Two-phase flow boiling heat transfer numerical simulation method, device and equipment based on deep learning and storage medium

Abstract

The invention provides a two-phase flow boiling heat transfer numerical simulation method, device and equipment based on deep learning and a storage medium, and relates to the technical field of heat transfer science, the method comprises the following steps: inputting a target input parameter into a target deep neural network model to obtain a bubble separation characteristic key parameter, and constructing a wall surface heat flow distribution model according to the key parameter; the bubble separation characteristic key parameters comprise bubble separation diameter, separation frequency, slippage speed, slippage distance, rising diameter and nucleation density; embedding the bubble wall surface heat flow distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework comprises a mass conservation equation, a momentum conservation equation and an energy conservation equation of a liquid phase and a vapor phase; simulation of a two-phase flow boiling heat transfer numerical value is carried out through the two embedded fluid frames, and liquid-phase and vapor-phase temperature field and velocity field distribution is obtained. According to the method, the applicability of numerical simulation on the macroscopic boiling phenomenon is high.

Description

Technical Field

[0001] The present invention relates to the technical field of heat transfer, and particularly relates to a numerical simulation method, device, equipment and storage medium for two-phase flow boiling heat transfer based on deep learning. Background Art

[0002] Two-phase flow boiling heat transfer involves both microscopic-scale bubble dynamics behaviors (such as bubble nucleation, growth, slip and detachment), and macroscopic-scale physical field characteristics (such as flow field and temperature field distributions), which is a typical multi-scale coupling and a strongly non-linear process. The numerical simulation process of two-phase flow boiling heat transfer is very important.

[0003] The traditional two-phase flow numerical simulation method is the interface tracking (Volume of Fluid, VOF) method. The VOF method is one of the core methods in Computational Fluid Dynamics (CFD) for simulating the interface dynamics of multiphase flows, and is particularly suitable for the interface tracking of immiscible fluids such as gas-liquid and liquid-liquid. It can be used to study the motion behavior of a single bubble and is suitable for the study of microscopic-scale boiling mechanisms, but its applicability to macroscopic boiling phenomena is poor. Summary of the Invention

[0004] Aiming at the above deficiencies existing in the prior art, the present invention provides a numerical simulation method, device, equipment and storage medium for two-phase flow boiling heat transfer based on deep learning, which improves the applicability of numerical simulation for macroscopic boiling phenomena.

[0005] In a first aspect, the present invention provides a numerical simulation method for two-phase flow boiling heat transfer based on deep learning, and the method includes the following steps: Input the collected target input parameters into a target deep neural network model to obtain key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, tilt angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; Embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes mass conservation equations, momentum conservation equations, and energy conservation equations for the liquid phase and the vapor phase; Perform numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework to obtain the temperature field and velocity field distributions of the liquid phase and the vapor phase.

[0006] According to a two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention, the target deep neural network model is constructed by the following steps, including: Collecting bubble dynamics data sets of multiple working conditions and multiple working fluids, and determining input parameters of each sample according to the bubble dynamics data sets; Performing dimensionless processing on the sample input parameters to obtain preprocessed sample input parameters; the sample input parameters include fluid flow, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle and fluid properties corresponding to the sample data; Inputting the preprocessed sample input parameters into the initial deep neural network model to obtain output parameters corresponding to the sample input parameters; Determining a total cost function based on the output parameters corresponding to the sample input parameters; According to the total cost function, the weight coefficient of the initial deep neural network model is optimized by the adaptive moment estimation Adam algorithm until the errors of the training set and the validation set are both lower than the preset threshold, so as to obtain the optimal weight coefficient; the total cost function is the weighted sum of the loss functions of the output parameters; Based on the optimal weight coefficients, the target deep neural network model is determined.

[0007] According to a two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention, the bubble dynamics data set of multiple working conditions and multiple working fluids is collected, and the input parameters of each sample are determined according to the bubble dynamics data set, including: Determine data selection conditions; the data selection conditions include the selected working fluid type, the coverage pressure range, the coverage flow range, the inclination angle and the wall superheat; the selected working fluid type includes at least water, liquid nitrogen and a new cooling working fluid; the coverage pressure range is 0.1 MPa to 20 MPa, the coverage flow range is 0.1 m / s to 10 m / s, and the inclination angle is 0 to 90 degrees; Determine input data based on data selection conditions; The input data is optimized through feature screening and redundant parameters are eliminated to obtain the input parameters of each sample.

[0008] According to a two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention, the target deep neural network model includes an input layer, a hidden layer, and an output layer, the number of nodes in the input layer is consistent with the dimension of the target input parameter, and the number of nodes in the output layer is consistent with the number of key parameters of the bubble detachment characteristic to be predicted; the collected target input parameters are input into the target deep neural network model to obtain the key parameters of the bubble detachment characteristic, including: Collect the target input parameters, perform normalization processing on the target input parameters, and obtain the preprocessed target input parameters; Input the preprocessed target input parameters into the input layer to obtain a feature vector matrix; Input the feature vector matrix into the hidden layer to obtain a weighted feature vector matrix; Input the weighted feature vector matrix into the output layer to obtain an output vector, and map the output vector to specific key parameters of the bubble detachment characteristics.

[0009] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, the bubble wall heat flux distribution model includes heat flux distribution sub-models corresponding to evaporation heat flux, wall quenching heat flux, sliding bubble heat flux, and convective heat flux respectively.

[0010] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, constructing a bubble wall heat flux distribution model according to the key parameters of the bubble detachment characteristics includes: Based on the bubble detachment frequency, the nucleation density, and the rising diameter, construct a heat flux distribution sub-model corresponding to the evaporation heat flux; According to the bubble detachment diameter and the wall temperature gradient, construct a heat flux distribution sub-model corresponding to the wall quenching heat flux; Based on the swept area and time of the sliding bubble, calculate a heat flux distribution sub-model corresponding to the sliding bubble heat flux; Calculate a heat flux distribution sub-model corresponding to the convective heat flux through the remaining heating area.

[0011] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, embedding the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain an embedded two-fluid framework further includes: Determine the target mass exchange and target energy exchange generated by the bubble wall heat flux distribution model; Determine the target mass exchange as the mass exchange source term in the Euler-Euler two-fluid model, and determine the target energy exchange as the energy exchange source term in the Euler-Euler two-fluid model to obtain the embedded two-fluid framework.

