Information processor, glass plate manufacturing apparatus, information processing method, and manufacturing method of glass plate

Abstract

To provide an information processor to enable quality maintenance of a plate thickness of a glass plate after manufacturing by quickly performing temperature control, etc., when the plate thickness of the glass plate after manufacturing is likely to be deviated from a certain range.SOLUTION: An information processor includes a determination part where, about a glass plate manufactured while flown in a predetermined flow direction, a manufacturing condition information on manufacturing the glass plate is input to a learned model of a machine learning and an output result from the learned model is obtained as information relating to an estimated result of a plate thickness of the glass plate.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to an information processing device, a glass sheet manufacturing device, an information processing method, and a glass sheet manufacturing method. [Background technology]

[0002] Glass sheets are produced by the float method or the like (see, for example, Patent Document 1). In the manufacture of glass sheets, maintaining the thickness of the manufactured glass sheets within a certain range is an important aspect of quality control.

[0003] Conventionally, the thickness of a glass sheet has been constantly monitored downstream of a glass sheet manufacturing apparatus, and the temperature in the float bath and the like have been adjusted based on the thickness, thereby attempting to maintain the thickness within a certain range. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2016-98161 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, in the conventional technology, the thickness of the glass sheet was monitored in a state where the temperature of the glass sheet was lowered downstream of the glass sheet manufacturing apparatus, and the temperature in the float bath upstream of the monitoring was adjusted, so that there was a large time difference (e.g., 30 minutes) between these. In the conventional technology, because of this time difference, for example, when the temperature adjustment was incomplete for some reason, the correction may be significantly delayed, and a situation may occur in which the decrease in yield continues for a long time. Moreover, similar problems exist not only in the float method but also in the fusion method and the like.

[0006] The present disclosure has been made in consideration of the above circumstances, and has an objective of providing an information processing device, a glass plate manufacturing device, an information processing method, and a glass plate manufacturing method that are capable of maintaining the quality of the thickness of the manufactured glass plate by quickly adjusting the temperature, etc., when the thickness of the manufactured glass plate is likely to deviate from a certain range. [Means for solving the problem]

[0007] One aspect is an information processing device that includes a determination unit that inputs information about manufacturing conditions during the manufacture of a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information about an estimated result of the thickness of the glass sheet.

[0008] One aspect is a glass plate manufacturing apparatus including an information processing device that inputs information about manufacturing conditions during the manufacture of a glass plate, the glass plate being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information about an estimation result of a thickness of the glass plate, and an adjustment unit that adjusts the manufacturing conditions based on the estimation result by the determination unit.

[0009] One aspect is an information processing method in which a judgment unit inputs information about manufacturing conditions during the manufacture of a glass plate, which is manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information about an estimated result of the thickness of the glass plate.

[0010] One aspect is a glass plate manufacturing method in which a judgment unit inputs information about manufacturing conditions during the manufacture of a glass plate, the glass plate being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information about an estimated result of the thickness of the glass plate, and an adjustment unit adjusts the manufacturing conditions based on the estimated result by the judgment unit. Effect of the Invention

[0011] According to the information processing device, information processing device, glass sheet manufacturing device, information processing method, and glass sheet manufacturing method disclosed herein, it is possible to maintain the quality of the thickness of the glass sheet after manufacture by quickly adjusting the temperature, etc., when the thickness of the glass sheet after manufacture is likely to deviate from a certain range. [Brief description of the drawings]

[0012] [Figure 1] 1 is a diagram showing a schematic configuration example of a glass sheet manufacturing apparatus according to an embodiment. FIG. [Diagram 2] FIG. 1 is a diagram illustrating an example of a configuration of an information processing device according to an embodiment. [Diagram 3] FIG. 13 is a diagram illustrating an example of input / output information of a trained model according to an embodiment. [Figure 4] FIG. 4 is a diagram showing an example of manufacturing conditions according to the embodiment. [Diagram 5] FIG. 2 is a diagram showing an example of a state of a glass plate according to the embodiment. [Figure 6] FIG. 4 is a diagram showing an example of a sheet thickness characteristic in a width direction of a glass sheet according to the embodiment. [Figure 7A] FIG. 4 is a diagram showing an example of characteristics of a change in thickness of a glass sheet over time according to the embodiment. [Figure 7B] FIG. 4 is a diagram showing an example of characteristics of a change in thickness of a glass sheet over time according to the embodiment. [Figure 7C] FIG. 4 is a diagram showing an example of characteristics of a change in thickness of a glass sheet over time according to the embodiment. [Figure 7D] FIG. 4 is a diagram showing an example of characteristics of a change in thickness of a glass sheet over time according to the embodiment. [Figure 7E] FIG. 4 is a diagram showing an example of characteristics of a change in thickness of a glass sheet over time according to the embodiment. [Figure 8] 3 is a diagram showing an example of a procedure of a process performed in the glass sheet manufacturing apparatus according to the embodiment. FIG. [Figure 9] FIG. 13 is a diagram illustrating another example of input / output information of a trained model according to an embodiment. [Figure 10] FIG. 2 is a cross-sectional view showing a portion of a float type glass sheet manufacturing apparatus. [Figure 11] FIG. 2 is a plan view showing a float bath. [Figure 12] FIG. 2 is a plan view showing a float bath. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0013] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings. In addition, various pieces of information in the following description of the embodiments may be referred to as data or the like.

[0014] [Glass plate manufacturing equipment] FIG. 1 is a diagram showing a schematic configuration example of a glass sheet manufacturing apparatus 1 according to an embodiment. For convenience of explanation, FIG. 1 shows an XYZ orthogonal coordinate system, which is a three-dimensional orthogonal coordinate system. In this embodiment, the positive direction of the Z axis represents the upward direction, and the negative direction of the Z axis represents the downward direction, and the XY plane represents a plane that is horizontal in the vertical direction.

[0015] The glass sheet manufacturing apparatus 1 includes a melting furnace 11, a float bath 12, an annealing furnace 13, and a cleaning / inspection / cutting section 14. The glass sheet manufacturing apparatus 1 further includes an information processing device 101 , a sheet thickness detection unit 121 , a manufacturing condition detection unit 122 , and an adjustment unit 123 . 1 also shows an operator 141 who operates the glass sheet manufacturing apparatus 1. The operator 141 is, for example, a person. Note that the operator may be called a user or the like.

[0016] In the example of Figure 1, in order to make the figure easier to see, the plate thickness detection unit 121, the manufacturing condition detection unit 122, and the adjustment unit 123 are illustrated separated from the glass plate manufacturing process parts (the melting kiln 11, the float bath 12, the annealing furnace 13, and the cleaning, inspection, and cutting unit 14). However, for example, one or more of the plate thickness detection unit 121, the manufacturing condition detection unit 122, and the adjustment unit 123 may be integrated with the glass plate manufacturing process parts.

[0017] In the melting furnace 11, the melting process is carried out. In the melting process, glass raw materials prepared by mixing a plurality of raw materials are melted to obtain molten glass. The glass raw materials are charged into the melting kiln 11, and then melted by heating means such as radiant heat from a flame sprayed from a burner or electric melting to become molten glass.

