Information processing apparatus, glass manufacturing system, information processing method, program, and correction parameter estimation method

Abstract

The information processing apparatus of the embodiment is provided with a state estimation unit that performs state estimation of an actual machine using measurement data stored in a measurement data storage unit that stores the measurement data of a plurality of sensors provided inside or outside a glass melting furnace in order to measure values related to molten glass, and a glass melting furnace simulation model obtained by modeling the glass melting furnace.

Description

Technical Field

[0001] This disclosure relates to information processing apparatus, glass manufacturing system, information processing method, procedure, and method for estimating correction parameters. Background Technology

[0002] A glass melting furnace is used in the manufacturing apparatus for glass articles.

[0003] The glass melting furnace includes, for example, a melting tank having a bottom and side walls, and an upper structure covering the top of the melting tank.

[0004] The convection generated in the molten glass obtained by melting glass raw materials will affect the quality of the final glass product, specifically the density of bubbles, the density of foreign matter, and uniformity.

[0005] Therefore, in the technology described in Patent Document 1, the following solution is proposed: the flow near the surface of the molten glass in the upstream circulating flow formed upstream of the bubbler is designated as the upstream surface flow, and the flow near the surface of the molten glass in the downstream circulating flow formed downstream of the bubbler is designated as the downstream surface flow, while adjusting the average flow velocity of the upstream surface flow and the downstream surface flow (see Patent Document 1).

[0006] Existing technical documents

[0007] Patent documents

[0008] Patent Document 1: International Publication No. 2015 / 033931 Summary of the Invention

[0009] The problem that the invention aims to solve

[0010] During the glass melting process, it is necessary to improve the quality of the produced glass, increase production efficiency, reduce costs, and reduce environmental impact.

[0011] However, because the glass melting process is a high-temperature process, it is difficult to have a detailed understanding of the internal conditions of the glass melting furnace, and sometimes it is difficult to consistently achieve the required results.

[0012] Numerical simulation may be used to understand the internal state of a glass melting furnace or to explore its operation; however, its use is limited in the prior art for the following reasons: The model adjustment work (the workflow for performing the simulation) required to obtain sufficiently accurate simulation results is time-consuming; and the number of personnel capable of performing the simulation (e.g., dedicated analysis personnel) is limited because the execution of the simulation requires specialized knowledge and skills. Therefore, simulations are sometimes difficult to execute in a timely manner and cannot fully meet the requirements of resolving manufacturing site problems in a short period.

[0013] This disclosure was made in view of the following situation, and its object is to provide, in relation to glass manufacturing, an information processing device, a glass manufacturing system, an information processing method, a program, and a correction parameter estimation method that can achieve simulation-based assistance and efficiency improvement compared with the prior art by realizing automated simulation.

[0014] Methods for solving problems

[0015] One aspect of this disclosure is an information processing apparatus comprising a state estimation unit that uses measurement data stored in a measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform state estimation of the actual machine. The measurement data storage unit stores the measurement data from a plurality of sensors, which are located inside or outside the glass melting furnace for measuring values ​​related to molten glass.

[0016] One aspect of this disclosure is a glass manufacturing system comprising a glass manufacturing apparatus and an information processing apparatus. The glass manufacturing apparatus includes a glass melting furnace. A measurement data storage unit is provided inside or outside the information processing apparatus. The measurement data storage unit stores measurement data from multiple sensors located inside or outside the glass melting furnace for measuring values ​​related to molten glass. The information processing apparatus includes a state estimation unit that uses a glass melting furnace simulation model obtained by modeling the glass melting furnace and the measurement data stored in the measurement data storage unit to perform a state estimation of the actual system.

[0017] One aspect of this disclosure is an information processing method in which an information processing device uses measurement data stored in a measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform a state estimation of the actual machine. The measurement data storage unit stores the measurement data from a plurality of sensors, which are located inside or outside the glass melting furnace for measuring values ​​related to the molten glass.

[0018] One aspect of this disclosure is a program for enabling a computer to perform a state estimation function, which uses measurement data stored in a measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform a state estimation of the actual machine. The measurement data storage unit stores the measurement data from a plurality of sensors located inside or outside the glass melting furnace for measuring values ​​related to the molten glass.

[0019] One aspect of this disclosure is a correction parameter estimation method, wherein the operating conditions and correction parameters of the glass melting furnace are used as set variables, and multiple simulations are performed on a glass melting furnace simulation model based on the experimental planning method, within a scope that comprehensively includes the conditions envisioned in the actual glass manufacturing process. The simulation results are then used for machine learning to generate a predictive model, which evaluates the sensitivity of the operating conditions and correction parameters of the glass melting furnace to changes in the simulation results of at least one of the following at multiple observation points of the glass melting furnace: molten glass temperature, molten glass resistivity, current or voltage or resistance values ​​between heating electrodes, molten glass flow rate, furnace material temperature, and the temperature of the atmosphere inside and outside the furnace.

[0020] Invention Effects

[0021] According to the information processing apparatus, glass manufacturing system, information processing method, program, and correction parameter estimation method disclosed herein, in relation to glass manufacturing, by realizing automated simulation, simulation-based assistance can be achieved to improve efficiency compared to the prior art. Attached Figure Description

[0022] Figure 1 This is a diagram illustrating a schematic structural example of a glass manufacturing system according to an embodiment.

[0023] Figure 2A This is a diagram showing an outline of an online simulation of the implementation method.

[0024] Figure 2B This is a diagram showing an outline of a digital prototype of the implementation method.

[0025] Figure 3A This is a diagram showing an example of a cross-section of a glass melting furnace according to an embodiment.

[0026] Figure 3B This is a diagram showing an example of another cross-section of the glass melting furnace in an embodiment.

[0027] Figure 4 This diagram illustrates an example of the processing steps performed in the information processing apparatus of an embodiment.

[0028] Figure 5 This diagram illustrates an example of the steps involved in the correction parameter estimation process performed in the information processing apparatus of the embodiment.

[0029] Figure 6 This is a diagram illustrating another example of the correction parameter estimation process performed in the information processing apparatus of the embodiment.

[0030] Figure 7 This is a diagram showing an example table illustrating the correction parameters of an implementation method.

[0031] Figure 8 This is a diagram schematically illustrating the first process in a specific example of the correction parameter estimation process of the implementation method.

[0032] Figure 9A This is a diagram schematically illustrating the result of the first process in a specific example of the correction parameter estimation process of the implementation method.

[0033] Figure 9B This is a diagram schematically illustrating the second process in a specific example of the correction parameter estimation process of the implementation method.

[0034] Figure 10 This is a graph showing an example of a table representing the estimation results of the correction parameters of an implementation method.

[0035] Figure 11 This is a graph showing a comparison between the characteristics of the simulation results estimated using the correction parameters of the implementation method and the characteristics of the actual machine data.

[0036] Figure 12 This is a diagram that schematically illustrates a comparison between the characteristics based on simulation results and the characteristics based on actual machine data without the correction parameter estimation process.

[0037] Figure 13 This is a diagram illustrating an example of the aging changes in heat dissipation from the furnace wall of a glass melting furnace. Detailed Implementation

[0038] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings.

[0039] (First Implementation)

[0040] The first embodiment will be described.

[0041] [Information processing device]

[0042] Figure 1 This is a diagram illustrating a schematic structural example of the glass manufacturing system 1 according to the embodiment.

[0043] The glass manufacturing system 1 includes an information processing device 111, a database 112, and a glass manufacturing device 131.

[0044] In this embodiment, the information processing device 111 is communicatively connected to the database 112, and the database 112 is communicatively connected to the glass manufacturing device 131.

[0045] The glass manufacturing apparatus 131 includes a glass melting furnace 311.

[0046] In addition, in this embodiment, the glass manufacturing apparatus 131 includes a sensor unit 331 disposed inside the glass melting furnace 311 and a sensor unit 351 disposed outside the glass melting furnace 311.

[0047] Here, the sensor unit 331 includes one or more sensors.

[0048] In addition, the sensor section 351 includes more than one sensor.

[0049] Furthermore, in this embodiment, a sensor unit 331 disposed inside the glass melting furnace 311 and a sensor unit 351 disposed outside the glass melting furnace 311 are shown, but it is not necessary to have both of them; for example, only one of them may be provided.

[0050] In the case where only one sensor unit (sensor unit 331 or sensor unit 351) is provided, the sensor unit includes multiple sensors.

[0051] Database 112 stores the measurement data R1 obtained by sensor unit 331 and sensor unit 351.

[0052] In this embodiment, a case is shown where a database 112 is separate from the information processing device 111. However, as another example, the function of the database 112 can also be integrated into the information processing device 111. In this case, the database 112 is not provided. For example, the information processing device 111 and the glass manufacturing device 131 are connected in a communicable manner.

[0053] <Information Processing Device>

[0054] In this embodiment, the information processing device 111 is configured using a computer. The information processing device 111 includes a processor such as a CPU (Central Processing Unit), which executes prescribed programs to perform various processes and controls.

[0055] The information processing device 111 includes an input unit 211, an output unit 212, a communication unit 213, a storage unit 214, and a control unit 215.

[0056] The input unit 211 has an operation unit 231.

[0057] The output unit 212 has a display unit 251.

[0058] The control unit 215 includes a state estimation unit 271, a correction parameter estimation unit 272, a model calculation unit 273, an experimental planning method execution unit 274, a prediction model generation unit 275, a change sensitivity evaluation unit 276, and an observation point data acquisition unit 277.

[0059] Furthermore, in this embodiment, the structure of the correction parameter estimation unit 272 is shown, which includes an experiment planning execution unit 274, a prediction model generation unit 275, a change sensitivity evaluation unit 276, and an observation point data acquisition unit 277, but it is not necessarily limited to this structure.

