Method and device for controlling the temperature of the crown of a glass melting furnace

Abstract

The application discloses a kind of control method and device for balancing glass melting furnace dome top temperature, method includes: obtaining sample data, including the gas flow data of each branch input to glass melting furnace;And the dome top temperature output data of the glass melting furnace;Establish temperature control neural network model, according to the sample data training the temperature control neural network model, obtain temperature control best model;According to the temperature control best model, optimize the input value of each branch gas flow data distribution;After collecting the input value of each branch gas flow data after optimized distribution, the dome top temperature of the glass melting furnace in real time, and compared with the dome top temperature data required by process, obtain comparison data;According to the comparison data, balance multiple dome top temperatures in the glass melting furnace.The control method and device for balancing glass melting furnace dome top temperature disclosed by the application can balance the dome top temperature of glass melting furnace.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of glass manufacturing, in particular to a control method and device for balancing the temperature of the dome top of a glass melting furnace. BACKGROUND

[0002] For a glass melting furnace of a new material glass, the melting quality of the glass liquid is crucial, directly determining the glass quality and yield. However, the glass liquid melting process is a process that consumes a lot of energy. For a new material glass, the energy consumption cost of the glass liquid melting accounts for about 70% of the glass production cost. Because the melting temperature of the new material glass is much higher than that of the traditional float glass and the calendaring glass, the energy consumption and the combustion method are also quite different. In addition, the stability of the temperature control of each interval of the new material glass is very important. In the case of large temperature changes, the pipe drawing is prone to quality defects such as nodules, inclusions, and bubbles. However, the space of the glass melting furnace is very large, and different tonnage furnaces need 4-8 dome top temperature measuring points, and the temperature control range is also different. The traditional control method mostly adopts single-loop PID (Proportional Integral Derivative) control. The input of the fuel gas and the output of the dome top temperature are in a one-to-one correspondence, and there is no internal relationship between them. It must be observed and adjusted manually several times to obtain a relatively stable value. However, the reality is that the input of the fuel gas and the output of the dome top temperature are in a many-to-many relationship, and there is a certain correlation between the parameters. For example, when the flame of a small furnace abnormally changes, adjusting the fuel gas flow of the corresponding small furnace will cause changes in other temperature points. Through manual adjustment, it is not conducive to the stability of the entire melting process, and it is easy to cause continuous disturbance of melting, unstable temperature and bubble boundary line, and other shortcomings.

[0003] In the prior art, the patent application with the application publication number CN111377595A discloses a method and system for real-time control of the fuel gas supply amount of a glass melting furnace. The method includes: first, obtaining the space temperature in the glass melting tank at the current time; second, inputting the space temperature in the glass melting tank into the fuel gas supply chain model of the DCS combustion system, and outputting the fuel gas supply amount at the current time; and third, the data center sends a command to the adjusting valve at the fuel gas pipe of the glass melting furnace according to the fuel gas supply amount at the current time, and adjusts the opening degree of the adjusting valve. In the prior art, when a certain dome top temperature fluctuates, the gas flow corresponding to the change of the dome top temperature needs to be adjusted to ensure the stability of the dome top temperature. However, the change of the gas flow will cause fluctuations in other dome top temperatures. This results in fluctuations in one dome top temperature, which requires manual observation and adjustment of the gas flow at different positions several times to finally achieve the stability of each dome top temperature. SUMMARY

[0004] The technical problem solved by the present application is to solve the problem that when the temperature of the glass furnace dome top fluctuates, the input amount of each gas needs to be observed and adjusted manually multiple times to balance the temperature of each dome top.

[0005] To solve the above technical problems, the present application provides the following technical solutions:

[0006] A control method for balancing the temperature of the glass furnace dome top, comprising:

[0007] Obtaining sample data, including the gas flow data input into the glass furnace by each branch, and the dome top temperature output data of the glass furnace;

[0008] Establishing a temperature control neural network model, training the temperature control neural network model according to the sample data, and obtaining a temperature control optimal model;

[0009] According to the temperature control optimal model, optimizing the input value of the gas flow data of each branch;

[0010] After the input value of the gas flow data of each branch is optimized, the real-time dome top temperature of the glass furnace is collected and compared with the process required dome top temperature data to obtain comparison data;

[0011] According to the comparison data, balancing the multiple dome top temperatures in the glass furnace.