[0018] In the float bath 12, the molding process is carried out. In the forming process, the molten glass obtained in the melting process is continuously supplied onto the molten tin in a bath, and the molten glass is caused to flow and formed on the molten tin to obtain a glass sheet (so-called glass ribbon) which is a plate-shaped glass. The glass sheet is cooled while flowing in a predetermined direction, and is pulled up from the molten tin.

[0019] In the annealing furnace 13, a annealing process is carried out. In the annealing step, the glass sheet obtained in the forming step is annealed inside the annealing furnace 13. Inside the annealing furnace 13, the glass sheet is annealed while being transported horizontally on rolls from the entrance to the exit of the annealing furnace 13.

[0020] In the cleaning, inspection and cutting section 14, the cleaning and inspection process and the cutting process are carried out. In the cleaning inspection step, a cleaning inspection is performed on the glass plate that has been cooled in the annealing step. In the cutting process, the glass plate that has been cooled in the cooling process is cut to a predetermined size by a cutting machine. In the cutting process, both edges (so-called edge portions) in the width direction of the glass plate are cut off. Both edges in the width direction of the glass plate are cut off because they become thicker due to the influence of surface tension, etc. After the cutting process, the glass sheets are shipped.

[0021] The plate thickness detection unit 121 detects the plate thickness of the glass plate (plate thickness). In this embodiment, the plate thickness is the thickness in the up-down direction (in the example of FIG. 1, the direction parallel to the Z axis). In this embodiment, the sheet thickness detection unit 121 is disposed at a position near the downstream outlet of the glass sheet manufacturing apparatus 1, and detects the thickness of the glass sheet at that position. This position is downstream of the annealing furnace 13 and is in the vicinity of the cleaning, inspection, and cutting unit 14.

[0022] The plate thickness detection unit 121 may detect the plate thickness at a plurality of positions in a width direction (a direction parallel to the Y axis in the example of FIG. 1) perpendicular to the conveying direction of the glass plate (a direction parallel to the X axis from the negative side to the positive side of the X axis in the example of FIG. 1). These positions may be arranged at regular intervals in the width direction, for example, or may be arranged at other intervals.

[0023] In this embodiment, the portion of the glass sheet whose thickness is to be detected by the sheet thickness detection unit 121 is in a state after a predetermined time has elapsed since the glass sheet passed through the float bath 12. The predetermined time is, for example, about 30 minutes, but may be another time.

[0024] The manufacturing condition detection unit 122 detects predetermined manufacturing conditions. In this embodiment, the production condition detection unit 122 detects the production conditions in the float bath 12 . As another example, the manufacturing condition detection unit 122 may detect manufacturing conditions in a portion other than the float bath 12 in the glass sheet manufacturing apparatus 1.

[0025] The adjustment unit 123 has a function of adjusting the manufacturing conditions. In this embodiment, the adjustment unit 123 has a function of adjusting the production conditions in the float bath 12. As another example, the adjustment unit 123 may have a function of adjusting the manufacturing conditions in a portion other than the float bath 12 in the glass sheet manufacturing apparatus 1.

[0026] As an example, the adjustment unit 123 may have an operation unit operated by the operator 141, and adjust the manufacturing conditions according to the contents of the operation received by the operation unit. As another example, the adjustment unit 123 may adjust the manufacturing conditions in response to an instruction input from an external device (the information processing device 101 in this embodiment).

[0027] [Information processing device] FIG. 2 is a diagram showing an example of the configuration of the information processing device 101 according to the embodiment. The information processing device 101 is configured, for example, by a computer. The information processing device 101 includes an input unit 211 , an output unit 212 , a communication unit 213 , a storage unit 214 , and a control unit 215 . The input unit 211 includes an operation unit 231 . The output unit 212 includes a display unit 251 . The control unit 215 includes a determination unit 271 .

[0028] The input unit 211 has a function of inputting information from the outside. For example, the operation unit 231 has a function of inputting information in response to an operation performed by the operator 141. The output unit 212 has a function of outputting information to the outside. For example, the display unit 251 has a function of displaying information to be displayed on a screen. The communication unit 213 has a function of communicating with an external device.

[0029] The storage unit 214 stores various types of information. In this embodiment, the storage unit 214 stores a trained model 311 and the like. The trained model 311 may be set in advance in the memory unit 214, or may be set from outside at any timing. Here, any machine learning model may be used as the learning model of the trained model 311, for example, a neural network model may be used.

[0030] The control unit 215 performs various types of processing and control. The judgment unit 271 inputs predetermined information to be judged to the trained model 311, and obtains the output from the trained model 311 as judgment result information.

[0031] The control unit 215 may, for example, output the acquired information on the judgment result by the output unit 212, and as one example, may display and output display content related to the acquired information on the judgment result on the screen of the display unit 251. In this case, for example, the operator 141 may control the adjustment of the manufacturing conditions by the adjustment unit 123 by looking at the display content and operating the adjustment unit 123 according to the display content (manual control by the operator 141).

[0032] Furthermore, the control unit 215 may have a function of controlling (automatically controlling) the adjustment of the manufacturing conditions by the adjustment unit 123 based on, for example, information on the acquired determination result. As a specific example, the control unit 215 may control the adjustment of the manufacturing conditions by the adjustment unit 123 by outputting the acquired information on the determination result or other information based on the information to the adjustment unit 123 via the communication unit 213 as a control signal for the adjustment unit 123. In this case, the adjustment unit 123 adjusts the manufacturing conditions in response to the control signal received from the control unit 215.

[0033] Here, the judgment unit 271 may, for example, directly adopt the output (estimated result) from the trained model 311 as the judgment result by the judgment unit 271, or may perform further judgment or analysis, etc. based on the output (estimated result) from the trained model 311, and use the result as the judgment result by the judgment unit 271. Instead of estimation, words such as prediction or forecast may be used.

[0034] In this embodiment, the information processing device 101 has a function of communicating with the plate thickness detection unit 121 via the communication unit 213. This allows the information processing device 101 to receive and acquire information on the detection result of the plate thickness detected by the plate thickness detection unit 121. In this embodiment, the information processing device 101 has a function of communicating with the manufacturing condition detection unit 122 via the communication unit 213. This allows the information processing device 101 to receive and acquire information on the detection result of the manufacturing conditions detected by the manufacturing condition detection unit 122. In this embodiment, the information processing device 101 has a function of communicating with the adjustment unit 123 via the communication unit 213. This allows the information processing device 101 to control the adjustment of the manufacturing conditions by the adjustment unit 123 by outputting (transmitting) an instruction to the adjustment unit 123.

[0035] Here, each communication may be, for example, a wired communication or a wireless communication. In this embodiment, the information processing device 101 is configured to communicate with each of the plate thickness detection unit 121, the manufacturing condition detection unit 122, and the adjustment unit 123, but the exchange of information between the two different devices may be performed by any method, for example, using a portable external storage medium. Furthermore, in this embodiment, a configuration in which the information processing device 101 can control the adjustment unit 123 is shown, but as another example, a configuration in which the information processing device 101 does not control the adjustment unit 123 may be used.

[0036] <Example of input / output information for a trained model> FIG. 3 is a diagram illustrating an example of input / output information of a trained model 311 according to an embodiment. Information on the detection results of manufacturing conditions is input as input information to the trained model 311. As the information, information on the detection results of manufacturing conditions detected by the manufacturing condition detection unit 122 is used. The example in FIG. 3 illustrates a case in which information on the detection results of the thickness of the glass plate is not used as input information to the trained model 311. Here, as the information on the detection result of the manufacturing conditions, for example, one piece of information may be used, or multiple pieces of information may be used.

