Temperature control method based on dynamic simulation of whole furnace water-cooled wall

Abstract

The present invention provides a temperature control method based on dynamic simulation of the water-cooled wall of the entire furnace, which relates to the technical field of water-cooled wall safety protection. The method includes collecting the structural parameters and circulation parameters of the water-cooled wall of the entire furnace; establishing a heat exchange model, simulating the temperature distribution of the water-cooled wall, and constructing a dynamic simulation model; identifying risk areas; constructing a prediction model to predict the predicted temperature of the risk area; setting adjustment thresholds and early warning thresholds, performing dynamic adjustment of circulation parameters and triggering alarms. The present invention transcends the limitations of traditional methods by identifying risk areas of the entire furnace water-cooled wall, predicting the temperature of the risk area, and setting thresholds, demonstrating the potential of modern automation control technology in industrial applications, while improving the scientificity and intelligence level of the temperature control of the water-cooled wall of the entire furnace, and providing strong support for improving the operating efficiency of the enterprise and ensuring the safe and stable operation of equipment.

Description

Technical Field

[0001] The present invention relates to the technical field of water-cooled wall safety protection, and in particular to a temperature control method based on full-furnace water-cooled wall dynamic simulation. Background Art

[0002] Furnace water walls are commonly used heat exchange equipment in high-temperature, high-pressure processes such as power generation and petrochemicals. Their primary function is to efficiently transfer heat generated by the combustion flames to the cooling water, thereby ensuring safe and stable operation of the furnace. Water walls typically use water as the cooling medium, and the water flow within them removes heat, preventing damage to the wall material due to overheating. In power plants, heat generated by fuel combustion flows through the water walls to heat circulating water, which is converted into steam in the boiler and then used to drive the steam turbine to generate electricity. Furnace water walls are typically constructed of welded steel pipes, offering excellent thermal conductivity and strength, capable of withstanding high temperatures and high pressures. The design and operation of furnace water walls are influenced by a variety of factors, including cooling water flow rate, inlet water temperature, and water flow conditions. To maximize heat exchange efficiency, water wall design typically requires excellent fluid dynamics, ensuring turbulent cooling water flow within the pipes for optimal heat transfer. However, the harsh operating environment and complex fluid dynamics present numerous challenges in temperature control within furnace water walls.

[0003] Traditional temperature control systems typically rely on a single temperature sensor for monitoring, resulting in inaccurate and limited data. For water-cooled walls spanning a large area, local temperature fluctuations often cannot be reflected by a single sensor, leading to information lags and misjudgments, thus compromising the cooling efficiency of the entire system. Existing technologies fail to fully account for the complex nonlinear characteristics of fluid dynamics and heat conduction. This results in slow response and low prediction accuracy when dealing with dynamically changing operating conditions, making it difficult to meet the demands of real-time control. Summary of the Invention

[0004] The present invention provides a temperature control method based on dynamic simulation of the entire furnace water-cooled wall, so as to solve the defects in the prior art.

[0005] The present invention provides a temperature control method based on dynamic simulation of the entire furnace water-cooled wall, comprising:

[0006] The structural parameters and circulation parameters of the water-cooled wall of the entire furnace are collected in real time, and the structural parameters and circulation parameters are preprocessed to obtain preprocessed data.

[0007] Based on the preprocessed data, combined with fluid dynamics and heat conduction theory, a heat exchange model is established to simulate the temperature distribution of the water-cooled wall, and a dynamic simulation model of the water-cooled wall of the entire furnace is constructed.

[0008] Based on the temperature distribution of the water-cooled wall and the structural parameters, areas prone to temperature anomalies are identified as risk areas.

[0009] The temperature data of the risk area is collected in real time, a prediction model based on convolutional neural network is constructed, and the hyperparameters of the prediction model are optimized using the Bayesian optimization algorithm. The preprocessed data and the real-time temperature data of the risk area are input, and the predicted temperature data of the risk area at the next moment is output.

[0010] Adjustment and warning thresholds for risk areas are set based on the performance parameters of the entire furnace water-cooled wall. When the predicted temperature reaches the adjustment threshold, an adaptive adjustment mechanism is introduced to dynamically adjust the cycle parameters. When the predicted temperature reaches the warning threshold, an alarm is triggered.

[0011] According to the temperature control method based on dynamic simulation of the water-cooled wall of the entire furnace provided by the present invention, the structural parameters include geometric structure data and thermophysical parameters; the thermophysical parameters include thermal conductivity and specific heat capacity; and the circulation parameters include flow rate data, flow rate data and cooling water temperature data of the water circulation of the water-cooled wall.

[0012] According to the temperature control method based on full furnace water wall dynamic simulation provided by the present invention, the process of preprocessing the structural parameters and cycle parameters includes:

[0013] The Kalman filter algorithm was used to remove random noise in the structural and circulatory parameters, and the Z-score method was used to identify and remove outliers in the structural and circulatory parameters.