[0060] The input unit 211 receives information from external devices or users (e.g., people). For example, the operation unit 231 receives information corresponding to an operation performed by the user.

[0061] The output unit 212 outputs information to external devices or users. For example, the display unit 251 has a screen, and information that is the object of display is displayed and output to the screen. This information can be information in any format, such as image information or numerical information.

[0062] The communication unit 213 communicates with external devices. This communication can be wired or wireless.

[0063] Furthermore, in this embodiment, for ease of explanation, the communication unit 213 is shown as a functional unit different from the input unit 211 and the output unit 212. However, as another example, the function of the communication unit 213 may be considered to be included in the input unit 211 and the output unit 212.

[0064] Storage unit 214 stores various types of information.

[0065] In this embodiment, the storage unit 214 stores, for example, a glass melting furnace simulation model R11, a correction parameter estimation model R12, a prediction model R13, and simulation results R14.

[0066] Furthermore, this embodiment illustrates a method of using a prediction model R13 when estimating correction parameters using the correction parameter estimation model R12. However, in cases where other methods are used, the prediction model R13 may not be used. In this embodiment, for example, the prediction model R13 may also be considered as a model that is generalized to the correction parameter estimation model R12.

[0067] Here, the information stored in the database 112 and the storage unit 214 can be arbitrary.

[0068] In this embodiment, various types of information can be stored in either the database 112 or the storage unit 214, or in both.

[0069] In addition, this embodiment illustrates information stored in database 112 and storage unit 214, but other arbitrary information may also be stored.

[0070] Furthermore, in this embodiment, various types of data can also be referred to as information.

[0071] Database 112 has the function of a measurement data storage unit, which stores measurement data from multiple sensors installed inside or outside the glass melting furnace 311 for measuring values ​​related to molten glass.

[0072] In this embodiment, the plurality of sensors are a total of multiple sensors included in sensor section 331 and sensor section 351.

[0073] The state estimation unit 271 uses the measurement data R1 stored in the measurement data storage unit (database 112 in this embodiment) and the glass melting furnace simulation model R11 obtained by modeling the glass melting furnace 311 to perform state estimation of the actual machine.

[0074] In this embodiment, the actual machine is the glass melting furnace 311 in the actual glass manufacturing apparatus 131.

[0075] Furthermore, estimation can also be referred to as, for example, prediction or speculation.

[0076] The calibration parameter estimation unit 272 uses the calibration parameter estimation model R12, which estimates the calibration parameters of the glass melting furnace simulation model R11, and the measurement data R1 stored in the measurement data storage unit to estimate the calibration parameters.

[0077] In this case, the state estimation unit 271 uses the correction parameters estimated by the correction parameter estimation unit 272 to perform state estimation of the actual machine.

[0078] Furthermore, in this embodiment, for ease of explanation, the parameters used to correct the glass melting furnace simulation model R11 will be referred to as correction parameters, but they may also be called by other names or simply parameters.

[0079] The model calculation unit 273 performs optimization calculations or sensitivity analyses based on the operation change guidance scheme, using the glass melting furnace simulation model R11 or a summary model based on the glass melting furnace simulation model R11. The operation change guidance scheme is based on the state estimation results of the state estimation unit 271.

[0080] In this embodiment, the summary model refers to a model that summarizes the phenomenon, such as a reduced model or a machine learning model, or both.

[0081] Here, the operational change guidance scheme can be provided by the user, or it can be automatically generated and instructed by the device that performs the predetermined process.

[0082] In addition, both optimization calculations and sensitivity analyses can be performed. For example, the results of sensitivity analysis can be used for optimization calculations.

[0083] The output unit 212 outputs information related to at least one of the state estimation results of the state estimation unit 271 and the calculation results of the model calculation unit 273.

[0084] Here, a structural example for estimating the correction parameters is shown. Furthermore, any other arbitrary method can be used as a method for estimating the correction parameters.

[0085] The experimental planning method execution unit 274 uses the operating conditions and calibration parameters of the glass melting furnace 311 as set variables, and performs multiple simulations on the glass melting furnace simulation model R11 based on the experimental planning method, within a range that comprehensively covers the conditions envisioned in the actual glass manufacturing process.

[0086] The storage unit 214 has the function of a simulation result storage unit, storing the simulation result R14 obtained by the experiment planning method execution unit 274.

[0087] The prediction model generation unit 275 uses the simulation result R14 stored in the simulation result storage unit (storage unit 214 in this embodiment) to perform machine learning and generate a prediction model R13.

[0088] The variation sensitivity evaluation unit 276 evaluates the variation sensitivity of the operating conditions and correction parameters of the glass melting furnace 311 to at least one of the simulation results of the molten glass temperature and the molten glass flow rate at multiple observation points of the glass melting furnace 311.

[0089] The observation point data acquisition unit 277 acquires the operating conditions at a predetermined time point for multiple observation points in the glass melting furnace 311, and acquires observation point data, which is data of at least one of the molten glass temperature and molten glass flow rate at the predetermined time point.

[0090] The correction parameter estimation unit 272 estimates the correction parameters based on the evaluation results of the change sensitivity evaluation unit 276 in a manner that the deviation between the observation point data obtained by the observation point data acquisition unit 277 for each observation point and the results obtained through simulation is reduced.

[0091] The correction parameter estimation unit 272 can also limit the range of correction parameter values ​​based on the change in heat dissipation of the furnace material caused by the aging of the glass melting furnace 311.

[0092] [Example of digital twin technology in the glass melting process]

[0093] Reference Figure 2A and Figure 2B This illustrates an example of digital twin technology in the glass melting process.

[0094] In addition, digital twins refer to the virtual reproduction of the real-world environment based on information from the real-world space.

[0095] The online simulator serves as a trigger for the execution of actions. For example, it can have a function that triggers automatically by the online simulator (automatic function), a function that triggers based on user input (manual function), or both.

[0096] When triggers are generated automatically, the online simulator executes actions at predetermined time intervals (e.g., a fixed time each day). Alternatively, when triggers are generated manually, the online simulator executes actions based on input information (trigger information) from the user. In this case, the user can, for example, instruct the online simulator to execute actions at any time and can specify any simulation period.

[0097] In this embodiment, an online simulator is used as an example to illustrate a system that automatically generates and updates simulation models in real time within a virtual space. However, as another example, an online simulator can also be a system that generates and updates simulation models that reproduce the state at any point in time through user interaction.

[0098] In addition, digital prototyping tools are used for sensitivity analysis and preliminary exploration of condition change simulations.

[0099] As an example of digital twin technology in the glass melting process, a system integrating online simulators and digital prototyping tools has been developed. This system allows for applications such as validation in float glass furnaces. This technology enables rapid and detailed monitoring of the glass melting process, which was previously difficult to achieve, as well as advance discussions of production conditions.

[0100] In general, this system uses data from the actual manufacturing process to perform simulations based on a simulation model via an online simulator. Parameter estimation can also be performed at this time.

[0101] Furthermore, this system utilizes digital prototypes to achieve optimal action, design, information, and added value.

[0102] Here, online simulators and digital prototyping tools can operate either synchronously or separately.

[0103] Specific examples of the application flow in this system are as follows (processing L1) to (processing L4).

[0104] (Processing L1) In this system, operational data and actual machine status data are acquired. Operational data includes, for example, data based on the heating distribution of the burner or electric flux. Actual machine status data includes, for example, data on temperature, flow rate, resistivity, or quality.

[0105] (Processing L2) In this system, an online simulator is used to automatically generate a simulation model of the current kiln. As another example, in this system, an online simulator and user input information (trigger information) can also be used to generate a simulation model that reproduces the state of the kiln at any point in time.

[0106] Here, for example, if both automatic and manual functions are possible, it can also be considered that (processing L2) includes (processing L2a) automatic functions and (processing L2b) manual functions.

[0107] (Processing L3) In this system, with reference to the simulation model, a case study simulation with arbitrarily changed operating conditions is executed.

[0108] (Process L4) In this system, the optimal operating conditions derived from the simulation are reflected in the actual machine and returned to the process described above (Process L1).

[0109] The internal temperature of a glass melting furnace or the convection of molten glass varies daily due to various factors such as the raw materials and the condition of the refractory bricks, affecting the quality of the produced glass. Whenever these factors change, it is necessary to re-establish optimal operating conditions, but this adjustment takes time and results in reduced production during the process.

[0110] The interior of a glass melting furnace is at a high temperature of 1600°C, making it difficult to understand the internal state in detail. Therefore, simulation technology of the glass melting process has been used for a long time. However, collecting the required data is time-consuming and it is difficult to execute simulations in a timely manner.

[0111] To address this issue, a simulation tool was developed that can identify changes in internal temperature distribution and convection of molten glass from simulation models generated using operational data from glass melting furnaces.

[0112] As a result, process technicians can easily perform the detailed understanding of the internal state of the glass melting process and the prior exploration of production conditions that previously required time and effort from dedicated personnel.

[0113] Figure 2A This is a diagram showing an outline of an online simulation of the implementation method.

[0114] This online simulation constructs a simulation model C1 that automatically (or manually triggered) reproduces the actual machine state.

[0115] In the actual machine manufacturing process A1, actual machine data B2 is obtained regarding actual machine B1. This data includes operational information such as production volume, gas flow rate, power consumption, pull speed, or batch ratio. Additionally, actual machine status data includes temperature, quality, or composition. The integrated data platform A3 saves (stores) actual machine data B2.

[0116] The online simulation process A2 automatically executes the processing flow when a trigger occurs.

[0117] Specifically, in simulation process A2, a data request (e.g., simulation execution request) is sent to the integrated data platform A3, data is obtained from the integrated data platform A3 and an input file is created, and the simulation is executed using simulation model C1.

