An automated control system and method for a full oxy-combustion glass furnace

Abstract

The application relates to an automatic control system and method of a full-oxygen combustion glass kiln, which comprises: a monitoring data acquisition module for acquiring glass kiln basic data and kiln real-time state data, and obtaining multi-dimensional monitoring data by using a preset soft measurement model; an analysis processing module for establishing a combustion control model with the multi-dimensional monitoring data as input based on a deep learning neural network, and obtaining a control scheme and multi-dimensional prediction results of the glass kiln by using the model; and an instruction output module for generating a control instruction according to the glass kiln control scheme, and adjusting the running state of the glass kiln system. The system of the application acquires multi-dimensional monitoring data which is difficult to directly measure in real time by using a preset soft measurement model, designs a combustion control model based on a deep learning neural network, and outputs a control scheme and multi-dimensional prediction results, so that accurate control of the kiln process is realized, and the glass production quality and efficiency are stabilized and improved.

Description

Technical Field

[0001] This invention relates to the field of glass furnace control technology, and in particular to an automated control system and method for an all-oxygen combustion glass furnace. Background Technology

[0002] The float glass manufacturing process is a widely used method for producing flat glass sheets. The process involves introducing molten glass at approximately 1100°C into a tin bath filled with protective gas. The molten glass floats on the surface of the molten tin, and under the influence of gravity and surface tension, it spreads, flattens, and hardens, forming a uniform glass strip—the glass sheet—during its flow. Compared to other forming methods, the float glass manufacturing process can efficiently produce high-quality flat glass with uniform thickness and smooth top and bottom surfaces.

[0003] Traditional glass melting processes use air as the combustion-supporting gas, but nitrogen, which constitutes the largest proportion of air, is of no benefit to the combustion process. Furthermore, nitrogen, after being heated to high temperatures, is discharged as flue gas, carrying away a significant amount of heat; moreover, nitrogen reacts with oxygen at high temperatures to produce toxic gases such as NO, NO2, and N2O, causing environmental pollution. Oxygen-based combustion technology, on the other hand, is an effective means to achieve energy conservation, environmental protection, and high melting quality. Oxygen-based combustion technology results in higher flame temperatures and more uniform heat transfer, improving the penetration efficiency of shorter-wavelength thermal radiation in the molten glass, thereby increasing the thermal efficiency of the float glass furnace. Simultaneously, it reduces the heat carried away by exhaust gases, achieving the goal of waste heat recovery.

[0004] Stable and precise control of the oxy-fuel combustion system in float glass furnaces is a key factor directly affecting glass quality and production efficiency. However, due to the nonlinear, time-varying, and strongly coupled characteristics of the combustion process in the furnace, existing glass furnace combustion systems often lack suitable control methods. This results in problems such as inaccurate control of the fuel-oxygen ratio during combustion, slow response speed, and difficulty in achieving ideal temperature field control, thus impacting glass production efficiency and quality.

[0005] Therefore, there is a need for an automated control system and method for an all-oxygen combustion glass furnace, which can provide comprehensive and precise control over the furnace process, thereby improving glass quality and output while reducing energy consumption and environmental pollution. Summary of the Invention

[0006] In view of this, the present invention provides an automated control system and method for an all-oxygen combustion glass furnace, which solves the technical problems of low control accuracy, large parameter fluctuations, low glass production efficiency and low production quality caused by the lack of advanced control algorithms and sensor technology in existing glass furnace systems.

[0007] To achieve the above-mentioned technical objectives, the present invention adopts the following technical solution:

[0008] In a first aspect, the present invention provides an automated control system for an all-oxygen combustion glass furnace, comprising:

[0009] The monitoring data acquisition module is used to acquire basic data and real-time status data of the glass kiln, and to obtain multi-dimensional monitoring data based on the basic data and real-time status data of the glass kiln using a preset soft measurement model.

[0010] The analysis and processing module is used to establish a combustion control model based on the multidimensional monitoring data as input, and to obtain the control scheme and multidimensional prediction results of the glass furnace using the combustion control model.

[0011] The instruction output module is used to generate control instructions according to the glass furnace control scheme, adjust the operating state of the glass furnace system, compare the multidimensional prediction results output by the combustion control model with the actual measurement results after the operating state adjustment, obtain the fitting error, and feed the fitting error back to the analysis and processing module.