[0012] In a second aspect, the present invention further provides a numerical simulation device for two-phase flow boiling heat transfer based on deep learning. The device includes the following modules: A building module for inputting the collected target input parameters into a target deep neural network model to obtain key parameters of bubble detachment characteristics, and constructing a bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat degree, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; Embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; A simulation module for simulating the two-phase flow boiling heat transfer numerically through the embedded two-fluid framework to obtain the temperature field and velocity field distributions of the liquid phase and the vapor phase.

[0013] In a third aspect, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the computer program, it implements the numerical simulation method of two-phase flow boiling heat transfer based on deep learning as described in any one of the above.

[0014] In a fourth aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the numerical simulation method of two-phase flow boiling heat transfer based on deep learning as described in any one of the above.

[0015] In a fifth aspect, the present invention also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the numerical simulation method of two-phase flow boiling heat transfer based on deep learning as described in any one of the above.

[0016] The numerical simulation method, device, equipment, and storage medium for two-phase flow boiling heat transfer based on deep learning provided by the present invention first input the collected target input parameters into the target deep neural network model to obtain the key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model based on the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; then, embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework, and the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; furthermore, perform numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework to obtain the temperature field and velocity field distribution of the liquid phase and the vapor phase.

[0017] In the present invention, the bubble characteristic parameters (key parameters of bubble detachment characteristics) are predicted by a deep neural network, a model of two-phase wall boiling phenomenon is established based on the bubble characteristic parameters, and the Euler-Euler two-phase flow model is coupled to obtain a cross-scale and wide-applicability two-phase boiling heat transfer calculation model. That is, the macroscopic two-fluid Euler-Euler model is used as the framework, and the deep neural network model is used to replace the microscopic bubble detachment model, and this model has high applicability for numerical simulation of macroscopic boiling phenomena. Brief Description of the Drawings

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0019] Figure 1 It is one of the flow diagrams of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention.

[0020] Figure 2 It is the second flow diagram of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention.

[0021] Figure 3 It is the third flow diagram of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention.

[0022] Figure 4 It is the fourth flow diagram of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention.

[0023] Figure 5 It is the fifth flow schematic diagram of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention.

[0024] Figure 6 It is the structural schematic diagram of the device for numerical simulation of two-phase flow boiling heat transfer based on deep learning provided by the present invention.

[0025] Figure 7 It is the structural schematic diagram of the electronic device provided by the present invention. Detailed implementation manners

[0026] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below with reference to the accompanying drawings in the present invention. Apparently, the described embodiments are some but not all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in the present invention without creative efforts shall fall within the protection scope of the present invention.

[0027] To more clearly understand the various embodiments provided by the present invention, the technical content related to the present invention is introduced as follows: The traditional two-phase flow numerical simulation method also includes the two-fluid (Euler-Euler) method. The two-fluid (Euler-Euler) method is based on the statistical characteristics of bubbles and often combines a wall heat flux distribution model to describe the boiling phenomenon. It is widely used. However, its bubble detachment characteristic sub-model depends on empirical relations. Due to the limitations of the empirical relations themselves, the application scope of the model is also limited. For example, the sub-model applicable to water is no longer applicable to liquid nitrogen, and there is a lack of corresponding experimental relations for new working fluids, resulting in limited application of the model.

[0028] Aiming at the above deficiencies in the prior art, the present invention proposes a numerical simulation method for two-phase flow boiling heat transfer with a wide application scope, less computing resources and high computing accuracy by using the macroscopic Euler-Euler two-fluid model as a framework, replacing the microscopic bubble detachment model with a deep neural network model to predict the key parameters of bubble detachment characteristics, and then constructing a multi-scale coupling framework.

[0029] The following combines Figures 1 - 7 to describe the numerical simulation method, device, equipment and storage medium for two-phase flow boiling heat transfer based on deep learning of the present invention.

[0030] Figure 1 It is one of the flow schematic diagrams of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention. As Figure 1 shown, the method includes the following: Step 101: Input the collected target input parameters into the target deep neural network model to obtain the key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model based on the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density. First of all, it should be noted that the execution subject of the present invention is an electronic device, which is used to realize the numerical simulation of two-phase flow boiling heat transfer based on deep learning to improve the applicability of the numerical simulation of phase flow boiling heat transfer.

[0031] Among them, the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties. In fluid mechanics and heat transfer, parameters such as fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties have important influences on flow and heat transfer characteristics. That is to say, the target input parameters are variables that affect the bubble detachment characteristic parameters. The fluid flow rate (Flow Rate) is the volume of fluid passing through the flow channel per unit time (volume flow rate, cubic meters per second m³ / s) or mass (mass flow rate, kilograms per second kg / s). The wall superheat (Wall Superheat) is the difference between the wall temperature T w and the fluid saturation temperature T sat difference , which is particularly crucial in boiling heat transfer. The fluid pressure (Fluid Pressure) is the hydrostatic pressure of the fluid or the system operating pressure (Pascal Pa or megapascal MPa). The contact angle (Contact Angle) is the angle between the liquid-solid interface and the gas-liquid interface when a liquid droplet contacts the solid surface, which characterizes wettability. The wall roughness (Wall Roughness) is the average height of the surface micro-irregularities (μm level), usually represented by Ra or Rz. The inclination angle (Inclination Angle) is the angle of inclination of the flow channel or wall relative to the horizontal direction.

[0032] In this embodiment, the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density.

[0033] Among them, the bubble departure diameter refers to the equivalent diameter when the bubble detaches from the wall surface (usually the diameter under the assumption of a spherical shape). The influencing factors of the bubble departure diameter include the wall superheat, surface tension, and contact angle. The bubble departure frequency is the number of bubbles generated by a single nucleation site per unit time. The slip velocity is the upward velocity of the bubble relative to the liquid, determined by the balance between buoyancy and drag. The sliding distance is the horizontal distance that the bubble slides along the wall surface after detaching from the wall. The bubble rising diameter is the dynamic diameter of the bubble in the free flow after detaching from the wall, affected by coalescence or breakup. The nucleation density is the number of active vaporization nuclei per unit area.

[0034] In traditional numerical simulation methods such as the two-fluid (Euler-Euler) method, the sub-model of bubble detachment characteristics relies on empirical relationships. Due to the limitations of the empirical relationships themselves, the application range of the model is also restricted. For example, the sub-model applicable to water is no longer applicable to liquid nitrogen, and there is a lack of corresponding experimental relationships for new working fluids, resulting in limited model application. Therefore, in the present invention, a deep neural network model is used to predict the key parameters of bubble detachment characteristics, and the collected target input parameters are input into the target deep neural network model to obtain the key parameters of bubble detachment characteristics.