[0118] At this point, for example, parameter estimation D1 can be performed based on data from the integrated data platform A3, and the estimated parameters can be used for simulation.

[0119] Furthermore, in simulation process A2, the timing for obtaining data through data requests can be arbitrary; for example, the data can be obtained at any time by issuing data requests at certain time intervals.

[0120] Additionally, in simulation process A2, the user can also specify any date and time to issue a data request. In this case, the processing following the user's data request (e.g., triggering information) will be executed automatically.

[0121] Figure 2B This is a diagram showing an outline of a digital prototype of the implementation method.

[0122] This digital prototype makes it easy to perform sensitivity analysis and preliminary simulations.

[0123] Here, we use simulation model C1 obtained through online simulation.

[0124] User E1 uses computer F1 to set simulation conditions. Server G1 then performs the simulation based on these conditions.

[0125] At this point, for example, a regression model based on real machine data can also be generated and utilized.

[0126] In addition, server G1 can be a cloud server or a locally deployed server.

[0127] User E2 confirms the simulation results via computer F2. Furthermore, user E2 and computer F2 may be the same as, for example, user E1 and computer F1, or they may be different.

[0128] Here, in server G1, simulations can be performed to estimate the current state of the kiln, simulations of condition changes, or predictions of changes in a summary model (such as a scaled-down model or a machine learning model). Furthermore, for example, a scaled-down model can be generated based on multiple simulation results using machine learning.

[0129] Such online simulation and digital prototyping technologies can solve problems on the glass manufacturing site in a short period of time.

[0130] In this embodiment, an online simulation is shown, but as another example, a structure in which the simulation is performed by an offline device (e.g., a computer) can also be used.

[0131] [Glass melting furnace]

[0132] Reference Figure 3A and Figure 3B As an example of a glass melting furnace, a glass melting kiln 100 is shown. In addition, kilns with other structures can also be used as glass melting kilns.

[0133] In addition, Figure 3A and Figure 3B Although the illustration is omitted, multiple sensors can be installed inside or outside the glass melting furnace 100 as needed. Figure 1 The example is a sensor included in one or both of sensor section 331 or sensor section 351.

[0134] Figure 3A This is a diagram showing an example of a cross-section of the glass melting furnace 100 according to an embodiment.

[0135] Figure 3B This is a diagram showing an example of another cross-section of the glass melting furnace 100 of the embodiment.

[0136] exist Figure 3A and Figure 3B For ease of explanation, the XYZ orthogonal coordinate system is shown as a three-dimensional orthogonal coordinate system.

[0137] Figure 3A The example shows a cross-sectional view of the glass melting furnace 100 in the XZ plane. Figure 3B The example shows a cross-sectional view of the glass melting furnace 100 in the YZ plane.

[0138] In addition, Figure 3A and Figure 3B In this context, the XY plane is a horizontal plane. For example... Figure 3A and Figure 3BAs shown, the glass melting furnace 100 includes: a melting tank 10 (glass melting furnace) into which glass material 70 is supplied (for ease of explanation, glass material 70 refers to glass raw material before melting, and molten glass formed by melting the glass raw material after melting, and may also include these mixed states. The same applies below); a plurality of energized electrodes 50; a plurality of thermocouples 40; an upper structure 20 covering the melting tank 10; and a plurality of burners 60.

[0139] The glass melting furnace 100 melts the glass material 70 by combustion in the burner 60 and by applying voltage to the energized electrodes 50. Figure 3B The molten glass material 70 supplied from the left end flows in the positive direction along the Y-axis, from... Figure 3B The right end is removed. Multiple burners 60 are arranged along the Y-axis (the flow direction of the glass material 70) on both sides of the upper structure 20. Each burner 60 heats the surface of the glass material 70 near its location through combustion. Therefore, by adjusting the amount of gas supplied to each of the multiple burners 60, the distribution of the heating amount on the surface of the glass material 70 (distribution in the flow direction of the glass material 70 and the width direction of the melting tank 10) can be obtained.

[0140] The energized electrodes 50 are arranged such that multiple pairs of electrodes 50 are applied along the Y-axis (flow direction of the glass material 70), one or more pairs are applied along the X-axis (width direction of the melting tank 10), and protrude from the bottom of the melting tank 10. Each pair of energized electrodes 50 heats the glass material 70 between the pairs of energized electrodes 50 by applying a voltage. Therefore, by applying voltage to each pair of energized electrodes 50, the distribution of heating amount inside the glass material 70 (distribution in the flow direction of the glass material 70 and the width direction of the melting tank 10) can be obtained. Furthermore, by adjusting the amount of gas supplied to the burner 60 and the voltage applied to the pairs of energized electrodes 50, the distribution of heating amount in the depth direction of the melting tank 10 can be obtained. Additionally, for the sake of simplicity, in... Figure 3A The example shows only one pair of energized electrodes 50, but multiple pairs can also be configured within the XZ section.

[0141] Each thermocouple 40 disposed at the bottom of the melting tank 10 measures (determines) the temperature of the glass material 70 at the bottom of the melting tank 10. In addition, each thermocouple 40 disposed on the upper structure 20, instead of measuring the surface temperature of the glass material 70, measures one or both of the upper atmosphere temperature inside the glass melting furnace 100 and the temperature of the upper structure 20.

[0142] Multiple thermocouples 40 are arranged at the bottom of the melting tank 10 along the Y-axis (the flow direction of the glass material 70) and multiple thermocouples 40 are arranged along the X-axis (the width direction of the melting tank 10). Additionally, multiple thermocouples 40 are arranged on the upper structure 20 along the Y-axis. Therefore, by using the multiple thermocouples 40 arranged at the bottom of the melting tank 10, the temperature distribution of the glass material 70 at the bottom of the melting tank 10 (the distribution in the flow direction of the glass material 70 and the width direction of the melting tank 10) can be measured. Furthermore, by using the multiple thermocouples 40 arranged on the upper structure 20, the temperature distribution of the surface of the glass material 70 (the distribution in the flow direction of the glass material 70) can be measured. Here, the temperature distribution of the surface of the glass material 70 includes not only the temperature distribution of the surface of the molten glass material 70, but may also include, for example, the temperature distribution from the surface of the molten glass material 70. Figure 3B The example shows the temperature distribution on the surface of the glass raw material supplied on the left before melting. The temperature distribution on the surface of the glass raw material before melting can be determined by configuring the glass material 70 near the supply point. Figure 3B Thermocouple 40 on the left end is used for measurement.

[0143] By detecting the current (or resistance value) between the pairs of energized electrodes 50, the temperature of the glass material 70 between those pairs can be measured. Therefore, by using multiple pairs of energized electrodes 50, the temperature distribution of the glass material 70 in the middle of the melting tank 10 (the distribution in the flow direction of the glass material 70 and the width direction of the melting tank 10) can be measured. Furthermore, by combining two or more of the thermocouples 40 disposed at the bottom of the melting tank 10, the thermocouples 40 disposed on the upper structure 20, and the pairs of energized electrodes 50, the temperature distribution in the depth direction of the melting tank 10 can be obtained. Additionally, by analyzing the video footage captured by a camera (not shown), for example, the flow velocity distribution on the surface of the glass material 70 can be obtained. Here, such analysis is performed, for example, by detecting the movement speed of bubbles or the like inside the glass material 70.

[0144] [Example of information processing]

[0145] Reference Figures 4-6 This illustrates an example of the processing steps performed in the information processing apparatus 111. In this example, a case is shown where time-series data is obtained as measurement data R1 of a physical machine.

[0146] In this example, it is assumed that the measurement data R1, the glass melting furnace simulation model R11, the calibration parameter estimation model R12, and the prediction model R13 are obtained in advance.

[0147] Figure 4 This diagram illustrates an example of the processing steps performed in the information processing apparatus 111 of the embodiment.

[0148] (Step S1)

[0149] In the information processing device 111, the control unit 215 acquires the actual machine data. Then, it moves to the processing in step S2.

[0150] In this example, the actual data is time-series measurement data R1.

[0151] (Step S2)

[0152] In the information processing device 111, the correction parameters are estimated by the correction parameter estimation unit 272. Then, the process moves to step S3.

[0153] In this example, the correction parameter estimation unit 272 uses the measurement data R1 and the correction parameter estimation model R12 to estimate the correction parameter.

[0154] (Step S3)

[0155] In the information processing device 111, the state estimation unit 271 performs a state estimation of the actual machine. Then, the process moves to step S4.

[0156] In this example, the state estimation unit 271 uses the measurement data R1, the glass melting furnace simulation model R11, and the estimated correction parameters to perform state estimation of the actual machine.

[0157] As a state estimate, for example, the state is estimated for a specified period (e.g., the date of year, month, or day, or the date and time). In this case, the estimation of the correction parameter is also performed by estimating the value of that period (e.g., the date of year, month, or day, or the date and time). Furthermore, the period can also be the time since the actual machine began to be used (e.g., years).

[0158] Alternatively, as another example, the specified period may not be directly specified, but rather indirectly specified through other conditions. For example, the specified temperature of a given location (e.g., the temperature of molten glass) could be a specified value.

[0159] (Step S4)

[0160] In the information processing device 111, an operational change guidance plan is obtained based on the results of the actual machine's state estimation. Then, the process moves to step S5.

[0161] In this example, the user explores operational change guidance schemes based on the results of the actual machine's state estimation, and inputs the determined operational change guidance schemes into the information processing device 111 through the operation unit 231.

[0162] As another example, the structure for processing operation change guidance schemes can also be explored by using the results of state estimation based on the actual machine automatically performed by the control unit 215 and the like in the information processing device 111.