[0014] The moving average method was used to smooth the structural and cyclic parameters, and data normalization was used to convert the structural and cyclic parameters to a uniform range.

[0015] Interpolation method was used to process missing values ​​in structural parameters and circulation parameters to obtain a complete data set, which was used as preprocessed data.

[0016] According to the temperature control method based on dynamic simulation of the entire furnace water wall provided by the present invention, the process of simulating the temperature distribution of the water wall includes:

[0017] Use the fluid dynamics equations to obtain the velocity field and pressure field of the cooling water flowing inside the water wall.

[0018] A heat conduction model is established based on the velocity field and pressure field combined with Fourier's law.

[0019] According to the heat exchange principle, the heat conduction model is combined with the fluid dynamics model to establish the heat exchange model of the water-cooled wall.

[0020] Set boundary conditions and initial conditions. The boundary conditions include the temperature and flow rate at the water wall inlet and the temperature and flow rate at the outlet.

[0021] The finite difference method is used to solve the heat exchange model and obtain the temperature distribution of the water-cooled wall.

[0022] According to the temperature control method based on full furnace water wall dynamic simulation provided by the present invention, the process of constructing a full furnace water wall dynamic simulation model includes:

[0023] Set up the 3D simulation basic model framework based on the preprocessed data.

[0024] Combining fluid dynamics and heat conduction theory, a set of equations describing the change of water-cooled wall temperature with time is established. The equations include flow equations and heat conduction equations.

[0025] The temperature distribution of the water-cooled wall is used as the initial standard of the three-dimensional simulation basic model, and the flow rate and flow rate of the water circulation in the water-cooled wall are used as the boundary standards.

[0026] Set the time step to perform dynamic simulation on the 3D simulation base model.

[0027] The equations are solved by the finite element method to generate time series data of water-cooled wall temperature and obtain the dynamic simulation model of the water-cooled wall of the whole furnace.

[0028] According to the temperature control method based on full furnace water wall dynamic simulation provided by the present invention, the process of identifying risk areas includes:

[0029] The dynamic simulation model of the entire furnace water-cooled wall is divided into regions.

[0030] The temperature data of each area of ​​the dynamic simulation model of the whole furnace water-cooled wall is collected in real time to generate the temperature time series curve of each area and the spatial distribution map of the whole furnace water-cooled wall.

[0031] According to the fluid dynamics combined with the spatial distribution map, the flow state data in each area is obtained.

[0032] A risk coefficient calculation model is established to obtain the risk coefficient of each area based on the average temperature, peak temperature, flow state data and structural parameters of each area.

[0033] The area whose risk factor reaches the preset risk factor threshold is regarded as the risk area.

[0034] According to the temperature control method based on full furnace water wall dynamic simulation provided by the present invention, the process of constructing a prediction model based on a convolutional neural network includes:

[0035] Collect the historical structural parameters, historical circulation parameters and historical temperature data of the risk area of ​​the entire furnace water-cooled wall, as well as the temperature data of the risk area at the next moment corresponding to the historical temperature data.

[0036] The collected historical structural parameters, historical cycle parameters, historical temperature data of risk areas and temperature data of risk areas at the next moment are preprocessed and divided into training sets and test sets.

[0037] A convolutional neural network (CNN) basic model was constructed, consisting of an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. The input layer receives preprocessed input data, including historical structural parameters, historical circulation parameters, and historical temperature data for risk areas. The convolutional layer, consisting of multiple convolution kernels, extracts features from the input data and generates feature maps. The pooling layer reduces the spatial dimensionality of the feature maps. The fully connected layers integrate the extracted features. The output layer outputs predictions based on the integrated results.

[0038] The convolutional neural network basic model is trained using the training set, and the model parameters that meet the prediction accuracy are retained to obtain the prediction model.

[0039] According to the temperature control method based on full furnace water wall dynamic simulation provided by the present invention, the process of optimizing the hyperparameters of the prediction model using the Bayesian optimization algorithm includes:

[0040] Define the hyperparameter space, take the learning rate, number of convolutional layers, convolution kernel size, and regularization parameter as the hyperparameters that need to be optimized, and set the value range for each hyperparameter.

[0041] A Gaussian process is used to randomly select several initial sample points in the hyperparameter space, each of which represents a set of hyperparameter combinations.

[0042] Each initial sample point is used as a hyperparameter of the prediction model, and the corresponding objective function value is calculated. The objective function value is the loss function.

[0043] Based on the initial sample points and their corresponding objective function values, a proxy model is constructed.

[0044] According to the surrogate model, find the next sample point in the hyperparameter space.

[0045] The prediction model is trained on the hyperparameter combination represented by the new sample point, the objective function is evaluated, and the feedback is fed back to the surrogate model.