[0037] As another example, information that has been processed based on information on the detection result of the manufacturing conditions may be used as the input information. In other words, the input information does not necessarily have to be the information on the detection result of the manufacturing conditions itself, and may be converted into other information.

[0038] Information on the estimation result of the plate thickness is output as output information from the trained model 311. The information is, for example, information on the plate thickness estimated to be detected by the plate thickness detection unit 121.

[0039] As another example, information processed based on the information on the plate thickness estimation result may be used as the output information. In other words, the output information does not necessarily have to be the information on the plate thickness estimation result itself, and may be converted into other information. As a specific example, the output information may be information indicating whether the estimated plate thickness satisfies a predetermined condition. The condition may be, for example, a condition indicating that the plate thickness (estimated plate thickness) is abnormal if it is not within a predetermined range (between a lower limit and an upper limit). In addition, for example, estimates of the plate thickness at multiple positions in the width direction of the glass plate may be obtained, and machine learning may be performed to ensure that there is no position in the width direction where the estimate of the plate thickness deviates significantly (for example, deviates by more than a predetermined threshold value).

[0040] Here, in this embodiment, the trained model 311 may be generated by any learning method. As an example, the trained model 311 may be generated by machine learning by using information on the detection results of manufacturing conditions as input information for the pre-trained model, and using information on the plate thickness detected by the plate thickness detection unit 121 as teacher data (true value data) for the output information of the pre-trained model.

[0041] In addition, in this embodiment, a case is shown in which the trained model 311 that has been trained before the judgment is made by the judgment unit 271 is used, but as another example, a training model in which the already trained model contents are sequentially updated (corrected) in a time series (in this embodiment, this is also considered as an example of the trained model 311) may be used. As such a training model, for example, a state space model may be used. In this way, as the trained model 311, a learning model for which training has already been completed and the model content is fixed may be used, or a learning model in which the model content is further updated sequentially over time may be used.

[0042] [Example of manufacturing conditions] FIG. 4 is a diagram showing an example of manufacturing conditions 511 according to the embodiment. The manufacturing conditions 511 include, for example, temperature conditions 531, pressure conditions 532, tool conditions 533, heater conditions 534, humidity conditions 535, and plate thickness conditions 536. It is not necessary to use all the conditions shown in FIG. 4 as the manufacturing conditions 511, and any one or more of the conditions may be used. Moreover, the manufacturing conditions 511 may be different from those exemplified in this embodiment. Here, the specific example of the manufacturing conditions 511 shown in FIG. 4 is an example, and is not necessarily limited to this specific example. For example, instead of using the manufacturing conditions shown in FIG. 4 (this specific example), other manufacturing conditions may be used. Each of the manufacturing conditions may be detected in real time by the manufacturing condition detection unit 122 during the manufacturing process of the glass plate, for example.

[0043] <Temperature conditions> As the manufacturing conditions, conditions related to temperature (temperature conditions 531) may be used. As the temperature condition 531, for example, the temperature inside the float bath 12 may be used, or another temperature may be used. For example, the temperature at one location may be used as the temperature condition 531, or the temperatures at multiple different locations may be used. These multiple different locations may be distributed in the conveying direction of the glass plate (in this embodiment, a direction parallel to the X-axis), or in the width direction of the glass plate (in this embodiment, a direction parallel to the Y-axis), or in both directions. Furthermore, as the temperature condition 531, for example, a temperature at one time may be used for each location, or temperatures at multiple different times may be used. The temperature at the predetermined location may be detected by, for example, a temperature sensor provided at the predetermined location (or in the vicinity thereof). In addition, a value obtained by averaging temperatures at a plurality of locations (spatial average value) may be used, or a value obtained by averaging temperatures at a plurality of times (temporal average value) may be used.

[0044] <Pressure conditions> As the manufacturing conditions, conditions related to pressure (pressure conditions 532) may be used. As the pressure condition 532, for example, the pressure inside the float bath 12 may be used, or another pressure may be used. For example, the pressure at one location may be used as the pressure condition 532, or the pressure at multiple different locations may be used. These multiple different locations may be distributed in the conveying direction of the glass plate (in this embodiment, a direction parallel to the X-axis), or in the width direction of the glass plate (in this embodiment, a direction parallel to the Y-axis), or in both directions. Furthermore, as pressure condition 532, for example, the pressure at one time may be used for each location, or the pressure at a plurality of different times may be used. The pressure at the predetermined location may be detected by, for example, a pressure sensor provided at the predetermined location (or in the vicinity thereof). In addition, a value obtained by averaging pressure at a plurality of locations (spatial average value) may be used, or a value obtained by averaging pressure at a plurality of times (temporal average value) may be used.

[0045] <Tweel conditions> As the manufacturing conditions, conditions related to the twill (twill conditions 533) may be used. In this embodiment, the tweel is a portion that controls the flow rate of the molten glass supplied from the upstream side of the float bath 12 . As the tweel condition 533, for example, a value indicating the degree of opening and closing of the tweel (for example, an angle) may be used, or other values ​​may be used. Furthermore, as tool condition 533, for example, a single time value may be used, or a plurality of different time values ​​may be used. Note that the value representing the degree of opening and closing of the tweel (for example, the angle) may be specified by, for example, a function that controls the tweel (for example, it may be a function of adjustment unit 123). Also, a value obtained by averaging values ​​at a plurality of times (a time average value) may be used.

[0046] <Heater conditions> As the manufacturing conditions, conditions related to the heater (heater conditions 534) may be used. In this embodiment, the heater is a part that adjusts the temperature inside the float bath 12. As the heater, a plurality of heaters may be provided. As the heater condition 534, for example, a value indicating the heating state by the heater (for example, the temperature of the heater) may be used, or another value may be used. For example, a value related to a heater at one location may be used as the heater condition 534, or values ​​related to heaters at multiple different locations may be used. These multiple different locations may be distributed, for example, in the conveying direction of the glass sheet (in this embodiment, a direction parallel to the X-axis), or in the width direction of the glass sheet (in this embodiment, a direction parallel to the Y-axis), or in both directions. Furthermore, as the heater condition 534, for example, a single value at a time may be used for each location, or multiple values ​​at different times may be used. The value indicating the heating state of the heater may be determined, for example, by a function that controls the heater (for example, it may be a function of the adjustment unit 123). Also, a value obtained by averaging values ​​relating to heaters at multiple locations (spatial average value) may be used, or a value obtained by averaging values ​​at multiple times (temporal average value) may be used.

[0047] <Humidity conditions> As the manufacturing conditions, conditions related to humidity (humidity conditions 535) may be used. As the humidity condition 535, for example, the humidity inside the float bath 12 may be used, or another humidity may be used. For example, the humidity at one location may be used as the humidity condition 535, or the humidity at multiple different locations may be used. These multiple different locations may be distributed, for example, in the conveying direction of the glass plate (in this embodiment, a direction parallel to the X-axis), or in the width direction of the glass plate (in this embodiment, a direction parallel to the Y-axis), or in both directions. Furthermore, as the humidity condition 535, for example, the humidity at one time may be used for each location, or the humidity at a number of different times may be used. The humidity at the predetermined location may be detected by, for example, a humidity sensor provided at the predetermined location (or in the vicinity thereof). In addition, a value obtained by averaging the humidity at a number of locations (spatial average value) may be used, or a value obtained by averaging the humidity at a number of times (temporal average value) may be used.

