Multi-physics field intelligent calculation method for rectangular flow channel

Abstract

The invention belongs to the technical field of nuclear reactor engineering, and particularly relates to a rectangular flow channel multi-physics field intelligent calculation method. Comprising the following steps: step 1, determining input-output matched data pairs related to two-phase flow of a rectangular channel and a heat transfer process; step 2, acquiring input-output matched data pairs related to rectangular channel two-phase flow and a heat transfer process required by a deep learning algorithm, and establishing a data pool; 3, a loss function is constructed, supervised regression learning is carried out on output matching data after machine learning training, and a rectangular channel two-phase parameter intelligent calculation model is obtained; and 4, calculating input parameters related to the two-phase flow and heat transfer process of the rectangular channel to obtain corresponding output parameters so as to realize rapid calculation of the parameters in the two-phase flow and heat transfer process of the rectangular channel. The method has the beneficial effects that input-output characteristics can be well approached, and the calculation precision of an intelligent calculation model can be effectively ensured.

Description

Technical Field

[0001] The invention belongs to the technical field of nuclear reactor engineering, and in particular relates to a multi-physical field intelligent calculation method for a rectangular flow channel. Background Art

[0002] Gas-liquid two-phase flow phenomenon is widely present in various industrial equipment of nuclear power system. Research on gas-liquid two-phase flow phenomenon is helpful to improve the safety and operational reliability of evaluation system equipment. As the most complex two-phase flow phenomenon, the gas-liquid two-phase flow process is accompanied by heat and mass transfer and phase change between the two phases, which makes the research on gas-liquid two-phase flow face huge challenges. At present, a lot of research work has been carried out on the multi-physical field characteristics such as rectangular channel flow field and temperature field. Most of them are based on experimental methods to study the development and change of flow form, phase distribution characteristics, pressure drop change, temperature change, etc. in the flow channel. However, experimental research is time-consuming and labor-intensive, costly, and has high requirements for test bench use and measurement personnel. On the other hand, in terms of numerical simulation, the two-phase flow and heat transfer process of rectangular channels is refined modeled and analyzed mainly based on experimental data through three-dimensional high-precision CFD fluid software or core multi-channel thermal hydraulic software. When the channel structure is expanded to the full core level, the consumption of computing resources is large, the computing speed is slow, and the computing efficiency is not high. Even on ultra-large-scale high-performance servers, the multi-physical field calculation of the core rectangular channel also requires a lot of time. Therefore, it is necessary to study a multi-physics intelligent prediction method for rectangular flow channels that consumes less computing resources and can achieve fast calculations. Summary of the invention

[0003] The purpose of the present invention is to provide a multi-physical field intelligent calculation method for a rectangular flow channel, which occupies very little computing resources and can realize rapid prediction and analysis of parameters in the two-phase flow and heat transfer process of a core rectangular channel.

[0004] The technical solution of the present invention is as follows: A multi-physical field intelligent calculation method for a rectangular flow channel comprises the following steps:

[0005] Step 1: Analyze the key influencing factors of multi-physics field characteristics in the rectangular channel two-phase flow and heat transfer process, and determine the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process;

[0006] Step 2: Obtain input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process required by the deep learning algorithm through high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software calculations, or from rectangular channel two-phase flow and heat transfer test data, and establish a data pool;

[0007] Step 3: Based on the deep learning algorithm and the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process, a loss function is constructed, and supervised regression learning of the output matching data is realized after machine learning training to obtain the rectangular channel two-phase parameter intelligent calculation model;

[0008] Step 4: Through the rectangular channel two-phase parameter intelligent calculation model, calculate the input parameters related to the rectangular channel two-phase flow and heat transfer process to obtain the corresponding output parameters, and realize the rapid calculation of parameters in the rectangular channel two-phase flow and heat transfer process.

[0009] The step 1 specifically includes the following:

[0010] Step 10: By analyzing the key influencing factors of multi-physics field characteristics in the two-phase flow and heat transfer process in the rectangular channel, the input parameters related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm are selected;

[0011] Step 11: By analyzing the multi-physics field characteristics of the two-phase flow and heat transfer process in the rectangular channel, the output parameters related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm are selected;

[0012] Step 12: Based on the multi-physics field computing requirements, determine the input-output matching data pairs related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm.

[0013] The deep learning algorithm described in step 10 includes a convolutional neural network CNN, a recurrent neural network RNN, and a long short-term memory neural network LSTM.

