Glass melting furnace current phase predictive regulation system and method based on space-time four-dimensional coupling analysis and glass melting furnace

Abstract

The application provides a glass melting furnace current phase predictive regulation system based on space-time four-dimensional coupling analysis, a regulation method and a glass melting furnace. The glass melting furnace current phase predictive regulation system based on space-time four-dimensional coupling analysis comprises a data acquisition module, a space-time four-dimensional analysis module, a prediction model module and a control module. The method is applied to the glass melting furnace current phase predictive regulation system based on space-time four-dimensional coupling analysis. The glass melting furnace comprises a melting furnace body and a plurality of electrodes arranged on the melting furnace body, and further comprises the glass melting furnace current phase predictive regulation system based on space-time four-dimensional coupling analysis. The application can predict the evolution trend of the mismatch region by performing high-precision four-dimensional coupling analysis on the temperature field, the current field and the flow field in the melting furnace, and can adjust the electrode current phase in advance, so that the problems of local overheating, uneven glass liquid quality and high energy consumption caused by the space-time mismatch between the current field and the temperature field in the glass melting furnace are solved.

Description

Technical Field

[0001] This invention relates to the field of electric melting technology, specifically to a glass melting furnace current phase prediction control system, control method, and glass melting furnace based on spatiotemporal four-dimensional coupling analysis. Background Technology

[0002] Glass melting furnaces are core thermal equipment in the glass production process, and the uniformity of their internal temperature field directly affects the clarification quality of the molten glass, melting efficiency, and furnace lifespan. Electric-assisted melting technology, which uses electrodes placed inside the furnace to utilize the Joule heat generated by the current flowing through the molten glass for auxiliary heating, has become a standard technology in large-scale glass melting furnaces. To ensure the stability of the internal temperature field, existing technologies typically employ distributed temperature measurement networks to collect temperature data, combined with digital twin systems for real-time temperature field simulation and prediction, and a controller to adjust the electrode power, achieving closed-loop feedback control of the temperature field. This approach has been widely used in industrial practice, enabling real-time monitoring and basic adjustment of the temperature field.

[0003] However, existing technologies still have the following shortcomings: First, existing systems are typical feedback control systems, which calculate the temperature field and make adjustments based on currently collected temperature data. Due to the large thermal inertia inside glass melting furnaces, with time constants reaching several minutes, feedback control has an inherent lag and is difficult to cope with rapid temperature fluctuations. Second, existing technologies mainly focus on the electrical conductivity of molten glass, neglecting the influence of the physical flow of the molten glass itself on heat transfer. In actual melting furnaces, molten glass flows continuously under the influence of natural and forced convection, and heat migrates with the fluid; relying solely on local temperature detection cannot predict the dynamic migration path of heat. Third, existing technologies cannot predict the development direction and rate of abnormal temperature regions. Summary of the Invention

[0004] To address the technical problems existing in the background art, this invention proposes a glass melting furnace current phase prediction control system, control method, and glass melting furnace based on spatiotemporal four-dimensional coupling analysis.

[0005] The glass melting furnace current phase predictive control system based on spatiotemporal four-dimensional coupling analysis proposed in this invention includes: a data acquisition module, a spatiotemporal four-dimensional analysis module, a prediction model module, and a control module, wherein: The data acquisition module includes a first optical fiber sensor network, a second optical fiber sensor network, a third optical fiber sensor network, and an embedded data processing module. The first optical fiber sensor network is embedded in the wall and bottom of the glass melting furnace to collect temperature data at multiple spatial points inside the furnace in real time. The second optical fiber sensor network is positioned around each electrode inside the furnace to collect current distribution data of each electrode in real time. The third optical fiber sensor network is deployed in the charging and discharging areas of the furnace to simultaneously measure temperature and strain. The data processing module acquires the collected data and calculates the temperature gradient vector field based on the data collected by the first optical fiber sensor network, the current density vector field in the molten glass based on the data collected by the second optical fiber sensor network, and the flow velocity field of the molten glass based on the data collected by the third optical fiber sensor network. The spatiotemporal four-dimensional analysis module is connected to the data acquisition module to receive the temperature gradient vector field, current density vector field and flow velocity field, and to calculate the time change rate of the temperature gradient, the time change rate of the current density, the angle field between the temperature gradient and the current density and its time change rate, as well as the convection-diffusion coupling coefficient. The prediction model module is connected to the spatiotemporal four-dimensional analysis module and is used to predict the location of the mismatch zone after a future time Δt based on the included angle field, the time change rate of the included angle field, the convection-diffusion coupling coefficient, and the flow velocity field at the current moment. The control module is connected to the prediction model module and the power supply of each electrode installed in the furnace, and is used to adjust the current phase of the corresponding electrode in advance according to the predicted mismatch zone location.