[0012] Furthermore, the basic data of the glass furnace includes: glass furnace structure and size data; the real-time status data of the furnace includes: real-time fuel supply, real-time oxygen supply, and real-time glass melt viscosity and surface tension; the monitoring data acquisition module includes:

[0013] The real-time acquisition unit is used to acquire the fuel supply, oxygen supply, glass melt viscosity, and surface tension of the glass furnace in real time.

[0014] The soft measurement analysis unit is used to calculate the furnace pressure, glass liquid temperature and glass liquid level of the glass furnace based on the real-time fuel supply, oxygen supply, glass liquid viscosity and surface tension using a preset soft measurement model.

[0015] The kiln database unit is used to store the basic data, real-time status data and multi-dimensional monitoring data of the glass kiln.

[0016] Furthermore, the soft measurement analysis unit includes a temperature soft measurement module, a pressure soft measurement module, and a liquid level soft measurement module;

[0017] The temperature soft measurement module is used to establish a temperature assessment model with glass melt viscosity, glass melt surface tension, fuel supply and oxygen supply as inputs, and to output the real-time temperature data of the glass furnace using the temperature assessment model.

[0018] The pressure soft measurement module is used to establish a pressure assessment model with fuel supply, oxygen supply and temperature data as input, and to output real-time internal pressure data of the glass furnace using the pressure assessment model.

[0019] The liquid level soft measurement module is used to establish a liquid level assessment model with glass melt viscosity, glass melt surface tension and kiln structure and size data as inputs, and to output real-time glass melt liquid level data using the liquid level assessment model.

[0020] Furthermore, the temperature assessment model is established based on the heat conduction equation and the energy balance equation, and is used to calculate the temperature changes and heat distribution inside the glass furnace, expressed by the following formula:

[0021]

[0022] in, Indicates temperature change, The expression represents the change over time, α represents the thermal diffusivity, Q represents the heat source, ρ represents the density, and c represents the thermal diffusivity. p This indicates specific heat capacity.

[0023] Furthermore, the analysis and processing module also includes a preprocessing module;

[0024] The preprocessing module is used to perform noise reduction, outlier removal, and normalization on the real-time status data of the kiln.

[0025] Furthermore, the combustion control model is constructed based on an LSTM-RNN network; the analysis and processing module includes:

[0026] The data sequence construction module is used to organize the multidimensional monitoring data in chronological order, set a sliding window, and construct a data sequence with the multidimensional monitoring data at historical moments as input and the data at the current moment as output to obtain a training dataset.

[0027] The model training module is used to construct a basic model of a recurrent neural network with LSTM units. The training dataset is input into the basic model, and the gradient descent optimization algorithm is used to train the basic model with the goal of minimizing the loss function to obtain the combustion control model.

[0028] The optimization and update module is used to optimize the parameters of the combustion control model based on the fitting error fed back by the instruction output module.

[0029] Furthermore, the system also includes a waste heat recovery module;

[0030] The waste heat recovery module is used to monitor the temperature and pressure of the flue gas in the glass furnace, determine the emission channel of the flue gas based on the monitoring results, and realize the waste heat recovery of the all-oxygen combustion glass furnace.

[0031] Furthermore, based on the monitoring results, the emission channels for the flue gas are determined, including:

[0032] Flue gas with a monitored temperature below the first threshold is discharged into a preset first air duct for heating and atomization.

[0033] Flue gas with a monitored temperature higher than the second threshold and a pressure higher than the third threshold is discharged into the second gas duct to drive the steam turbine system.

[0034] Furthermore, the instruction output module includes a PID controller, which is used to generate control instructions according to the glass furnace control scheme.

[0035] Secondly, the present invention also provides an automated control method for an oxygen-fired glass furnace, comprising:

[0036] The basic data and real-time status data of the glass kiln are acquired through the monitoring data acquisition module, and multi-dimensional monitoring data are obtained based on the basic data and real-time status data of the glass kiln using a preset soft measurement model.

[0037] The analysis and processing module establishes a combustion control model based on deep learning neural network, using the multidimensional monitoring data as input. The control scheme and multidimensional prediction results of the glass furnace are obtained using the combustion control model.

