Method and system for predicting icing time of fire-fighting pipeline in converter station low-temperature environment

Abstract

The invention provides a method and a system for predicting freezing time of a fire-fighting pipeline in a converter station low-temperature environment. The method comprises the following steps: collecting low-temperature freezing experiment data of the fire-fighting pipeline of the converter station; establishing a fire-fighting pipeline low-temperature environment CFD numerical simulation model, and verifying a convective heat transfer coefficient based on experimental data; based on the CFD numerical simulation model, a convective heat transfer coefficient empirical formula is obtained; the temperature of the low-temperature environment is changed, and icing simulation data under different working conditions are output; based on the simulation data, establishing and optimizing a neural network model by taking environment temperature and time as input parameters; and based on the neural network model, predicting and verifying the icing time of the fire-fighting pipeline under different working conditions. The method has the advantages that changes of convective heat transfer coefficients under different temperature differences are considered, and complexity of a traditional prediction model is overcome; a method of combining numerical simulation and a neural network is adopted, so that the prediction time is shortened; and the prediction process is simplified and is clear and simple.

Description

Technical Field

[0001] The invention relates to the technical field of data processing methods, and in particular to a method for predicting freezing time of a fire protection pipeline. Background Art

[0002] Due to the extremely low temperature in winter, the fire-fighting pipes in the converter station often freeze, which seriously restricts the effective implementation of fire-fighting and rescue work. The master's thesis "Thermal Environment Characteristics of the Entrance Section of Underground Tunnel in Cold Regions and Antifreeze Insulation of Fire-fighting Water Pipes, Luo Wenhao, Xi'an University of Architecture and Technology, 2023" conducted an antifreeze test on the fire-fighting pipe, with a simulation time of up to 6 hours. The master's thesis "Numerical Analysis and Experimental Research on Electric Heating of Low-temperature Deck and Pipe, Zhang Chen, Harbin Engineering University, 2022" conducted a numerical simulation study on the temperature distribution of pipes under different low-temperature environments, but did not consider the changes in the convective heat transfer coefficient caused by changes in ambient temperature. In summary, the time cost of the freezing experiment is high, and the model conditions of the numerical simulation study are complex; at the same time, traditional prediction methods often find it difficult to accurately capture the complex relationship between various factors, resulting in inaccurate prediction results, which brings new challenges to the prediction of the freezing time of fire-fighting pipes under low-temperature environments.

[0003] Therefore, it is necessary to develop a method to predict the freezing time of fire protection pipes in low-temperature environment converter stations that can comprehensively consider the influence of multiple factors. Summary of the invention

[0004] The technical problem to be solved by the present invention is how to comprehensively consider the influence of various factors so as to accurately predict the freezing time of fire protection pipes in a low-temperature environment converter station.

[0005] The present invention solves the above technical problems through the following technical means: A method for predicting the freezing time of fire protection pipelines in a low temperature environment of a converter station comprises the following steps:

[0006] S1, collect low temperature icing experimental data of fire protection pipelines in converter stations;

[0007] S2, establish a CFD numerical simulation model for low-temperature environment of fire-fighting pipelines, and verify the convective heat transfer coefficient based on experimental data.

[0008] In the established CFD numerical simulation model, a temperature sensor is set on the outer wall of the fire protection pipe to obtain temperature changes, and a liquid fraction sensor is set in the diameter direction to obtain the icing situation. The CFD simulation results are compared with the experimental data by trial and error method to determine the convective heat transfer coefficient.

[0009] S3, based on the CFD numerical simulation model, obtain the empirical formula of the convective heat transfer coefficient;

[0010] S4, changing the low temperature environment temperature and outputting icing simulation data under different working conditions;

[0011] S5, based on the simulation data, establish and optimize the neural network model with ambient temperature and time as input parameters;

[0012] S6, based on the neural network model, predicts and verifies the freezing time of fire pipes under different working conditions.

[0013] As a further optimized technical solution, in step S1: the experimental data includes pipeline type, initial temperature, experimental environment temperature, and data on changes in pipe wall temperature over time.

[0014] As a further optimized technical solution, in step S3: the empirical formula is as follows:

[0015] h=aΔT b

[0016] h is the convective heat transfer coefficient, △T is the difference between the ambient temperature and the initial temperature, a and b are two parameters that need to be determined, and the least squares method is used to fit the data to determine the convective heat transfer coefficient formula.