[0046] The process of finding the next sample point, training, and evaluation is repeated until the preset number of repetitions is reached, and the hyperparameter combination represented by the sample point with the smallest objective function is used as the hyperparameter combination of the prediction model.

[0047] According to the temperature control method based on dynamic simulation of the entire furnace water-wall system, the process of setting adjustment thresholds and warning thresholds for risk areas includes: collecting performance parameters of the entire furnace water-wall system, including the maximum safe temperature, the minimum safe temperature, and the expected normal operating temperature range; using the expected normal operating temperature range as the adjustment threshold, and using the maximum safe temperature and the minimum safe temperature as the warning thresholds.

[0048] The temperature control method based on dynamic simulation of the entire furnace water wall, provided by the present invention, dynamically adjusts circulation parameters by increasing the cooling water flow rate through the water wall to improve heat exchange capacity and reduce the local temperature of the water wall. The cooling water circulation pressure is also adjusted to increase the flow rate through the water wall to optimize the cooling effect.

[0049] The temperature control method, device, and storage medium based on dynamic simulation of the entire furnace water-cooled wall, provided by the present invention, provide comprehensive and accurate system status information by collecting structural and circulation parameters of the entire furnace water-cooled wall in real time, laying the foundation for subsequent data analysis and modeling. Combining fluid dynamics and heat conduction theory, a heat exchange model is established to simulate the temperature distribution of the entire furnace water-cooled wall. This allows for more accurate identification of potential temperature anomalies, enhancing understanding of cooling system behavior and enabling operators to intervene before potential risks arise, thereby reducing the likelihood of accidents. By monitoring temperature data in risk areas in real time, combining it with a constructed convolutional neural network prediction model and incorporating a Bayesian optimization algorithm for hyperparameter optimization, the accuracy and stability of temperature predictions are ensured. This deep learning-based prediction method can capture complex patterns in temperature changes, providing a more sensitive and effective response in a real-time changing environment. When the temperature data output by the prediction model approaches or exceeds a set adjustment threshold, an adaptive adjustment mechanism is automatically introduced to dynamically adjust the circulation parameters to ensure that the water-cooled wall temperature remains within a safe range. This not only increases the flexibility and intelligence of the cooling system but also significantly enhances the overall safety of the equipment, effectively preventing equipment damage or safety accidents caused by temperature anomalies. When the predicted temperature in a risk area reaches the warning threshold, an alarm is immediately triggered, quickly notifying operators to take appropriate emergency measures. This extends temperature control beyond the analysis of historical and current data to provide a scientific basis for real-time processing and response, forming a closed loop of data monitoring, analysis, and response. This surpasses the limitations of traditional methods and demonstrates the potential of modern automated control technology in industrial applications. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0051] Figure 1 1 is a flow chart of a temperature control method based on dynamic simulation of the entire furnace water wall provided by an embodiment of the present invention;

[0052] Figure 2 1 is a schematic diagram of a process for simulating the temperature distribution of a water-cooled wall according to an embodiment of the present invention;

[0053] Figure 3 It is a schematic diagram of the process of constructing a dynamic simulation model of the entire furnace water-cooled wall in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0055] The following combination Figure 1-Figure 3 The temperature control method based on dynamic simulation of the whole furnace water-cooled wall of the present invention is described.

[0056] Figure 1 It is a structural schematic diagram of a temperature control method based on dynamic simulation of the entire furnace water-cooled wall provided by an embodiment of the present invention.

[0057] like Figure 1 As shown, the temperature control method based on dynamic simulation of the whole furnace water wall provided by the embodiment of the present invention includes:

[0058] The structural parameters and circulation parameters of the water-cooled wall of the entire furnace are collected in real time, and the structural parameters and circulation parameters are preprocessed to obtain preprocessed data.

[0059] Structural parameters include geometric structure data and thermophysical parameters. Thermophysical parameters include thermal conductivity and specific heat capacity. Circulation parameters include flow rate data, flow rate data, and cooling water temperature data for the water wall water circulation.

[0060] The process of preprocessing structural parameters and cycle parameters includes:

[0061] The Kalman filter algorithm was used to remove random noise in the structural and circulatory parameters, and the Z-score method was used to identify and remove outliers in the structural and circulatory parameters.

[0062] The moving average method was used to smooth the structural and cyclic parameters, and data normalization was used to convert the structural and cyclic parameters to a uniform range.

[0063] Interpolation method was used to process missing values ​​in structural parameters and circulation parameters to obtain a complete data set, which was used as preprocessed data.

[0064] Various sensors and monitoring devices installed on the water wall accurately capture cooling water flow rate, temperature, and pressure parameters, as well as the material properties and piping structure of the water wall. Real-time monitoring provides detailed operational status data. Data cleaning, removal of outliers, and missing values ​​ensure data validity and accuracy. Standardization and normalization techniques are employed to reduce dimensional differences between parameters, making model construction more robust. This preprocessed data fully reflects the actual operating conditions of the water wall, laying a solid foundation for developing the heat exchange model.