[0038] The command output module generates control commands based on the glass furnace control scheme to adjust the operating state of the glass furnace system. The multidimensional prediction results output by the combustion control model are compared with the actual measurement results after the operating state adjustment to obtain the fitting error, and the fitting error is fed back to the analysis and processing module.

[0039] This invention provides an automated control system and method for an oxy-fuel glass furnace. The system acquires basic and real-time status data of the glass furnace through a monitoring data acquisition module, and obtains multi-dimensional monitoring data using a pre-set soft sensor model. An analysis and processing module establishes a combustion control model based on a deep learning neural network, and obtains control schemes and multi-dimensional prediction results from the combustion control model. An instruction output module generates control instructions based on the control scheme to adjust the operating state of the glass furnace system. The adjusted measured results are compared with the model prediction results to optimize the combustion control model. This invention utilizes a pre-set soft sensor model to acquire multi-dimensional monitoring data that is difficult to measure directly in real time. It designs a combustion control model based on neural networks and deep learning algorithms, thereby outputting control schemes and multi-dimensional prediction results. This achieves precise control of the furnace process, enabling online adjustment of the glass furnace's process parameters, effectively suppressing fluctuations in furnace parameters, stabilizing and improving glass production quality and efficiency, reducing energy consumption, and providing a fast response speed to adapt to changes in furnace conditions, bringing direct economic benefits to actual production. Attached Figure Description

[0040] Figure 1 A schematic diagram of an embodiment of the automated control system for an all-oxygen combustion glass furnace provided by the present invention;

[0041] Figure 2 This is a schematic diagram of the structure of an embodiment of the monitoring data acquisition module provided by the present invention. Detailed Implementation

[0042] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings, which form part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, but are not intended to limit the scope of the present invention.

[0043] This invention provides an automated control system and method for an oxygen-fired glass furnace, which will be described below.

[0044] Combination Figure 1 As shown in the figure, a specific embodiment of the present invention discloses an automated control system 100 for an all-oxygen combustion glass furnace, comprising:

[0045] The monitoring data acquisition module 101 is used to acquire basic data and real-time status data of the glass kiln, and to obtain multi-dimensional monitoring data based on the basic data and real-time status data of the glass kiln using a preset soft measurement model.

[0046] The analysis and processing module 102 is used to establish a combustion control model based on the multidimensional monitoring data as input, and to obtain the control scheme and multidimensional prediction results of the glass furnace using the combustion control model.

[0047] The instruction output module 103 is used to generate control instructions according to the glass furnace control scheme, adjust the operating state of the glass furnace system, compare the multidimensional prediction results output by the combustion control model with the actual measurement results after the operating state adjustment, obtain the fitting error, and feed the fitting error back to the analysis and processing module.

[0048] Compared to existing technologies, the system in this embodiment acquires basic data and real-time status data of the glass furnace through a monitoring data acquisition module, and obtains multi-dimensional monitoring data using a preset soft sensor model. An analysis and processing module establishes a combustion control model based on a deep learning neural network, and obtains control schemes and multi-dimensional prediction results through this model. An instruction output module generates control instructions based on the control scheme to adjust the operating state of the glass furnace system. The adjusted measured results are compared with the model prediction results to optimize the combustion control model. This embodiment utilizes a preset soft sensor model to acquire multi-dimensional monitoring data that is difficult to measure directly in real time. A combustion control model is designed based on neural networks and deep learning algorithms, thereby outputting control schemes and multi-dimensional prediction results. This achieves precise control of the furnace process, enabling online adjustment of the glass furnace's process parameters. It effectively suppresses fluctuations in parameters within the melting furnace, stabilizes and improves glass production quality and efficiency, reduces energy consumption, and has a fast response speed, allowing for quicker adaptation to changes in furnace operating conditions, bringing direct economic benefits to actual production.

[0049] As a preferred embodiment, the basic data of the glass furnace includes: glass furnace structure and size data; the real-time status data of the furnace includes: real-time fuel supply, real-time oxygen supply, and real-time glass melt viscosity and surface tension; such as Figure 2 The monitoring data acquisition module 101 shown includes:

[0050] The real-time acquisition unit 201 is used to acquire the fuel supply, oxygen supply, glass melt viscosity and surface tension of the glass furnace in real time.