[0163] Here, as a guideline for operational changes, various guidance schemes can be adopted, such as considering one or more of the following: increasing or decreasing the amount of gas used, increasing or decreasing the amount of electricity used, adjusting the convection conditions by changing the flow rate of the bubbler, or improving the glass quality.

[0164] (Step S5)

[0165] In the information processing device 111, the model calculation unit 273 performs optimization calculations or sensitivity analysis based on the operation change guidance scheme, using the glass melting furnace simulation model R11 or a summary model based on the glass melting furnace simulation model R11. Then, the process moves to step S6.

[0166] Here, the calculation is performed in a manner consistent with the operational change guidance scheme.

[0167] Alternatively, when using a summary model, for example, the model calculation unit 273 generates the summary model based on the simulation result R14 of the glass melting furnace simulation model R11.

[0168] Furthermore, in the first round of processing steps S4 to S7, for example, change prediction can be performed using condition change simulation, or change prediction can be performed using a summary model (e.g., a reduced model or a machine learning model).

[0169] (Step S6)

[0170] In the information processing device 111, the output unit 212 outputs information related to the calculation results of the optimization calculation or sensitivity analysis. Then, the process moves to step S7.

[0171] In this example, the information is displayed on the screen of display unit 251. The user can then evaluate the information.

[0172] Furthermore, as another example, if the information processing device 111 performs the evaluation automatically, the information may not be output.

[0173] (Step S7)

[0174] In the information processing device 111, the control unit 215, based on the evaluation results, moves to step S8 if it determines that the goal has been achieved, and moves to step S4 if it determines that the goal has not been achieved.

[0175] In this example, the user inputs whether the goal has been achieved into the information processing device 111 by operating the operation unit 231, and the control unit 215 determines whether the goal has been achieved.

[0176] As another example, the information processing device 111 can also be used to automatically determine whether the target has been achieved.

[0177] (Step S8)

[0178] Once the operation is judged, the process ends.

[0179] In addition, in this example, the operation is determined by the user, but in other examples, the information processing device 111 may perform the same processing as the operation determination.

[0180] Reference Figure 5 and Figure 6 This shows an example of the correction parameter estimation process.

[0181] Figure 5 The example is an example using a function (non-machine learning).

[0182] Figure 6 The example is an example of using machine learning.

[0183] <An example of correction parameter estimation processing>

[0184] Figure 5 This is a diagram illustrating an example of the steps in the correction parameter estimation process performed in the information processing apparatus 111 of the embodiment.

[0185] In general, the correction parameter estimation model is constructed in steps S31 to S32, and the correction parameter estimation model is used to perform correction parameter estimation in steps S33 to S34.

[0186] (Step S31)

[0187] In the information processing device 111, the control unit 215 acquires the actual machine data. Then, it proceeds to the processing in step S32.

[0188] In this example, time-series measurement data R1 is used as the actual machine data.

[0189] (Step S32)

[0190] In the information processing device 111, the control unit 215 constructs a correction parameter estimation model using a function.

[0191] Here, in the model for estimating the correction parameters using a function, the correction parameters are obtained through the function, for example, based on the simulation conditions. That is, the function takes the conditions as variables and the correction parameters as solutions.

[0192] (Step S33)

[0193] In the information processing device 111, the control unit 215 inputs information that constitutes the conditions for simulation. Then, the process proceeds to step S34.

[0194] Here, the information can be input to the information processing device 111 by the user of the operation unit 231, or it can be input to the information processing device 111 by other devices.

[0195] This condition can be, for example, date and time (or year, month, day, or month, day, etc.), or temperature, etc.

[0196] (Step S34)

[0197] In the information processing device 111, the correction parameter estimation unit 272 performs correction parameter estimation processing based on the correction parameter estimation model using the function.

[0198] As a result, the estimated result of the correction parameter is obtained in the information processing device 111.

[0199] <Other examples of correction parameter estimation processing>

[0200] Figure 6 This is a diagram illustrating another example of the correction parameter estimation process performed in the information processing apparatus 111 of the embodiment.

[0201] In general, the correction parameter estimation model is constructed in steps S71 to S73, and the correction parameter estimation model is used to perform correction parameter estimation in steps S74 to S76.

[0202] (Step S71)

[0203] In the information processing device 111, the control unit 215 (e.g., the state estimation unit 271) performs a simulation for building a database to create a correction parameter estimation model. Then, the process moves to step S72.

[0204] (Step S72)

[0205] In the information processing device 111, the simulation results are obtained by the control unit 215. Then, the process moves to step S73.

[0206] (Step S73)

[0207] In the information processing device 111, the control unit 215 constructs a correction parameter estimation model using machine learning.

[0208] Here, in the calibration parameter estimation model using machine learning, the value of the calibration parameter is determined based on the output value from the model, for example, by inputting specified information into the machine learning model.

[0209] In this example, the model takes the simulated conditions and the actual data (in this example, the time series measurement data R1) as input, and outputs either the value of the correction parameter itself or related information used to obtain the value of the correction parameter.

[0210] (Step S74)

[0211] In the information processing device 111, the control unit 215 inputs information that constitutes the conditions for simulation. Then, the process proceeds to step S75.

[0212] Here, the information can be input to the information processing device 111 by the user of the operation unit 231, or it can be input to the information processing device 111 by other devices.

[0213] This condition can be, for example, date and time (or year, month, day, or month, day, etc.), or temperature, etc.

[0214] (Step S75)

[0215] In the information processing device 111, the control unit 215 collects (acquires) the actual machine data. Then, the process moves to step S76.

[0216] In this example, time-series measurement data R1 is used as the actual machine data.

[0217] (Step S76)

[0218] In the information processing device 111, the correction parameter estimation unit 272 performs correction parameter estimation processing based on a correction parameter estimation model using machine learning.

[0219] As a result, the estimated result of the correction parameter is obtained in the information processing device 111.

[0220] <Specific examples of correction parameter estimation processing>

[0221] Reference Figure 7 , Figure 8 ,and Figure 9A Figure 9C shows a specific example of the correction parameter estimation process.

[0222] Figure 7This is a diagram of Table 3011 showing examples of correction parameters for an implementation method. Here, in this example, examples of correction parameters are shown in tabular form, but tabular form is not required.

[0223] Table 3011 shows a summary of the correction parameters.

[0224] Table 3011 stores the items of the correction parameters, the minimum value (Min) related to the limit, the maximum value (Max) related to the limit, and the units accordingly.

[0225] Furthermore, the stored information in Table 3011 is an example to illustrate a summary and is not actual information. In this example, multiple calibration parameters are set. Moreover, these multiple calibration parameters may include items that are difficult to measure in real time (actual measurement).

[0226] exist Figure 7 In the example, for instance, the correction parameters for the project "Heat Dissipation AAA" are stored as Min = "..." ("..." represents the corresponding numerical value, the same below), Max = "...", and the unit is "[W / m 2 ·K]”.

[0227] Additionally, for example, the correction parameters for the project "Heat Transfer AA" are stored with Min as "...", Max as "...", and the unit as "[W / m"]. 2 ·K]”. Additionally, for example, regarding the correction parameters for the project “air volume aaa”, it stores Min as “…”, Max as “…”, and the unit as “[Nm]”. 3 Similarly, in Figure 7 In the example, information about the correction parameters for other items is stored.

[0228] In this example, the correction parameter estimation unit 272 estimates (determines) the value contained in the range between the minimum (Min) and maximum (Max) involved in each limit, as the estimated value of each correction parameter.

[0229] In addition, as a limitation, for example, only the minimum value (Min) or only the maximum value (Max) can be set.

[0230] Such restrictions can be set based on past performance, for example.

[0231] Furthermore, it is not necessary to set such a restriction. Without setting such a restriction, the correction parameter estimation unit 272 can also estimate (determine) any value as the estimated value of each correction parameter.

[0232] Figure 8 This is a diagram schematically illustrating the first process in a specific example of the correction parameter estimation process of the implementation method.

[0233] Figure 9A This is a diagram schematically illustrating the result of the first process in a specific example of the correction parameter estimation process of the implementation method.

[0234] Figure 9B This is a diagram schematically illustrating the second process in a specific example of the correction parameter estimation process of the implementation method.

[0235] The names "first processing" and "second processing" are used here for ease of explanation, and any other names may be used.

[0236] (First treatment)

[0237] In the first process, the information processing device 111 performs a simulation under comprehensive operating conditions and generates a machine learning model based on the simulation results.

[0238] In this example, we show the case where the machine learning model generates temperature predictions.

[0239] In addition, as another example, machine learning models can also be generated to predict physical quantities other than temperature.

[0240] In the first process, the experimental planning method execution unit 274 uses the experimental planning method to perform a simulation of the glass melting furnace simulation model R11 under comprehensive conditions.

[0241] These conditions include, for example, the operating conditions of the glass melting furnace 311, as well as the conditions of multiple correction parameters (in this example, the values ​​of multiple correction parameters).

[0242] The operating conditions of the glass melting furnace 311 are, for example, conditions related to heat input. Conditions related to heat input include, for example, the amount of gas used, the output of the electric flux, or the control method of the bubbler.

[0243] Alternatively, as a way to achieve comprehensive conditions, one could use random numbers to mechanically determine the combination of multiple parameters within a predetermined range (between the upper and lower limits), thereby covering (or roughly covering) that range. Another approach is to use other conditions in addition to such random numbers.

[0244] exist Figure 8 In the diagram, the overall set of conditions U1 is shown in boxes, and each of the multiple conditions (simulated conditions) contained in this overall set U1 is shown in dots. Figure 8 In the example, to simplify the illustration, a label is added to represent a single condition V1.

[0245] Here, since the total number of conditions contained in the overall set U1 may be infinite, the experimental planning execution unit 274, for example, performs simulations under a finite number of conditions. A simulation is performed once for each condition.