[0014] The input parameters related to the two-phase flow and heat transfer process in the rectangular channel described in step 10 include the gas phase reduced velocity j g , liquid phase conversion velocity j f , channel cross-sectional dimensions including radial height H and width G, channel inlet and outlet fluid pressure P, channel inlet and outlet fluid temperature T, channel wall roughness ε, channel inlet fluid flow rate W, channel heating power Q or heat flux density q", fluid physical properties parameters, and one or more combinations of motion freedom parameters.

[0015] The output parameters related to the two-phase flow and heat transfer process in the rectangular channel described in step 11 include the position z of the measuring points in different length directions. i , Width direction measuring point position x i and the measuring point position L in the axial direction i One or more combinations of two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soter average diameter, two-phase temperature, pressure, flow rate, and velocity.

[0016] The data pairs described in step 12 include input parameters related to the two-phase flow and heat transfer process in the rectangular channel, including the gas phase reduced velocity j g , liquid phase conversion velocity j f , channel cross-sectional dimensions including radial height H and width G, channel inlet and outlet fluid pressure P, channel inlet and outlet fluid temperature T, channel wall roughness ε, channel inlet fluid flow rate W, channel heating power Q or heat flux density q", fluid physical properties, one or more combinations of motion freedom parameters, and corresponding output parameters including two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soth average diameter parameter, one or more combinations of two-phase temperature, pressure, flow rate, and velocity parameters.

[0017] The step 2 comprises the following steps:

[0018] Step 20: Divide the rectangular channel into multiple spatial grids in radial and axial directions;

[0019] Step 21: Calculate the initial multi-physical field state parameters of the two-phase flow and heat transfer process of each spatial grid by high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software, or measure the corresponding parameters by experimental means to obtain the two-phase parameters that characterize the multi-physical field characteristics of the two-phase flow and heat transfer process, including interface parameters such as two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soth average diameter, and phase parameters such as two-phase temperature, pressure, flow rate, and velocity;

[0020] Step 22: Select the phase parameters of the rectangular channel two-phase flow and heat transfer process in the spatial grid where the test measurement points are located, establish the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process required by the deep learning algorithm, and form the data pool required for the multi-physical field.

[0021] One expression of the loss function in step 3 is expressed in the form of mean square error, as shown below:

[0022]

[0023] The step 3 includes building a deep learning neural network structure based on a deep learning algorithm and input-output matching data pairs related to the two-phase flow and heat transfer process in a rectangular channel, constructing loss functions corresponding to different physical fields, and realizing supervised regression learning of the output matching data after machine learning training. Based on the K-fold cross-validation method, when the calculation result error is less than the set value, the learning stops, otherwise it continues to relearn, thereby obtaining an intelligent calculation model of the two-phase parameters of the rectangular channel.

[0024] The deep learning neural network described in step 3 includes a combination of one or more structures of a fully connected neural network, a long short-term memory neural network or a recurrent neural network, and the combination is selected according to the multi-physical field computing requirements.

[0025] In step 3, the input-output matching data related to the two-phase flow and heat transfer process in the rectangular channel is processed by dimensionless parameter processing to eliminate the influence of different magnitudes of phase parameters.

[0026] The neuron connection weights and biases of the neural network described in step 3 are used as adjustable optimization parameters for deep machine learning to adjust the loss function of partial spatial sample data one by one until the loss function of all spatial sample data is adjusted.

[0027] The beneficial effects of the present invention are:

[0028] 1) The calculation accuracy of the two-phase parameter intelligent calculation model of the rectangular flow channel based on deep learning established by the present invention can well approximate the input-output characteristics, and can effectively ensure the calculation accuracy of the intelligent calculation model.

[0029] 2) The intelligent computing model obtained by the present invention has good generalization ability for measured data. When the deep learning algorithm is consistent, the intelligent computing model can be quickly corrected and optimized by re-learning a large amount of measured data.