[0051] The soft measurement analysis unit 202 is used to calculate the furnace pressure, glass liquid temperature and glass liquid level of the glass furnace based on the real-time fuel supply, oxygen supply, glass liquid viscosity and surface tension using a preset soft measurement model.

[0052] The kiln database unit 203 is used to store the basic data, real-time status data and multi-dimensional monitoring data of the glass kiln.

[0053] As a specific example, soft sensing technology is a technique that uses computer technology and data processing methods to estimate physical quantities that are difficult to measure directly. Specifically, in the process control of an oxy-fuel glass furnace, a pre-set soft sensing model is used to estimate parameters that are difficult to measure directly, such as temperature, pressure, and liquid level within the furnace, thereby achieving precise control of the furnace process. The implementation of soft sensing technology first requires defining the measurement target, that is, clarifying the process parameters to be estimated and their related variables. For example, in temperature soft sensing, it is necessary to estimate the temperature of the molten glass and its related variables, such as viscosity and surface tension; in pressure soft sensing, it is necessary to estimate the pressure within the furnace and its related variables, such as fuel supply, air supply, and temperature; and in liquid level soft sensing, it is necessary to estimate the liquid level of the molten glass and its related variables, such as viscosity and surface tension.

[0054] Furthermore, after determining the measurement target, it is necessary to select appropriate measuring instruments based on the target. Different process parameters require different measuring instruments. For example, for temperature measurement, infrared thermometers, thermocouples, and other temperature measuring instruments can be selected; for pressure measurement, pressure sensors, pressure gauges, and other pressure measuring instruments can be selected; for liquid level measurement, ultrasonic level gauges, radar level gauges, and other liquid level measuring instruments can be selected.

[0055] To obtain more accurate monitoring data, it is necessary to select appropriate installation locations and methods based on the structure and process requirements of the furnace. For example, temperature measuring instruments need to be installed in designated locations inside the furnace to monitor the temperature of the molten glass in real time; pressure measuring instruments need to be installed in critical parts of the furnace to monitor the pressure inside the furnace in real time; and level measuring instruments need to be installed in designated locations within the glass melting furnace to monitor the level of the molten glass in real time.

[0056] After the measuring instruments are installed, the monitoring data acquisition module collects measurement data of various process parameters in real time. The monitoring data acquisition module needs to possess characteristics such as high precision, high reliability, and real-time performance. The collected data includes parameters such as fuel supply, air supply, temperature, and pressure.

[0057] In a preferred embodiment, the soft measurement analysis unit includes a temperature soft measurement module, a pressure soft measurement module, and a liquid level soft measurement module;

[0058] The temperature soft measurement module is used to establish a temperature assessment model with glass melt viscosity, glass melt surface tension, fuel supply and oxygen supply as inputs, and to output the real-time temperature data of the glass furnace using the temperature assessment model.

[0059] The pressure soft measurement module is used to establish a pressure assessment model with fuel supply, oxygen supply and temperature data as input, and to output real-time internal pressure data of the glass furnace using the pressure assessment model.

[0060] The liquid level soft measurement module is used to establish a liquid level assessment model with glass melt viscosity, glass melt surface tension and kiln structure and size data as inputs, and to output real-time glass melt liquid level data using the liquid level assessment model.

[0061] In a preferred embodiment, the temperature assessment model is established based on the heat conduction equation and the energy balance equation, and is used to calculate the temperature changes and heat distribution inside the glass furnace, expressed by the following formula:

[0062]

[0063] in, Indicates temperature change, The expression represents the change over time, α represents the thermal diffusivity, Q represents the heat source, ρ represents the density, and c represents the thermal diffusivity. p This indicates specific heat capacity.

[0064] As a specific example, the estimation model for the liquid glass level can be designed based on the principle of mass conservation:

[0065]

[0066] Where h represents the molten pool level, Q represents the mass flow rate into and out of the molten pool, ρ represents the molten pool density, and A represents the molten pool cross-sectional area.