[0065] Based on the preprocessed data, combined with fluid dynamics and heat conduction theory, a heat exchange model is established to simulate the temperature distribution of the water-cooled wall, and a dynamic simulation model of the water-cooled wall of the entire furnace is constructed.

[0066] Figure 2 3 is a schematic diagram of a process for simulating the temperature distribution of a water-cooled wall in an embodiment of the present invention.

[0067] like Figure 2 As shown in Figure 2, the process of simulating the temperature distribution of the water-cooled wall includes:

[0068] Use the fluid dynamics equations to obtain the velocity field and pressure field of the cooling water flowing inside the water wall.

[0069] A heat conduction model is established based on the velocity field and pressure field combined with Fourier's law.

[0070] According to the heat exchange principle, the heat conduction model is combined with the fluid dynamics model to establish the heat exchange model of the water-cooled wall.

[0071] Set boundary conditions and initial conditions. The boundary conditions include the temperature and flow rate at the water wall inlet and the temperature and flow rate at the outlet.

[0072] The finite difference method is used to solve the heat exchange model and obtain the temperature distribution of the water-cooled wall.

[0073] Figure 3 It is a schematic diagram of the process of constructing a dynamic simulation model of the entire furnace water-cooled wall in an embodiment of the present invention.

[0074] like Figure 3 As shown in Figure 2, the process of building a dynamic simulation model of the entire furnace water-cooled wall includes:

[0075] Set up the 3D simulation basic model framework based on the preprocessed data.

[0076] Combining fluid dynamics and heat conduction theory, a set of equations describing the change of water-cooled wall temperature with time is established. The equations include flow equations and heat conduction equations.

[0077] The temperature distribution of the water-cooled wall is used as the initial standard of the three-dimensional simulation basic model, and the flow rate and flow rate of the water circulation in the water-cooled wall are used as the boundary standards.

[0078] Set the time step to perform dynamic simulation on the 3D simulation base model.

[0079] The equations are solved by the finite element method to generate time series data of water-cooled wall temperature and obtain the dynamic simulation model of the water-cooled wall of the whole furnace.

[0080] Combining fluid dynamics and heat conduction theory, a heat exchange model was constructed to simulate the temperature distribution throughout the furnace's water-cooled walls. By analyzing the cooling water's flow characteristics and heat transfer processes, the model depicts temperature trends at various locations. This not only provides important reference data for actual operation but also instantly identifies areas of potential temperature anomalies. Through continuously updated preprocessed data and model output, the temperature distribution throughout the furnace's water-cooled walls is captured in real time, enabling the system to proactively identify and warn of potential risks. This provides a scientific basis for subsequent risk area definition and temperature prediction.

[0081] Based on the temperature distribution of the water-cooled wall and the structural parameters, areas prone to temperature anomalies are identified as risk areas.

[0082] The process of identifying risk areas includes:

[0083] The dynamic simulation model of the entire furnace water-cooled wall is divided into regions.

[0084] The temperature data of each area of ​​the dynamic simulation model of the whole furnace water-cooled wall is collected in real time to generate the temperature time series curve of each area and the spatial distribution map of the whole furnace water-cooled wall.

[0085] According to the fluid dynamics combined with the spatial distribution map, the flow state data in each area is obtained.

[0086] A risk coefficient calculation model is established to obtain the risk coefficient of each area based on the average temperature, peak temperature, flow state data and structural parameters of each area.

[0087] The area whose risk factor reaches the preset risk factor threshold is regarded as the risk area.

[0088] Based on the simulation results of the heat exchange model, the system can identify areas prone to temperature anomalies and define them as risk zones. By collecting real-time temperature data from these risk areas, local temperature conditions can be immediately understood, improving sensitivity to temperature fluctuations. This effectively focuses attention on key risk areas, avoiding indiscriminate monitoring of the entire water wall, saving resources and time. By focusing on these areas, responses can be accelerated, improving the effectiveness and targeted nature of temperature control.

[0089] The temperature data of the risk area is collected in real time, a prediction model based on convolutional neural network is constructed, and the hyperparameters of the prediction model are optimized using the Bayesian optimization algorithm. The preprocessed data and the real-time temperature data of the risk area are input, and the predicted temperature data of the risk area at the next moment is output.

[0090] The process of building a prediction model based on a convolutional neural network includes:

[0091] Collect the historical structural parameters, historical circulation parameters and historical temperature data of the risk area of ​​the entire furnace water-cooled wall, as well as the temperature data of the risk area at the next moment corresponding to the historical temperature data.

[0092] The collected historical structural parameters, historical cycle parameters, historical temperature data of risk areas and temperature data of risk areas at the next moment are preprocessed and divided into training sets and test sets.