[0035] Furthermore, based on the key parameters of bubble detachment characteristics predicted by the deep neural network, the two-phase wall boiling phenomenon can be modeled to obtain a bubble wall heat flux distribution model. The wall heat flux distribution model is used to describe the wall boiling heat transfer. The bubble wall heat flux distribution model includes multiple sub-models. Among them, the enhancement of boiling heat transfer during forced convection can be attributed to the presence of both slip bubbles and stationary bubbles. It is mainly through two mechanism principles: 1) latent heat transfer due to the evaporation of the micro-liquid layer; 2) instantaneous heat conduction during the reconstruction process of the disrupted thermal boundary layer during the waiting period (i.e., during the formation of the next bubble at the same nucleation site).

[0036] Step 102: Embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and vapor phase. Specifically, after obtaining the bubble wall heat flux distribution model through modeling, the bubble wall heat flux distribution model is further embedded into the Euler-Euler two-fluid framework, that is, the bubble wall heat flux distribution model and the Euler-Euler two-fluid model are coupled to obtain the embedded two-fluid framework. Among them, the embedded two-fluid framework can be understood as a two-phase flow boiling model based on a neural network, which is used for numerical simulation of two-phase flow boiling heat transfer, that is, to simulate flow boiling.

[0037] Among them, the Euler-Euler two-fluid model mainly solves the mass conservation equation, momentum conservation equation, and energy conservation equation of the vapor phase and liquid phase. The wall boiling model mainly calculates the wall heat flux distribution. The wall heat flux distribution includes sub-models for the distribution of quenching heat flux, convective heat flux, evaporation heat flux, and slip heat flux. Key parameters such as the bubble departure diameter, bubble departure frequency, bubble rising diameter, and nucleation density of the bubble in these heat flux distribution sub-models are predicted using a deep neural network.

[0038] Among them, the momentum conservation equation is as follows:

[0039] Among them, represents time (seconds, s), represents the void fraction of phase q, represents the density of phase q (kilograms per cubic meter, kg / m 3 ) represents the velocity of phase q (meters per second, m / s), is the interphase mass transfer from phase p to phase q (kg / m 3 ) represents the interphase mass transfer from phase q to phase p (kg / m 3 ) represents phase p, such as the vapor phase, and q represents phase q, such as the liquid phase, represents the mass source term (kg / m 3 ) represents the gradient operator.

[0040] The mass conservation equation is as follows:

[0041] Among them, represents the void fraction of phase q, represents the density of phase q (kg / m 3 ) represents the pressure (Pa), represents the gradient operator, represents the gravitational constant, represents the velocity of phase q (m / s), represents the inter-phase mass transfer from phase p to phase q (kg / m 3 ), represents the inter-phase mass transfer from phase q to phase p (kg / m 3 ), is the inter-phase velocity from phase p to phase q (meters per second, m·s -1 ), is the inter-phase velocity from phase q to phase p (meters per second, m·s -1 ), is the inter-phase force (Newtons per cubic meter, N·m -3 ), is the external volume force (N·m -3 ), is the shear lift force (N·m -3 ); is the virtual mass force (N·m -3 ), is the wall lubrication force (N·m -3 ), is the turbulent dissipation force (N·m -3 ), represents the gradient operator, represents the stress tensor of phase q (Pascals, Pa).

[0042] The energy conservation equation is as follows:

[0043] represents the void fraction of phase q, represents the density of phase q (kg / m 3 ), represents the enthalpy of phase q (kilojoules per kilogram, kJ / kg), represents the velocity of phase q (m / s), represents the thermal conductivity of phase q (watts per meter per Kelvin, W / (m·K)), represents the temperature of phase q (Kelvin, K), represents the inter-phase heat transfer (kilojoules per cubic meter, kJ / m 3 ), represents the inter-phase mass transfer from phase p to phase q (kg / m 3 ), represents the inter-phase saturation enthalpy of phase p (kJ / kg), represents the inter-phase mass transfer from phase q to phase p (kg / m 3 ), represents the inter-phase saturation enthalpy of phase q (kJ / kg), represents the energy source term (kJ / m 3).

[0044] Step 103: Perform numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework to obtain the temperature fields and velocity field distributions of the liquid phase and the vapor phase.

[0045] Specifically, performing numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework, that is, simulating flow boiling, to obtain, for example, the temperature fields and velocity field distributions of the liquid phase and the vapor phase.

[0046] For the method provided in this embodiment, first, input the collected target input parameters into the target deep neural network model to obtain the key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model based on the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, tilt angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; then, embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework, and the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; furthermore, perform numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework to obtain the temperature fields and velocity field distributions of the liquid phase and the vapor phase.

[0047] In the present invention, the bubble characteristic parameters (key parameters of bubble detachment characteristics) are predicted based on a deep neural network, a model of two-phase wall boiling phenomenon is built based on the bubble characteristic parameters, and the Euler-Euler two-phase flow model is coupled to obtain a cross-scale and wide-applicability two-phase boiling heat transfer calculation model, that is, using the macroscopic two-fluid Euler-Euler model as the framework and replacing the microscopic bubble detachment model with a deep neural network model, and this model has high applicability for numerical simulation of macroscopic boiling phenomena.

[0048] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, the target deep neural network model is constructed through the following steps, including: Collect a bubble dynamics data set of multiple working conditions and multiple working fluids, and determine each sample input parameter according to the bubble dynamics data set; Perform dimensionless processing on each sample input parameter to obtain the preprocessed each sample input parameter; each sample input parameter includes the fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, tilt angle, and fluid physical properties corresponding to the sample data; Input the preprocessed each sample input parameter into the initial deep neural network model to obtain the output parameter corresponding to each sample input parameter; Determine the total cost function based on the output parameters corresponding to each sample input parameter; According to the total cost function, optimize the weight coefficients of the initial deep neural network model through the Adaptive Moment Estimation (Adam) algorithm until the errors of both the training set and the validation set are lower than the preset threshold to obtain the optimal weight coefficients; the total cost function is the weighted sum of the loss functions of each output parameter; Determine the target deep neural network model based on the optimal weight coefficients.