[0067] In addition, the liquid level can be estimated based on pressure sensors. Specifically, pressure sensors installed at the bottom of the molten pool are used to measure the static pressure of the molten pool, thereby calculating the liquid level height. Assuming that the molten pool has a uniform density, that is, the density is equal at all points in the molten pool, the liquid level height can be calculated based on the relationship between the pressure and density of the molten pool.

[0068] In practical applications, the liquid level estimation results can be adjusted based on experience or advanced control algorithms, taking into account factors such as molten pool temperature, redox state, and molten pool fluidity. Generally, the higher the molten pool temperature, the lower the density and viscosity of the molten glass, resulting in a lower liquid level. A molten pool with strong reducing properties leads to increased surface tension, while a molten pool with strong oxidizing properties reduces surface tension. Good molten pool fluidity, meaning significant convection and stirring, results in a relatively uniform liquid level distribution. Conversely, poor molten pool fluidity can cause stagnant areas to form, leading to uneven liquid levels.

[0069] By collecting real-time data on temperature, redox state, and fluidity, and combining this data with a liquid level estimation model, the estimation results are corrected through calculation and comparison to improve the accuracy and stability of the estimation.

[0070] To ensure the normal operation and accuracy of the system, regular maintenance and upkeep are required. This includes regularly checking the operating status and accuracy of measuring instruments, and performing timely calibration and adjustments; regularly cleaning and maintaining the data acquisition system to ensure data accuracy and reliability; and regularly backing up and organizing measurement data to prevent data loss and corruption. The soft measurement model should be regularly updated and optimized to improve its estimation accuracy and response speed.

[0071] In a preferred embodiment, the analysis and processing module further includes a preprocessing module;

[0072] The preprocessing module is used to perform noise reduction, outlier removal, and normalization on the real-time status data of the kiln.

[0073] As a specific example, in order to ensure the accuracy and reliability of the data, it is necessary to standardize the data, including removing outliers and noise, and converting data of different dimensions into comparable standardized data.

[0074] In a preferred embodiment, the combustion control model is constructed based on an LSTM-RNN network; the analysis and processing module includes:

[0075] The data sequence construction module is used to organize the multidimensional monitoring data in chronological order, set a sliding window, and construct a data sequence with the multidimensional monitoring data at historical moments as input and the data at the current moment as output to obtain a training dataset.

[0076] The model training module is used to construct a basic model of a recurrent neural network with LSTM units. The training dataset is input into the basic model, and the gradient descent optimization algorithm is used to train the basic model with the goal of minimizing the loss function to obtain the combustion control model.

[0077] The optimization and update module is used to optimize the parameters of the combustion control model based on the fitting error fed back by the instruction output module.

[0078] LSTM-RNN (Long Short-Term Memory Recurrent Neural Network) is a type of recurrent neural network used for processing and predicting time series data. It addresses the vanishing and exploding gradient problems found in traditional RNNs by using Long Short-Term Memory (LSTM) units. The basic structure of an LSTM-RNN network includes an input layer, LSTM units, hidden layers, and an output layer. The LSTM unit is the core component of LSTM-RNN, and each LSTM unit consists of a forget gate, an input gate, an output gate, and a cell state.

[0079] The forget gate determines which information from the cell state in the previous time step needs to be forgotten. Using the inputs from the previous and current time steps as inputs, a sigmoid activation function is applied, outputting a value between 0 and 1.

[0080] The input gate determines which information from the current time step's input needs to be stored in the cell state. Similarly, using the inputs from the previous and current time steps as inputs, a sigmoid activation function outputs a value between 0 and 1.

[0081] The cell state update is calculated using a tanh activation function based on the cell state of the previous time step, the input gate, and the input of the current time step.

[0082] The output gate determines the output of the LSTM unit at the current time step. Based on the inputs of the previous and current time steps, the input is passed through a sigmoid activation function and outputs a value between 0 and 1.