[0093] A convolutional neural network (CNN) basic model was constructed, consisting of an input layer, convolutional layers, pooling layers, fully connected layers, and an output layer. The input layer receives preprocessed input data, including historical structural parameters, historical circulation parameters, and historical temperature data for risk areas. The convolutional layer, consisting of multiple convolution kernels, extracts features from the input data and generates feature maps. The pooling layer reduces the spatial dimensionality of the feature maps. The fully connected layers integrate the extracted features. The output layer outputs predictions based on the integrated results.

[0094] The convolutional neural network basic model is trained using the training set, and the model parameters that meet the prediction accuracy are retained to obtain the prediction model.

[0095] The process of optimizing the hyperparameters of a predictive model using the Bayesian optimization algorithm involves:

[0096] Define the hyperparameter space, take the learning rate, number of convolutional layers, convolution kernel size, and regularization parameter as the hyperparameters that need to be optimized, and set the value range for each hyperparameter.

[0097] A Gaussian process is used to randomly select several initial sample points in the hyperparameter space, each of which represents a set of hyperparameter combinations.

[0098] Each initial sample point is used as a hyperparameter of the prediction model, and the corresponding objective function value is calculated. The objective function value is the loss function.

[0099] Based on the initial sample points and their corresponding objective function values, a proxy model is constructed.

[0100] According to the surrogate model, find the next sample point in the hyperparameter space.

[0101] The prediction model is trained on the hyperparameter combination represented by the new sample point, the objective function is evaluated, and the feedback is fed back to the surrogate model.

[0102] The process of finding the next sample point, training, and evaluation is repeated until the preset number of repetitions is reached, and the hyperparameter combination represented by the sample point with the smallest objective function is used as the hyperparameter combination of the prediction model.

[0103] A temperature prediction model based on a convolutional neural network leverages historical and real-time data to learn complex patterns in temperature fluctuations. By incorporating a Bayesian optimization algorithm to optimize the model's hyperparameters, the model maintains high prediction accuracy across a wide range of operating conditions. This prediction capability facilitates early warning and adjustments before temperature thresholds are reached. This enables early prediction of temperature changes, significantly improving response speed compared to traditional methods. When predictions indicate that a set adjustment threshold is about to be reached, the system proactively takes action to mitigate potential risks.

[0104] Adjustment and warning thresholds for risk areas are set based on the performance parameters of the entire furnace water-cooled wall. When the predicted temperature reaches the adjustment threshold, an adaptive adjustment mechanism is introduced to dynamically adjust the cycle parameters. When the predicted temperature reaches the warning threshold, an alarm is triggered.

[0105] The process of setting adjustment and warning thresholds for risk areas involves collecting performance parameters for the entire furnace water-cooled wall, including the maximum safe temperature, minimum safe temperature, and expected normal operating temperature range. The expected normal operating temperature range serves as the adjustment threshold, while the maximum safe temperature and minimum safe temperature serve as the warning thresholds.

[0106] The process of dynamically adjusting circulation parameters includes increasing the cooling water flow rate in the water wall to improve heat exchange capacity and reduce the local temperature of the water wall. It also adjusts the cooling water circulation pressure to increase the cooling water flow through the water wall and optimize the cooling effect.

[0107] When the predicted temperature reaches the set adjustment threshold, an adaptive adjustment mechanism is introduced to dynamically adjust circulation parameters. This enables the system to flexibly adjust key parameters such as flow rate and pressure based on real-time data, taking timely action in the event of temperature anomalies, significantly enhancing system safety and stability. Furthermore, the set early warning threshold ensures that an alarm is triggered when the temperature is too high, quickly notifying system operators to take emergency measures. Through proactive temperature control and risk management mechanisms, this solution effectively prevents equipment damage or safety accidents, thereby improving the operational reliability and lifespan of the entire furnace water-cooled wall.

[0108] Example 1: Target water wall A, geometry data: Water wall height: 15m, Water wall diameter: 0.5m. Thermophysical parameters: Thermal conductivity: 50W / (m·K), Specific heat capacity: 4200J / (kg·K). Circulation parameters: Water circulation velocity: 2.5m / s, Flow rate: 30m³ / h, Cooling water temperature: 85°C.

[0109] The Kalman filter algorithm was used to remove random noise and obtain smooth flow velocity data.

[0110] The Z-score method was used to identify and remove outliers. It was found that there was an outlier (3.5 m / s) in the velocity data, which was removed.

[0111] After moving average processing, the flow velocity data becomes: 2.4m / s, 2.5m / s, 2.6m / s.

[0112] After data normalization, the flow rate data range is adjusted to [0, 1], and after conversion it is: 0.0, 0.5, 1.0.

[0113] Missing values ​​were handled using interpolation to obtain a complete dataset.