[0246] This generates a database containing multiple simulation results.

[0247] The prediction model generation unit 275 generates a prediction model R13 based on the results of simulations performed for each of the multiple conditions.

[0248] Here, in this example, the predictive model R13 is used to generate a machine learning model.

[0249] exist Figure 8 The diagram illustrates how, within a set of simulation results (simulation result group W1), machine learning is used to generate predictions between the simulation results (in...). Figure 8 The example is a machine learning model that simulates (pseudo-simulations) between points (point-to-point).

[0250] The generated prediction model R13 can predict simulation results (pseudo-simulation results) corresponding to any conditions.

[0251] exist Figure 8 The diagram shows the relationship between the value (predicted value) and the actual value (measured value) of the specified temperature of the glass melting furnace 311, as predicted by the prediction model R13 (in this example, a machine learning model for temperature prediction).

[0252] In addition, Figure 8 The example illustrates the case for a given temperature, but predictions for multiple temperatures can also be made using the prediction model R13. In this case, for each of the multiple temperatures, the following is obtained: Figure 8 The chart shown.

[0253] Thus, in this example, a high-precision prediction model R13 is generated.

[0254] Then, by using such a prediction model R13, it is possible to quantify the sensitivity of the manipulation variables and correction parameters to the simulation results.

[0255] Furthermore, this sensitivity can be a value like the weight in a weighted average, or it can also be referred to as the weight.

[0256] (Second Processing)

[0257] In the second process, the actual conditions (conditions of glass melting furnace 311) at any time point are imported into the prediction model R13 to estimate the actual values.

[0258] exist Figure 9A The diagram shows a glass melting furnace model (glass melting furnace model M1), three heat sources H1 to H3, three temperatures I1 to I3, and two correction parameters J1 to J2. Although represented in two dimensions in the diagram, the glass melting furnace model M1 is a three-dimensional model (e.g., a model in an XYZ orthogonal coordinate system). For example, the temperatures I1 to I3 are represented by multiple locations (observation points) arranged in a three-dimensional configuration.

[0259] Furthermore, the number of heat sources, temperatures, and correction parameters in this example are merely examples and are not limited to this.

[0260] Based on the results of the first processing, the change sensitivity evaluation unit 276 can quantify the change sensitivity values ​​(sensitivity coefficients) of each variable according to the simulation results.

[0261] The observation point data acquisition unit 277 acquires data from each observation point (in this example, the temperature data of the molten glass).

[0262] exist Figure 9A In the example, sensitivity w1 is the sensitivity of heat source H1 to temperature I1, sensitivity w2 is the sensitivity of heat source H1 to temperature I2, and sensitivity w3 is the sensitivity of heat source H1 to temperature I3.

[0263] In addition, for other heat sources (heat source H2, heat source H3), there is also sensitivity for each of the multiple temperatures I1 to I3, but the method for calculating the sensitivity is the same as that for heat source H1. In this example, the illustration and detailed explanation are omitted.

[0264] In addition, Figure 9A In the example, sensitivity w11 is the sensitivity of correction parameter J1 to temperature I1, sensitivity w12 is the sensitivity of correction parameter J1 to temperature I2, and sensitivity w13 is the sensitivity of correction parameter J1 to temperature I3.

[0265] In addition, there is sensitivity for each of the multiple temperatures I1 to I3 for other correction parameters (correction parameter J2), but the method for determining the sensitivity is the same as that for correction parameter J1. In this example, the illustration and detailed description are omitted.

[0266] Here, in Figure 9A In the example, the heat of heat sources H1 to H3, temperatures I1 to I3, and correction parameters J1 to J2 are known in the experimental planning method; only the sensitivity is unknown. Therefore, the sensitivity needs to be determined.

[0267] Figure 9B The text shows that in Figure 9A The diagram shown illustrates the values ​​of each sensitivity level within the structure.

[0268] exist Figure 9B In the example, the sensitivity w1 is 0.6, the sensitivity w2 is 0.3, the sensitivity w3 is 0.1, the sensitivity w11 is 0.4, the sensitivity w12 is 0.4, the sensitivity w13 is 0.2, and other sensitivities are also shown in the figure.

[0269] In this practical test, the heat and temperature of heat sources H1 to H3 are known, only the correction parameters J1 to J2 are unknown. Therefore, the correction parameters J1 to J2 are calculated using the obtained sensitivity.

[0270] Therefore, the correction parameter estimation unit 272 estimates (presumably calculates) the correction parameters J1 to J2.

[0271] A specific example of the correction parameter estimation process in the implementation method will be described.

[0272] For ease of explanation, the explanation will be divided into the first mathematical expression part K1 and the second mathematical expression part K2.

[0273] The following steps will proceed in the order of (processing Q1) to (processing Q4).

[0274] (Handling Q1)

[0275] In this example, it is assumed that simulations under dozens to hundreds of conditions were performed using the experimental planning method. These conditions include the operating conditions of the glass melting furnace 311 and the conditions of the correction parameters, i.e., the values ​​of the correction parameters are also used as variables.

[0276] In the first mathematical expression K1, the symbols (x1, x2, x3, ...) with numerical values ​​appended to x represent the operational variables, and their coefficients (a1, a2, ..., b1, b2, ..., c1, c2, ..., etc.) represent their respective sensitivities.

[0277] In addition, as operational variables, there may be variables related to gas, EB output (electro-assisted melting output), or bubbler flow rate.

[0278] The symbols (p1, p2, p3, ...) appended to p represent correction parameters, and their coefficients (α1, α2, ..., β1, β2, ..., γ1, γ2, ..., etc.) represent their respective sensitivities.

[0279] In addition, as correction parameters, there may be factors such as furnace wall heat dissipation or tertiary air volume.

[0280] Additionally, at this stage, the values ​​used in the experimental planning method are substituted in as correction parameters.

[0281] Additionally, the numerical symbols appended to T (T1, T2, T3, ...) indicate the temperature of the observation point to be fitted to the actual results.

[0282] In addition, this temperature may include, for example, the furnace bottom temperature or the furnace top temperature.

[0283] (Handling Q2)

[0284] Here, the prediction model R13 is generated in a way that allows for the creation of the same number of formulas as the number of observation points (the number of temperatures) that are to be fitted in the first mathematical formula K1.

[0285] In the first mathematical formula K1, there is the following formula, and for combinations of multiple coefficients, there is such a formula: the sum of the product of each operand and each coefficient and the product of each correction parameter and each coefficient equals the temperature of the observation point.

[0286] In the first mathematical expression K1, the sensitivities (coefficients) are unknown. These sensitivities are determined through regression analysis using a dataset obtained through simulation. The same number of regression analyses are performed for the locations of the temperatures to be fitted. That is, as a specific example, if there are 5 locations of the temperatures to be fitted, 5 regression analyses are performed.

[0287] Furthermore, in this example, linear models were used for the expressions that form the first mathematical part K1, but nonlinear models could also be used in other examples.

[0288] Alternatively, other analytical methods can be used instead of regression analysis. For example, Bayesian estimation can be used instead of maximum likelihood estimation.

[0289] (Handling Q3)

[0290] Next, after calculating each sensitivity in the first mathematical expression K1, values ​​other than the correction parameters (p1, p2, p3, ...) are substituted with values ​​based on actual machine data. In this case, the terms concerning the operation variables (x1, x2, x3, ...) become constant terms, and the correction parameters (p1, p2, p3, ...) become unknown.

[0291] Then, as the left side of the second mathematical expression K2, imagine the product of a matrix containing all coefficients related to multiple correction parameters and a vector of multiple correction parameters, and as the right side of the second mathematical expression K2, imagine the term (ΔT1, ΔT2, ΔT3, ...).

[0292] Therefore, the problem boils down to solving a system of equations using the matrix shown in the second mathematical expression K2. Here, in the second mathematical expression K2, the terms on the right-hand side (ΔT1, ΔT2, ΔT3, ...) are constants.

[0293] Furthermore, in this example, the data used as actual machine data, whether or not it uses a period (such as a date in years, months, or months, or a date and time), can be arbitrary.

[0294] (Handling Q4)

[0295] Then, in this example, the unknown correction parameters are estimated using either maximum likelihood estimation or Bayesian estimation.

[0296] In this embodiment, the case where the temperature of the molten glass is used as a reference for adjusting the correction parameter (i.e., adjusting it to an appropriate value) is shown. However, as other examples, the resistivity of the molten glass, the current value between the heating electrodes, the voltage value between the heating electrodes, the resistance value between the heating electrodes, the flow rate of the molten glass, etc., may also be used, or any combination of two or more items may be used.

[0298] <Example of the estimation results for the correction parameters>

[0299] Figure 10 The figure in Table 3211 shows an example of the estimation results of the correction parameters representing the implementation method.

[0300] Here, in this example, an example of the correction parameters is shown in tabular form, but it is not necessary to use tabular form.

[0301] Table 3211 shows a summary of the estimation results for the correction parameters.

[0302] Table 3211 stores the symbols (p1, p2, ...) representing the correction parameters, the items of the correction parameters, and the estimated values ​​of the correction parameters.

[0303] Furthermore, the storage information in Table 3211 is an example to show a summary and is not actual information.

[0304] exist Figure 10 In the example, for instance, the correction parameter for the item "Heat Dissipation AAA" represented by the symbol "p1" is stored as an estimated value of "..." ("..." represents the corresponding numerical value. The same applies below).

[0305] Additionally, for example, the correction parameter for the item "Heat Transfer AA" represented by the symbol "p11" is stored with an estimated value of "...".

[0306] Additionally, for example, the correction parameter for the item "air volume aaa" represented by the symbol "p21" has an estimated value of "...".