[0006] Preferably, the control module is configured to: when the predicted mismatch zone migrates downstream of the furnace, adjust the current phase of the downstream electrode in advance so that its current density vector field direction points to the forward direction of the predicted mismatch zone migration; when the predicted mismatch zone migrates upstream of the furnace, adjust the current phase of the upstream electrode in advance; when the predicted mismatch zone has no obvious migration trend, adjust the current phase of the electrode at the corresponding position of the predicted mismatch zone.

[0007] Preferably, the control module adopts a multi-electrode collaborative pre-adjustment strategy: when the predicted mismatch zone migrates downstream of the furnace, the current phase of the upstream electrode is adjusted with lag, so that its current density vector points behind the predicted mismatch zone, forming a "pull forward and push backward" synergistic effect.

[0008] Preferably, the prediction model module uses a linear extrapolation algorithm or a long short-term memory neural network for prediction, wherein the long short-term memory neural network is trained with historical time series data as input features and the temperature gradient field and the included angle field at future time moments as output labels.

[0009] Preferably, it also includes a feedback correction module, which is connected to both the control module and the data acquisition module, and is used to correct the prediction error of the prediction model module based on the actual acquired temperature field data at future times.

[0010] Preferably, the feedback correction module adopts a rolling time-domain optimization method, using the prediction error as a feedback quantity to dynamically correct the model parameters of the prediction model module.

[0011] Preferably, the first optical fiber sensing network includes horizontal and vertical optical fibers arranged in an orthogonal grid to form a three-dimensional temperature measurement grid; the horizontal optical fibers are arranged along the length and width of the melting furnace, and the vertical optical fibers are arranged along the height of the melting furnace.

[0012] Preferably, the second fiber optic sensing network is a ring-shaped fiber optic current sensor, which is arranged around the outer periphery of the electrode.

[0013] Preferably, the third fiber optic sensing network has at least two parallel measurement lines arranged along the flow direction of the molten glass, and multiple femtosecond grating measurement points are set on each measurement line; the flow velocity field is obtained by measuring the strain difference and time delay between adjacent measurement points and inverting the calculation using a cross-correlation algorithm.

[0014] The present invention proposes a method for predictive control of glass melting furnace current phase based on spatiotemporal four-dimensional coupling analysis. This method is applied to the aforementioned glass melting furnace current phase predictive control system based on spatiotemporal four-dimensional coupling analysis, and includes the following steps: S1: Data Acquisition and Feature Field Calculation The first fiber optic sensor network in the data acquisition module collects temperature data at multiple spatial points inside the furnace, the second fiber optic sensor network collects current distribution data at each electrode, and the third fiber optic sensor network collects temperature and strain data at the feed port and discharge port. The data processing module then calculates the temperature gradient vector field, current density vector field, and flow velocity field. S2: Spatiotemporal Four-Dimensional Coupling Analysis The spatiotemporal four-dimensional analysis module calculates the rate of change of time, the rate of change of current density, the angle field between temperature gradient and current density and its rate of change of time, and the convection-diffusion coupling coefficient. S3: Mismatch Zone Prediction The prediction model module predicts the angle field after a future time Δt based on the current angle field, the time rate of change of the angle field, the convection-diffusion coupling coefficient, and the flow velocity field, and identifies the region where the angle field value is within a preset range as the prediction mismatch zone. S4: Current phase pre-adjustment The control module adjusts the current phase of the corresponding electrode in advance based on the predicted spatial location and evolution direction of the mismatch region.