[0114] The velocity field and pressure field of the cooling water flowing inside the water wall are obtained using the fluid dynamics equation. At a certain moment, the velocity field is:

[0115] v(x)=2.5−0.1x(m / s), where x is the height of the water-cooled wall.

[0116] A heat conduction model is established based on the velocity field and pressure field combined with Fourier's law.

[0117] Set the boundary conditions as follows: inlet temperature: 85°C, flow rate: 30 m³ / h, outlet temperature: 75°C, flow rate: 30 m³ / h.

[0118] The finite difference method is used to solve the heat exchange model, and the temperature distribution of the water-cooled wall is obtained as follows:

[0119] T(x)=85−10(x / 15), where T is temperature (°C) and x is height (m).

[0120] The dynamic simulation model of the entire furnace water-cooled wall is divided into five areas (each 3m is an area).

[0121] The temperature data of each area is collected in real time, and the generated temperature time series curve shows: Area 1: 82°C, Area 2: 80°C, Area 3: 78°C, Area 4: 76°C, Area 5: 75°C.

[0122] Based on the fluid dynamics combined with the spatial distribution map, the flow state data was obtained and it was found that the flow state in area 3 was weak.

[0123] The risk coefficient calculation model shows: risk coefficient for area 1: 0.2, risk coefficient for area 2: 0.3, risk coefficient for area 3: 0.6 (exceeding the threshold and marked as a risk area), risk coefficient for area 4: 0.4, and risk coefficient for area 5: 0.2.

[0124] Historical structural parameters (data from the past week): Geometric data: Waterwall height: 20m, Waterwall diameter: 0.6m. Thermophysical properties: Thermal conductivity: 55W / (m·K), specific heat capacity: 3900J / (kg·K). Historical circulation parameters (flow rate and flow velocity): Flow rate data (m / s): [2.8, 3.0, 3.1, 2.9, 2.7, 3.2, 3.0], Flow rate data (m³ / h): [35, 40, 38, 37, 40, 41, 39], Historical temperature data for the risk area (°C): [87, 86, 85, 88, 87, 89, 90], Forecasted temperature data for the risk area at the next moment (°C): [87, 88, 90, 91, 92, 90, 89].

[0125] The historical data was preprocessed to remove noise, check for outliers, and normalize. The results are as follows:

[0126] Normalized historical flow rate: [0.0, 0.08, 0.1, 0.04, -0.08, 0.16, 0.08] (illustrative data).

[0127] Normalized historical temperature data: [0.0, 0.02, 0.04, 0.06, 0.05, 0.07, 0.08].

[0128] Split the data into training and testing sets:

[0129] Training set: the first 5 records of historical flow rate and temperature data.

[0130] Test set: the last 2 records of historical flow rate and temperature data.

[0131] Constructing the basic model of convolutional neural network:

[0132] Network Structure: Input Layer: Receives preprocessed data, including historical structural parameters, circulation parameters, and historical temperature data for risk areas. Convolution Layer: Uses three convolution kernels to extract features for detecting the relationship between flow velocity and temperature. Pooling Layer: Uses max pooling to reduce feature dimensionality. Fully Connected Layer: Integrates the extracted features and outputs the predicted temperature for the risk area at the next moment. Output Layer: Generates a predicted temperature for the risk area at the next moment.

[0133] The convolutional neural network model is trained using the training set, retaining the model parameters that achieve a prediction accuracy of 95% or higher. Example output from the trained model: Predicted values ​​(°C): [89, 90], corresponding to the risk zone temperatures in the test set.

[0134] Define the hyperparameter space: The hyperparameters that need to be optimized include: learning rate: range (0.001, 0.1), number of convolutional layers: 2 to 5 layers, convolution kernel size: 3 to 7, regularization parameter: range (0.01, 0.1).

[0135] Randomly select several hyperparameter combinations for preliminary evaluation:

[0136] Sample point 1: learning rate 0.01, number of convolutional layers 3, convolution kernel size 5, regularization parameter 0.05.

[0137] Sample point 2: learning rate 0.005, number of convolutional layers 4, convolution kernel size 7, regularization parameter 0.01.

[0138] Train the model corresponding to each sample point and calculate the loss function, which is recorded as follows:

[0139] The loss value of sample point 1 is 0.025.

[0140] The loss value of sample point 2 is 0.020.

[0141] A Gaussian process model is constructed based on the initial sample points and their objective function values ​​to search the hyperparameter space.

[0142] Find a new combination point and train it, repeat the process until the set number of training rounds is reached, and finally determine the hyperparameter combination.

[0143] Collect the performance parameters of the water-cooled wall of the entire furnace: maximum safe temperature: 97°C, minimum safe temperature: 75°C, normal operating temperature range: 80°C to 90°C.

[0144] The adjustment threshold (normal range) is set at 80°C to 90°C.