[0012] Advantages: The temperature control neural network model is trained by sample data to find the correlation between the input value of the gas flow data and the dome top temperature, and a temperature control optimal model is obtained. The temperature control optimal model optimizes the input value of the gas flow data of each branch to balance the dome top temperature.

[0013] In an embodiment of the present application, the temperature control optimal model comprises the following steps:

[0014] The gas flow data is used as the input amount, the input amount is normalized to form an input matrix;

[0015] An implicit layer output matrix from the input layer to the implicit layer is obtained;

[0016] According to the implicit layer output matrix, an output layer output matrix is obtained;

[0017] According to the implicit layer output matrix, the output layer output matrix and the actual output matrix, a loss function is obtained;

[0018] According to the loss function, an optimal weight matrix is obtained;

[0019] According to the optimal weight matrix, the temperature control optimal model is obtained.

[0020] In an embodiment of the present application, the optimal weight matrix is obtained by the following formula:

[0021]

[0022] wherein, w (l) represents the optimal weight matrix from the l-1th layer to the lth layer, represents the optimal loss function, and E represents the total error.

[0023] In an embodiment of the present application, the balancing of the multiple crown top temperatures in the glass melting furnace comprises:

[0024] According to the comparison data, it is determined whether the gas flow input value input to each branch of the glass melting furnace needs to be corrected. If the comparison data is within the range of -1℃ to 1℃, the gas flow input value does not need to be corrected. At this time, the controller controls the gas regulating valve on each branch of the glass melting furnace, so that the gas flow input value of each branch is consistent with the gas flow input value of each branch output by the temperature control optimal model.

[0025] In an embodiment of the present application, the balancing of the multiple crown top temperatures in the glass melting furnace further comprises:

[0026] If the comparison data is within the range of -1℃ to -3℃ and 1℃ to 3℃, the input value of the gas flow data needs to be corrected. The comparison data is input as a traditional PID, the input value of the gas flow output by the traditional PID is combined with the gas flow input value distributed by the temperature control optimal model, and the adjustment data value of the gas flow is obtained. The controller adjusts the opening degree of the gas regulating valve according to the adjustment data value.

[0027] In an embodiment of the present application, the balancing of the multiple crown top temperatures in the glass melting furnace further comprises: if the comparison data is outside the range of -3℃ to 3℃, the gas flow input value is manually adjusted according to artificial experience at this time.

[0028] In an embodiment of the present application, the input value of the gas flow output by the traditional PID is obtained by the following formula:

[0029]

[0030] wherein, F flow represents the input value of the gas flow output by the traditional PID, G ain represents the gain coefficient, T S represents the integral time, T D represents the differential time, and D iffGain represents the action delay time, s represents the complex number of Laplace transform, and ΔT1 represents the comparison data.

[0031] In an embodiment of the present application, the adjustment data value of the gas flow is obtained by the following formula:

[0032] MV∝F flow +F ac ;

[0033] In the formula, MV represents the adjustment data value, F ac represents the gas flow input value allocated by the temperature control optimal model.

[0034] The present application also provides a device for balancing the crown temperature of a glass melting furnace, comprising:

[0035] an input data control device for obtaining the gas flow data input into the glass melting furnace by each branch;

[0036] a thermocouple for obtaining the crown temperature output data of the glass melting furnace; sample data is obtained by the input data control device and the thermocouple;

[0037] a server in communication connection with the input data control device and the thermocouple, for establishing a temperature control neural network model, training the temperature control neural network model according to the sample data, obtaining a temperature control optimal model; according to the temperature control optimal model, optimizing the input value of the gas flow data of each branch; after the input value of the gas flow data of each branch is optimized, collecting the real-time crown temperature of the glass melting furnace, and comparing it with the crown temperature data required by the process, to obtain comparison data;

[0038] a controller in communication connection with the input data control device, the thermocouple and the server, for balancing the multiple crown temperatures in the glass melting furnace according to the comparison data.

[0039] In an embodiment of the present application, the input data control device comprises a gas regulating valve, which is in communication connection with the controller, and the opening degree of the gas regulating valve is controlled by the controller to adjust the gas input flow of each branch of the glass melting furnace.