[0017] As a further optimized technical solution, in step S4: the freezing thickness H is defined as the ratio of the ice thickness h to the inner diameter R of the pipe according to the liquid fraction cloud map of the freezing simulation data results, and the liquid fraction of different nodes is obtained according to the liquid fraction sensor in the diameter direction. A liquid fraction less than 0.02 is defined as a solid state, i.e., freezing.

[0018] As a further optimization technical solution, in step S5: the number of hidden layers, training function and transfer function of the neural network are optimized by trial and error method, with the accuracy R 2 The correlation coefficient and mean square error (MSE) are used as the criteria for model establishment and optimization.

[0019] The present invention also provides a fire protection pipeline freezing time prediction system in a low temperature environment of a converter station, comprising the following modules:

[0020] Data collection module, used to collect low-temperature icing experimental data of fire protection pipelines in converter stations;

[0021] CFD numerical simulation model building module, used to build the CFD numerical simulation model of the low-temperature environment of the fire protection pipeline and verify the convection heat transfer coefficient based on experimental data;

[0022] In the established CFD numerical simulation model, a temperature sensor is set on the outer wall of the fire protection pipe to obtain temperature changes, and a liquid fraction sensor is set in the diameter direction to obtain the icing situation. The CFD simulation results are compared with the experimental data by trial and error method to determine the convective heat transfer coefficient.

[0023] An empirical formula acquisition module is used to obtain an empirical formula for convection heat transfer coefficient based on a CFD numerical simulation model;

[0024] Numerical simulation module, used to output icing simulation data under different working conditions by changing the low temperature environment temperature;

[0025] A neural network module is used to establish and optimize a neural network model with ambient temperature and time as input parameters;

[0026] The prediction and verification module is used to predict and verify the freezing time of fire pipes under different working conditions based on the neural network model.

[0027] As a further optimized technical solution, in the data collection module: the experimental data includes pipeline type, initial temperature, experimental environment temperature, and pipe wall temperature change data over time.

[0028] As a further optimized technical solution, in the empirical formula acquisition module: the empirical formula is as follows:

[0029] h=aΔT b

[0030] h is the convective heat transfer coefficient, △T is the difference between the ambient temperature and the initial temperature, a and b are two parameters that need to be determined, and the least squares method is used to fit the data to determine the convective heat transfer coefficient formula.

[0031] As a further optimized technical solution, in the numerical simulation module: the freezing thickness H is defined as the ratio of the ice thickness h to the inner diameter R of the pipe according to the liquid fraction cloud map of the freezing simulation data results, and the liquid fraction of different nodes is obtained according to the liquid fraction sensor in the diameter direction. A liquid fraction less than 0.02 is defined as a solid state, i.e. freezing.

[0032] As a further optimization technical solution, in the neural network module, the number of hidden layers, training function and transfer function of the neural network are optimized by trial and error method, with an accuracy of R 2 The correlation coefficient and mean square error (MSE) are used as the criteria for model establishment and optimization.

[0033] The advantages of the present invention are as follows: the present invention takes into account the change of the convective heat transfer coefficient under different temperature differences, overcoming the complexity of the traditional prediction model; combining heat transfer analysis, numerical simulation and neural network, it can save the time cost of prediction while considering the convective heat transfer coefficient, quickly process and analyze, and extract useful data information, thereby improving the accuracy of the prediction; at the same time, the method also simplifies the prediction process, providing a new way to predict the freezing time of fire protection pipes in the low temperature environment of the converter station. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1This is a flow chart of a method for predicting freezing time of fire protection pipes in a low temperature environment of a converter station according to the present invention;

[0035] Figure 2 A schematic diagram of the frozen thickness (H) defined by the liquid fraction cloud diagram of the numerical simulation results according to an embodiment of the present invention, that is, the ratio of the ice thickness (h) to the inner diameter (R) of the pipe;

[0036] Figure 3 It is a fitting diagram of the empirical formula of the convective heat transfer coefficient of the embodiment of the present invention, the line is the fitting formula, the symbol is the convective heat transfer coefficient determined at different ambient temperatures, and the shaded part is the confidence interval;

[0037] Figure 4 It is a comparison diagram between the prediction results and numerical simulation of an embodiment of the present invention, where the symbols are numerical simulation values ​​and the lines are prediction values. DETAILED DESCRIPTION

[0038] Please refer to Figure 1 , which is a flow chart of a method for predicting the freezing time of fire pipes in a low temperature environment of a converter station according to the present invention. The method for predicting the freezing time of fire pipes in a low temperature environment of a converter station according to the present invention comprises the following steps:

[0039] S1, collect low temperature icing experimental data of fire protection pipelines in converter stations.