[0083] LSTM-RNN networks, through the memory mechanism of LSTM units, can effectively capture and utilize long-term dependencies in time-series data, making them suitable for processing time-series tasks with long-term dependencies. A combustion control model for an all-oxygen combustion glass furnace is established based on the LSTM-RNN network structure, expressed by the following formula:

[0084] y(t)=f(x(t),h(t-1),c(t-1))

[0085] Where y(t) represents the output at time step t, x(t) represents the input at time step t, which includes the actual variables of the glass furnace, including temperature, pressure and liquid level obtained through the soft sensor model, h(t-1) is the hidden state (memory) at time step t-1, used to capture long-term dependencies in the sequence, c(t-1) is the cell state at time step t-1, used to control the transmission and forgetting of information, and the function f(x) is the operation of the LSTM unit. By training and optimizing the LSTM-RNN network, it can learn the complex nonlinear relationship between the input and output variables, and is used to predict the operating state variables of the all-oxygen combustion glass furnace.

[0086] LSTM layers are used to model time series data. Multiple LSTM layers can be selected to increase the expressive power of the model. Fully connected layers can be used for output prediction. The control scheme output by the combustion control model can be the amount of cold air supply and the amount of fuel supply.

[0087] The mean squared error (MSE) is used to measure the difference between the model output and the true value. The Adam optimizer is used to optimize the model parameters to minimize the loss function. By constructing a neural network model based on LSTM-RNN, it is possible to predict the cold air supply and fuel supply of an all-oxygen combustion glass furnace, optimize the furnace control strategy, thereby improving production efficiency and achieving energy conservation and emission reduction.

[0088] In practical applications, oxy-fuel combustion technology results in higher flame temperatures and, due to its uniform heat transfer, improves the penetration efficiency of shorter wavelength thermal radiation into the molten glass. This technology can improve the thermal efficiency of float glass furnaces while reducing heat carried away by waste, thus achieving waste heat recovery.

[0089] In a preferred embodiment, the system further includes a waste heat recovery module;

[0090] The waste heat recovery module is used to monitor the temperature and pressure of the flue gas in the glass furnace, determine the emission channel of the flue gas based on the monitoring results, and realize the waste heat recovery of the all-oxygen combustion glass furnace.

[0091] As a preferred embodiment, determining the emission channel of the flue gas based on monitoring results includes:

[0092] Flue gas with a monitored temperature below the first threshold is discharged into a preset first air duct for heating and atomization.

[0093] Flue gas with a monitored temperature higher than the second threshold and a pressure higher than the third threshold is discharged into the second gas duct to drive the steam turbine system.

[0094] As a specific implementation example, the actual selection of flue gas emission channels is as follows: Flue gas exiting the furnace regenerator and heat exchanger (or heat exchanger) is generally below 500℃. This flue gas can be used to generate steam through a heat pipe waste heat boiler. The steam can be used for heating and atomizing heavy oil, pipe insulation, and domestic heating. For flue gas with a large volume and high temperature, high-pressure steam (3.5MPa) can be generated through a heat pipe waste heat boiler for power generation in a steam turbine, or directly to drive turbine air compressors, fans, water pumps, and other machinery. For flue gas exiting from the working pool and feeding duct, which has a small volume but high temperature, a small number of high-temperature heat pipes (operating temperature > 650℃) can be used to preheat the air. When the flue gas temperature is 1000–1200℃, the air preheating temperature can reach 400–500℃, resulting in oil savings of up to 20%.

[0095] In a preferred embodiment, the instruction output module includes a PID controller, which is used to generate control instructions according to the glass furnace control scheme.

[0096] PID controllers have the advantages of being simple to use and robust. By adjusting the proportional coefficient, integral time, and derivative time, they can regulate different types of systems, such as continuous, discrete, linear, and nonlinear systems. They can also regulate and control various equipment in the combustion system of glass furnaces.

[0097] As a specific implementation, control parameters are adjusted based on the output of the combustion control model and actual conditions. For example, if the actual temperature is higher than the estimated temperature, the supply of cooling air can be increased appropriately; if the actual pressure is lower than the estimated pressure, the fuel supply can be increased appropriately. By adjusting the control parameters, the combustion process in the furnace can be optimized, thereby improving the quality and yield of glass.