[0049] Specifically, in some embodiments, the steps for constructing the target deep neural network model are as follows, including: First, collect the bubble dynamics data sets of multiple working conditions and multiple working fluids, and determine each sample input parameter according to the bubble dynamics data sets. Further, perform dimensionless processing on each sample input parameter to obtain the preprocessed sample input parameters.

[0050] The data set mainly includes sample input parameters such as the fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, tilt angle, and fluid physical properties corresponding to the sample data, as well as key parameters of bubble detachment characteristics such as the bubble detachment diameter, bubble detachment frequency, bubble slip distance, and bubble nucleation density. The key parameters are used as label data to guide the model to learn the internal relationship between the key parameters and the input parameters. The data source can come from experimental data (such as particle image velocimetry, infrared thermometry), simulation data, and publicly available literature data sets. The process of determining the sample input parameters according to the data set is a preprocessing process, for example, performing feature screening, optimizing the sample input parameters, dividing the training set and the validation set, and so on.

[0051] Further, input the preprocessed sample input parameters into the initial deep neural network model to obtain the output parameters corresponding to each sample input parameter. Based on the output parameters corresponding to each sample input parameter, determine the total cost function. Furthermore, according to the total cost function, optimize the weight coefficients of the initial deep neural network model through the Adaptive Moment Estimation (Adam) algorithm until the errors of both the training set and the validation set are lower than the preset threshold to obtain the optimal weight coefficients. Among them, the weight coefficients of the initial deep neural network model are not optimized, and the initial deep neural network model is used for prediction. Then, based on the prediction results (the output parameters corresponding to each sample input parameter) and the label data, calculate the total cost function. The total cost function is the weighted sum of the loss functions of each output parameter, and is expressed as follows:

[0052] Wherein, represents the total cost function, w is the coefficient matrix composed of inputs and outputs, is the bias term, usually set to a constant. i represents the i-th group of data, and j represents the j-th component. represents the loss function of the j-th component of the i-th group of data. represents the predicted value of the j-th component of the i-th group of data. represents the measured value (label data) of the j-th component of the i-th group of data, and m is the total of m groups of data.

[0053] The input and output of the neural network satisfy the following relationship: . In the formula, w is the coefficient matrix composed of the input and output, b is the bias term, is the input term, and Y is the output term.

[0054] Among them, is expressed as follows:

[0055] Among them, represents the loss function of the j-th component of the i-th group of data, i represents the i-th group of data, and j represents the j-th component. represents the predicted value of the j-th component of the i-th group of data. represents the measured value (label data) of the j-th component of the i-th group of data.

[0056] After that, in order to minimize the total cost function, the optimal weight coefficients w and b are found according to the Adaptive Moment Estimation Adam algorithm, that is, the optimal coefficient matrix w composed of the input and output and the optimal bias term b are obtained.

[0057] Finally, based on the optimal weight coefficients, the target deep neural network model can be determined. That is, the optimal weight coefficients are used as the updated model parameters to obtain the updated deep neural network model, that is, the target deep neural network model.

[0058] The method provided in this embodiment trains the target deep neural network model through the collected bubble dynamics data set. After that, the key parameters of the bubble detachment characteristics are predicted based on the target input parameters through the target deep neural network model. Then, based on the key parameters, the two-phase wall boiling phenomenon is modeled, and the Euler-Euler two-phase flow model is coupled to obtain a cross-scale and wide-applicability two-phase boiling heat transfer calculation model, improving the applicability of the model.

[0059] According to a two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention, a bubble dynamics data set of multiple working conditions and multiple working fluids is collected, and the input parameters of each sample are determined according to the bubble dynamics data set, including: Determine the data selection conditions; the data selection conditions include the selected working fluid type, the covered pressure range, the covered flow rate range, the inclination angle, and the wall superheat degree; the selected working fluid type includes at least water, liquid nitrogen, and a new type of cooling working fluid; the covered pressure range includes 0.1 MPa to 20 MPa, the flow rate range includes 0.1 m / s to 10 m / s, and the inclination angle includes 0° to 90°; Based on the data selection conditions, determine the input data; Optimize the input data through feature screening and eliminate redundant parameters to obtain the input parameters of each sample.

[0060] Specifically, in some embodiments, the specific implementation process of collecting the bubble dynamics data sets of multiple working conditions and multiple working fluids and determining the input parameters of each sample based on the bubble dynamics data sets includes the following steps: First, determine the data selection conditions, where the data selection conditions include the selected working fluid type, the covered pressure range, the covered flow rate range, the inclination angle, and the wall superheat degree. Further, based on the data selection conditions, determine the input data.

[0061] To make the model have a wider applicable range, the working condition range included in the selected data should be as wide as possible. For example, select data of different working fluids, data under a wide pressure range and a wide flow rate range, data of different inclination angles, and experimental data under different wall superheat degrees, so that the trained model can have a wide application range. For example, the selected working fluid type includes at least water, liquid nitrogen, and a new type of cooling working fluid; the covered pressure range includes 0.1 MPa to 20 MPa, the flow rate range includes 0.1 m / s to 10 m / s, and the inclination angle includes 0° to 90°.

[0062] Further, optimize the input data through feature screening and eliminate redundant parameters to obtain the input parameters of each sample. For example, find important feature parameters through feature screening to optimize the input data; furthermore, eliminate redundant parameters to obtain the sample input parameters.

[0063] It is also possible to divide the sample input parameters into a training set and a validation set. The training set is mainly used to train the weight coefficients of the neural network to make the error between the network prediction value and the true value within an acceptable range. The validation set data is mainly used to verify the generalization characteristics of the neural network. When both the training set error and the validation set error meet the requirements, it can be considered that the performance of the neural network is good.

[0064] Exemplarily, Figure 2 is the second flow chart of the two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention. As Figure 2 shown, the method includes: Obtain the original data pool: including experimental data, numerical simulation data, and literature data; Representative data screening: including coverage check, outlier removal, feature selection, and balance processing; Dataset verification: physical reasonableness, extrapolation test, and sensitivity analysis; Generate a representative dataset: The representative dataset evenly covers all key working conditions, and the distribution of bubble parameters conforms to physical laws.

[0065] For the method provided in this embodiment, the working condition range included in the data selected for the dataset is as wide as possible. After preprocessing the data in the dataset, the initial deep neural network model is trained using the preprocessed input data, enabling the model to better learn the associations between the data, and the trained target deep neural network model has a better prediction effect.