[0145] The warning thresholds are set at 75°C (minimum safety) and 97°C (maximum safety).

[0146] When the predicted temperature reaches the adjustment threshold (90°C), the water wall circulation parameters are dynamically adjusted:

[0147] Increase cooling water flow rate from 2.5m / s to 3.0m / s to improve heat exchange capacity. Increase circulation pressure to increase flow and optimize cooling effect.

[0148] Set up an alarm mechanism. When the predicted temperature data value approaches the warning threshold (97°C), the alarm is triggered: the alarm message is output: "Alarm! The temperature of water wall area 3 is approaching the safety limit. Please check and adjust the cooling system immediately."

[0149] In summary, this embodiment provides a temperature control method based on dynamic simulation of the entire furnace water wall. By collecting structural and circulation parameters of the entire furnace water wall in real time, it can provide comprehensive and accurate system status information, laying the foundation for subsequent data analysis and modeling. Combining fluid dynamics and heat conduction theory, a heat exchange model is established to simulate the temperature distribution of the entire furnace water wall. This method can more accurately identify areas of potential temperature anomalies, enhance understanding of cooling system behavior, and help operators intervene before potential risks arise, thereby reducing the likelihood of accidents. By monitoring temperature data in risk areas in real time, combining it with a constructed convolutional neural network prediction model and incorporating a Bayesian optimization algorithm for hyperparameter optimization, the accuracy and stability of temperature predictions are ensured. This deep learning-based prediction method can capture complex patterns in temperature changes, providing a more sensitive and effective response in a real-time changing environment. When the temperature data output by the prediction model approaches or exceeds a set adjustment threshold, an adaptive adjustment mechanism is automatically introduced to dynamically adjust the circulation parameters to ensure that the water wall temperature remains within a safe range. This not only increases the flexibility and intelligence of the cooling system but also significantly enhances the overall safety of the equipment, effectively preventing equipment damage or safety accidents caused by temperature anomalies. When the predicted temperature in a risk area reaches the warning threshold, an alarm is immediately triggered, quickly notifying operators to take appropriate emergency measures. This extends temperature control beyond the analysis of historical and current data to provide a scientific basis for real-time processing and response, forming a closed loop of data monitoring, analysis, and response. This surpasses the limitations of traditional methods and demonstrates the potential of modern automated control technology in industrial applications.

[0150] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0151] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0152] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A temperature control method based on dynamic simulation of the entire furnace water wall is characterized in that: include: Real-time collection of structural parameters and circulation parameters of the entire furnace water-cooled wall, and pre-processing of the structural parameters and the circulation parameters to obtain pre-processed data; The structural parameters include geometric structure data and thermophysical parameters; the thermophysical parameters include thermal conductivity and specific heat capacity; the circulation parameters include flow rate data, flow rate data and cooling water temperature data of the water-cooled wall water circulation; Based on the pre-processed data, combined with fluid dynamics and heat conduction theory, a heat exchange model is established to simulate the temperature distribution of the water-cooled wall and a dynamic simulation model of the water-cooled wall of the entire furnace is constructed; The process of simulating the temperature distribution of the water wall includes: Use fluid dynamics equations to obtain the velocity field and pressure field of cooling water flowing inside the water wall; According to the velocity field and the pressure field, a heat conduction model is established in combination with Fourier's law; According to the heat exchange principle, the heat conduction model is combined with the fluid dynamics model to establish a heat exchange model of the water wall; Setting boundary conditions and initial conditions, wherein the boundary conditions include the temperature and flow rate of the water wall inlet and the temperature and flow rate of the water wall outlet; The heat exchange model is solved using the finite difference method to obtain the temperature distribution of the water-cooled wall; Based on the temperature distribution of the water-cooled wall and the structural parameters, areas prone to temperature anomalies are identified as risk areas; Collect temperature data of the risk area in real time, build a prediction model based on a convolutional neural network, optimize hyperparameters of the prediction model using a Bayesian optimization algorithm, input the preprocessed data and the real-time temperature data of the risk area, and output the predicted temperature data of the risk area at the next moment; The adjustment threshold and warning threshold of the risk area are set according to the performance parameters of the water-cooled wall of the entire furnace. When the predicted temperature data reaches the adjustment threshold, an adaptive adjustment mechanism is introduced to dynamically adjust the circulation parameters; when the predicted temperature reaches the warning threshold, an alarm is triggered.

2. The temperature control method based on dynamic simulation of the whole furnace water wall according to claim 1 is characterized in that: The process of preprocessing the structural parameters and the cycle parameters includes: A Kalman filter algorithm is used to remove random noise in the structural parameters and the circulation parameters, and a Z-score method is used to identify and remove outliers in the structural parameters and the circulation parameters; The structural parameters and the cyclic parameters are smoothed by a moving average method, and the structural parameters and the cyclic parameters are converted to a uniform range by data normalization; An interpolation method is used to process missing values ​​in the structural parameters and the circulation parameters to obtain a complete data set, and the complete data set is used as the preprocessed data.