[0015] Preferably, in step S3, the angle field at future moments is predicted using a time-series prediction model, and the training data of the time-series prediction model includes historical angle field sequences, historical flow velocity field sequences, and corresponding actual temperature field evolution data.

[0016] Preferably, in step S4, the current phase of the corresponding electrode is adjusted using the following optimization model:

[0017] The constraints include: phase adjustment of each electrode ≤ 30°, and total power variation ≤ 10% of rated power; Where N is the number of electrodes involved in the regulation, wi is the weighting coefficient of each electrode, θ_{pred,i} is the predicted angle of the region where electrode i is located, and θ_{target} is the target angle.

[0018] The present invention proposes a glass melting furnace, which includes a melting furnace body and multiple electrodes disposed on the melting furnace body; it also includes the glass melting furnace current phase prediction control system based on spatiotemporal four-dimensional coupling analysis as described above.

[0019] This invention, through high-precision four-dimensional coupled analysis of the temperature, current, and flow fields within the melting furnace, predicts the evolution trend of the mismatch zone and adjusts the electrode current phase in advance. This solves problems such as localized overheating, uneven glass melt quality, and high energy consumption caused by the spatiotemporal mismatch between the current and temperature fields in glass melting furnaces. It achieves a shift from a "passive response" to an "active prediction" control mode, offering the following advantages compared to existing technologies: 1. It can predict the evolution of the temperature field and the migration trend of the mismatch zone, upgrading the traditional feedback control to feedforward predictive control, and overcoming the control lag problem caused by the large thermal inertia of the glass melting furnace. 2. When a mismatch zone is predicted to appear in the future, the current phase of the relevant electrodes is adjusted in advance so that the energy compensation and the glass melt flow arrive at the target area synchronously, thus shortening the response time and reducing the temperature fluctuation amplitude. 3. By using current phase regulation as the main means (rather than power regulation), hot spots are eliminated and temperature is balanced by changing the spatial distribution direction of Joule heat instead of increasing the total energy input, thus improving energy utilization efficiency by 15%-20%. Attached Figure Description

[0020] Figure 1 This is a block diagram of the glass melting furnace current phase prediction control system based on spatiotemporal four-dimensional coupling analysis proposed in this invention. Detailed Implementation

[0021] Reference Figure 1The glass melting furnace current phase prediction control system proposed in this invention, based on spatiotemporal four-dimensional coupling analysis, includes: a data acquisition module, a spatiotemporal four-dimensional analysis module, a prediction model module, and a control module, wherein: The data acquisition module includes a first optical fiber sensor network, a second optical fiber sensor network, a third optical fiber sensor network, and an embedded data processing module.

[0022] The first fiber optic sensing network is embedded inside the walls and bottom of the glass melting furnace to collect temperature data from multiple spatial points within the furnace in real time. As a preferred embodiment, this network is deployed in an orthogonal grid configuration, including horizontal and vertical fiber groups. The horizontal fiber groups are spaced along the length and width of the furnace, forming a two-dimensional planar grid; the vertical fiber groups are spaced along the height of the furnace, penetrating different depths through the walls and bottom. This deployment method creates a three-dimensional temperature measurement grid covering the entire furnace space, providing high spatial resolution data support for subsequent gradient field calculations.

[0023] A second fiber optic sensing network is positioned around each electrode within the melting furnace to collect real-time current distribution data for each electrode. In this embodiment, the second fiber optic sensing network is preferably a ring-shaped fiber optic current sensor, which is densely arranged around the outer periphery of the electrodes. This sensor, based on the Faraday magneto-optical effect, can non-contactly and with high precision measure the current density vector in the molten glass surrounding the electrodes, avoiding measurement errors caused by electromagnetic interference in traditional current transformers.

[0024] The third fiber optic sensing network is deployed in the charging and discharging areas of the melting furnace, as well as along the centerline of the main glass flow channel, for simultaneous measurement of temperature and strain. As a preferred embodiment, the third fiber optic sensing network employs femtosecond grating fibers: the network has at least two parallel measurement lines along the glass flow direction, with multiple femtosecond grating measurement points on each line. Femtosecond gratings possess high-temperature resistance and corrosion resistance, making them suitable for the harsh environment inside glass melting furnaces.