[0040] Compared with the prior art, the beneficial effects of the present application are: the input values of the branch gas flow data are optimized by the temperature control optimal model to balance the multiple crown top temperatures in the glass melting furnace. If part of the corresponding output crown top temperatures of the input values of the branch gas flow data distributed by the temperature control optimal model does not meet the process requirements, the gas regulating valve is adjusted through the joint action of the traditional PID and the temperature control optimal model. The comparative data are taken as the input quantity of the traditional PID, the traditional PID outputs the input value of the gas flow, and the input value of the gas flow is combined with the input value of the gas flow distributed by the temperature control optimal model to obtain the adjustment data value, and the controller controls the opening degree of the gas regulating valve according to the adjustment data value, so that the input value of the gas flow is consistent with the adjustment data value. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 A control method flow chart for balancing the crown top temperature of a glass melting furnace is provided for an embodiment of the present application.

[0042] Figure 2 An input data control device schematic diagram is provided for an embodiment of the present application.

[0043] Figure 3 A thermocouple schematic diagram is provided for an embodiment of the present application.

[0044] Figure 4 A flow chart for obtaining a temperature control optimal model is provided for the present application.

[0045] Figure 5 A neural network model schematic diagram is provided for an embodiment of the present application.

[0046] Figure 6 A device schematic diagram for balancing the crown top temperature of a glass melting furnace is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0047] In order to facilitate those skilled in the art to understand the technical solutions of the present application, the technical solutions of the present application will be further described in conjunction with the drawings in the specification.

[0048] The terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0049] Please refer to Figure 1 As shown in the drawings, the present application provides a control method for balancing the crown top temperature of a glass melting furnace, which comprises:

[0050] S100, obtaining sample data, including the gas flow data input to the glass furnace by each branch, and the crown top temperature output data of the glass furnace;

[0051] S200, establishing a temperature control neural network model, training the temperature control neural network model according to the sample data, and obtaining a temperature control optimal model;

[0052] S300, optimizing the input value of the gas flow data of each branch according to the temperature control optimal model;

[0053] S400, collecting the real-time crown top temperature of the glass furnace after the input value of the gas flow data of each branch is optimized, and comparing the crown top temperature with the process required crown top temperature data to obtain comparison data;

[0054] S500, balancing the multiple crown top temperatures in the glass furnace according to the comparison data.

[0055] Please refer to Figures 1 to 3 In an embodiment of the present application, in step S100, the glass liquid 1100 is located in the glass furnace 1000, and the input data control device 110 is located at the side of the glass furnace 1000. The input data control device 110 obtains the gas flow data input to the glass furnace by each branch. The number of input data control devices 110 is 8, i.e. 1# to 8#, and each input data control device 110 includes a gas control device 111 composed of a gas regulating valve 1111, a gas gun 1112 and a gas pressure transmitter 11113, and an oxygen control device 112 composed of an oxygen regulating valve 1121, an oxygen gun 1122 and an oxygen pressure transmitter 1123, as shown in Figure 6 . Among them, the gas gun 1112 delivers natural gas to the glass furnace 1000, the gas regulating valve 1111 controls the input natural gas flow of the gas gun 1112, and the gas pressure transmitter 11113 adjusts the pressure of the input natural gas of the gas gun 1112. The oxygen gun 1122 delivers oxygen to the glass furnace 1000, the oxygen regulating valve 11121 controls the input oxygen flow of the oxygen gun 1122, and the oxygen pressure transmitter 1123 adjusts the pressure of the input oxygen of the oxygen gun 1122. In this embodiment, it is assumed that the data of the oxygen control device 112 is constant, and the number of input data control devices 110 is set according to actual needs. The thermocouple 120 is located at the top of the glass furnace 1000 for measuring the crown top temperature of the glass furnace 1000 as output data. The gas control device 111 obtains the input gas flow data, and the thermocouple 120 obtains the output crown top temperature, and the sample data can be queried according to historical data.

[0056] Please refer to Figure 1 、 5 and Figure 6As shown, in an embodiment of the present application, step S200 further comprises the following steps:

[0057] S210, taking the gas flow data as input, normalizing the input and forming an input matrix.