[0066] According to a two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention, the target deep neural network model includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is the same as the dimension of the target input parameters, and the number of nodes in the output layer is the same as the number of key parameters of the bubble detachment characteristics to be predicted; the collected target input parameters are input into the target deep neural network model to obtain the key parameters of the bubble detachment characteristics, including: Collect the target input parameters, perform normalization processing on the target input parameters to obtain the preprocessed target input parameters; Input the preprocessed target input parameters into the input layer to obtain a feature vector matrix; Input the feature vector matrix into the hidden layer to obtain a weighted feature vector matrix; Input the weighted feature vector matrix into the output layer to obtain an output vector, and map the output vector to specific key parameters of the bubble detachment characteristics.

[0067] Specifically, in some embodiments, the target deep neural network model includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is the same as the dimension of the target input parameters, and the number of nodes in the output layer is the same as the number of key parameters of the bubble detachment characteristics to be predicted. For example, if the input is pressure, temperature, fluid viscosity, etc. in a time series, the number of nodes is equal to the number of features.

[0068] Correspondingly, the specific implementation process of inputting the collected target input parameters into the target deep neural network model in step 101 to obtain the key parameters of the bubble detachment characteristics includes the following steps: First, collect the target input parameters, perform normalization processing on the target input parameters to obtain the preprocessed target input parameters. Further, input the preprocessed target input parameters into the input layer to obtain a feature vector matrix. For example, collect the target input parameters to form a feature vector X = [x1, x2,... x n .

[0069] Normalizing the target input parameters helps the network converge quickly. The normalization formula is:

[0070] where represents the normalized target input parameter, is the target input parameter, is the minimum value of the target input parameter, represents the maximum value of the input parameter.

[0071] Then, input the feature vector matrix into the hidden layer to obtain the weighted feature vector matrix.

[0072] Furthermore, input the weighted feature vector matrix into the output layer to obtain the output vector, and map the output vector to specific key parameters of bubble detachment characteristics. (Such as Y1 = bubble detachment diameter, Y2 = bubble detachment frequency, Y3 = bubble slip distance, Y4 = bubble nucleation density).

[0073] Exemplarily, Figure 3 is the third schematic diagram of the process of the two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention. As Figure 3 shown, the input layer inputs fluid wall parameters (including: fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, wall inclination angle, fluid physical properties). Then, input the feature vector output by the input layer into the hidden layer. The hidden layer includes a multi-head attention mechanism, including H1_4, H1_3, H1_2, H1_1, H2_2, H2_1. After the hidden layer processes the feature vector, a weighted feature vector matrix is obtained. Then, input the weighted feature vector matrix into the output layer to obtain the finally predicted bubble characteristic parameters (including: bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density).

[0074] The method provided in this embodiment models the two-phase wall boiling phenomenon based on the bubble characteristic parameters predicted by the target deep neural network model, and couples the Euler-Euler two-phase flow model to obtain a cross-scale and wide-applicability two-phase boiling heat transfer calculation model, and the applicability of this model is relatively high.

[0075] According to a two-phase flow boiling heat transfer numerical simulation method based on deep learning provided by the present invention, the bubble wall heat flux distribution model includes heat flux distribution sub-models corresponding to evaporation heat flux, wall quenching heat flux, sliding bubble heat flux, and convective heat flux respectively.

[0076] Specifically, in some embodiments, the bubble wall heat flux distribution model includes heat flux distribution sub-models corresponding to evaporation heat flux, wall quenching heat flux, sliding bubble heat flux, and convective heat flux respectively.

[0077] These heat flux distribution sub-models are then used as source terms in equations such as the energy equation to couple the bubble wall heat flux distribution model and the Euler-Euler two-fluid model, obtaining a two-phase flow boiling heat transfer model based on a deep learning network for simulating boiling heat transfer.

[0078] In the method provided in this embodiment, the bubble wall heat flux distribution model includes heat flux distribution sub-models corresponding to evaporation heat flux, wall quenching heat flux, sliding bubble heat flux, and convective heat flux respectively. After modeling, the heat flux distribution sub-models are coupled with the Euler-Euler two-fluid model to realize numerical simulation of two-phase flow boiling heat transfer based on deep learning, with high applicability.

[0079] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, a bubble wall heat flux distribution model is constructed based on key parameters of bubble detachment characteristics, including: Constructing a heat flux distribution sub-model corresponding to evaporation heat flux based on bubble detachment frequency, nucleation density, and rising diameter; Constructing a heat flux distribution sub-model corresponding to wall quenching heat flux according to bubble detachment diameter and wall temperature gradient; Calculating a heat flux distribution sub-model corresponding to sliding bubble heat flux based on the swept area and time of the sliding bubble; Calculating a heat flux distribution sub-model corresponding to convective heat flux through the remaining heating area.

[0080] Specifically, in some embodiments, the process of constructing the bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics in step 101 is implemented through the following steps, including: First, construct a heat flux distribution sub-model corresponding to evaporation heat flux based on bubble detachment frequency, nucleation density, and rising diameter. For example, the evaporation heat flux is calculated through the following formula:

[0081] where q e represents the evaporation heat flux density (kW / m 2 ), R f represents the reduction factor, represents the bubble nucleation density (1 / m 2 ), that is, the nucleation density, is the bubble rising diameter (m), is the vapor density (kg / m 3 ), is the latent heat of vaporization (kJ / kg), is the bubble detachment frequency (1 / s), is the bubble volume (m 3 ).

[0082] Furthermore, based on the bubble detachment diameter and the wall temperature gradient, a sub-model for heat flux distribution corresponding to the wall quenching heat flux is constructed. For example, the wall quenching heat flux is calculated by the following formula:

[0083] where is the wall quenching heat flux density (kW / m 2 ), is the liquid phase thermal conductivity (watts per meter per degree, W / (m·K)), is the liquid phase density (kg / m 3 ), is the liquid phase specific heat capacity (J / (kg·K)), is the temperature of the heater surface (K), is the temperature of the liquid (K), D d is the bubble detachment diameter (m), R f represents the reduction factor, is the bubble nucleation density (1 / m 2 ), K is the bubble influence factor, is the bubble waiting time (s), is the bubble detachment frequency (1 / s).