[0048] <Plate thickness conditions> As the manufacturing conditions, conditions related to the thickness of the glass sheet (sheet thickness conditions 536) may be used. As the sheet thickness condition 536, for example, the sheet thickness at a predetermined location of the glass sheet in a state in which it is present inside the float bath 12 may be used, or the sheet thickness of the glass sheet in another state may be used. For example, the thickness at one location may be used as the thickness condition 536, or the thickness at multiple different locations may be used. These multiple different locations may be distributed in the conveying direction of the glass plate (in this embodiment, a direction parallel to the X-axis), or in the width direction of the glass plate (in this embodiment, a direction parallel to the Y-axis), or in both directions. Furthermore, as the plate thickness condition 536, for example, for each location, the plate thickness at one time may be used, or plate thicknesses at a plurality of different times may be used. The plate thickness at the predetermined location may be detected by, for example, a plate thickness sensor provided at the predetermined location (or in the vicinity thereof). When the predetermined location is the same as the location where the plate thickness is detected by the plate thickness detection unit 121, the function of the plate thickness detection unit 121 may be used as the function for detecting the plate thickness at the predetermined location. In addition, a value obtained by averaging the plate thickness at a plurality of locations (spatial average value) may be used, or a value obtained by averaging the plate thickness at a plurality of times (temporal average value) may be used.

[0049] Here, the temperature, pressure, and humidity may change due to disturbances such as the weather or the time of day (such as the rising and setting of the sun), and it may be difficult to control them to be constant. Furthermore, the degree of expansion in the width direction during the production of the glass sheet may change due to disturbances, and it may be difficult to control it to a constant degree. In this embodiment, the tweezers and heaters are controllable.

[0050] <Example of glass plate appearance> FIG. 5 is a diagram showing an example of the state of a glass plate 611 according to the embodiment. For convenience of explanation, FIG. 5 shows an XYZ orthogonal coordinate system similar to that shown in FIG. Moreover, the shape of the glass plate 611 shown in FIG. 5 is a schematic one for the purpose of explanation.

[0051] The glass plate 611 is distributed in the conveying direction (a direction parallel to the X-axis in the example of FIG. 5) and expands in the width direction (a direction parallel to the Y-axis in the example of FIG. 5) toward the downstream. In the example of FIG. 5, the width of the glass plate 611 is shown to increase symmetrically with respect to the center of the glass plate 611 in the width direction.

[0052] In the example of FIG. 5, six positions (position x0 to position x5) are indicated on the glass plate 611 in the conveying direction of the glass plate 611. For example, a glass component that was at position x0 at time t0 moves to position x1 at time t1, and similarly moves to positions x0, x2, x3, x4, and x5 at times t2, t3, t4, and t5, respectively. Here, the larger the numbers are for positions x0 to x5, the more downstream the positions are represented, and the larger the numbers are for times t0 to t5, the more advanced the times are represented (that is, later times).

[0053] In the example of FIG. 5, five positions (position y0, position yR1, position yR2, position yL1, position yL2) are shown on the glass plate 611 in the width direction. The position y0 represents the center position of the glass plate 611 in the width direction. The positions yR1 and yL1 are symmetrical in the width direction with respect to the position y0 and are spaced the same distance from the position y0 in the respective directions (upward and downward in the example of FIG. 5). Similarly, positions yR2 and yL2 are symmetrical in the width direction with respect to position y0 and are spaced the same distance from position y0 in the respective directions (upward and downward in the example of FIG. 5). Here, positions yR2 and yL2 are spaced farther from position y0 than positions yR1 and yL1.

[0054] FIG. 6 is a diagram showing an example of the sheet thickness characteristics in the width direction of the glass sheet according to the embodiment. It should be noted that the characteristics shown in the example of FIG. 6 are merely illustrative and are not necessarily strict. In the graph shown in FIG. 6, the horizontal axis represents the width direction of the glass plate (in this embodiment, the direction parallel to the Y axis), and the vertical axis represents the plate thickness.

[0055] FIG. 6 shows the sheet thickness characteristic 2011 of the glass sheet in the width direction. The horizontal axis shows the center position h0 of the glass plate in the width direction, positions hR1 and hL1 which are symmetrical with respect to the center position h0, and positions hR2 and hL2 which are symmetrical with respect to the center position h0. Here, position hR1 is a position that is a predetermined distance away from center position h0 in a predetermined direction (positive direction of the Y axis in the example of FIG. 6), and position hR2 is a position that is further away. Conversely, position hL1 is a position that is a predetermined distance away from center position h0 in the opposite direction to the predetermined direction (negative direction of the Y axis in the example of FIG. 6), and position hL2 is a position that is further away.

[0056] For example, the thickness characteristic of the glass sheet to be detected by the thickness detection unit 121 may be as shown in FIG. Further, for example, even at a position upstream of the thickness detection unit 121, such as the float bath 12, the thickness characteristic of the glass sheet may be as shown in FIG. However, the characteristic 2011 shown in FIG. 6 is an example for the purpose of explanation, and in reality, the thickness characteristic of the glass sheet may change as the glass sheet moves from the upstream to the downstream.

[0057] 7A to 7E are diagrams showing examples of characteristics of changes in thickness over time of a glass plate according to an embodiment. In each of the graphs shown in FIGS. 7A to 7E, the horizontal axis represents time (hours), and the vertical axis represents plate thickness.

[0058] FIG. 7A shows a characteristic 2110 of the change in thickness of the glass sheet over time at the position h0 shown in FIG. FIG. 7B shows a characteristic 2121 of the change in thickness of the glass sheet over time at the position hR1 shown in FIG. FIG. 7C shows a characteristic 2122 of the change in thickness of the glass sheet at the position hR2 shown in FIG. FIG. 7D shows a characteristic 2131 of the change in thickness of the glass sheet at the position hL1 shown in FIG. FIG. 7E shows a characteristic 2132 of the change in thickness of the glass sheet at the position hL2 shown in FIG.

[0059] For example, as shown in FIGS. 7A to 7E, plate thickness estimation (determination by the trained model 311) may be performed for a plurality of positions in the width direction of the glass plate. In this example, five positions in the width direction of the glass plate are shown, but any number of positions in the width direction of the glass plate may be used, for example, 10 positions may be used.

[0060] FIG. 8 is a diagram showing an example of a procedure of a process performed in the glass sheet manufacturing apparatus 1 according to the embodiment.

[0061] (Step S1) In the glass sheet manufacturing apparatus 1, the determination unit 271 acquires input information (information on the detection results of the manufacturing conditions in this embodiment) to the trained model 311. Then, the process proceeds to step S2.

[0062] (Step S2) In the glass sheet manufacturing apparatus 1, the determination unit 271 inputs the acquired input information to the trained model 311. Then, the process proceeds to step S3.

[0063] (Step S3) In the glass sheet manufacturing apparatus 1, the determination unit 271 acquires output information (information on the estimation result of the sheet thickness in this embodiment) from the trained model 311. Then, the process proceeds to step S4.