[0040] The experimental data mainly include pipeline type, initial temperature, experimental environment temperature, and data on the change of pipe wall temperature over time.

[0041] S2, establish a CFD numerical simulation model for low-temperature environment of fire-fighting pipelines, and verify the convective heat transfer coefficient based on experimental data.

[0042] In the established CFD numerical simulation model, a temperature sensor is set on the outer wall of the fire protection pipe to obtain temperature changes, and a liquid fraction sensor is set in the diameter direction to obtain the freezing situation. The trial and error method is used to compare the CFD simulation results with the experimental data to determine the convective heat transfer coefficient.

[0043] S3, based on the CFD numerical simulation model, obtain the empirical formula of the convective heat transfer coefficient.

[0044] The empirical formula for convective heat transfer coefficient is as follows:

[0045] h=aΔT b

[0046] h is the convective heat transfer coefficient, △T is the difference between the ambient temperature and the initial temperature, a and b are two parameters that need to be determined, and the least squares method is used to fit the data to determine the convective heat transfer coefficient formula.

[0047] S4, changes the low temperature environment temperature and outputs icing simulation data under different working conditions.

[0048] According to the liquid fraction cloud diagram of the freezing simulation data results, the freezing thickness H is defined as the ratio of the ice thickness h to the inner diameter R of the pipe. The liquid fraction of different nodes is obtained by the liquid fraction sensor in the diameter direction. A liquid fraction less than 0.02 is defined as a solid state, i.e., freezing.

[0049] S5, based on the simulation data, establish and optimize the neural network model with ambient temperature and time as input parameters;

[0050] The number of hidden layers, training function and transfer function of the neural network are optimized by trial and error method with an accuracy of R 2 The correlation coefficient and mean square error (MSE) are used as the criteria for model establishment and optimization.

[0051] S6, based on the neural network model, predicts and verifies the freezing time of fire pipes under different working conditions.

[0052] The following is a further explanation of the above steps using the fire protection pipeline of a converter station as an example:

[0053] In step S1, the data of the outer wall temperature variation over time of a fire protection pipe of a converter station with a model of DN150, an initial temperature of 5°C, and ambient temperatures of -7°C, -10°C, and -12.8°C are collected.

[0054] In step S2, a CFD numerical simulation model is established according to the actual size parameters and physical parameters of the fire protection pipeline (see Table 1), a temperature sensor is set on the outer wall of the pipeline to obtain temperature changes, and a liquid fraction sensor is set in the diameter direction to obtain the freezing situation. For specific working conditions (7°C, -10°C and -12.8°C), the convective heat transfer coefficient in the boundary conditions is changed, and the trial and error method is used to make the outer wall temperature data of the numerical simulation model consistent with the collected experimental data, and the convective heat transfer coefficient under the ambient temperature of the working condition is determined.

[0055] Table 1 Example CFD numerical simulation model parameters

[0056]

[0057] In step S3, according to the empirical formula of convective heat transfer coefficient and temperature difference:

[0058] h=aΔT b

[0059] h is the convective heat transfer coefficient, △T is the difference between the ambient temperature and the initial temperature, a and b are two parameters that need to be determined, and the least squares method is used to fit the data to determine the convective heat transfer coefficient formula. Please refer to the fitting results Figure 2 , the fitting formula is:

[0060] h=0.366ΔT0.895

[0061] In step S4, the ambient temperature is changed (-5°C to -60°C), and the convective heat transfer coefficient is input into the numerical simulation according to the empirical formula of the convective heat transfer coefficient and the temperature difference determined in step S3 to obtain the simulation results under different working conditions. The frozen thickness H is defined as the ratio of the ice thickness h to the inner diameter R of the pipe according to the liquid fraction cloud map of the freezing simulation data results. For the schematic diagram, please refer to Figure 3 The liquid fraction sensor in the diameter direction obtains the liquid fraction at different nodes, and a liquid fraction less than 0.02 is defined as a solid state, i.e., ice.