[0307] Similarly, in Figure 10 In the example, information about the correction parameters for other items is stored.

[0308] exist Figure 10 In the example, the estimated values ​​of each correction parameter are estimated within the actual performance range, and are included in... Figure 7 The values ​​shown are within the range of minimum (Min) and maximum (Max) values ​​associated with each constraint.

[0309] Figure 11 This is a graph comparing the characteristics of the simulation results estimated using the correction parameters of the implementation method with the characteristics of the actual machine data. This comparison is not the actual result itself, but is schematically fabricated for illustrative purposes.

[0310] exist Figure 11 The example shows a comparative example of the temperature at a specified location.

[0311] exist Figure 11 In the chart shown, the horizontal axis represents the temperature measurement points, and the vertical axis represents the temperature itself. Furthermore, the specific scale markings for both the horizontal and vertical axes are omitted.

[0312] In this chart, points represent the temperature characteristics of the simulation results, and lines represent the temperature characteristics of the actual machine (characteristic 3521).

[0313] In addition, Figure 11 In the example, to simplify the illustration, only one point (characteristic 3511) among the points of multiple simulation results is labeled.

[0314] exist Figure 11 In the example, the simulation results, relative to the actual temperature characteristic 3521, can simultaneously and roughly reproduce the temperature levels and distribution of multiple observation points, which is a characteristic with good accuracy.

[0315] As described above, in the glass manufacturing system 1 of the embodiments, for example, regarding glass manufacturing, by realizing automated simulation, it is possible to achieve simulation-based assistance efficiency compared to the prior art (e.g., the case where simulation is performed by process technicians and analysis specialists).

[0316] For example, in the existing technology (e.g., simulations performed by process technicians and analysis specialists), heat dissipation is set based on actual kiln information through a comparison of actual results over about 1 to 2 months, and then the basic flow results of the glass melting furnace are estimated by performing a simulation over about 1 to 2 weeks, which takes a considerable amount of time.

[0317] In contrast, in the information processing apparatus 111 of the embodiment, based on information about the actual kiln, the heat dissipation and other parameters can be set through a short-term (e.g., about tens of minutes) comparison of actual results, and then the basic flow results of the glass melting furnace can be estimated through a short-term (e.g., about a few hours) simulation execution. Furthermore, these required times are illustrative examples and are not necessarily limited to these times.

[0318] Therefore, in the glass manufacturing system 1 of the embodiment, anyone can obtain the evaluation results of the cyclic state of the molten glass in a timely manner.

[0319] In the glass manufacturing system 1 of the embodiment, when an automated simulation is performed in the information processing device 111, the accuracy of the simulation can be improved, for example, even if there are aging changes in the glass melting furnace, by estimating the value of the correction parameter of the simulation and setting it to an appropriate value.

[0320] In this embodiment, for example, when manufacturing glass using a glass melting furnace via the float glass process, the operation of the glass manufacturing process can be assisted. Generally, the state control and management of the glass melting furnace is important in the raw material melting process of glass manufacturing, and this embodiment can provide such assistance.

[0321] In the glass manufacturing system 1 of the embodiment, for example, a time series data management process, a real machine data acquisition process, a correction parameter estimation process, and a state estimation simulation process are performed.

[0322] Furthermore, in this embodiment, the time-series data management process is performed by the glass manufacturing apparatus 131 and the database 112, while the other processes are performed by the information processing apparatus 111.

[0323] In the time-series data management process, to measure various quantities related to molten glass, measurement data from multiple sensors installed inside or outside the glass melting furnace (e.g., melting tank) are collected, and the collected measurement data is stored as time-series data in a database. Alternatively, other data formats can be used instead of time-series data (time-series measurement data). Measurement data can be, for example, measurements taken during glass manufacturing processes in an actual machine (an actual glass manufacturing apparatus).

[0324] In the process of acquiring real machine data, the data (real machine data) required for subsequent processing is obtained from the time series data stored in the time series data management process.

[0325] In the calibration parameter estimation process, the calibration parameters of the simulation model (a simulation model that models the molten furnace) are estimated using the calibration parameter estimation model and the actual machine data obtained in the actual machine data acquisition process.

[0326] In the state estimation simulation process, the simulation model, the correction parameters estimated in the correction parameter estimation process, and the time series data stored in the time series data management process (e.g., the actual machine data obtained in the actual machine data acquisition process) are used to perform the state estimation of the actual machine.

[0327] In the glass manufacturing system 1 of the embodiment, the process of constructing a correction parameter estimation model can also be performed.

[0328] In the calibration parameter estimation model construction process, a calibration parameter estimation model is constructed to estimate the calibration parameters of the simulation model (a simulation model that models the molten furnace).

[0329] In the glass manufacturing system 1 of the embodiment, an optimization calculation process can also be performed. In the optimization calculation process, based on the operational change guidance scheme explored according to the results of the state estimation simulation process (simulation results), optimization calculations or sensitivity analyses are performed using a simulation model or a summary model constructed based on the analysis results of the simulation model.

[0330] In the glass manufacturing system 1 of the embodiment, visualization processes can also be performed.

[0331] In the visualization process, the specified information is visualized (e.g., displayed).

[0332] The specified information may be, for example, data generated as the result of a state estimation simulation process, or data generated as the result of an optimization calculation process, or both, or other data.

[0333] Here, various types of data can be used as data that can be used for estimation (e.g., time series data), such as temperature data measured by a standing thermocouple or data during the operation of the melting furnace.

[0334] For example, in this embodiment, the correction parameters can be estimated even if the input information is only date and time information, during the correction parameter estimation process.

[0335] In addition, various data can be used as correction parameters for the estimated object, such as data related to the heat dissipation of the furnace wall, or data related to the erosion of the furnace material (e.g., the amount of soil, throat, etc.).

[0336] In addition, any method can be used as an estimation method, such as a regression model that utilizes kiln survey data.

[0337] Furthermore, the simulation model (method) can be adopted in various ways, such as using a model (method) that couples two or more elements together (forming an analytical model). Moreover, this model (method) can be, for example, two-dimensional or three-dimensional, and can be a steady-state analysis or a non-steady-state analysis.

[0338] This element may include, for example, a raw material melting simulation that models the melting behavior of glass raw materials, a thermal fluid simulation of molten glass, a combustion simulation of the upper combustion space, a simulation of the potential and current distribution within the molten glass based on an electric heating device, and a steady-state analysis of heat transfer simulation inside the furnace material.

[0339] Alternatively, for example, a model obtained by summarizing the simulation model (a summary model) can be used instead of the original simulation model.

[0340] As an example of reduction, the Reduced Order Model (ROM) can also be used.

[0341] In addition, as a simulation model (method), for example, 1D-CAE, circuit network model, or system model can be used.

[0342] Alternatively, as a model (method) for simulation, deep learning methods such as PINNS (Physics Informed Neural Nets) can also be used.

[0343] In addition, various execution environments can be used as the simulated execution environment, such as client PCs and workstations, cluster machines, cloud services, etc.

[0344] In the glass manufacturing system 1, for example, a predictive model generation process can also be performed to generate a predictive model of the simulation results by using experimental planning and machine learning.

[0345] In the predictive model generation process, the operating conditions of the glass melting furnace and the simulation correction parameters are used as set variables. Within the range of conditions envisioned in the actual glass manufacturing process, multiple simulations are performed based on the experimental planning method.

[0346] Furthermore, in the predictive model generation process, a machine learning model is constructed by using the simulation results data to evaluate the sensitivity (variation sensitivity) of the influence of the operating conditions of the glass melting furnace and the simulation correction parameters on the simulation results of specified values ​​at multiple observation points of the glass melting furnace. Moreover, the specified values ​​can be, for example, the molten glass temperature, or the molten glass flow rate, or a combination of both.

[0347] Alternatively, in the predictive model generation process, the simulation results data can be stored in a database. In this case, the data stored in the database can be used for machine learning to build a machine learning model.

[0348] In the glass manufacturing system 1 of the embodiment, for example, a process of determining the correction parameters (unknown variables) by inputting actual machine data into the prediction model can also be performed.

[0349] Here, data related to an existing glass melting furnace are used. This existing glass melting furnace, for example, has the same shape as the hypothetical glass melting furnace in the simulation (or a similar shape that the simulation can approximate), and is equipped with multiple sensors for measuring various quantities related to the molten glass.

[0350] In the calibration parameter determination process, the operating conditions at any time point and the specified value at that time point are obtained at multiple observation points (in this embodiment, at least one of the molten glass temperature and the molten glass flow rate).

[0351] Furthermore, in the calibration parameter determination process, based on the results of the sensitivity evaluation, the calibration parameters are optimized to the most probable values ​​by minimizing the deviation between the specified values ​​obtained at the aforementioned observation points and the values ​​obtained through simulation (the values ​​corresponding to the specified values). Thus, the calibration parameters are determined.

[0352] The parameter estimation method in this process can use Bayesian estimation instead of maximum likelihood estimation. The estimation algorithm can utilize not only machine learning methods, but also computational methods that apply approximations such as polynomials or exponential functions formulated based on prior knowledge of the relevant process.

[0353] In the glass manufacturing system 1 of the embodiment, a limited range can be set for the possible values ​​of the estimated results of the correction parameters. For example, the range of optimized values ​​can be limited by time to calculate or reflect the changes in heat dissipation of the furnace material caused by the aging of the glass melting furnace, which are formulated from experimental or past measurement results.

[0354] This constraint could be, for example, a minimum value, a maximum value, or both.

[0355] (Second Implementation)

[0356] The second embodiment will be described.

[0357] [Variation: Example of omitting the estimation of correction parameters]

[0358] Here, in the first embodiment, the process is shown. Figure 4The process shown is step S2 (processing for estimating correction parameters). However, as a variation, a structure that does not perform step S2 (processing for estimating correction parameters) can also be adopted. In this case, the information processing device 111 may not have the function and data for estimating correction parameters.

