering, ensuring that the air pressure adjustment of the nozzle array precisely matches the actual process requirements. For example, when the monitoring point predicts that local stress deviation will occur in the third second, the MPC algorithm increases the airflow output of the corresponding nozzle area in advance in the first second, eliminating the accumulation of deviation through forward control. This model-based closed-loop control method significantly improves the uniformity and stability of the internal stress distribution of the glass insulator, thereby improving product quality and reducing potential risks such as "zero-value spontaneous explosion."

[0055] This invention further proposes a scheme for generating optimization and adjustment instructions using a fuzzy control algorithm model.

[0056] The fuzzy control algorithm model includes a fuzzification processing module, a fuzzy rule base, a fuzzy inference engine, and a defuzzification module. The fuzzification processing module converts the numerical values ​​of stress deviations into fuzzy linguistic variables. The fuzzy rule base stores expert experience rules defined in the form of conditional statements. The fuzzy inference engine performs logical operations based on the input variables and the rule base. The defuzzification module converts the fuzzy output into deterministic control commands through mathematical calculations. The fuzzy linguistic variables include seven levels: negative large, negative medium, negative small, zero, positive small, positive medium, and positive large. The fuzzy control rules adopt a mapping relationship of "if the deviation is negative large, then the adjustment amount is positive large." The defuzzification calculation uses the centroid method, determining the final control quantity through weighted averaging.

[0057] Specifically, during the tempering process, stress deviations are converted into fuzzy linguistic variables through fuzzification. For example, when the deviation is +5MPa, the membership function calculates a membership degree of 0.8 for positive deviations and 0.2 for neutral deviations. The fuzzy inference engine calls 25 preset rules, such as activating the rule "if the deviation is positive, then adjust the airflow to negative," and generates fuzzy control outputs through weighted synthesis. The defuzzification module performs integral calculations on the output membership distribution, for example, calculating an airflow adjustment of -12.3kPa. This instruction drives the solenoid valve of the corresponding nozzle array through the execution controller, reducing the airflow by 12.3kPa. Since the fuzzy control algorithm does not rely on precise mathematical models, but only generates control instructions through rule matching and membership calculation, the single calculation time is shortened to less than 10ms, enabling real-time response to sudden temperature field changes and effectively suppressing further expansion of stress deviations.

[0058] As a preferred embodiment, the solution of the present invention is implemented as follows: The control decision model adopts a fuzzy control algorithm model. When generating optimization adjustment instructions based on stress deviation, the stress deviation is first fuzzified and converted into fuzzy linguistic variables. For example, the stress deviation is divided into seven levels: "negative large", "negative medium", "negative small", "zero", "positive small", "positive medium", and "positive large".

[0059] Furthermore, a set of preset fuzzy control rules are applied. These rules take the form of "if the stress deviation is positive, then the airflow is adjusted to negative," etc. Specifically, 25 rules can be set to cover control strategies under different stress deviation conditions.

[0060] Therefore, by processing fuzzy linguistic variables and fuzzy control rules through a fuzzy inference engine, a fuzzified control output is obtained. The fuzzy inference adopts the Mamdani inference method and uses the max-min composition operator.

[0061] Finally, the fuzzy control output is defuzzified to obtain deterministic optimized adjustment commands for each nozzle array of the tempered glass air grating device. The defuzzification uses the centroid method to calculate specific adjustment values ​​for air pressure, air volume, or nozzle angle.

[0062] The present invention further proposes that the optimization adjustment command specifically includes the instruction parameter for adjusting at least one of the air pressure, air volume or air nozzle angle of a specific air nozzle array of the tempered air grid device; the execution controller is a programmable logic controller (PLC) or an industrial personal computer; after receiving the optimization adjustment command, the execution controller converts it into a drive signal and sends it to the solenoid valve or variable frequency fan driver corresponding to the specific air nozzle array.

[0063] As a preferred embodiment, the present invention is implemented as follows: During the tempering process of the glass insulator, the optimization adjustment command is configured to include composite adjustment parameters for air pressure and air volume in the nozzle array. The execution controller adopts an industrial-grade programmable logic controller, which integrates an analog output module. When the optimization adjustment command is transmitted to the execution controller via Ethernet, the air pressure parameter in the command is converted into a 4-20mA current signal and transmitted to the variable frequency fan driver of the corresponding nozzle array via a dedicated cable; the air volume parameter is converted into a pulse width modulation signal to drive the solenoid valve to adjust its opening. The adjustment command for a specific nozzle array preferentially acts on the area with the largest temperature difference from the surface of the glass insulator blank, and the cooling rate is precisely matched by independently controlling the air supply intensity of each nozzle.