[0062] In step S5, the simulation results under different ambient temperatures are obtained according to step S4, and the ambient temperature and ice thickness are used as inputs and the freezing time is used as output to construct a data set (refer to Table 2), and the neural network model is established and optimized with the internal data as the training set. The number of hidden layers, training function and transfer function of the neural network are optimized by trial and error method.

[0063] Table 2 Example data set structure

[0064]

[0065] In step S6, the neural network model in step S5 is used to perform prediction verification on an external data set. For the prediction results, please refer to Figure 4 , the symbol is the numerical simulation value, and the line is the predicted value. It can be found that the prediction accuracy of this embodiment is high, and the prediction speed of the neural network model is extremely fast, which greatly reduces the prediction time.

[0066] The technical solution provided by the present invention has the following beneficial effects: it takes into account the change of the convective heat transfer coefficient under different temperature differences, overcomes the complexity of the traditional prediction model; adopts a method combining numerical simulation and neural network to reduce the prediction time; simplifies the prediction process, making it clear and simple, and opens up a new way to predict the freezing time of fire protection pipes in the low temperature environment of the converter station.

[0067] The present invention also provides a fire protection pipeline freezing time prediction system in a low temperature environment of a converter station, comprising:

[0068] Data collection module, used to collect low-temperature icing experimental data of fire protection pipelines in converter stations;

[0069] The experimental data mainly include pipeline type, initial temperature, experimental environment temperature, and data on the change of pipe wall temperature over time.

[0070] CFD numerical simulation model building module, used to build the CFD numerical simulation model of the low-temperature environment of the fire protection pipeline and verify the convection heat transfer coefficient based on experimental data;

[0071] In the established CFD numerical simulation model, a temperature sensor is set on the outer wall of the fire protection pipe to obtain temperature changes, and a liquid fraction sensor is set in the diameter direction to obtain the freezing situation. The trial and error method is used to compare the CFD simulation results with the experimental data to determine the convective heat transfer coefficient.

[0072] An empirical formula acquisition module is used to obtain an empirical formula for convection heat transfer coefficient based on a CFD numerical simulation model;

[0073] The empirical formula for convective heat transfer coefficient is as follows:

[0074] h=aΔT b

[0075] h is the convective heat transfer coefficient, △T is the difference between the ambient temperature and the initial temperature, a and b are two parameters that need to be determined, and the least squares method is used to fit the data to determine the convective heat transfer coefficient formula.

[0076] Numerical simulation module, used to output icing simulation data under different working conditions by changing the low temperature environment temperature;

[0077] According to the liquid fraction cloud diagram of the freezing simulation data results, the freezing thickness H is defined as the ratio of the ice thickness h to the inner diameter R of the pipe. The liquid fraction of different nodes is obtained by the liquid fraction sensor in the diameter direction. A liquid fraction less than 0.02 is defined as a solid state, i.e., freezing.

[0078] A neural network module is used to establish and optimize a neural network model with ambient temperature and time as input parameters;

[0079] The number of hidden layers, training function and transfer function of the neural network are optimized by trial and error method with an accuracy of R 2 The correlation coefficient and mean square error (MSE) are used as the criteria for model establishment and optimization.

[0080] The prediction and verification module is used to predict and verify the freezing time of fire pipes under different working conditions based on the neural network model.

[0081] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. 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 embodiments of the present invention.

Claims

1. A method for predicting freezing time of fire protection pipelines in a low temperature environment of a converter station, characterized in that: The following steps are involved: S1, collect low temperature icing experimental data of fire protection pipelines in converter stations; S2, establish a CFD numerical simulation model for low-temperature environment of fire protection pipelines, and verify the convective heat transfer coefficient based on experimental data. In the established CFD numerical simulation model, a temperature sensor is set on the outer wall of the fire protection pipe to obtain temperature changes, and a liquid fraction sensor is set in the diameter direction to obtain the icing situation. The CFD simulation results are compared with the experimental data by trial and error method to determine the convective heat transfer coefficient. S3, based on the CFD numerical simulation model, obtain the empirical formula of the convective heat transfer coefficient; S4, changing the low temperature environment temperature and outputting icing simulation data under different working conditions; S5, based on the simulation data, establish and optimize the neural network model with ambient temperature and time as input parameters; S6, based on the neural network model, predicts and verifies the freezing time of fire pipes under different working conditions.