[0058] The gas control device 111 collects data of gas flow, with a sampling interval of Δt, and after cleaning up abnormal data, the gas flow data is normalized by the following data.

[0059]

[0060] In the formula, x(n) represents the gas data collected by the nth group of gas control devices, minmax[x(n)] represents the normalized gas flow data, max[x(n)] represents the maximum value in the gas flow data, and min[x(n)] represents the minimum value in the gas flow data.

[0061] The input layer input matrix formed by normalizing the input is represents the input value of the ith neuron in the input layer. In this embodiment, the maximum value of i is 8, that is, there are 8 neuron inputs.

[0062] In this embodiment, the initial weight matrix expression from the input layer to the hidden layer is ω 1 , the hidden layer is two layers, including the first layer hidden layer and the second layer hidden layer, wherein the initial weight matrix expression from the first layer hidden layer to the second layer hidden layer is ω 2 , and the initial weight matrix expression from the second layer hidden layer to the output layer is ω 3 . The initial weight can be set by experience.

[0063] S220, obtaining the initial weight from the input layer to the hidden layer and the hidden layer output matrix.

[0064] wherein,

[0065]

[0066] That is, the output matrix expression of the first layer hidden layer is:

[0067] In the formula, F() is a transfer function, a 2 1~a 2 j represents the output value of the neuron in the first layer hidden layer, and Z ja represents the state of the first layer of hidden layers; j represents the number of nodes of the first layer of hidden layers, in this embodiment, the number of nodes of the first layer of hidden layers is 10-16, that is, the first layer of hidden layers has 10-16 neuron output values, and the output matrix expression is a (2) ω ij represents the weight from the first layer of hidden layers to the second layer of hidden layers, and the matrix expression is ω 2 , and ∑ represents a summation formula.

[0068] wherein,

[0069]

[0070] that is, the output expression of the second layer of hidden layers is:

[0071] wherein, a 3 1-a 3 k represents the output value of the neuron of the second layer of hidden layers, and Z k represents the state of the neuron of the second layer of hidden layers. K represents the number of nodes of the second layer of hidden layers, which can also be set to 6-10, and the output matrix expression is a (3) ω jk represents the weight from the second layer of hidden layers to the output layer, and the matrix expression is ω (3) .

[0072] S230, obtaining an output layer output matrix according to the hidden layer output matrix.

[0073] wherein,

[0074]

[0075] that is, the output expression of the output layer output matrix is:

[0076] wherein, a 4 1-a 4 m is the output value of the neuron of the output layer, that is, the actual obtained top temperature value. m represents the neuron output value, in this embodiment, m is 4, that is, there are 4 top temperature neuron output values, and the output matrix expression is a (4) . represents the weight from the second layer of hidden layers to the output layer, and the matrix expression is ω 3 .

[0077] S240, obtaining a loss function according to the hidden layer output matrix, the output layer output matrix and the actual output matrix.

[0078] The output layer output matrix and the actual output matrix, and obtain the output layer error.

[0079] The actual output value matrix is measured by the transmitter, and the output layer error is obtained by the following formula:

[0080] δ (4) = a (4) -y;

[0081] In the formula, δ (4) represents the output layer error, a (4) represents the output layer output matrix, and y represents the actual output matrix.

[0082] Specifically, represents the error of the mth output value, represents the output value of the mth neuron in the output layer, and y m represents the actual output value of the output value of the mth neuron in the output layer.

[0083] Correspondingly, the second layer hidden layer error is:

[0084]

[0085]

[0086] In the formula, represents the output error of the second layer hidden layer, and the matrix form is:

[0087] δ (3) = (ω (3) ) T δ 4 *f'(z (3) ) ;

[0088] Similarly, the first layer hidden layer output error matrix is:

[0089] δ (2) = (ω (2) ) T δ 3 *f'(z (3) ) ;

[0090] The output error matrix of the input layer is: (1) = a (1) .

[0091] Then the total error E = {δ (1) , δ (2) , δ (3) , δ (4)}.

[0092] Then, partial derivatives of the total error E are obtained to obtain the loss function.

[0093]

[0094] In the formula, w is the loss function, and w l is the weight matrix from the l-1 layer to the l layer, and E is the total error.

[0095] S250, according to the loss function, the optimal weight matrix is obtained.