[0359] In this variation, Figure 4 In the processing flow shown, step S3 is performed after step S1.

[0360] In this variation, no correction parameter estimation is performed; for example, the value equivalent to the correction parameter is kept constant.

[0361] Figure 12 This is a diagram schematically illustrating an example of characteristics based on simulation results without corrected parameter estimation processing. Furthermore, Figure 12 The examples are not necessarily actual examples, but rather examples that illustrate the general idea.

[0362] exist Figure 12 In the example, for a specific temperature, characteristics 1011 based on simulation results and characteristics 1021 based on actual machine data are shown.

[0363] exist Figure 12 In the chart shown, the horizontal axis represents time (over time), and the vertical axis represents temperature. Detailed scale markings for both the vertical and horizontal axes are omitted.

[0364] like Figure 12 As shown in the example, without correction parameter estimation, the deviation relative to characteristic 1021 based on real machine data becomes larger compared to the case with correction parameter estimation.

[0365] That is, for example, even if the most recent actual conditions are input into a glass melting furnace simulation model with calibration parameters fixed at past time points, it is still impossible to reproduce the temperature and other parameters of the most recent actual machine with high accuracy. This is considered to be mainly due to the significant impact of aging changes such as heat dissipation in the furnace. For example, due to aging changes, the surface temperature and cross-sectional temperature of the furnace increase, which not only reduces the temperature but also decreases the accuracy of simulation results such as flow rate.

[0366] Furthermore, in situations where the effects of such aging changes are minimal (e.g., close to...), Figure 12 Under the conditions of performance comparison shown, no comparison is made. Figure 4 The structure of the processing in step S2 (processing for correcting parameter estimation) shown is also valid.

[0367] As an example, for a period when the impact of aging changes is not significant, a period of several months or so can be considered from the time of performance comparison, but it is not necessarily limited to this.

[0368] Reference Figure 13 Examples of aging changes associated with glass melting furnaces are shown. Furthermore, these are examples illustrating the general outline, omitting detailed descriptions.

[0369] Figure 13 This is a diagram illustrating an example of the aging changes in heat dissipation from the furnace wall of a glass melting furnace.

[0370] exist Figure 13 In the chart shown, the horizontal axis represents the number of months of operation, and the vertical axis represents the heat dissipation from the furnace wall.

[0371] The graph shows the average characteristic 1221 representing the characteristics of multiple kilns (e.g., more than 5 kilns) consisting of glass melting furnaces.

[0372] exist Figure 13 In the example, it can be seen that the heat dissipation of the furnace wall increases with aging.

[0373] Here, by formulating such aging changes, the furnace wall heat dissipation at any given time can be calculated and used as a correction parameter. This method of setting correction parameters is not limited to furnace wall heat dissipation; it can also be applied to other aspects.

[0374] As described above, in this embodiment, the accuracy of the simulation may be reduced compared to the structure that performs correction parameter estimation processing as in the first embodiment. However, in this embodiment, with respect to glass manufacturing, by realizing automated simulation, it is possible to achieve simulation-based assistance efficiency compared to the prior art (e.g., the case where simulation is performed by process technicians and analysis specialists).

[0375] For example, in the existing technology (e.g., simulations performed by process technicians and analysis specialists), the heat dissipation is set based on information from the actual kiln by comparing it with actual results for about 1 to 2 months, and then the basic roughness result (basic flow result) of the glass melting furnace is estimated by performing a simulation for about 1 to 2 weeks, which takes a considerable amount of time.

[0376] In contrast, in this embodiment, based on information from the actual kiln, the heat dissipation, etc., can be set through short-term (e.g., around tens of minutes) performance comparisons. Then, the basic roughness result (basic flow result) of the glass melting furnace can be estimated through short-term (e.g., around several hours) simulation executions. Furthermore, these required times are illustrative examples and are not necessarily limited to these times.

[0377] Therefore, in this embodiment, anyone can obtain timely evaluation results of the cyclic state of the molten glass.

[0378] (Regarding the above implementation methods)

[0379] Furthermore, a program for implementing the functions of any component in any of the devices described above can be recorded on a computer-readable recording medium, and the program can be read into a computer system and executed. In addition, the term "computer system" as used here includes hardware such as an operating system or peripheral devices. Furthermore, "computer-readable recording medium" refers to portable media such as floppy disks, optical disks, ROMs, CDs (Compact Discs)-ROMs (Read Only Memory), and hard disks built into a computer system. Moreover, "computer-readable recording medium" also includes media that retain programs for a certain period of time, such as volatile memory within a computer system that acts as a server or client when the program is transmitted via a network such as the Internet or a communication line such as a telephone line. This volatile memory can be, for example, RAM (Random Access Memory). The recording medium can be, for example, a non-transitory recording medium.

[0380] Furthermore, the aforementioned program can also be transmitted from a computer system storing the program in a storage device or the like to other computer systems via a transmission medium or through transmission waves in the transmission medium. Here, "transmission medium" for transmitting the program refers to a medium with information transmission capabilities, such as a network like the Internet or a communication line like a telephone line.

[0381] Alternatively, the program described above can be a part of a program used to implement the aforementioned functions. Furthermore, the program described above can also be a program that can achieve the aforementioned functions by combining with programs already recorded in the computer system, i.e., a so-called differential file. A differential file can also be called a differential program.

[0382] Furthermore, the functions of any component in any of the devices described above can also be implemented by a processor. For example, each process in the embodiments can be implemented by a processor that operates based on information such as programs, and a computer-readable recording medium that stores information such as programs. Here, the processor may be implemented by individual hardware for each part, or by integrated hardware for each part. For example, the processor includes hardware that may include at least one of circuitry for processing digital signals and circuitry for processing analog signals. For example, the processor may also be constructed using one or more circuit devices or one or more circuit elements mounted on a circuit board. As circuit devices, ICs (Integrated Circuits) may be used, and as circuit elements, resistors or capacitors may be used.

[0383] Here, the processor can be, for example, a CPU. However, the processor is not limited to a CPU; various processors such as GPUs (Graphics Processing Units) or DSPs (Digital Signal Processors) can also be used. Additionally, the processor can be, for example, an ASIC (Application Specific Integrated Circuit) based hardware circuit. Furthermore, the processor can be composed of multiple CPUs, or multiple ASIC-based hardware circuits. Additionally, the processor can be composed of a combination of multiple CPUs and multiple ASIC-based hardware circuits. Furthermore, the processor can include, for example, one or more amplifier circuits or filter circuits for processing analog signals.

[0384] The embodiments of this disclosure have been described in detail above with reference to the accompanying drawings, but the specific configuration is not limited to these embodiments, and also includes designs that do not depart from the spirit of this disclosure.

[0385] [Postscript]

[0386] Examples of structure 1 to 11 are shown.

[0387] (Structure Example 1)

[0388] An information processing device includes a state estimation unit that uses measurement data stored in a measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform state estimation of the actual machine. The measurement data storage unit stores the measurement data from a plurality of sensors, which are installed inside or outside the glass melting furnace to measure values ​​related to molten glass.

[0389] (Structure Example 2)

[0390] According to the information processing apparatus described in (Structural Example 1), wherein, The information processing device includes a calibration parameter estimation unit, which estimates the calibration parameters using a calibration parameter estimation model that estimates the calibration parameters of the glass melting furnace simulation model and the measurement data stored in the measurement data storage unit. The state estimation unit uses the correction parameters estimated by the correction parameter estimation unit to perform the state estimation of the actual machine.

[0391] (Structure Example 3)

[0392] According to the information processing apparatus described in (Structure Example 1) or (Structure Example 2), wherein, The information processing device includes a model calculation unit, which performs optimization calculations or sensitivity analyses based on the operation change guidance scheme, using the glass melting furnace simulation model or a summary model based on the glass melting furnace simulation model. The operation change guidance scheme is based on the state estimation results performed by the state estimation unit.

[0393] (Structure Example 4)

[0394] According to the information processing device described in (Structural Example 3), wherein, The information processing device includes an output unit that outputs information related to at least one of the result of the state estimation performed by the state estimation unit and the result of the calculation performed by the model calculation unit.

[0395] (Structure Example 5)

[0396] According to the information processing apparatus described in (Structural Example 2), wherein, The correction parameter estimation unit includes: The experimental planning method execution unit uses the operating conditions of the glass melting furnace and the correction parameters as set variables, and performs multiple simulations on the glass melting furnace simulation model based on the experimental planning method, within a scope that comprehensively includes the conditions envisioned in the actual glass manufacturing process. The simulation result storage unit stores the simulation results of the experimental planning method execution unit; A prediction model generation unit uses the simulation results stored in the simulation result storage unit to perform machine learning and generate a prediction model; and The sensitivity evaluation unit evaluates the sensitivity of the glass melting furnace's operating conditions and correction parameters to the changes caused by the simulation results of at least one of the following at multiple observation points of the glass melting furnace: molten glass temperature, molten glass resistivity, current or voltage or resistance value between heating electrodes, molten glass flow rate, furnace material temperature, and furnace internal and external atmosphere temperature.

[0397] (Structure Example 6)

[0398] According to the information processing apparatus described in (Structural Example 5), wherein, The correction parameter estimation unit further includes an observation point data acquisition unit. This unit acquires the operating conditions at a predetermined time point for multiple observation points within the glass melting furnace, and obtains observation point data. The observation point data includes at least one of the following at the predetermined time point: the temperature of the molten glass, the resistivity of the molten glass, the current, voltage, or resistance value between the heating electrodes, the flow rate of the molten glass, the temperature of the furnace materials, and the temperature of the atmosphere inside and outside the furnace. The correction parameter estimation unit estimates the correction parameter based on the evaluation result of the change sensitivity evaluation unit, in a manner that the deviation between the observation point data obtained by the observation point data acquisition unit for each observation point and the result obtained through the simulation is reduced.