[0064] The present invention further proposes that, before the control method begins, the model and specification information of the glass insulator blank to be processed is input; the method automatically loads the corresponding ideal target stress field from the ideal stress model database according to the model and specification information, and selects a matching thermo-mechanical coupling physical model from the model library.

[0065] The model specification information includes the geometric dimensions, material properties, and process standards of the glass insulator blank; the ideal stress model database stores the actual optimal residual stress distribution data corresponding to different models, which is obtained through physical tempering experiments and digitally processed; the model library pre-stores multiple parameterized thermo-mechanical coupling physical models, each model corresponding to the heat transfer and stress response characteristics of a specific model of glass insulator blank; the loading process triggers database retrieval and model matching through the identification code in the model specification information to ensure the consistency of the benchmark for subsequent calculations and comparisons.

[0066] Specifically, the operator inputs the model and specifications of the current glass insulator blank into the human-machine interface of the tempering equipment, such as entering the identification code " The system uses this identifier to search the ideal stress model database, retrieves the corresponding ideal target stress field data file, and simultaneously filters out high-precision finite element model parameters that match the model number from the model library. For example, model number " In the corresponding thermo-mechanical coupling physical model, material parameters include density. thermal conductivity Coefficient of thermal expansion Therefore, in subsequent steps, the matching model is driven by real-time temperature field data to perform stress calculation, and compared with the ideal target stress field of the same model to ensure that the reference data in the closed-loop control loop strictly corresponds to the current processing object, eliminating control deviations caused by model mismatch.

[0067] This invention further proposes that the data communication interface adopts the high-speed industrial Ethernet protocol, and the single cycle time of the closed-loop control loop is less than a preset time threshold, so as to ensure effective tracking and real-time response to the rapid dynamic changes of the stress field during the tempering process.

[0068] As a preferred embodiment, the specific implementation of the present invention is as follows: The data communication interface adopts the EtherCAT industrial Ethernet protocol, which is built based on the IEEE 802.3 standard and supports a periodic communication time synchronization mechanism. The single cycle time of the closed-loop control loop is strictly controlled within 50 milliseconds, and deterministic task scheduling is implemented through a real-time operating system kernel. The data acquisition module and the execution controller use a master-slave architecture for synchronous communication. The execution controller is equipped with a hardware timestamp processing unit to ensure that control commands are transmitted within the communication cycle, and the port forwarding delay of the industrial switch is optimized to the microsecond level. The preset time threshold is set to 50 milliseconds through the human-machine interface. When the real-time monitoring shows that the single cycle time exceeds the threshold, the system automatically triggers an alarm and activates the redundant communication channel.

[0069] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0070] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A control method for an intelligent production line of glass insulators based on digital twins, characterized in that, The method includes: S1. Obtain real-time temperature field data of the surface of a glass insulator blank to be tempered, which is placed in the tempering station. S2. Based on the real-time temperature field data, drive the preset thermo-mechanical coupling physical model to calculate the stress distribution inside the glass insulator blank in real time, and thereby obtain the virtual measured stress field. S3. Retrieve the ideal target stress field corresponding to the model and specifications of the glass insulator blank from the preset ideal stress model database; S4. Compare the virtual measured stress field with the ideal target stress field across the entire domain or at key points to calculate the stress deviation characterizing the difference between the two. S5. Based on the stress deviation, generate optimized adjustment instructions for the tempered air grid device under zone control through a preset control decision model; S6. The optimization and adjustment command is sent to the execution controller through the data communication interface, and the execution controller drives the multiple independently adjustable air nozzle arrays of the tempered air grid device to perform zoned differential cooling on the glass insulator blank. The method involves repeatedly executing steps such as acquiring real-time temperature field data, solving the virtual measured stress field, calculating stress deviation, and generating and executing optimization adjustment instructions to form a closed-loop control loop until the tempering process of the glass insulator blank is completed.

2. The method according to claim 1, characterized in that, Acquiring real-time temperature field data of the surface of the glass insulator blank includes: At least one infrared thermal imager is used to perform continuous non-contact scanning of all or key surfaces of the glass insulator blank; The infrared thermal imager will send the raw thermal imaging data, which includes multiple pixels and their corresponding temperature values, to the data acquisition and synchronization module. The data acquisition and synchronization module preprocesses the raw thermal imaging data, including noise filtering and coordinate registration, to generate structured real-time temperature field data that can be directly used by the thermo-mechanical coupling physical model.