[0096] The weight parameters are iteratively updated by the following formula. According to experience, the training times are 10000, and the step length is 0.1. The mature weight value parameters are trained.

[0097]

[0098] In the formula, w (l) is the optimal weight matrix from the l-1 layer to the l layer, is the optimal loss function.

[0099] S260, according to the optimal weight matrix, the temperature control optimal model is obtained.

[0100] The above steps are trained by sample data, and are executed multiple times to finally find the optimal weight matrix and obtain the temperature control optimal model.

[0101] Please refer to Figure 1 and Figure 6 In an embodiment of the present application, in steps S300 and S400, the server 200 is connected to the input data control device 110 and the thermocouple 120 through the switch 234 to obtain the gas flow data input to the glass furnace from each branch and the temperature output data of the top of the glass furnace. The temperature control optimal model obtained after the neural network is trained is established on the server 200. According to the optimal weight matrix, the input value of the gas flow data of each branch is optimized and distributed. The real-time top temperature of the glass furnace after the input value of the gas data of each branch is optimized and distributed is collected, and compared with the process required top temperature data to obtain comparison data, so as to balance the multiple top temperatures in the glass furnace.

[0102] Please refer to Figure 1 and Figure 6 In an embodiment of the present application, in step S500, the controller 300 is connected to the input data control device 110, the thermocouple 120, the server 200 and the monitoring server 400 through the switch 234. The controller 300 is used to balance the multiple top temperatures in the glass furnace according to the comparison data.

[0103] The balance of the glass furnace includes:

[0104] According to the comparison data, it is determined whether the gas flow input value of each branch input to the glass furnace needs to be corrected. If the comparison data is within the range of -1℃ to 1℃, the gas flow input value does not need to be corrected, and at this time the controller controls the gas regulating valve on each branch of the glass furnace, so that the gas flow input value of each branch is consistent with the gas flow input value of each branch output by the temperature control optimal model.

[0105] In this embodiment, when the comparison data is within the range of -1℃ to 1℃, it means that the input value of the gas data of each branch allocated by the temperature control optimal model can be directly applied. The controller 300 can control the gas regulating valve 1111 to make the input value of the gas data of each branch consistent with the input value of the gas data of each branch allocated by the temperature control optimal model.

[0106] If the comparison data is within the range of -1℃ to -3℃ and 1℃ to 3℃, the input value of the gas flow data needs to be corrected. The comparison data is input to the traditional PID, the traditional PID outputs the input value of the gas flow, and the input value of the gas flow allocated by the temperature control optimal model is combined to obtain the adjustment data value of the gas flow. The controller adjusts the opening of the gas regulating valve according to the adjustment data value.

[0107] In this embodiment, when the comparison data is within the range of -1℃ to -3℃ and 1℃ to 3℃, it means that the comparison data of the real-time glass furnace crown temperature after the input value of the gas data of each branch allocated by the temperature control optimal model and the crown temperature data required by the process cannot meet all the requirements. The gas flow corresponding to the crown temperature with large deviation needs to be fine-tuned. The fine-tuning method is to use the traditional PID and the temperature control optimal model to jointly act on the gas regulating valve. For example, among the four crown temperatures, three meet the requirement of being within 1℃, and one is between 1℃ and 3℃. For the three crown temperatures that meet the process requirements, the input value of the gas can be set according to the allocation value of the temperature control optimal model. For the crown temperature between 1℃ and 3℃, we need to increase the adjusting means. The comparison data is input to the traditional PID, and the gas flow is output. The sum of the gas flow output by the traditional PID and the gas flow obtained by the temperature control optimal model is the adjustment data value. The controller 300 controls the opening of the gas regulating valve 1111 according to the adjustment data value, so as to achieve the effect of adjusting the gas flow. The adjustment data value and the corresponding crown temperature are fed back to the server 200, so as to train the temperature control optimal model again.

[0108] The input value of the gas flow output by the traditional PID is obtained by the following formula:

[0109]

[0110] wherein, F flow represents the input value of the gas flow of the traditional PID output, G ain represents a gain coefficient, T S represents an integral time, T D represents a differential time, D iffGain represents an action delay time, s represents a complex number of Laplace transform, and ΔT1 represents comparative data.