[0030] 3) The intelligent computing model obtained by the present invention consumes relatively little computing resources and can replace the two-phase flow test measurement or simulation method to a certain extent, thereby reducing the test or simulation cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 A flow chart of a multi-physical field intelligent calculation method for a rectangular flow channel provided by the present invention;

[0032] Figure 2 It is the top view of the rectangular flow channel;

[0033] Figure 3 Schematic diagram of the neural network structure. DETAILED DESCRIPTION

[0034] The present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0035] like Figure 1 As shown, a multi-physics field intelligent calculation method for a rectangular flow channel includes the following steps:

[0036] Step 1: Analyze the key influencing factors of multi-physics field characteristics in the rectangular channel two-phase flow and heat transfer process, and determine the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process;

[0037] Step 1 specifically includes the following steps:

[0038] Step 10: By analyzing the key influencing factors of multi-physical field characteristics in the two-phase flow and heat transfer process of the rectangular channel, the input parameters related to the two-phase flow and heat transfer process of the rectangular channel required by the deep learning algorithm (such as convolutional neural network CNN, recurrent neural network RNN, long short-term memory neural network LSTM, etc.) are selected;

[0039] The input parameters related to the two-phase flow and heat transfer process in the rectangular channel include the gas phase conversion velocity j g , liquid phase conversion velocity j f , channel cross-sectional dimensions (radial height H and width G), channel inlet and outlet fluid pressure P, channel inlet and outlet fluid temperature T, channel wall roughness ε, channel inlet fluid flow rate W, channel heating power Q or heat flux density q", fluid physical properties (such as fluid density ρ, etc.), motion freedom parameters (such as swing angle α, position, etc.) and other parameters or a combination thereof.

[0040] Step 11: By analyzing the multi-physical field characteristics in the two-phase flow and heat transfer process of the rectangular channel (for the different physical fields such as the flow field, temperature field, and pressure field of the rectangular channel, the physical factors related to the different physical fields are determined, so that the parameters of the corresponding physical factors can be further analyzed), the output parameters related to the two-phase flow and heat transfer process of the rectangular channel required by the deep learning algorithm are selected;

[0041] The output parameters related to the two-phase flow and heat transfer process in the rectangular channel include the position of the measuring point z in different length directions. i , Width direction measuring point position x i and the measuring point position L in the axial direction i Interface parameters such as two-phase interface concentration (such as taking the channel entrance position as the starting point), cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soter average diameter, and one or more combinations of two-phase temperature, pressure, flow rate, velocity and other phase parameters.

[0042] Step 12: According to the multi-physics field calculation requirements, determine the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process required by the deep learning algorithm, wherein the data pairs include input parameters related to the rectangular channel two-phase flow and heat transfer process (including the gas phase conversion velocity j g , liquid phase conversion velocity j f, channel cross-sectional dimensions (radial height H and width G), channel inlet and outlet fluid pressure P, channel inlet and outlet fluid temperature T, channel wall roughness ε, channel inlet fluid flow W, channel heating power Q or heat flux density q", fluid physical properties (such as fluid density ρ, etc.), motion freedom parameters (such as swing angle α, position, etc.), etc. What parameters are included in one or more combinations of parameters, and the corresponding output parameters (at different length direction measuring point positions z i , Width direction measuring point position x i and the measuring point position L in the axial direction i What parameters are included in the interface parameters such as the two-phase interface concentration (taking the channel entrance position as the starting point), cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soter average diameter, and one or more combinations of phase parameters such as two-phase temperature, pressure, flow rate, velocity, etc.

[0043] Step 2: Calculate through high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software, or obtain the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process required by the deep learning algorithm (different combinations of the above input parameters and output parameters are determined according to different physical fields), and establish a data pool;

[0044] Step 2 specifically includes the following steps:

[0045] Step 20: Divide the rectangular channel into multiple spatial grids in radial and axial directions;

[0046] Step 21: Calculate the initial multi-physical field state parameters of the two-phase flow and heat transfer process of each spatial grid by high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software (such as CTF, etc.), or measure the corresponding parameters by experimental means (such as flow meter to obtain flow; differential pressure gauge to obtain pressure; probe to obtain interface concentration and velocity; cavitation meter to obtain cavitation fraction; thermometer to obtain temperature, etc.), and obtain the two-phase parameters that characterize the multi-physical field characteristics of the two-phase flow and heat transfer process, including interface parameters such as two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soth average diameter, and phase parameters such as two-phase temperature, pressure, flow, velocity, etc.;

[0047] Step 22: Select the phase parameters of the two-phase flow and heat transfer process in the rectangular channel of the spatial grid where the test measuring points are located (interface parameters such as two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soth average diameter, and two-phase temperature, pressure, flow rate, velocity and other phase parameters), and establish the input-output matching data pairs related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm (different combinations of the above input parameters and output parameters are determined according to different physical fields, that is, the input combination corresponds to the output combination), to form the data pool required for multiple physical fields.