2. The method for predicting freezing time of fire protection pipes in a low temperature environment of a converter station according to claim 1, characterized in that: In the step S1: the experimental data includes pipeline type, initial temperature, experimental environment temperature, and data on the change of pipe wall temperature over time.

3. A method for predicting freezing time of fire protection pipes in a low temperature environment of a converter station according to claim 1, characterized in that: In step S3: the empirical formula is as follows: h=aΔT b h is the convective heat transfer coefficient, △T is the difference between the ambient temperature and the initial temperature, a and b are two parameters that need to be determined, and the least squares method is used to fit the data to determine the convective heat transfer coefficient formula.

4. A method for predicting freezing time of fire protection pipes in a low temperature environment of a converter station according to claim 1, characterized in that: In step S4: the freezing thickness H is defined as the ratio of the ice thickness h to the inner diameter R of the pipe according to the liquid fraction cloud map of the freezing simulation data result, and the liquid fraction of different nodes is obtained according to the liquid fraction sensor in the diameter direction. The liquid fraction less than 0.02 is defined as a solid state, i.e. freezing.

5. The method for predicting freezing time of fire protection pipes in a low temperature environment of a converter station according to claim 1, characterized in that: In step S5: the number of hidden layers, training function and transfer function of the neural network are optimized by trial and error method, and the accuracy R 2 The correlation coefficient and mean square error (MSE) are used as the criteria for model establishment and optimization.

6. A fire protection pipeline freezing time prediction system in a low temperature environment of a converter station, characterized by: Includes the following modules: Data collection module, used to collect low-temperature icing experimental data of fire protection pipelines in converter stations; CFD numerical simulation model building module, used to build the CFD numerical simulation model of the low-temperature environment of the fire protection pipeline and verify the convection heat transfer coefficient based on experimental data; In the established CFD numerical simulation model, a temperature sensor is set on the outer wall of the fire protection pipe to obtain temperature changes, and a liquid fraction sensor is set in the diameter direction to obtain the icing situation. The CFD simulation results are compared with the experimental data by trial and error method to determine the convective heat transfer coefficient. An empirical formula acquisition module is used to obtain an empirical formula for convection heat transfer coefficient based on a CFD numerical simulation model; Numerical simulation module, used to output icing simulation data under different working conditions by changing the low temperature environment temperature; A neural network module is used to establish and optimize a neural network model with ambient temperature and time as input parameters; The prediction and verification module is used to predict and verify the freezing time of fire pipes under different working conditions based on the neural network model.

7. The fire protection pipeline freezing time prediction system in the low temperature environment of the converter station according to claim 6, characterized in that: In the data collection module: the experimental data includes pipeline type, initial temperature, experimental environment temperature, and pipe wall temperature change data over time.

8. A fire protection pipeline freezing time prediction system in a converter station low temperature environment as claimed in claim 6, characterized in that: In the empirical formula acquisition module: the empirical formula is as follows: h=aΔT b h is the convective heat transfer coefficient, △T is the difference between the ambient temperature and the initial temperature, a and b are two parameters that need to be determined, and the least squares method is used to fit the data to determine the convective heat transfer coefficient formula.

9. A fire protection pipeline freezing time prediction system in a converter station low temperature environment according to claim 6, characterized in that: In the numerical simulation module: the freezing thickness H is defined as the ratio of the ice thickness h to the inner diameter R of the pipe according to the liquid fraction cloud map of the freezing simulation data result, and the liquid fraction of different nodes is obtained according to the liquid fraction sensor in the diameter direction. If it is less than 0.02, it is defined as a solid state, i.e. freezing.

10. A fire protection pipeline freezing time prediction system in a converter station low temperature environment according to claim 6, characterized in that: In the neural network module, the number of hidden layers, training function and transfer function of the neural network are optimized by trial and error method with an accuracy of R 2 The correlation coefficient and mean square error (MSE) are used as the criteria for model establishment and optimization.

Images