[0098] Furthermore, combustion control strategies for glass furnaces also include optimizing fuel resources. Based on prediction results, the difference between actual combustion effects and predicted values ​​is compared to select high-calorific-value, high-efficiency fuels, such as natural gas and oil, to reduce fuel consumption and energy loss. In addition, the rational utilization and allocation of fuel resources can also improve energy efficiency and reduce energy consumption. By adjusting parameters such as air supply, burner settings, and temperature control, goals such as flame stability, complete combustion, and high energy efficiency can be achieved. Furthermore, adopting new, high-efficiency burners and technologies can improve flame temperature and thermal efficiency, reduce heat loss, and lower energy consumption.

[0099] As a specific example, optimizing the production process can reduce the energy consumption of oxy-fuel glass furnaces. Energy savings and increased production efficiency can be achieved through measures such as optimizing material ratios, improving production processes, and enhancing operational control. Furthermore, adopting advanced production technologies and processes can improve product quality and reduce defect rates, thereby further reducing energy consumption.

[0100] In summary, energy consumption optimization and control of oxy-fuel glass furnaces encompasses aspects such as fuel selection, combustion optimization, molten glass temperature control, waste gas treatment and utilization, equipment maintenance and improvement, production process optimization, furnace sealing enhancement, and automation and intelligent control. By comprehensively applying these technologies, energy consumption in oxy-fuel glass furnaces can be significantly reduced, energy utilization efficiency and production efficiency can be improved, providing strong support for the sustainable development of the glass manufacturing industry.

[0101] This invention also provides an automated control method for an all-oxygen combustion glass furnace, comprising:

[0102] The basic data and real-time status data of the glass kiln are acquired through the monitoring data acquisition module, and multi-dimensional monitoring data are obtained based on the basic data and real-time status data of the glass kiln using a preset soft measurement model.

[0103] The analysis and processing module establishes a combustion control model based on deep learning neural network, using the multidimensional monitoring data as input. The control scheme and multidimensional prediction results of the glass furnace are obtained using the combustion control model.

[0104] The command output module generates control commands based on the glass furnace control scheme to adjust the operating state of the glass furnace system. The multidimensional prediction results output by the combustion control model are compared with the actual measurement results after the operating state adjustment to obtain the fitting error, and the fitting error is fed back to the analysis and processing module.

[0105] This invention provides an automated control system and method for an oxy-fuel glass furnace. The system acquires basic data and real-time status data of the glass furnace through a monitoring data acquisition module, and obtains multi-dimensional monitoring data using a preset soft sensor model. An analysis and processing module establishes a combustion control model based on a deep learning neural network, and obtains a control scheme and multi-dimensional prediction results from the combustion control model. An instruction output module generates control instructions based on the control scheme to adjust the operating state of the glass furnace system. The adjusted measured results are compared with the model prediction results to optimize the combustion control model.

[0106] The system in this embodiment utilizes a preset soft measurement model to acquire multidimensional monitoring data that is difficult to measure directly in real time. Based on neural networks and deep learning algorithms, a combustion control model is designed to output control schemes and multidimensional prediction results, thereby achieving precise control of the kiln process. It can adjust the process parameters of the glass kiln online, effectively suppressing fluctuations in parameters within the melting furnace, stabilizing and improving the quality and efficiency of glass production, reducing energy consumption, and having a fast response speed, enabling it to adapt to changes in kiln operating conditions more quickly, bringing direct economic benefits to actual production.