[0084] where R f The reduction factor depicts the ratio of the actual number of detached bubbles per unit heater surface area to the bubble activation nuclei per unit area, i.e.: R f = 1 / (l s / s + 1) where Rf represents the reduction factor, l s represents the sliding distance (meters, m); s represents the spacing between two activation nuclei (meters, m). It is assumed that the nucleation sites are distributed in a square grid and the bubbles only slide in the fluid flow direction. Therefore, the spacing between the activation nuclei can be approximated as , is the bubble nucleation density (1 / m 2 ). Note that for the sliding distance l s less than the activation nucleus spacing s, .

[0085] Furthermore, based on the swept area and time of the sliding bubbles, a sub-model for heat flux distribution corresponding to the sliding bubble heat flux is calculated.

[0086] For a sliding bubble, the slip heat flux density is:

[0087] where represents the slip heat flux density (kilowatts per square meter, kW / m 2 ), represents the sliding distance (meters, m), the average bubble diameter D avg = (D d + D l ) / 2, D d is the bubble detachment diameter / m, D l is the bubble rising diameter (m), R f represents the reduction factor, is the liquid-phase thermal conductivity (W / (m K)), is the liquid-phase density (kg / m 3 ), is the liquid-phase specific heat capacity (J / (kg·K)), is the temperature of the heater surface (kelvins, K), is the temperature of the liquid (K), K is the bubble influence factor, is the bubble detachment frequency (1 / s), represents the slip time (seconds, s), is the bubble nucleation density (1 / m 2 ).

[0088] Furthermore, a heat flux distribution sub-model corresponding to the convective heat flux is calculated through the remaining heating area. For example, the convective heat flux is calculated by the following formula: where forced convection always exists in the heating surface area not affected by stationary bubbles and slip bubbles, and the forced convection heat transfer area ratio is:

[0089] where, is the forced convection heat transfer area ratio, represents the reduction factor, is the bubble nucleation density (1 / m 2 ), is the sliding displacement (m), D avg is the average bubble diameter (m), D d is the bubble detachment diameter (m), t w is the bubble waiting time (s), is the bubble detachment frequency (1 / s), K is the bubble influence factor, represents the slip time (seconds, s).

[0090] The forced convection heat flux (i.e., the wall convective heat flux density) q c can be expressed as:

[0091] represents the temperature of the heater surface (K), represents the temperature of the liquid (K), is the convective heat transfer coefficient (watts per square meter per kelvin, W / m 2 ·K), is the forced convection heat transfer area ratio.

[0092] Furthermore, the total wall heat flux (i.e., the total wall heat flux density) can be determined based on the above heat flux components:

[0093] wherein, is the total wall heat flux density, is the wall evaporation heat flux density, is the wall quenching heat flux density, is the wall slip heat flux density, is the wall convective heat flux density.

[0094] The method provided in this embodiment uses a deep neural network to predict key parameters such as the bubble departure diameter, bubble departure frequency, bubble rise diameter, and bubble nucleation density. Furthermore, based on these key parameters, a wall heat flux distribution model is constructed. Then, the wall heat flux distribution model is coupled with the Euler-Euler two-fluid model to achieve numerical simulation of two-phase flow boiling heat transfer, with strong applicability.

[0095] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, embedding the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework, the obtained embedded two-fluid framework includes: Determine the target mass exchange and target energy exchange generated by the wall boiling model; Set the target mass exchange as the mass exchange source term in the Euler-Euler two-fluid model, and set the target energy exchange as the energy exchange source term in the Euler-Euler two-fluid model, to obtain the embedded two-fluid framework.

[0096] Specifically, in some embodiments, step 102 can be implemented through the following steps, including: The process steps of coupling the bubble wall heat flux distribution model and the Euler-Euler two-fluid model are as follows: First, calculate the mass exchange generated by the wall boiling model, and the calculation formula is:

[0097] wherein, represents the mass exchange during boiling (kilograms per cubic meter, kg / m 3 ), is the wall surface evaporation heat flux density (kilowatts per square meter, kW / m 2 ), is the latent heat of vaporization (kilojoules per kilogram, kJ / kg), is the liquid specific heat capacity (joules per kilogram per kelvin, J / (kg·K)), is the subcooling (kelvin, K).

[0098] The heat source generated by boiling is:

[0099] is the energy source term for boiling to occur (kilojoules per cubic meter, kJ / m 3 ), is the mass exchange for boiling to occur (kg / m 3 ), is the latent heat of vaporization (kJ / kg).

[0100] Furthermore, the mass exchange and energy exchange S H calculated above are substituted into the mass exchange and energy source terms of the three conservation equations in the Euler-Euler two-fluid model, thereby realizing the coupling of the Euler-Euler two-fluid and the wall boiling model.

[0101] The method provided in this embodiment embeds the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework.

[0102] Figure 4 is the fourth flow chart of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention. As Figure 4 shown, the method includes: First, the key parameters of bubble detachment characteristics are predicted by a deep neural network: The deep neural network includes an input layer, a hidden layer, and an output layer.

[0103] Then, based on the key parameters of bubble detachment characteristics, a bubble wall heat flux distribution model is constructed. The total heat flux distribution in the bubble wall heat flux distribution model includes evaporation heat flux, quenching heat flux, convective heat flux, and slip heat flux.

[0104] Furthermore, the wall heat flux contribution output by the bubble wall heat flux distribution model is used as the energy exchange source term and the mass exchange source term, and coupled to the Euler-Euler two-fluid model. The Euler-Euler two-fluid model includes a mass conservation equation, a momentum conservation equation, and an energy conservation equation.

[0105] Figure 5This is the fifth flow schematic diagram of the numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention. As Figure 5 shown, the method includes: Obtain experimental data, numerical simulation data, and literature data, as well as fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, tilt angle, and fluid physical properties; Perform dimensionless and feature screening processing on the input parameters in the obtained dataset.

[0106] Input the processed dataset into the initial deep neural network model, and train to obtain the target deep neural network model. Use the target deep neural network model to predict the output parameters of bubble characteristics. The output parameters of bubble characteristics include bubble departure diameter, departure frequency, slip velocity, slip distance, rising diameter, and nucleation density.

[0107] After that, perform CFD coupling based on the output parameters of bubble characteristics, and predict the numerical value of boiling heat transfer in real time.

[0108] The method provided in this embodiment dynamically couples the bubble slip velocity and slip distance through the interfacial force model, thereby updating the two-phase flow field parameters.