[0025] The data processing module is used to preprocess and calculate features from the raw data of the three types of sensor networks mentioned above. The specific operations are as follows: Based on the temperature data collected by the first fiber optic sensor network, the data processing module calculates the temperature gradient vector field ∇T(x,y,z,t) using three-dimensional spatial interpolation and difference algorithms. The magnitude and direction of the temperature gradient reflect the trend of heat transfer in space.

[0026] Based on the current data collected by the second fiber optic sensor network, combined with the geometric position of the electrodes and the conductivity model of the molten glass, the current density vector field J(x,y,z,t) in the molten glass is calculated. This vector field determines the spatial distribution of Joule heat within the melting furnace.

[0027] Based on strain data acquired using a third fiber optic sensor network, the flow velocity field v(x,y,z,t) of the molten glass is calculated by measuring the strain difference and time delay between adjacent measurement points and employing a cross-correlation algorithm. The velocity field is a key parameter characterizing the convective transport capability of the molten glass.

[0028] The spatiotemporal four-dimensional analysis module receives data from the data acquisition module. We analyze J and v, and perform deep coupling analysis on them to calculate the following key characteristic parameters: ① Rate of change of temperature gradient over time This parameter reflects the intensity of temperature field evolution and is an important indicator for judging the trend of thermal stability changes.

[0029] ② Rate of change of current density over time This parameter reflects changes in the electrode power supply status and can be used to identify abnormal current distribution caused by power fluctuations or electrode erosion.

[0030] ③ The angle between the temperature gradient and the current density and its rate of change over time Angle Calculated using the following formula:

[0031] This included angle is a core indicator for measuring the matching degree between the current field and the temperature field. When When the temperature is close to 0°, it indicates that the direction of the current density is consistent with the direction of heat conduction, and Joule heating is mainly used to enhance heat transfer along the temperature gradient; when When the angle approaches 90°, it indicates that the direction of the current is orthogonal to the direction of heat conduction, which may lead to local heat accumulation or a decrease in heat transfer efficiency.

[0032] ④ Convection-diffusion coupling coefficient C(x,y,z,t): This coefficient is calculated by the following formula:

[0033] This coefficient characterizes the contribution of the physical flow of the molten glass to heat transfer. The value of C ranges from 0 to 1. When C is close to 1, it indicates that heat is mainly carried and transferred by the flow of the molten glass (convection-dominated); when C is close to 0, it indicates that heat is mainly transferred through thermal conduction and diffusion (diffusion-dominated). This parameter provides a quantitative basis for predicting the migration of the mismatch region by the coupling of the flow field and the temperature field.

[0034] The prediction model module is connected to the spatiotemporal four-dimensional analysis module and is used to determine the angle field at the current moment. , the rate of change of the angle field over time Using the convection-diffusion coupling coefficient C(t) and the flow velocity field v(t), we can predict the future. The location of the mismatch zone after time. The optimal value range is 30 to 300 seconds. If it is too short, the control effect is limited, and if it is too long, the prediction accuracy will decrease.

[0035] Specifically, the prediction model module first predicts the future. Angle field after time Then identify the included field angle. The region where the value falls within a preset range is designated as the "predicted mismatch zone". This preset angle range is preferably between 75 degrees and 105 degrees. When the angle is close to 90 degrees, the orthogonality between the current field and the temperature field is strongest, making it most likely to induce local thermal imbalance, which is the main target area for regulation.

[0036] As a basic implementation method, the prediction model module can use a linear extrapolation algorithm to linearly calculate the location of the mismatch zone based on the current angle field and its rate of change, combined with the flow velocity field.