3. The temperature control method based on dynamic simulation of the whole furnace water wall according to claim 1 is characterized in that: The process of building a dynamic simulation model of the entire furnace water wall includes: Setting a three-dimensional simulation basic model framework according to the preprocessed data; Combining fluid dynamics and heat conduction theory, a set of equations describing the change of water wall temperature over time is established, wherein the set of equations includes a flow equation and a heat conduction equation; The temperature distribution of the water wall is used as the initial standard of the three-dimensional simulation basic model, and the flow rate and flow rate of the water circulation of the water wall are used as the boundary standard; Setting a time step to perform dynamic simulation on the three-dimensional simulation basic model; The equations are solved by the finite element method to generate time series data of the water-cooled wall temperature and obtain a dynamic simulation model of the water-cooled wall of the entire furnace.

4. The temperature control method based on dynamic simulation of the whole furnace water wall according to claim 1 is characterized in that: The process of identifying the risk areas includes: Divide the dynamic simulation model of the entire furnace water-cooled wall into regions; Real-time collection of temperature data from each area of ​​the dynamic simulation model of the entire furnace water-cooled wall, generating temperature time series curves for each area and spatial distribution diagrams of the entire furnace water-cooled wall; Obtaining flow state data in each area based on fluid dynamics combined with the spatial distribution map; Establish a risk coefficient calculation model and obtain the risk coefficient of each area based on the average temperature, peak temperature, flow state data and structural parameters of each area; The area whose risk factor reaches the preset risk factor threshold is regarded as the risk area.

5. The temperature control method based on dynamic simulation of the whole furnace water wall according to claim 1 is characterized in that: The process of building a prediction model based on a convolutional neural network includes: Collect historical structural parameters, historical circulation parameters and historical temperature data of the risk area of ​​the entire furnace water-cooled wall, as well as the temperature data of the risk area at the next moment corresponding to the historical temperature data; The collected historical structural parameters, historical cycle parameters, historical temperature data of risk areas, and temperature data of risk areas at the next moment are preprocessed and divided into training sets and test sets; Construct a convolutional neural network basic model, including an input layer, a convolution layer, a pooling layer, a fully connected layer, and an output layer; the input layer is used to receive preprocessed input data, which includes historical structural parameters, historical circulation parameters, and historical temperature data of risk areas; the convolution layer includes multiple convolution kernels, which are used to extract features from the input data and generate feature maps; the pooling layer is used to reduce the spatial dimension of the feature map; the fully connected layer is used to integrate the extracted features; and the output layer is used to output prediction results based on the integration results; The convolutional neural network basic model is trained using the training set, and the model parameters that meet the prediction accuracy are retained to obtain a prediction model.

6. The temperature control method based on dynamic simulation of the whole furnace water wall according to claim 1 is characterized in that: The process of optimizing the hyperparameters of the prediction model using the Bayesian optimization algorithm includes: Define the hyperparameter space, taking the learning rate, number of convolutional layers, convolution kernel size, and regularization parameter as the hyperparameters that need to be optimized, and set the value range for each hyperparameter; Using a Gaussian process to randomly select a number of initial sample points in the hyperparameter space, each of the initial sample points represents a set of hyperparameter combinations; Taking each of the initial sample points as a hyperparameter of the prediction model, calculating a corresponding objective function value, wherein the objective function value is a loss function; Constructing a proxy model based on the initial sample points and their corresponding objective function values; According to the proxy model, find the next sample point in the hyperparameter space; Training the prediction model on the hyperparameter combination represented by the new sample point, evaluating the objective function, and feeding back to the surrogate model; The process of finding the next sample point, training, and evaluating is repeated until a preset number of repetitions is reached, and the hyperparameter combination represented by the sample point with the minimum objective function is used as the hyperparameter combination of the prediction model.

7. The temperature control method based on dynamic simulation of the whole furnace water wall according to claim 1 is characterized in that: The process of setting the adjustment threshold and warning threshold of the risk area includes: collecting performance parameters of the water-cooled wall of the entire furnace, the performance parameters including the maximum safe temperature, the minimum safe temperature, and the expected normal operating temperature range; using the expected normal operating temperature range as the adjustment threshold, and using the maximum safe temperature and the minimum safe temperature as the warning threshold.

8. The temperature control method based on dynamic simulation of the whole furnace water wall according to claim 1 is characterized in that: The process of dynamically adjusting the circulation parameters includes: increasing the cooling water flow rate of the water-cooled wall, improving the heat exchange capacity, and reducing the local temperature of the water-cooled wall; adjusting the circulation pressure of the cooling water, increasing the flow rate of the cooling water through the water-cooled wall, and optimizing the cooling effect.

Images