[0064] (Step S4) In the glass sheet manufacturing apparatus 1, the control unit 215 controls the adjustment of the manufacturing conditions based on the output information acquired by the determination unit 271. Then, the process proceeds to step S5.

[0065] (Step S5) In the glass sheet manufacturing apparatus 1, the control unit 215 determines whether or not to end the processing of this flow. As a result of this determination, when it is determined that the process of this flow is to be ended (step S5: YES), control unit 215 ends the process of this flow. On the other hand, as a result of this determination, when the control unit 215 determines not to end the processing of this flow (step S5: NO), the process proceeds to step S1.

[0066] Here, in the example of Figure 8, for ease of explanation, a case is shown in which in the processing of step S5, control unit 215 determines whether or not to terminate the processing of this flow, but any configuration may be used as a configuration for determining whether or not to terminate the processing of this flow.

[0067] In addition, in the example of Figure 8, a processing flow including a processing step (step S4) of controlling the adjustment of manufacturing conditions by adjustment unit 123 is shown, but as another example, a processing flow that does not include such a processing step (step S4) may be executed.

[0068] <Other examples of input / output information for trained models> FIG. 9 is a diagram illustrating another example of input / output information of the trained model 311a according to the embodiment. Here, the trained model 311a is a modified version of the trained model 311. Trained model 311a is generally similar to trained model 311, except for the input information, and here we will explain in detail the differences and omit a detailed explanation of the similarities. The example of FIG. 9 illustrates a case where information on the detection result of the thickness of a glass plate is included in the input information to the trained model 311a.

[0069] Information on the detection results of manufacturing conditions is input as input information to the trained model 311a. The input information may be various information, and in this example, includes information on the detection results of plate thickness other than the purpose of estimation. Here, the plate thickness other than the estimation target is a plate thickness other than the target plate thickness to be estimated by the trained model 311a (i.e., the plate thickness itself that is desired to be obtained as the estimation result).

[0070] For example, if the thickness to be estimated is the thickness at a specific location on the glass plate at a specific timing (in the example of Figure 1, the thickness at the specific location on the glass plate detected by the thickness detection unit 121 at the time the specific location on the glass plate is actually manufactured), then as a thickness other than the thickness to be estimated, for example, the thickness at a different location on the glass plate (a location other than the specific location), or the thickness at the same location on the glass plate (the specific location) but at a different timing (a timing other than the specific timing) is used.

[0071] The trained model 311a outputs information on the estimation result of the plate thickness as output information. The information is, for example, information on the plate thickness estimated to be detected by the plate thickness detection unit 121. Note that other information (for example, converted information) may be used as the input information and the output information. Here, the trained model 311a may be generated using any learning method.

[0072] In addition, in this example, a case is shown in which the trained model 311a that has been trained before the judgment is made by the judgment unit 271 is used, but as another example, a trained model (which is also considered as an example of the trained model 311a in this embodiment) in which the already trained model contents are sequentially updated (corrected) in a time series may be used. As such a trained model, for example, a state space model may be used. In this way, as the trained model 311a, a training model in which training has already been completed and the model content is fixed may be used, or a training model in which the model content is further updated sequentially over time may be used.

[0073] <Time series information on sheet thickness during glass sheet manufacturing> The input information of the trained model 311a may include time-series information on the thickness of the glass sheet during its manufacture.

[0074] Here, as the time-series information on the sheet thickness during the manufacture of the glass sheet, for example, information on the sheet thickness at a plurality of times for a portion of the glass sheet (glass component) that is to be estimated for the sheet thickness may be used. These multiple times may be, for example, times at regular intervals, or may be times at other intervals.

[0075] The start and end times of these multiple times may vary. As an example, the time when the production of the glass component begins (or a time around that time) may be used as the start time, and the time when the production of the glass component ends (or a time around that time) may be used as the end time. As another example, focusing on a specific step in the glass manufacturing process (e.g., a manufacturing step in float bath 12), the time when the manufacturing of the glass component begins (or a time around that time) in that step may be used as the start time, and the time when the manufacturing of the glass component ends (or a time around that time) may be used as the end time.

[0076] As a specific example, if the production of a glass component part of a glass plate takes one hour from the start of production to the end of production, information on the thickness of the glass component part for each minute from the start of production to the end of production (i.e., information on the thickness at 60 different times) may be used as time-series information on the thickness of the glass plate during production.

[0077] In this embodiment, machine learning is performed under the assumption that each glass component part in the glass plate flows linearly along the transport direction of the glass plate (in this embodiment, a direction parallel to the X-axis). However, in reality, each glass component part may flow along a path that has a portion bent in the width direction (a path that curves in the width direction), so a configuration that performs machine learning taking such a path into account may be used. However, even if it is assumed, as in this embodiment, that each glass component part of the glass plate flows linearly along the transport direction of the glass plate (in this embodiment, a direction parallel to the X-axis), by performing machine learning under that assumption, it is possible to obtain an estimation result with sufficient accuracy for practical use as the output result from the trained model.

[0078] [Example of a heater in a float bath using the float method] 10 and 11, an example of a heater in a float bath 810 of a float type will be described. In this example, a configuration example of heater division is shown.

[0079] 10 and 11 may be applied to, for example, the float bath 12 according to the embodiment. In this case, the adjustment unit 123 shown in Fig. 1 may adjust the heating state of one or more of the heaters 830 shown in Fig. 10 as the adjustment of the manufacturing conditions. Also, with regard to the configurations other than the heater in the float bath 810 shown in FIGS. 10 and 11, some or all of them may be applied to the float bath 12 according to the embodiment.

[0080] FIG. 10 is a cross-sectional view showing a portion of a float type glass sheet manufacturing apparatus 1. 11 and 12 are plan views showing the float bath. For convenience of explanation, FIG. 10 shows an XYZ orthogonal coordinate system similar to that shown in FIG. The cross-sectional view shown in FIG. 10 is an XZ cross-sectional view.

[0081] As in FIG. 1, the Z-axis direction is the up-down direction, the X-axis direction is the length direction of the float bath 810, and the Y-axis direction is the width direction of the float bath 810. 10 to 12, and in this embodiment, the direction in which the glass ribbon GR (glass sheet) is transported. The width direction of the float bath 810 is the vertical direction in Fig. 10 to 12, and is perpendicular to the direction in which the glass ribbon GR is transported.

[0082] The conveying direction of the glass ribbon GR is the direction in which the glass ribbon GR is conveyed in a plan view. In addition, the upstream side and downstream side refer to the +X side (positive side of the X-axis) in the conveying direction (X-axis direction) of the glass ribbon GR inside the glass sheet manufacturing apparatus 1, which is the downstream side, and the -X side (negative side of the X-axis) is the upstream side.

[0083] The inside in the width direction (Y-axis direction) is the side on which the center of the width direction of the float bath 810 is located. The outside in the width direction is the side opposite to the side on which the center of the width direction of the float bath 810 is located. In the following description, unless otherwise specified, the width direction means the width direction of the float bath 810 and the width direction of the glass ribbon GR, and the conveying direction means the conveying direction of the glass ribbon GR.