[0107] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. An automated control system for an all-oxygen combustion glass furnace, characterized in that, include: The monitoring data acquisition module is used to acquire basic data and real-time status data of the glass kiln, and to obtain multi-dimensional monitoring data based on the basic data and real-time status data of the glass kiln using a preset soft measurement model. The basic data of the glass furnace includes: glass furnace structure and size data; the real-time status data of the furnace includes: real-time fuel supply, real-time oxygen supply, and real-time glass melt viscosity and surface tension; the monitoring data acquisition module includes: The real-time acquisition unit is used to acquire the fuel supply, oxygen supply, glass melt viscosity, and surface tension of the glass furnace in real time. The soft measurement analysis unit is used to calculate the furnace pressure, glass melt temperature, and glass melt level of the glass furnace based on the real-time fuel supply, oxygen supply, glass melt viscosity, and surface tension using a preset soft measurement model; the soft measurement analysis unit includes a temperature soft measurement module, a pressure soft measurement module, and a liquid level soft measurement module; The temperature soft measurement module is used to establish a temperature assessment model with glass melt viscosity, glass melt surface tension, fuel supply and oxygen supply as inputs, and output the real-time temperature data of the glass furnace using the temperature assessment model; the temperature assessment model is established based on the heat conduction equation and the energy balance equation. The pressure soft measurement module is used to establish a pressure assessment model with fuel supply, oxygen supply and temperature data as input, and to output real-time internal pressure data of the glass furnace using the pressure assessment model. The liquid level soft measurement module is used to establish a liquid level assessment model with glass melt viscosity, glass melt surface tension and kiln structure and size data as inputs, and outputs real-time glass melt liquid level data using the liquid level assessment model; the glass melt liquid level estimation model is designed based on the principle of mass conservation. The kiln database unit is used to store the basic data, real-time status data and multi-dimensional monitoring data of the glass kiln; The analysis and processing module is used to establish a combustion control model based on the multidimensional monitoring data as input, and to obtain the control scheme and multidimensional prediction results of the glass furnace using the combustion control model. The instruction output module is used to generate control instructions according to the glass furnace control scheme, adjust the operating state of the glass furnace system, compare the multidimensional prediction results output by the combustion control model with the actual measurement results after the operating state adjustment, obtain the fitting error, and feed the fitting error back to the analysis and processing module. The system also includes a waste heat recovery module; The waste heat recovery module is used to monitor the temperature and pressure of the flue gas in the glass furnace, determine the emission channel of the flue gas based on the monitoring results, and realize the waste heat recovery of the all-oxygen combustion glass furnace.

2. The automated control system for the all-oxygen combustion glass furnace according to claim 1, characterized in that, The temperature assessment model, based on the heat conduction equation and the energy balance equation, is used to calculate the temperature changes and heat distribution inside the glass furnace, and is expressed by the following formula: in, Indicates temperature change, Indicates changes over time. Indicates the thermal diffusivity. Indicates the heat source. Indicates density, This indicates specific heat capacity.

3. The automated control system for the all-oxygen combustion glass furnace according to claim 1, characterized in that, The analysis and processing module also includes a preprocessing module; The preprocessing module is used to perform noise reduction, outlier removal, and normalization on the real-time status data of the kiln.

4. The automated control system for the all-oxygen combustion glass furnace according to claim 1, characterized in that, The combustion control model is constructed based on an LSTM-RNN network; the analysis and processing module includes: The data sequence construction module is used to organize the multidimensional monitoring data in chronological order, set a sliding window, and construct a data sequence with the multidimensional monitoring data at historical moments as input and the data at the current moment as output to obtain a training dataset. The model training module is used to construct a basic model of a recurrent neural network with LSTM units. The training dataset is input into the basic model, and the gradient descent optimization algorithm is used to train the basic model with the goal of minimizing the loss function to obtain the combustion control model. The optimization and update module is used to optimize the parameters of the combustion control model based on the fitting error fed back by the instruction output module.

5. The automated control system for the all-oxygen combustion glass furnace according to claim 1, characterized in that, The emission channels for the flue gas are determined based on monitoring results, including: Flue gas with a monitored temperature below the first threshold is discharged into a preset first air duct for heating and atomization. Flue gas with a monitored temperature higher than the second threshold and a pressure higher than the third threshold is discharged into the second gas duct to drive the steam turbine system.

6. The automated control system for the all-oxygen combustion glass furnace according to claim 1, characterized in that, The instruction output module includes a PID controller, which is used to generate control instructions according to the glass furnace control scheme.

7. A method for controlling an automated control system as described in any one of claims 1-6, characterized in that, include: The basic data and real-time status data of the glass kiln are acquired through the monitoring data acquisition module, and multi-dimensional monitoring data are obtained based on the basic data and real-time status data of the glass kiln using a preset soft measurement model. The analysis and processing module establishes a combustion control model based on deep learning neural network, using the multidimensional monitoring data as input. The control scheme and multidimensional prediction results of the glass furnace are obtained using the combustion control model. The command output module generates control commands based on the glass furnace control scheme to adjust the operating state of the glass furnace system. The multidimensional prediction results output by the combustion control model are compared with the actual measurement results after the operating state adjustment to obtain the fitting error, and the fitting error is fed back to the analysis and processing module.

Images