[0399] (Structure Example 7)

[0400] According to the information processing apparatus described in (Structural Example 6), wherein, The correction parameter estimation unit limits the range of the correction parameter values ​​based on the changes in heat dissipation of the furnace material caused by the aging of the glass melting furnace.

[0401] A glass manufacturing system that can provide information processing devices as described above.

[0402] (Structure Example 8)

[0403] A glass manufacturing system includes a glass manufacturing apparatus and an information processing apparatus, wherein the glass manufacturing apparatus includes a glass melting furnace, wherein... The information processing device includes a measurement data storage unit, either inside or outside, which stores measurement data from multiple sensors located inside or outside the glass melting furnace for measuring values ​​related to the molten glass. The information processing device includes a state estimation unit, which uses a glass melting furnace simulation model obtained by modeling the glass melting furnace and the measurement data stored in the measurement data storage unit to perform state estimation of the actual machine.

[0404] It is possible to provide a method performed by such an information processing device.

[0405] (Structure Example 9)

[0406] An information processing method, wherein, The information processing device uses measurement data stored in the measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform state estimation of the actual machine. The measurement data storage unit stores the measurement data from multiple sensors, which are located inside or outside the glass melting furnace to measure values ​​related to the molten glass.

[0407] It is capable of providing a program (computer program) that can be executed by a computer that constitutes such an information processing device.

[0408] (Structure Example 10)

[0409] A program is provided for enabling a computer to perform a state estimation function, the state estimation function using measurement data stored in a measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform state estimation of the actual machine, the measurement data storage unit storing the measurement data of a plurality of sensors located inside or outside the glass melting furnace for measuring values ​​related to molten glass.

[0410] It can provide a method for estimating correction parameters that can be applied to information processing devices such as those described above.

[0411] (Structure Example 11)

[0412] A method for estimating correction parameters, wherein, Using the operating conditions and calibration parameters of the glass melting furnace as set variables, and comprehensively encompassing the conditions envisioned in actual glass manufacturing processes, multiple simulations were performed on the glass melting furnace simulation model based on the experimental planning method. Machine learning is performed using the simulation results to generate predictive models. Evaluate the sensitivity of the operating conditions and correction parameters of the glass melting furnace to the changes caused by the simulation results of at least one of the following at multiple observation points of the glass melting furnace: molten glass temperature, molten glass resistivity, current or voltage or resistance value between heating electrodes, molten glass flow rate, furnace material temperature, and the temperature of the atmosphere inside and outside the furnace.

[0413] Label Explanation

[0414] 1…Glass manufacturing system

[0415] 20…Superstructure

[0416] 40…Thermocouple

[0417] 50…Electrode

[0418] 60…burner

[0419] 70… Glass materials (glass raw materials or molten glass)

[0420] 100…glass melting furnace

[0421] 111…Information Processing Device

[0422] 112…database

[0423] 131… Glass Manufacturing Apparatus

[0424] 211…Input Section

[0425] 212… Output Department

[0426] 213…Ministry of Communications

[0427] 214… Storage Department

[0428] 215…Control Department

[0429] 231…Operations Department

[0430] 251… Display Department

[0431] 271…State Estimation Department

[0432] 272…Correction Parameter Estimation Section

[0433] 273… Model Calculation Department

[0434] 274…Experimental Planning Method Implementation Department

[0435] 275…Predictive Model Generation Department

[0436] 276…Sensitivity Evaluation Department

[0437] 277… Data Acquisition Department at Observation Points

[0438] 311…glass melting furnace

[0439] 331, 351… Sensor Department

[0440] Tables 3011, 3211…

[0441] 1011, 1021, 1221, 3111, 3511, 3521… characteristics

[0442] A1… Actual Manufacturing Process

[0443] A2…Simulation process

[0444] A3…Comprehensive Data Platform

[0445] B1…Actual Machine

[0446] B2… Actual machine data

[0447] C1…simulation model

[0448] D1…Parameter Estimation

[0449] E1, E2... users

[0450] F1, F2… Computer

[0451] G1… server

[0452] H1~H3…heat source

[0453] I1~I3…Temperature

[0454] J1~J2…Correction parameters

[0455] M1… Glass melting furnace model

[0456] R1… Measurement data

[0457] R11…Simulation model of glass melting furnace

[0458] R12…Correction parameter estimation model

[0459] R13…Predictive Model

[0460] R14…Simulation Results

[0461] U1…the entire set of conditions

[0462] V1…conditions

[0463] W1…Simulation Results Group

[0464] Sensitivity w1~w3, w11~w13…

Claims

1. An information processing apparatus comprising a state estimation unit, the state estimation unit performing state estimation of a real machine using measurement data stored in a measurement data storage unit and a glass melting furnace simulation model obtained by modeling a glass melting furnace, the measurement data storage unit storing the measurement data of a plurality of sensors, the plurality of sensors being disposed inside or outside the glass melting furnace for measuring values ​​related to molten glass.

2. The information processing apparatus according to claim 1, wherein, The information processing device includes a calibration parameter estimation unit, which estimates the calibration parameters using a calibration parameter estimation model that estimates the calibration parameters of the glass melting furnace simulation model and the measurement data stored in the measurement data storage unit. The state estimation unit uses the correction parameters estimated by the correction parameter estimation unit to perform the state estimation of the actual machine.

3. The information processing apparatus according to claim 1 or 2, wherein, The information processing device includes a model calculation unit, which performs optimization calculations or sensitivity analyses based on the operation change guidance scheme, using the glass melting furnace simulation model or a summary model based on the glass melting furnace simulation model. The operation change guidance scheme is based on the state estimation results performed by the state estimation unit.

4. The information processing apparatus according to claim 3, wherein, The information processing device includes an output unit that outputs information related to at least one of the result of the state estimation performed by the state estimation unit and the result of the calculation performed by the model calculation unit.

5. The information processing apparatus according to claim 2, wherein, The correction parameter estimation unit includes: The experimental planning method execution unit uses the operating conditions of the glass melting furnace and the correction parameters as set variables, and performs multiple simulations on the glass melting furnace simulation model based on the experimental planning method, within a scope that comprehensively includes the conditions envisioned in the actual glass manufacturing process. The simulation result storage unit stores the simulation results of the experimental planning method execution unit; The prediction model generation unit uses the simulation results stored in the simulation result storage unit to perform machine learning and generate a prediction model; as well as The sensitivity evaluation unit evaluates the sensitivity of the glass melting furnace's operating conditions and correction parameters to the changes caused by the simulation results of at least one of the following at multiple observation points of the glass melting furnace: molten glass temperature, molten glass resistivity, current or voltage or resistance value between heating electrodes, molten glass flow rate, furnace material temperature, and furnace internal and external atmosphere temperature.

6. The information processing apparatus according to claim 5, wherein, The correction parameter estimation unit further includes an observation point data acquisition unit. This unit acquires the operating conditions at a predetermined time point for multiple observation points within the glass melting furnace, and obtains observation point data. The observation point data includes at least one of the following at the predetermined time point: the temperature of the molten glass, the resistivity of the molten glass, the current, voltage, or resistance value between the heating electrodes, the flow rate of the molten glass, the furnace material temperature, and the temperature of the atmosphere inside and outside the furnace. The correction parameter estimation unit estimates the correction parameter based on the evaluation result of the change sensitivity evaluation unit, in a manner that the deviation between the observation point data obtained by the observation point data acquisition unit for each observation point and the result obtained through the simulation is reduced.

7. The information processing apparatus according to claim 6, wherein, The correction parameter estimation unit limits the range of the correction parameter values ​​based on the changes in heat dissipation of the furnace material caused by the aging of the glass melting furnace.

8. A glass manufacturing system comprising a glass manufacturing apparatus and an information processing apparatus, wherein the glass manufacturing apparatus includes a glass melting furnace, wherein, The information processing device includes a measurement data storage unit, either inside or outside, which stores measurement data from multiple sensors located inside or outside the glass melting furnace for measuring values ​​related to the molten glass. The information processing device includes a state estimation unit, which uses a glass melting furnace simulation model obtained by modeling the glass melting furnace and the measurement data stored in the measurement data storage unit to perform state estimation of the actual machine.

9. An information processing method, wherein, The information processing device uses measurement data stored in the measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform state estimation of the actual machine. The measurement data storage unit stores the measurement data from multiple sensors, which are located inside or outside the glass melting furnace to measure values ​​related to the molten glass.

10. A program for enabling a computer to perform a state estimation function, the state estimation function using measurement data stored in a measurement data storage unit and a glass melting furnace simulation model obtained by modeling the glass melting furnace to perform a real-world state estimation, the measurement data storage unit storing the measurement data from a plurality of sensors located inside or outside the glass melting furnace for measuring values ​​related to molten glass.

11. A method for estimating correction parameters, wherein, Using the operating conditions and calibration parameters of the glass melting furnace as set variables, and comprehensively encompassing the conditions envisioned in actual glass manufacturing processes, multiple simulations were performed on the glass melting furnace simulation model based on the experimental planning method. Machine learning is performed using the simulation results to generate predictive models. Evaluate the sensitivity of the operating conditions and correction parameters of the glass melting furnace to the changes caused by the simulation results of at least one of the following at multiple observation points of the glass melting furnace: molten glass temperature, molten glass resistivity, current or voltage or resistance value between heating electrodes, molten glass flow rate, furnace material temperature, and the temperature of the atmosphere inside and outside the furnace.

Images