[0109] Next, the numerical simulation device for two-phase flow boiling heat transfer based on deep learning provided by the present invention will be described. The numerical simulation device for two-phase flow boiling heat transfer based on deep learning described below can be mutually corresponding and referred to the numerical simulation method for two-phase flow boiling heat transfer based on deep learning described above.

[0110] Figure 6 This is the structural schematic diagram of the numerical simulation device for two-phase flow boiling heat transfer based on deep learning provided by the present invention. As Figure 6 shown, the numerical simulation device 600 for two-phase flow boiling heat transfer based on deep learning includes the following modules: A construction module 610, configured to input the collected target input parameters into the target deep neural network model to obtain the key parameters of bubble departure characteristics, and construct a bubble wall heat flux distribution model according to the key parameters of bubble departure characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, tilt angle, and fluid physical properties; the key parameters of bubble departure characteristics include bubble departure diameter, departure frequency, slip velocity, slip distance, rising diameter, and nucleation density; Embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; A simulation module 620, configured to perform a numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework, so as to obtain the temperature fields and velocity field distributions of the liquid phase and the vapor phase.

[0111] The device provided in this embodiment includes a construction module 610 and a simulation module 620. The construction module 610 is configured to input the collected target input parameters into a target deep neural network model to obtain key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; then, embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework, and the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; furthermore, the simulation module 620 is configured to perform a numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework, so as to obtain the temperature fields and velocity field distributions of the liquid phase and the vapor phase.

[0112] In the present invention, bubble characteristic parameters (key parameters of bubble detachment characteristics) are predicted according to a deep neural network, a two-phase wall boiling phenomenon is modeled based on the bubble characteristic parameters, and an Euler-Euler two-phase flow model is coupled to obtain a two-phase boiling heat transfer calculation model with cross-scale and wide applicability. That is, the macroscopic two-fluid Euler-Euler model is used as a framework, and a deep neural network model is used to replace the microscopic bubble detachment model. This model has high applicability for numerical simulation of macroscopic boiling phenomena.

[0113] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, the target deep neural network model is constructed through the following steps, including: Collect a bubble dynamics data set of multiple working conditions and multiple working fluids, and determine each sample input parameter according to the bubble dynamics data set; Perform dimensionless processing on each sample input parameter to obtain the preprocessed each sample input parameter; each sample input parameter includes the fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties corresponding to the sample data; Input the preprocessed each sample input parameter into an initial deep neural network model to obtain the output parameter corresponding to each sample input parameter; Determine the total cost function based on the output parameters corresponding to each sample input parameter; According to the total cost function, optimize the weight coefficients of the initial deep neural network model through the Adaptive Moment Estimation (Adam) algorithm until the errors of both the training set and the validation set are lower than a preset threshold to obtain the optimal weight coefficients; the total cost function is the weighted sum of the loss functions of the output parameters. Based on the optimal weight coefficients, determine the target deep neural network model.

[0114] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, collect the bubble dynamics data sets of multiple working conditions and multiple working fluids, and determine each sample input parameter according to the bubble dynamics data sets, including: Determine the data selection conditions; the data selection conditions include the selected working fluid types, the covered pressure range, the covered flow rate range, the inclination angle, and the wall superheat; the selected working fluid types include at least water, liquid nitrogen, and new cooling working fluids; the covered pressure range is from 0.1 MPa to 20 MPa, the covered flow rate range is from 0.1 m / s to 10 m / s, and the inclination angle is from 0° to 90°. Based on the data selection conditions, determine the input data. Optimize the input data through feature screening and eliminate redundant parameters to obtain each sample input parameter.

[0115] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, the target deep neural network model includes an input layer, a hidden layer, and an output layer. The number of nodes in the input layer is consistent with the dimension of the target input parameters, and the number of nodes in the output layer is consistent with the number of key parameters of the bubble detachment characteristics to be predicted. The construction module 610 is specifically used for: Collect the target input parameters, perform normalization processing on the target input parameters to obtain the preprocessed target input parameters. Input the preprocessed target input parameters into the input layer to obtain a feature vector matrix. Input the feature vector matrix into the hidden layer to obtain a weighted feature vector matrix. Input the weighted feature vector matrix into the output layer to obtain an output vector, and map the output vector to specific key parameters of the bubble detachment characteristics.

[0116] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, the bubble wall heat flux distribution model includes heat flux distribution sub-models corresponding to the evaporation heat flux, the wall quenching heat flux, the sliding bubble heat flux, and the convective heat flux respectively.

[0117] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, the construction module 610 is further configured to: Construct a heat flux distribution sub-model corresponding to the evaporation heat flux based on the bubble detachment frequency, the nucleation density, and the rising diameter; Construct a heat flux distribution sub-model corresponding to the wall quenching heat flux according to the bubble detachment diameter and the wall temperature gradient; Calculate a heat flux distribution sub-model corresponding to the sliding bubble heat flux based on the swept area and time of the sliding bubble; Calculate a heat flux distribution sub-model corresponding to the convective heat flux through the remaining heating area.

[0118] According to a numerical simulation method for two-phase flow boiling heat transfer based on deep learning provided by the present invention, the construction module 610 is specifically configured to: Determine the target mass exchange and the target energy exchange generated by the bubble wall heat flux distribution model; Determine the target mass exchange as the mass exchange source term in the Euler-Euler two-fluid model, and determine the target energy exchange as the energy exchange source term in the Euler-Euler two-fluid model, to obtain the embedded two-fluid framework.

[0119] Figure 7 An example of a schematic physical structure diagram of an electronic device is shown as Figure 7 shown. The electronic device may include: a processor 710, a communication interface 720, a memory 730, and a communication bus 740. Among them, the processor 710, the communication interface 720, and the memory 730 complete mutual communication through the communication bus 740. The processor 710 can call the logical instructions in the memory 730 to execute a numerical simulation method for two-phase flow boiling heat transfer based on deep learning, and the method includes: Input the collected target input parameters into the target deep neural network model to obtain key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, tilt angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; Embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; Perform numerical simulation of two-phase flow boiling heat transfer through the two-fluid framework after the embedding, and obtain the temperature fields and velocity field distributions of the liquid phase and the vapor phase.

[0120] In addition, when the logical instructions in the memory 730 described above can be implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The foregoing storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0121] On the other hand, the present invention also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the numerical simulation method of two-phase flow boiling heat transfer based on deep learning provided by the above-mentioned various methods. The method includes: Input the collected target input parameters into the target deep neural network model to obtain the key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; Embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; Perform numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework, and obtain the temperature fields and velocity field distributions of the liquid phase and the vapor phase.