[0037] As a preferred implementation, the prediction model module employs a Long Short-Term Memory (LSTM) neural network for prediction. The LSTM network uses historical time-series data as input features, including: the angle field sequence of the past N time points. Flow velocity field sequence Convection-diffusion coupling coefficient sequence The network is trained using the actual temperature gradient field and angle field at future moments as output labels, along with the electrode power sequence. Through offline training, the LSTM network can learn the nonlinear temporal evolution of multi-physics coupling within the furnace, achieving high-precision prediction of the mismatch zone location.

[0038] The control module is connected to the prediction model module and the power supply for each electrode located within the furnace. It adjusts the current phase of the corresponding electrode in advance based on the predicted location of the mismatch zone. Its core control logic involves changing the phase of the electrode current to rotate the current density vector field J, thereby reducing its interaction with the temperature gradient vector field when the predicted mismatch zone arrives. The included angle is used to achieve "pre-alignment" control. Specifically, the control module is configured as follows: When the mismatch zone is predicted to migrate downstream of the furnace (i.e., towards the discharge port), the current phase of the downstream electrode is adjusted in advance so that its current density vector field points in the predicted direction of the mismatch zone's migration. This "forward pull" effect can establish a matching current field in advance at the location where the mismatch zone is about to reach.

[0039] When the mismatch zone is predicted to migrate upstream of the furnace (i.e., towards the feed port), the current phase of the upstream electrode is adjusted in advance.

[0040] When the predicted mismatch region shows no obvious migration trend, the current phase of the electrode at the corresponding position of the predicted mismatch region is directly adjusted.

[0041] As a more preferred implementation, the control module employs a multi-electrode collaborative pre-adjustment strategy. For example, when the predicted mismatch zone migrates downstream of the furnace, in addition to "pulling forward" the downstream electrode, the current phase of the upstream electrode is also hysteret-adjusted so that its current density vector points behind the predicted mismatch zone, forming a synergistic "pull forward and push backward" effect, which effectively "pinch" and eliminates the mismatch zone during its migration.

[0042] When adjusting the current phase of the corresponding electrode, the control module uses the following optimization model to solve for the phase adjustment amount of each electrode. :

[0043] The constraints include: Condition 1: Phase adjustment amount of each electrode To avoid causing excessive impact on the power grid; Condition 2: Change in total power 10% of the rated power ensures the stability of the furnace's heat load.

[0044] Where N is the number of electrodes involved in the adjustment, and w_i is the weighting coefficient of each electrode (which can be set according to the spatial distance between the electrode and the predicted mismatch region). The predicted angle of the region where electrode i is located. The target angle is (preferably 0° or a small positive value).

[0045] As a preferred embodiment, the control system also includes a feedback correction module. This module is connected to both the control module and the data acquisition module, and is used to correct the prediction error of the prediction model module based on the actual acquired temperature field data for future times. Specifically, after one prediction cycle... Afterwards, the data acquisition module collects the actual temperature field and angle field data, and the feedback correction module compares the actual value with the predicted value and calculates the prediction error.

[0046] As a preferred implementation, the feedback correction module employs a rolling time-domain optimization method. This method uses the prediction error as feedback to dynamically correct the model parameters or initial values ​​of the linear extrapolation of the prediction model module (such as an LSTM network). Through continuous rolling optimization, the system's prediction accuracy and adaptability to changes in operating conditions are continuously improved, forming a closed-loop intelligent control system of "prediction-control-feedback-correction".

[0047] Furthermore, this embodiment also provides a method for predictive control of glass melting furnace current phase based on spatiotemporal four-dimensional coupling analysis. This method is applied to the above system and specifically includes the following steps: S1: Data Acquisition and Feature Field Calculation Temperature data from multiple spatial points inside the furnace is collected via a first fiber optic sensor network; current distribution data from each electrode is collected via a second fiber optic sensor network; and temperature and strain data from the feed and discharge ports are collected via a third fiber optic sensor network. The data processing module calculates the temperature gradient vector field. The current density vector field J(x,y,z,t) and the flow velocity field v(x,y,z,t).

[0048] S2: Spatiotemporal Four-Dimensional Coupling Analysis The spatiotemporal four-dimensional analysis module calculates the rate of change of time. , Calculate the angle field between the temperature gradient and the current density. and its rate of change over time And calculate the convection-diffusion coupling coefficient. .