[0111] The input value of the gas flow of the traditional PID output is combined with the input value of the gas flow distributed by the temperature control optimal model to obtain an adjustment data value of the gas flow, and the adjustment data value is obtained by the following formula:

[0112] MV∝F flow +F ac ;

[0113] wherein, MV represents an adjustment data value, F ac represents the input value of the gas flow distributed by the temperature control optimal model.

[0114] If the comparative data is outside the range of -3℃ to 3℃, the gas flow input value is manually adjusted according to artificial experience at this time.

[0115] In the embodiment, when the comparative data is outside the range of -3℃ to 3℃, it indicates that the setting of the gas flow by the temperature control optimal model does not adapt to the set process requirement of the dome temperature demand value, at this time, the gas flow needs to be adjusted by historical data import or artificial experience, and the final gas flow value and the corresponding dome temperature data are fed back to the server 200 to train the temperature control optimal model.

[0116] Please refer to Figure 6As shown, the application also provides a device for balancing the crown temperature of a glass melting furnace, comprising an input data control device 110, a thermocouple 120, a server 200, a controller 300, a monitoring server 400 and a switch 234. The input data control device 110 is used to obtain the gas flow data input into the glass melting furnace by each branch, and the thermocouple 120 is used to obtain the crown temperature output data of the glass melting furnace. Sample data is obtained through the input data control device and the thermocouple. The server 200 is in communication connection with the input data control device 110 and the thermocouple 120, and is used to establish a temperature control neural network model, train the temperature control neural network model according to the sample data, obtain a temperature control optimal model, and optimize the input value of the gas flow data of each branch according to the temperature control optimal model. After the input value of the gas flow data of each branch is optimized, the real-time crown temperature of the glass melting furnace is collected, and compared with the crown temperature data required by the process to obtain comparison data. The controller 300 is in communication connection with the input data control device 110, the thermocouple 120 and the server 200, and is used to balance the multiple crown temperatures in the glass melting furnace according to the comparison data. The input data control device 110 comprises a gas regulating valve 1111, which is in communication connection with the controller 300. The opening degree of the gas regulating valve 1111 is controlled by the controller 300 to adjust the gas input flow of each branch of the glass melting furnace. The monitoring server 400 is used to switch the PID system, and the switch 234 is used for data exchange.

[0117] It is apparent for those skilled in the art that the present application is not limited to the details of the above-described exemplary embodiments, but can be implemented in other concrete forms without departing from the spirit or essential characteristics of the present application. Therefore, the embodiments should be considered in all aspects as illustrative and not restrictive, and the scope of the present application is defined by the appended claims rather than the above description, and it is intended to encompass all changes falling within the meaning and range of equivalents of the claims, and no figure reference in the claims should be considered as limiting the claims.

[0118] The above-described embodiments only represent the implementation of the application, and the protection scope of the application is not limited to the above-described embodiments. For those skilled in the art, without departing from the concept of the application, a number of modifications and improvements can be made, which are all within the protection scope of the application.

Claims

1. A method for controlling the temperature of the crown of a glass melting furnace, characterized in that, include: Acquire sample data, including the gas flow rate data of each branch input to the glass melting furnace; and the crown temperature output data of the glass melting furnace; Establish a temperature control neural network model, train the temperature control neural network model based on the sample data, and obtain the optimal temperature control model. Based on the optimal temperature control model, optimize the allocation of input values ​​for gas flow data in each branch; After optimizing and allocating the input values ​​for the gas flow data of each branch, the real-time crown temperature of the glass melting furnace is collected and compared with the crown temperature data required by the process to obtain comparison data. Based on the comparative data, the temperatures of multiple arches within the glass melting furnace are balanced, including: If the comparison data is within the range of -1℃ to -3℃ and 1℃ to 3℃, the input value of the gas flow data needs to be corrected. The comparison data is used as the input of the traditional PID controller, and the traditional PID controller outputs the input value of the gas flow. Combined with the gas flow input value allocated by the optimal temperature control model, the adjustment data value of the gas flow is obtained. The controller adjusts the opening of the gas regulating valve according to the adjustment data value. The input value for the traditional PID output gas flow rate is obtained through the following: In the formula, F flow The input value, G, represents the gas flow rate output of a traditional PID controller. ain Represented as the gain coefficient, T S Let T be the integration time. D Represented as differential time, D iffGain Let represent the action delay time, s represent the complex number of the Laplace transform, and ΔT1 represent the comparison data; The adjusted data value for obtaining the gas flow rate is obtained using the following formula: MV∝F flow +F ac ; In the formula, MV represents the adjusted data value, and F ac This represents the gas flow input value assigned to the optimal temperature control model.