3. The method according to claim 1, characterized in that, The thermo-mechanical coupling physical model is a high-precision finite element model. The thermo-mechanical coupling physical model is driven by the real-time temperature field data, including: The real-time temperature field data is input into the finite element model as a thermal load boundary condition. Transient heat conduction analysis was performed using the finite element model to calculate the temperature gradient distribution inside the glass insulator blank. Transient heat conduction equation: ; in, Density of glass, unit: ), where c is the unit of specific heat capacity: kg·℃, where k is the thermal conductivity, unit: m·℃, The internal heat source is represented by a value of 0 here. T represents temperature in °C, and t represents time. ; Based on the temperature gradient distribution, thermal stress analysis and calculation are performed to solve the thermal stress and residual stress distribution at any position inside the glass insulator blank, and the distribution is used as the virtual measured stress field. thermal stress constitutive equation: , ;in, Stress tensor, unit: MPa For the elasticity matrix, For the total strain tensor, For thermal strain tensor, Coefficient of thermal expansion, unit: 1 / ℃ This refers to the change in temperature, in degrees Celsius (°C). It is an identity matrix.

4. The method according to claim 3, characterized in that, The method for constructing the ideal stress model database includes: For glass insulators of different models and specifications, multiple physical tempering tests were conducted in advance; In the physical tempering experiment, multiple groups of samples with optimal mechanical and electrical properties were obtained by optimizing process parameters. The sample was subjected to physical testing to obtain data on the actual optimal residual stress distribution inside it. The actual optimal residual stress distribution data is digitized and stored together with the corresponding model and specification information in the ideal stress model database to form the ideal target stress field.

5. The method according to claim 1, characterized in that, The virtual measured stress field is compared with the ideal target stress field to calculate the stress deviation, including: In the virtual measured stress field and the ideal target stress field, a set of preset stress monitoring points corresponding to spatial locations are determined; Extract the stress tensor or principal stress value of each stress monitoring point in the virtual measured stress field; Extract the target stress tensor or target principal stress value of each stress monitoring point in the ideal target stress field; The difference or vector difference between the virtual measured stress value and the target stress value is calculated to form a stress deviation matrix or deviation vector containing deviation information of each monitoring point.

6. The method according to claim 5, characterized in that, The control decision model is a Model Predictive Control (MPC) algorithm model. Based on the stress deviation, an optimization adjustment instruction is generated, including: The stress deviation matrix or deviation vector is provided as input to the Model Predictive Control (MPC) algorithm model. The Model Predictive Control (MPC) algorithm model performs rolling optimization calculations within a preset control time domain based on the input and the prediction of future states. Model Predictive Control (MPC) objective function formula: ; in, The stress deviation predicted at time t after k steps, in MPa. For prediction in the time domain, values ​​range from 5 to 10. The control quantity adjustment value at time t. To control the time domain Q and R are weight matrices for stress deviation and control quantity, respectively, and are diagonal matrices with positive elements. The objective of the rolling optimization calculation is to minimize the sum of predicted stress deviations over a future period of time. Output a control sequence for each nozzle array of the tempered air grating device that can achieve the stated objective, and use the first element of the control sequence as the optimization adjustment instruction for the current cycle.

7. The method according to claim 1, characterized in that, The control decision model is a fuzzy control algorithm model; Based on the stress deviation, an optimization adjustment instruction is generated, including: The stress deviation is fuzzified and converted into one or more fuzzy linguistic variables. A set of preset fuzzy control rules are applied, the form of which is "if the stress deviation is positive, then the air volume is adjusted to negative". The fuzzy linguistic variables and the fuzzy control rules are processed by the fuzzy inference engine to obtain the fuzzy control output; Fuzzy control membership function formula: Large burden Membership function: ; Zhengda Membership function: ; Where e is the stress deviation, and E is the quantification range of the stress deviation, in MPa, which is set according to the insulator model, such as 10MPa; The fuzzy control output is defuzzified to obtain deterministic optimization adjustment instructions for each nozzle array of the tempered air grating device. Fuzzy control defuzzification, centroid method formula: ; in, Let i be the output membership degree of the i-th fuzzy rule. Let m be the output of the i-th rule, and m be the number of rules, ranging from 9 to 25.

8. The method according to claim 1, characterized in that, The optimization and adjustment instructions specifically include instruction parameters for adjusting at least one of the air pressure, air volume, or air nozzle angle of a specific air nozzle array of the tempered air grid device. The execution controller is a programmable logic controller (PLC) or an industrial personal computer. After receiving the optimization and adjustment command, the execution controller converts it into a drive signal and sends it to the solenoid valve or variable frequency fan driver corresponding to the specific nozzle array.

9. The method according to claim 1, characterized in that, The method further includes: Before the control method begins, input the model and specification information of the glass insulator blank to be processed; The method automatically loads the corresponding ideal target stress field from the ideal stress model database based on the model specification information, and selects the matching thermo-mechanical coupling physical model from the model library.

10. The method according to claim 1, characterized in that, The data communication interface adopts the high-speed industrial Ethernet protocol; The single-cycle time of the closed-loop control loop is less than a preset time threshold to ensure effective tracking and real-time response to the rapid dynamic changes in the stress field during the tempering process.

Images