[0084] The roof 811b is omitted from illustration in Figures 11 and 12. Furthermore, the top rolls 820 to 829 are omitted from illustration in Figure 12. The glass sheet manufacturing apparatus 1 includes a float bath 810 having an internal space AR. An opening 810a of the float bath 810 is shown in FIG. The float bath 810 is a device that forms a glass ribbon GR inside an internal space AR. A melting furnace (melting furnace 11 shown in FIG. 1) is connected to the upstream side (-X side) of the float bath 810. The melting furnace supplies molten glass Gm onto the surface Ma of the molten metal M stored in the float bath 810 from the upstream side through a lip 819. The molten glass Gm is continuously supplied onto the surface Ma of the molten metal M while the flow rate is controlled by a tweezers 818.

[0085] An annealing furnace (annealing furnace 13 shown in FIG. 1) is connected to the downstream side (+X side) of the float bath 810. The annealing furnace cools the glass ribbon GR formed in the float bath 810. The annealing furnace is provided with, for example, a plurality of transport rolls. The glass ribbon GR is transported from the upstream side (-X side) to the downstream side by being pulled downstream by the transport rolls of the annealing furnace. That is, the glass sheet manufacturing apparatus 1 continuously supplies molten glass Gm onto a surface Ma of molten metal M stored in a float bath 810 to form a glass ribbon GR, and transports the glass ribbon GR in the transport direction (X-axis direction).

[0086] The float bath 810 has a bottom 811a, a roof 811b, a plurality of heaters 830, and a plurality of control devices 840. The float bath 810 also has a plurality of top rolls 820 to 829, as shown in FIG. 10, the molten metal M is stored in the bottom 811a. The roof 811b covers the upper side (+Z side) of the bottom 811a. A plurality of heaters 830 are provided on the roof 811b. Each heater 830 is controlled by a control device 840 and is capable of heating the glass ribbon GR formed on the molten metal M. Each part of the float bath 810 will be described in detail below.

[0087] <Bottom> The bottom 811 a has a bottom body 812 and a bottom casing 813 . A storage tank 812a that is concave on the lower side (-Z side) is formed on the upper side (+Z side) of the bottom body 812. Molten metal M is stored inside the storage tank 812a. The molten metal M is, for example, molten tin, molten tin alloy, etc. The material of the bottom body 812 is, for example, clay bricks, etc. The bottom casing 813 covers the outer surface of the bottom body 812. The bottom casing 813 is made of, for example, steel.

[0088] <Roof> The roof 811b is disposed on the upper side (+Z side) of the bottom 811a. The roof 811b has a roof casing 815 and a roof brick layer 816. The roof casing 815 is suspended from a superstructure (not shown), such as a beam of a building in which the float bath 810 is installed. The roof casing 815 is box-shaped and opens on the lower side (-Z side). The roof casing 815 is provided with a gas inlet (not shown). The roof casing 815 is made of, for example, steel.

[0089] A sidewall (not shown) is fixed to the inside of the lower side (-Z side) of the roof casing 815. The material of the sidewall is, for example, a heat-insulating brick or sillimanite. The roof brick layer 816 is provided on the inner side of the sidewall. The roof brick layer 816 is constructed by placing approximately rectangular parallelepiped brick blocks called PBA on a lattice-shaped framework (not shown). The lattice-shaped framework is suspended, for example, from the inner top surface of the roof casing 815. This allows the roof brick layer 816 to be positioned at a desired height.

[0090] The roof brick layer 816 divides the interior space AR of the float bath 810 into a lower space AR1 and an upper space AR2. The lower space AR1 is a space located between a bottom 811a and a roof 811b in the internal space AR of the float bath 810. The lower space AR1 is in contact with the molten metal M and the glass ribbon GR formed on the molten metal M.

[0091] The upper space AR2 is a space located inside the roof 811b of the internal space AR of the float bath 810. The upper space AR2 is located above (on the +Z side of) the roof brick layer 816 and the sidewall. The upper space AR2 communicates with the lower space AR1 via a through-hole provided in the roof brick layer 816. A side seal (not shown) is provided between the bottom 811a and the roof 811b in the vertical direction. The side seal seals the gap between the bottom 811a and the roof 811b in the vertical direction. This forms a substantially sealed internal space AR surrounded by the bottom 811a, the roof 811b, and the side seal.

[0092] <Heater> The heaters 830 are provided above the bottom 811a. For example, several thousand heaters 830 are provided in the float bath 810. The heaters 830 extend in the vertical direction (Z-axis direction). The heaters 830 are provided penetrating the roof brick layer 816. The heaters 830 are, for example, in the shape of a solid cylinder. The material of the heaters 830 is, for example, silicon carbide (SiC).

[0093] 12, the heater area ARH in which the multiple heaters 830 are arranged when the float bath 810 is viewed from above has multiple sections divided along the transport direction (X-axis direction) and the width direction (Y-axis direction) perpendicular to the transport direction of the glass ribbon GR formed on the molten metal M. In the example of Fig. 12, the heater area ARH is divided into rows A to H along the transport direction, and rows B to G are divided into five sections along the width direction.

[0094] Each compartment is provided with a plurality of heaters 830. The plurality of heaters 830 provided in one compartment are connected to the same control device 840. That is, the plurality of heaters 830 provided in one compartment are collectively controlled by one control device 840. The plurality of heaters 830 in one compartment are installed, for example, such that the amount of heat radiated from the plurality of heaters 830 toward the glass ribbon GR is approximately uniform within one compartment. The heater area ARH is an area located inside the storage tank 812a when viewed from above.

[0095] <Control device> The control device 840 controls the heater 830. The control device 840 is provided for each section of the heater area ARH. The control device 840 collectively controls the heaters 830 provided in the corresponding section. The configuration of the control device 840 is not particularly limited as long as it can collectively adjust the output of the heaters 830 in one section.

[0096] <Top roll> As shown in FIG. 11, the top rolls 820 to 829 are arranged to face each other on both sides of the width direction (Y-axis direction) of the glass ribbon GR. The top rolls 820 to 829 prevent the width of the glass ribbon GR from narrowing due to surface tension. The number of top rolls to be installed is appropriately set according to the forming conditions such as the type of glass or the target thickness. The number of top rolls to be installed is, for example, 4 pairs or more and 30 pairs or less, preferably 10 pairs or more and 30 pairs or less. In the example of FIG. 11, a total of five pairs of top rolls are provided, including top rolls 820, 825, top rolls 821, 826, top rolls 822, 827, top rolls 823, 828, and top rolls 824, 829. The number of top rolls to be installed tends to increase as the plate thickness of the float glass to be manufactured is reduced.

[0097] In addition, since the top rolls 820 to 829 have the same configuration except for the positions at which they are provided, in the following explanation, there are cases where only the top roll 824 is explained as a representative example. The top roll 824 is composed of a top roll body 824A that supports the ends of the glass ribbon GR in the width direction (Y-axis direction), and a rotation shaft 824B that is connected to the top roll body 824A.

[0098] These top rolls 820 to 829, more precisely, the top roll body 824A are provided in a forming area ARF for the glass ribbon GR. 4.5 dPa s or more, 10 7.5 The forming area ARF corresponds to the area where rows B to G are provided in the heater area ARH.