2. The method for controlling the temperature of the crown of a glass melting furnace according to claim 1, characterized in that, The process of obtaining the optimal temperature control model includes the following steps: The gas flow rate data is used as the input, and the input is normalized to form an input matrix; Obtain the hidden layer output matrix from the input layer to the hidden layer; Based on the hidden layer output matrix, obtain the output layer output matrix; The loss function is obtained based on the hidden layer output matrix, the output layer output matrix, and the actual output matrix; Based on the loss function, obtain the optimal weight matrix; The optimal temperature control model is obtained based on the optimal weight matrix.

3. The method for controlling the temperature of the crown of a glass melting furnace according to claim 2, characterized in that, The optimal weight matrix is ​​obtained using the following formula: In the formula, w (l) This is represented as the optimal weight matrix from layer l-1 to layer l. Let E represent the optimal loss function, and E represent the total error.

4. The method for controlling the temperature of the crown of a glass melting furnace according to claim 1, characterized in that, The temperatures of the multiple arches within the balanced glass melting furnace include: Based on the comparison data, it is determined whether the gas flow input value of each branch to the glass melting furnace needs to be corrected. If the comparison data is within the range of -1℃ to 1℃, then the gas flow input value does not need to be corrected. At this time, the controller controls the gas regulating valve on each branch of the glass melting furnace to make the gas flow input value of each branch consistent with the gas flow input value of each branch output by the optimal temperature control model.

5. The method for controlling the temperature of the crown of a glass melting furnace according to claim 1, characterized in that, The method for balancing the multiple arch temperatures within the glass melting furnace also includes: if the comparison data is outside the range of -3℃ to 3℃, then the gas flow input value is manually adjusted based on human experience.

6. A device for balancing the temperature at the top of a glass melting furnace, characterized in that, include: The input data control device is used to acquire the gas flow rate data of each branch supplying the glass melting furnace; Thermocouples are used to acquire the output data of the crown temperature of the glass melting furnace; sample data is acquired through the input data control device and the thermocouples. The server, communicatively connected to the input data control device and the thermocouple, is used to establish a temperature control neural network model, train the temperature control neural network model based on the sample data, and obtain the optimal temperature control model; based on the optimal temperature control model, optimize the allocation of input values ​​for the gas flow data of each branch; after the gas flow data of each branch has been optimized and allocated, the real-time arch temperature of the glass melting furnace is collected and compared with the arch temperature data required by the process to obtain comparison data; The controller, communicatively connected to the input data control device, the thermocouple, and the server, is used to balance the temperatures of multiple arches within the glass melting furnace based on the comparison data, including: If the comparison data is within the range of -1℃ to -3℃ and 1℃ to 3℃, the input value of the gas flow data needs to be corrected. The comparison data is used as the input of a traditional PID controller, and the traditional PID controller outputs the input value of the gas flow. Combined with the gas flow input value allocated by the optimal temperature control model, the gas flow adjustment data value is obtained. The controller adjusts the opening of the gas regulating valve according to the adjustment data value. The input value for the traditional PID output gas flow rate is obtained through the following: In the formula, F flow The input value, G, represents the gas flow rate output of a traditional PID controller. ain Represented as the gain coefficient, T S Let T be the integration time. D Represented as differential time, D iffGain Let represent the action delay time, s represent the complex number of the Laplace transform, and ΔT1 represent the comparison data; The adjusted data value for obtaining the gas flow rate is obtained using the following formula: MV∝F flow +F ac ; In the formula, MV represents the adjusted data value, and F ac This represents the gas flow input value assigned to the optimal temperature control model.

7. The apparatus for balancing the temperature at the top of a glass melting furnace according to claim 6, characterized in that, The input data control device includes a gas regulating valve, which is communicatively connected to the controller. The controller controls the opening of the gas regulating valve to adjust the gas input flow rate of each branch of the glass melting furnace.

Images