[0048] Step 3: Based on the deep learning algorithm and the input-output matching data pairs related to the two-phase flow and heat transfer process in the rectangular channel (different combinations of the above input parameters and output parameters are determined according to different physical fields, that is, the input combination corresponds to the output combination), a deep learning neural network structure (such as Figure 3 The corresponding loss function is constructed, and supervised regression learning of the output matching data is realized after machine learning training to obtain the two-phase parameter intelligent calculation model of the rectangular channel;

[0049] The loss function can be expressed in many forms, one of which can be expressed in the form of mean square error, as shown below:

[0050]

[0051] In the formula, y i is the predicted value; is the true value; N is the number of data.

[0052] Step 3 specifically includes the following steps: based on the deep learning algorithm and the input-output matching data pairs related to the two-phase flow and heat transfer process in the rectangular channel, a deep learning neural network structure is built, and the loss function corresponding to different physical fields is constructed. After machine learning training, supervised regression learning of the output matching data is realized. Based on the K-fold cross-validation method, when the calculation result error is less than the set value, the learning stops, otherwise, the learning continues to be re-learned, thereby obtaining an intelligent calculation model of the two-phase parameters of the rectangular channel.

[0053] In one possible implementation, in step 3, the deep learning neural network includes a combination of one or more structures of neural networks such as a fully connected neural network, a long short-term memory neural network or a recurrent neural network, and the combination is selected according to the multi-physical field computing requirements.

[0054] In a possible implementation, in step 3, the input-output matching data related to the two-phase flow and heat transfer process in the rectangular channel is processed by dimensionless parameters to eliminate the influence of different magnitudes of phase parameters.

[0055] In one possible implementation, in step 3, for multi-physical field calculation and analysis, parameters of different categories or changing trends can be classified, that is, changing trends of the same order of magnitude are placed in the same neural network for training and learning, and the classification is determined manually based on experience.

[0056] The process includes:

[0057] 1. Confirm the trend or type of parameter changes.

[0058] 2. Put parameters with the same magnitude of change trend or the same type of parameters in the same neural network. Construct different loss function terms, and in the process of deep learning with loss function, continuously and adaptively adjust the weight value of each part of the loss function until the loss function approaches the minimum value.

[0059] The loss function can be expressed in many forms, one of which can be expressed in the form of mean square error, as shown below:

[0060]

[0061] In one possible implementation, in step 3, different loss function terms may be trained in parallel on different GPUs.

[0062] In one possible implementation, in step 3, the neuron connection weights and biases of the neural network are used as adjustable optimization parameters for deep machine learning to successively adjust the loss functions of partial spatial sample data until the loss functions of all spatial sample data are adjusted.

[0063] Step 4: Through the rectangular channel two-phase parameter intelligent calculation model, calculate the input parameters related to the rectangular channel two-phase flow and heat transfer process to obtain the corresponding output parameters, and realize the rapid calculation of parameters in the rectangular channel two-phase flow and heat transfer process.

[0064] Example:

[0065] This embodiment is directed to a smooth rectangular channel under static conditions of a given geometric structure, such as Figure 2 As shown, the conditions are normal temperature and pressure, the inlet flow rate is known, the radial measuring point position is fixed, and the axial direction is divided into 3 measuring point positions. An intelligent calculation model of the two-phase parameters of the rectangular flow channel under certain specific working conditions is established.

[0066] A multi-physics field intelligent calculation method for a rectangular flow channel comprises the following steps:

[0067] Step 1: By analyzing the key influencing factors of the multi-physical field characteristics in the two-phase flow and heat transfer process of the rectangular channel, the input parameters related to the two-phase flow and heat transfer process of the rectangular channel required by the deep learning algorithm are selected; by analyzing the multi-physical field characteristics in the two-phase flow and heat transfer process of the rectangular channel, the output parameters related to the two-phase flow and heat transfer process of the rectangular channel required by the deep learning algorithm are selected; on this basis, according to the multi-physical field calculation requirements, the input-output matching data pairs related to the two-phase flow and heat transfer process of the rectangular channel required by the deep learning algorithm are determined.

[0068] In this embodiment, the key factors affecting the flow characteristics in the rectangular channel two-phase flow and heat transfer process include but are not limited to the gas phase conversion velocity j g , liquid phase conversion velocity j f , channel inlet and outlet fluid pressure P, channel inlet and outlet fluid temperature T, channel inlet fluid flow rate W, channel heating power Q or heat flux density q'', channel wall roughness ε, fluid physical properties (such as fluid density ρ, etc.) and other initial state parameters of the flow process and channel geometry information.