[0099] The forming zone ARF has an upstream forming zone ARF1 and a downstream forming zone ARF2. The upstream forming zone ARF1 is a zone located on the upstream side (−X side) of the forming zone ARF. The upstream forming zone ARF1 is a zone located on the upstream side (−X side) of the forming zone ARF. 4.5 dPa s or more and 10 5.7 dPa s. In this example, the upstream forming zone ARF1 corresponds to a zone in the heater zone ARH where rows B to E are provided. In the upstream forming zone ARF1, the output of the heaters 830 in each section of rows B to E is adjusted to adjust the sheet thickness of the glass ribbon GR.

[0100] The downstream forming zone ARF2 is a zone located on the downstream side (+X side) of the forming zone ARF. The downstream forming zone ARF2 is a zone located on the downstream side (+X side) of the forming zone ARF. 5.7 dPa s or more, 10 7.5 dPa s or less. That is, the viscosity of the glass ribbon GR in the downstream forming zone ARF2 is higher than the viscosity of the glass ribbon GR in the upstream forming zone ARF1. In this example, the downstream forming zone ARF2 corresponds to the zone where the rows F and G are provided in the heater zone ARH.

[0101] In the downstream forming zone ARF2, the output of the heaters 830 in each section of rows F and G is adjusted so that the temperature of the glass ribbon GR, whose thickness has been adjusted in the upstream forming zone ARF1, gradually decreases from the upstream side (-X side) to the downstream side (+X side).

[0102] 11 and 12, for convenience of explanation, a division line PL1 that divides two adjacent rows in the transport direction (X-axis direction) is shown. The division line PL1 is located approximately in the center between the heaters 830 adjacent to each other in the transport direction. The division line PL1 is, for example, a straight line extending in the width direction. 11 and 12, for convenience of explanation, a dividing line PL2 that divides two sections adjacent to each other in the width direction (Y-axis direction) is shown. The dividing line PL2 is located approximately in the center between the heaters 830 adjacent to each other in the width direction of the glass ribbon GR. The dividing line PL2 is, for example, a straight line extending in the conveying direction. The dividing line PL2 includes internal dividing lines PL2a to PL2d.

[0103] For convenience of explanation, flow lines FL1 and FL2 of the glass ribbon GR are shown in FIG. In the example of FIG. 12, the flow line FL1 has a portion where it bends so as to bulge to one side in the width direction (+Y side) with respect to the transport direction (X-axis direction). Furthermore, in the example of FIG. 12, the flow line FL2 has a portion where it bends so as to bulge toward the other side in the width direction (-Y side) with respect to the transport direction (X-axis direction). The multiple heaters 830 may be arranged, for example, along each flow line (e.g., flow line FL1, flow line FL2) of the glass ribbon GR and may reduce (e.g., minimize) the temperature difference in the direction of the flow line.

[0104] Here, when the heater configuration in the float bath 810 shown in Figures 10 and 11 is applied to the float bath 12 according to the embodiment, for example, the function of the control device 840 may be provided in the adjustment unit 123 shown in Figure 1.

[0105] As described above, in the glass sheet manufacturing apparatus 1 according to the present embodiment, the thickness of the glass sheet that is finally manufactured (or information related thereto) can be estimated by a machine learning model that inputs the manufacturing conditions of the glass sheet during manufacturing. As a result, in the glass sheet manufacturing apparatus 1, when the thickness of the manufactured glass sheet is likely to deviate from a certain range (a range having a width, including a case where the width is a constant value that can be regarded as 0), the quality of the thickness of the manufactured glass sheet can be maintained by quickly adjusting the temperature, etc.

[0106] Furthermore, in the glass sheet manufacturing apparatus 1 according to this embodiment, the manufacturing conditions of the glass sheet can be adjusted based on the results of this estimation. For example, when it is estimated that a thickness defect will occur, the temperature, etc. can be adjusted to correct the defect early, thereby minimizing the period of yield reduction.

[0107] Moreover, in the glass sheet manufacturing apparatus 1 according to the present embodiment, it is possible to estimate the change over time in the thickness of the glass sheet (or information relating thereto), thereby making it possible to continuously monitor in real time for the occurrence of thickness defects, for example. Moreover, in the glass sheet manufacturing apparatus 1 according to the present embodiment, the manufacturing conditions of the glass sheet can be continuously adjusted based on the results of the estimation, and the occurrence of defective sheet thickness can be suppressed over a long period of time.

[0108] As one configuration example, in a glass plate manufacturing apparatus 1, the information processing device 101 has a judgment unit 271 that inputs information on the manufacturing conditions during the manufacture of a glass plate, which is manufactured while being flowed in a predetermined flow direction (conveying direction), into a machine learning trained model 311, and obtains the output result from the trained model 311 as information on the estimated result of the plate thickness of the glass plate. Therefore, in the glass sheet manufacturing apparatus 1, the thickness of the glass sheet that is finally manufactured (or information related thereto) can be estimated by a machine learning model (trained model 311) that inputs the manufacturing conditions of the glass sheet during manufacturing.

[0109] As one configuration example, in the glass sheet manufacturing apparatus 1, in the information processing device 101, information on manufacturing conditions input to the trained model 311 includes information on the thickness of the glass sheet when the glass sheet is manufactured. Therefore, in the glass sheet manufacturing apparatus 1, by including information on the thickness of the glass sheet at the time of manufacture as information to be input to the trained model 311, the accuracy of the output (estimation result) from the trained model 311 can be improved.

[0110] As one configuration example, in the glass sheet manufacturing apparatus 1, in the information processing device 101, information on the thickness of the glass sheet during the manufacturing of the glass sheet includes time-series information on the thickness of the glass sheet during the manufacturing of the glass sheet. Therefore, in the glass sheet manufacturing apparatus 1, by including time series information of the thickness of the glass sheet at the time of manufacturing as information to be input to the trained model 311, the accuracy of the output (estimated result) from the trained model 311 can be further improved.

[0111] As one configuration example, in the glass sheet manufacturing apparatus 1, the adjustment unit 123 adjusts the manufacturing conditions based on the estimation result by the determination unit 271. Therefore, in the glass sheet manufacturing apparatus 1, the thickness of the manufactured glass sheet can be maintained within a constant range.

[0112] Here, the thickness of a glass sheet is an important quality in the production of glass sheets. In conventional manufacturing processes, the thickness of a glass sheet is constantly monitored and kept constant by adjusting the temperature in a glass sheet forming device, etc. Here, the thickness monitored in the conventional method is the thickness measured by an inspection machine after the temperature of the glass sheet has dropped downstream of the forming device. In other words, conventionally, the thickness is detected at a position downstream of the forming device (for example, a position near the exit) and judged to be non-defective, and the information is fed back to temperature adjustment in a float bath, thereby maintaining the thickness constant. However, in such a conventional monitoring method, there is a large time difference (for example, 30 minutes) between the position where the temperature adjustment is performed and the position where the quality is judged, so if the temperature adjustment is incomplete for some reason, the correction may be significantly delayed. In other words, during the time from when the inspection machine judges the thickness defect to when the temperature adjustment is corrected, glass sheets with thickness defects continue to be manufactured, resulting in a prolonged decrease in yield. Therefore, in the present embodiment, a machine learning model is used to estimate the distribution of thickness of the glass sheet in the float bath using parameters related to the manufacture of the glass sheet, thereby making it possible to correct defects in the thickness distribution at an early stage and minimize the period of yield reduction.