[0122] On another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the numerical simulation method of two-phase flow boiling heat transfer based on deep learning provided by the above-mentioned various methods. The method includes: Input the collected target input parameters into the target deep neural network model to obtain the key parameters of bubble detachment characteristics, and construct a bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; Embed the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation, and energy conservation equation of the liquid phase and the vapor phase; Perform numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework to obtain the temperature field and velocity field distributions of the liquid phase and the vapor phase.

[0123] The device embodiments described above are merely illustrative, where the units described as separation components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative labor.

[0124] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product, which can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to make a computer device (which can be a personal computer, server, or network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0125] 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 them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present invention.

Claims

1. A numerical simulation method for two-phase flow boiling heat transfer based on deep learning, characterized in that include: Input the collected target input parameters into the target deep neural network model to obtain key parameters of bubble detachment characteristics, and construct a bubble wall heat flow distribution model based on the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle and fluid properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rise diameter and nucleation density; The bubble wall heat flux distribution model is embedded in the Euler-Euler two-fluid framework to obtain the embedded two-fluid framework; the embedded two-fluid framework includes the mass conservation equation, momentum conservation equation and energy conservation equation of the liquid phase and the vapor phase; The embedded two-fluid framework is used to perform numerical simulation of two-phase flow boiling heat transfer, and the temperature field and velocity field distribution of the liquid phase and the vapor phase are obtained.

2. The numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to claim 1, characterized in that The target deep neural network model is constructed by the following steps, including: Collecting bubble dynamics data sets of multiple working conditions and multiple working fluids, and determining input parameters of each sample according to the bubble dynamics data sets; Performing dimensionless processing on the sample input parameters to obtain preprocessed sample input parameters; the sample input parameters include fluid flow, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle and fluid properties corresponding to the sample data; Inputting the preprocessed sample input parameters into the initial deep neural network model to obtain output parameters corresponding to the sample input parameters; Determining a total cost function based on the output parameters corresponding to the sample input parameters; According to the total cost function, the weight coefficient of the initial deep neural network model is optimized by the adaptive moment estimation Adam algorithm until the errors of the training set and the validation set are both lower than the preset threshold, so as to obtain the optimal weight coefficient; the total cost function is the weighted sum of the loss functions of the output parameters; Based on the optimal weight coefficients, the target deep neural network model is determined.

3. The numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to claim 2, characterized in that The collecting of bubble dynamics data sets of multiple working conditions and multiple working fluids, and determining input parameters of each sample according to the bubble dynamics data sets, includes: Determine data selection conditions; the data selection conditions include the selected working fluid type, the coverage pressure range, the coverage flow range, the inclination angle and the wall superheat; the selected working fluid type includes at least water, liquid nitrogen and a new cooling working fluid; the coverage pressure range is 0.1 MPa to 20 MPa, the coverage flow range is 0.1 m / s to 10 m / s, and the inclination angle is 0 to 90 degrees; Determine input data based on the data selection condition; The input data is optimized through feature screening and redundant parameters are eliminated to obtain the input parameters of each sample.

4. The numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to claim 1, characterized in that, The target deep neural network model includes an input layer, a hidden layer, and an output layer, the number of nodes in the input layer is consistent with the dimension of the target input parameter, and the number of nodes in the output layer is consistent with the number of key parameters of the bubble detachment characteristics to be predicted; Inputting the collected target input parameters into a target deep neural network model to obtain key parameters of bubble detachment characteristics, including: Collecting the target input parameters, performing normalization processing on the target input parameters to obtain preprocessed target input parameters; Inputting the preprocessed target input parameters into the input layer to obtain a feature vector matrix; Inputting the feature vector matrix into the hidden layer to obtain a feature vector matrix with weights assigned; Inputting the feature vector matrix with weights assigned into the output layer to obtain an output vector, and mapping the output vector to specific key parameters of bubble detachment characteristics.

5. The numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to claim 1, characterized in that, The bubble wall heat flux distribution model includes heat flux distribution sub-models corresponding to evaporation heat flux, wall quenching heat flux, sliding bubble heat flux, and convective heat flux respectively.

6. The numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to claim 5, characterized in that, Constructing the bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics, including: Based on the bubble detachment frequency, the nucleation density, and the rising diameter, constructing a heat flux distribution sub-model corresponding to the evaporation heat flux; According to the bubble detachment diameter and the wall temperature gradient, constructing a heat flux distribution sub-model corresponding to the wall quenching heat flux; Based on the swept area and time of the sliding bubble, calculating a heat flux distribution sub-model corresponding to the sliding bubble heat flux; Calculating a heat flux distribution sub-model corresponding to the convective heat flux through the remaining heating area.

7. The numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to claim 1, characterized in that Embedding the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain an embedded two-fluid framework, including: Determining the target mass exchange and target energy exchange generated by the bubble wall heat flux distribution model; Determining the target mass exchange as the mass exchange source term in the Euler-Euler two-fluid model, and determining the target energy exchange as the energy exchange source term in the Euler-Euler two-fluid model to obtain the embedded two-fluid framework.

8. A numerical simulation device for two-phase flow boiling heat transfer based on deep learning, characterized in that, Including: A construction module for inputting the collected target input parameters into a target deep neural network model to obtain key parameters of bubble detachment characteristics, and constructing a bubble wall heat flux distribution model according to the key parameters of bubble detachment characteristics; the target input parameters include fluid flow rate, wall superheat, fluid pressure, contact angle, wall roughness, inclination angle, and fluid physical properties; the key parameters of bubble detachment characteristics include bubble detachment diameter, detachment frequency, slip velocity, slip distance, rising diameter, and nucleation density; Embedding the bubble wall heat flux distribution model into the Euler-Euler two-fluid framework to obtain an embedded two-fluid framework; the embedded two-fluid framework includes mass conservation equations, momentum conservation equations, and energy conservation equations for the liquid phase and the vapor phase; A simulation module for performing numerical simulation of two-phase flow boiling heat transfer through the embedded two-fluid framework to obtain the temperature field and velocity field distributions of the liquid phase and the vapor phase.

9. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to any one of claims 1 to 7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the numerical simulation method for two-phase flow boiling heat transfer based on deep learning according to any one of claims 1 to 7.

Images