[0049] S3: Mismatch Zone Prediction The prediction model module is based on the current angle field θ(t) and the time rate of change of the angle field. The convection-diffusion coupling coefficient C(t) and the flow velocity field v(t) are used to predict future convection-diffusion coupling coefficients C(t) and flow velocity field v(t) using a trained LSTM time series prediction model. Angle field after time And identify the included angle field The region where the value is within a preset range (e.g., 75 degrees to 105 degrees) is the predicted mismatch region.

[0050] S4: Current phase pre-adjustment The control module determines the electrode groups requiring pre-adjustment based on the predicted spatial location and evolution direction of the mismatch region. The current phase adjustment amount for each electrode is calculated using the aforementioned optimization model. It also sends a phase adjustment command to the power supply of the target electrode in advance at the current moment to achieve active pre-alignment of the current density vector field.

[0051] In addition, this embodiment also provides a melting furnace, including a furnace body and multiple electrodes disposed on the furnace body, and further including the glass melting furnace current phase prediction control system based on spatiotemporal four-dimensional coupling analysis as described in any of the above embodiments. By integrating this system, the glass melting furnace can achieve high-precision sensing and advanced control of internal multi-physics fields, significantly improving melting efficiency, extending furnace life, and enhancing the uniformity of glass product quality.

[0052] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A glass furnace current phase predictive control system based on spatio-temporal four-dimensional coupling analysis, characterized in that, include: The system comprises a data acquisition module, a spatiotemporal four-dimensional analysis module, a prediction model module, and a control module, among which: The data acquisition module includes a first optical fiber sensor network, a second optical fiber sensor network, a third optical fiber sensor network, and an embedded data processing module. The first optical fiber sensor network is embedded in the wall and bottom of the glass melting furnace to collect temperature data at multiple spatial points inside the furnace in real time. The second optical fiber sensor network is positioned around each electrode inside the furnace to collect current distribution data of each electrode in real time. The third optical fiber sensor network is deployed in the charging and discharging areas of the furnace to simultaneously measure temperature and strain. The data processing module acquires the collected data and calculates the temperature gradient vector field based on the data collected by the first optical fiber sensor network, the current density vector field in the molten glass based on the data collected by the second optical fiber sensor network, and the flow velocity field of the molten glass based on the data collected by the third optical fiber sensor network. The spatiotemporal four-dimensional analysis module is connected to the data acquisition module to receive the temperature gradient vector field, current density vector field and flow velocity field, and to calculate the time change rate of the temperature gradient, the time change rate of the current density, the angle field between the temperature gradient and the current density and its time change rate, as well as the convection-diffusion coupling coefficient. The prediction model module is connected to the spatiotemporal four-dimensional analysis module and is used to predict the location of the mismatch zone after a future time Δt based on the included angle field, the time change rate of the included angle field, the convection-diffusion coupling coefficient, and the flow velocity field at the current moment. The control module is connected to the prediction model module and the power supply of each electrode installed in the furnace, and is used to adjust the current phase of the corresponding electrode in advance according to the predicted mismatch zone location.

2. The glass melter current phase predictive regulation system based on spatiotemporal four-dimensional coupled analysis of claim 1, wherein, The control module is configured to: when the predicted mismatch zone migrates downstream of the furnace, adjust the current phase of the downstream electrode in advance so that its current density vector field direction points to the forward direction of the predicted mismatch zone migration; when the predicted mismatch zone migrates upstream of the furnace, adjust the current phase of the upstream electrode in advance; when the predicted mismatch zone has no obvious migration trend, adjust the current phase of the electrode at the corresponding position of the predicted mismatch zone. Preferably, the control module adopts a multi-electrode collaborative pre-adjustment strategy: when the predicted mismatch zone migrates downstream of the furnace, the current phase of the upstream electrode is adjusted with lag, so that its current density vector points behind the predicted mismatch zone, forming a "pull-forward and push-back" collaborative effect.