[0113] In the above embodiment, a case where a glass sheet is manufactured by a float method has been described. However, the technology according to the above embodiment may be applied to a case where a glass sheet is manufactured by another method, for example, may be applied to a case where a glass sheet is manufactured by a fusion method. Furthermore, the techniques according to the above embodiments may be applied, for example, to the manufacture of various types of glass plates, and as one example, may be applied to the manufacture of glass plates used in flat panel displays (FPDs).

[0114] A program for implementing the functions of any of the components in any of the above-described devices may be recorded in a computer-readable recording medium, and the program may be read into a computer system and executed. The term "computer system" as used herein includes hardware such as an operating system or peripheral devices. The term "computer-readable recording medium" refers to portable media such as flexible disks, optical magnetic disks, ROMs, and CDs (Compact Discs)-ROMs (Read Only Memory), and storage devices such as hard disks built into computer systems. The term "computer-readable recording medium" also refers to storage devices that hold a program for a certain period of time, such as volatile memory in a computer system that serves as a server or client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. The volatile memory may be, for example, a RAM (Random Access Memory). The recording medium may be, for example, a non-transitory recording medium.

[0115] The above-mentioned program may be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium, or by a transmission wave in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium having a function of transmitting information, such as a network such as the Internet or a communication line such as a telephone line. The above program may be for implementing some of the above functions. Furthermore, the above program may be a so-called differential file that can implement the above functions in combination with a program already recorded in the computer system. The differential file may be called a differential program.

[0116] In addition, the function of any of the components in any of the above-described devices may be realized by a processor. For example, each process in the embodiment may be realized by a processor that operates based on information such as a program and a computer-readable recording medium that stores information such as a program. Here, the functions of each part of the processor may be realized by individual hardware, or the functions of each part may be realized by integrated hardware. For example, the processor may include hardware, and the hardware may include at least one of a circuit that processes a digital signal and a circuit that processes an analog signal. For example, the processor may be configured using one or more circuit devices mounted on a circuit board, or one or both of one or more circuit elements. An IC (Integrated Circuit) or the like may be used as the circuit device, and a resistor or a capacitor or the like may be used as the circuit element.

[0117] Here, the processor may be, for example, a CPU. However, the processor is not limited to a CPU, and various processors such as a GPU (Graphics Processing Unit) or a DSP (Digital Signal Processor) may be used. The processor may be, for example, a hardware circuit using an ASIC (Application Specific Integrated Circuit). The processor may be, for example, a plurality of CPUs, or a hardware circuit using a plurality of ASICs. The processor may be, for example, a combination of a plurality of CPUs and a hardware circuit using a plurality of ASICs. The processor may include, for example, one or more of an amplifier circuit or a filter circuit that processes an analog signal.

[0118] Although the embodiments of this disclosure have been described in detail above with reference to the drawings, the specific configuration is not limited to this embodiment, and designs that do not deviate from the gist of this disclosure are also included.

[0119] [Note] [Configuration Example 1] to [Configuration Example 6] are shown.

[0120] [Configuration example 1] A determination unit is provided that inputs information on manufacturing conditions during the manufacture of a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and acquires an output result from the trained model as information on an estimation result of a thickness of the glass sheet. Information processing device.

[0121] [Configuration example 2] The information on the manufacturing conditions input to the trained model includes information on the thickness of the glass sheet at the time of manufacturing the glass sheet. The information processing device according to [Configuration Example 1].

[0122] [Configuration example 3] The information on the thickness of the glass sheet at the time of manufacturing the glass sheet includes time-series information on the thickness of the glass sheet at the time of manufacturing the glass sheet. The information processing device according to [Configuration Example 2].

[0123] [Configuration example 4] An information processing device including a determination unit that inputs information on manufacturing conditions at the time of manufacturing a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and acquires an output result from the trained model as information on an estimation result of a thickness of the glass sheet; and an adjustment unit that adjusts the manufacturing conditions based on the estimation result by the determination unit. Glass sheet manufacturing equipment.

[0124] For example, a processing method performed in an information processing device can be implemented. [Configuration example 5] A determination unit inputs information on manufacturing conditions at the time of manufacturing a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information on an estimation result of a thickness of the glass sheet. Information processing methods.

[0125] For example, a method of processing carried out in a glass sheet manufacturing apparatus can be carried out. [Configuration example 6] A determination unit inputs information on manufacturing conditions at the time of manufacturing a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information on an estimation result of a thickness of the glass sheet; an adjustment unit adjusts the manufacturing conditions based on the estimation result by the determination unit; Glass sheet manufacturing method. [Explanation of symbols]

[0126] 1...glass sheet manufacturing apparatus, 11...melting furnace, 12, 810...float bath, 13...annealing furnace, 14...cleaning inspection and cutting section, 101...information processing device, 121...sheet thickness detection section, 122...manufacturing condition detection section, 123...adjustment section, 141...operator, 211...input section, 212...output section, 213...communication section, 214...storage section, 215...control section, 231...operation section, 251...display section, 271...determination section, 311, 311a...trained model, 511...manufacturing condition, 531...temperature condition, 532...pressure condition, 533...tweel condition, 534...heater condition, 535...humidity condition, 536...sheet thickness condition, 611...glass sheet, 810a...opening, 811a...bottom, 811b...roof, 812...bottom Main body, 812a...storage tank, 813...bottom casing, 815...roof casing, 818...tweel, 819...lip, 820-829...top roll, 824A...top roll main body, 824B...rotating shaft, 830...heater, 840...control device, 2011, 2110, 2121-2122, 2131-2132...characteristics, AR...internal space, AR1...lower space, AR2...upper space, ARF...forming area, ARF1...upstream forming area, ARF2...downstream forming area, ARH...heater area, FL1-FL2...flow lines, Gm...molten glass, GR...glass ribbon (glass sheet), M...molten metal, Ma...surface, PL1...partition line, PL2...partition line, PL2a-PL2d...inner partition line

Claims

1. A determination unit is provided that inputs information on manufacturing conditions during the manufacture of a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and acquires an output result from the trained model as information on an estimation result of a thickness of the glass sheet. Information processing device.

2. The information on the manufacturing conditions input to the trained model includes information on the thickness of the glass plate at the time of manufacturing the glass plate. The information processing device according to claim 1 .

3. The information on the thickness of the glass sheet at the time of manufacturing the glass sheet includes time-series information on the thickness of the glass sheet at the time of manufacturing the glass sheet. The information processing device according to claim 2 .

4. An information processing device including a determination unit that inputs information on manufacturing conditions at the time of manufacturing a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and acquires an output result from the trained model as information on an estimation result of a thickness of the glass sheet; and an adjustment unit that adjusts the manufacturing conditions based on the estimation result by the determination unit. Glass sheet manufacturing equipment.

5. A determination unit inputs information on manufacturing conditions at the time of manufacturing a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information on an estimation result of a thickness of the glass sheet. Information processing methods.

6. A determination unit inputs information on manufacturing conditions at the time of manufacturing a glass sheet, the glass sheet being manufactured while being flowed in a predetermined flow direction, into a machine learning trained model, and obtains an output result from the trained model as information on an estimation result of a thickness of the glass sheet; an adjustment unit adjusts the manufacturing conditions based on the estimation result by the determination unit; Glass sheet manufacturing method.

Images