3. The glass melter current phase predictive regulation system based on spatiotemporal four-dimensional coupled analysis of claim 1, wherein, The prediction model module uses a linear extrapolation algorithm or a long short-term memory neural network for prediction. The long short-term memory neural network is trained with historical time series data as input features and the temperature gradient field and the included angle field at future time moments as output labels.

4. The glass melter current phase predictive regulation system based on spatiotemporal four-dimensional coupled analysis of claim 1, wherein, It also includes a feedback correction module, which is connected to the control module and the data acquisition module respectively, and is used to correct the prediction error of the prediction model module based on the actual temperature field data of future times collected. Preferably, the feedback correction module adopts a rolling time-domain optimization method, using the prediction error as a feedback quantity to dynamically correct the model parameters of the prediction model module.

5. The spatiotemporal four-dimensional coupled analysis-based predictive regulation system of electric current phase of a glass melter of claim 1, wherein, The first optical fiber sensing network includes horizontal and vertical optical fibers arranged in an orthogonal grid to form a three-dimensional temperature measurement grid; the horizontal optical fibers are arranged along the length and width of the melting furnace, and the vertical optical fibers are arranged along the height of the melting furnace.

6. The spatiotemporal four-dimensional coupled analysis-based glass melter current phase predictive regulation system of claim 1, wherein, The second fiber optic sensing network is a ring-shaped fiber optic current sensor, which is arranged around the outer periphery of the electrodes.

7. The spatiotemporal four-dimensional coupled analysis-based glass melter current phase predictive regulation system of claim 1, wherein, The third fiber optic sensing network is arranged with at least two parallel measurement lines along the flow direction of the molten glass, and multiple femtosecond grating measurement points are set on each measurement line; the flow velocity field v(x,y,z,t) is obtained by measuring the strain difference and time delay between adjacent measurement points and inverting the calculation using a cross-correlation algorithm.

8. A method for predictive control of glass melting furnace current phase based on spatiotemporal four-dimensional coupling analysis, characterized in that, This method, applied to the glass melting furnace current phase prediction control system based on spatiotemporal four-dimensional coupling analysis as described in any one of claims 1-7, includes the following steps: S1: Data Acquisition and Feature Field Calculation The first fiber optic sensor network in the data acquisition module collects temperature data at multiple spatial points inside the furnace, the second fiber optic sensor network collects current distribution data at each electrode, and the third fiber optic sensor network collects temperature and strain data at the feed port and discharge port. The data processing module then calculates the temperature gradient vector field, current density vector field, and flow velocity field. S2: Spatiotemporal Four-Dimensional Coupling Analysis The spatiotemporal four-dimensional analysis module calculates the rate of change of time, the rate of change of current density, the angle field between temperature gradient and current density and its rate of change of time, and the convection-diffusion coupling coefficient. S3: Mismatch Zone Prediction The prediction model module predicts the angle field after a future time Δt based on the current angle field, the time rate of change of the angle field, the convection-diffusion coupling coefficient, and the flow velocity field, and identifies the region where the angle field value is within a preset range as the prediction mismatch zone. S4: Current phase pre-adjustment The control module adjusts the current phase of the corresponding electrode in advance based on the predicted spatial location and evolution direction of the mismatch region.

9. The method for predictive control of glass melting furnace current phase based on spatiotemporal four-dimensional coupling analysis according to claim 8, characterized in that, In step S3, the angle field at future moments is predicted using a time-series prediction model. The training data of the time-series prediction model includes historical angle field sequences, historical flow velocity field sequences, and corresponding actual temperature field evolution data. Preferably, in step S4, the current phase of the corresponding electrode is adjusted using the following optimization model: The constraints include: phase adjustment of each electrode ≤ 30°, and total power variation ≤ 10% of rated power; Where N is the number of electrodes involved in the regulation, w_i is the weighting coefficient of each electrode, θ_{pred,i} is the predicted angle of the region where electrode i is located, and θ_{target} is the target angle.

10. A glass melting furnace, comprising a furnace body and a plurality of electrodes disposed on the furnace body, characterized in that, It also includes the glass melting furnace current phase prediction control system based on spatiotemporal four-dimensional coupling analysis as described in any one of claims 1-7.

Images