[0069] Select the calculation input parameters related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm, including but not limited to the gas phase conversion velocity j g , liquid phase conversion velocity j f One or more combinations of parameters.

[0070] Select the calculation output parameters related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm, including but not limited to the measurement point positions z in different length directions i , Width direction measuring point position x i and the measuring point position L in the axial direction i Interface parameters such as two-phase interface concentration (such as taking the channel entrance position as the starting point), cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soter average diameter, and one or more combinations of two-phase temperature, pressure, flow rate, velocity and other phase parameters.

[0071] Step 2: Obtain the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process required by the deep learning algorithm through high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software calculations, or from the rectangular channel two-phase flow and heat transfer test data, and establish a data pool, which specifically includes the following steps:

[0072] The rectangular channel is divided into multiple spatial grids in radial and axial directions; the initial multi-physical field state parameters of the two-phase flow and heat transfer process of each spatial grid are calculated by high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software, or the corresponding parameters are measured by experimental means (such as flow meter to obtain flow; differential pressure gauge to obtain pressure; probe to obtain interface concentration and velocity; cavitation meter to obtain cavitation fraction; thermometer to obtain temperature, etc.), to obtain the two-phase parameters that characterize the multi-physical field characteristics of the two-phase flow and heat transfer process; the phase parameters of the two-phase flow and heat transfer process of the rectangular channel in the spatial grid where the test measuring point is located are selected, and the input-output matching data pairs related to the two-phase flow and heat transfer process of the rectangular channel required by the deep learning algorithm are established to form the data pool required for the multi-physical field.

[0073] In this embodiment, different gas phase conversion velocities j are selected. g and liquid phase conversion velocity j f , and the axial measuring point lengths L1, L2, and L3 of the channel, form data groups under different gas-liquid phase conversion velocity conditions, and are calculated through high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software, or through two-phase flow and heat transfer test measurements, to obtain the parameters in the two-phase flow and heat transfer process of the rectangular channel, including two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soth average diameter and other interface parameters, two-phase temperature, pressure, flow rate, velocity and other phase parameters, so as to obtain the input-output matching data pairs related to the two-phase and heat transfer flow process of the rectangular channel required by the deep learning algorithm.

[0074] Step 3: Based on the deep learning algorithm and the input-output matching data pairs related to the two-phase flow and heat transfer process in the rectangular channel, a deep learning neural network structure is built, and the corresponding loss function is constructed. After machine learning training, supervised regression learning of the output matching data is realized to obtain the intelligent calculation model of the two-phase parameters of the rectangular channel.

[0075] The establishment of data learning and intelligent computing models can be determined according to the actual multi-physics field computing requirements and application methods. For example, the gas phase conversion velocity j can be selected g , liquid phase conversion velocity j f As the calculation input parameter; for pressure field or velocity field calculation, select the measuring point position z in different length directions i , Width direction measuring point position x i and the measuring point position L in the axial direction i The combination of pressure and velocity (such as taking the channel entrance position as the starting point) is used as the calculation output parameter; a sufficient number of characteristic data are extracted from the data pool as the input-output matching data pairs related to the two-phase flow and heat transfer process in the rectangular channel required for learning the two-phase parameter intelligent calculation model of the rectangular flow channel.

[0076] On this basis, a fully connected neural network can be used, and hyperparameters such as the number of neural network layers, the number of neurons, the network weights and biases can be set. The activation function of the hidden layer nodes is set to sigmoid, etc., and a deep neural network structure is designed and established. The loss functions of pressure and speed are constructed for supervised regression learning. When the loss function is reduced to a stable minimum value, the neural network intelligent computing model can be obtained. Based on the K-fold cross-validation method, when the calculation result error is less than the set value, the learning stops, otherwise it continues to relearn until the intelligent computing model can be solidified.

[0077] Step 4: Combined with new physical field calculation requirements, such as temperature field, the rectangular channel two-phase parameter intelligent calculation model is used to calculate the input parameters related to the rectangular channel two-phase flow and heat transfer process to obtain the corresponding output parameters, thereby realizing the rapid calculation of parameters in the rectangular channel two-phase flow and heat transfer process.

[0078] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A multi-physics intelligent calculation method for rectangular flow channel, characterized in that: The steps include: Step 1: Analyze the key influencing factors of multi-physics field characteristics in the rectangular channel two-phase flow and heat transfer process, and determine the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process; Step 2: Obtain input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process required by the deep learning algorithm through high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software calculations, or from rectangular channel two-phase flow and heat transfer test data, and establish a data pool; Step 3: Based on the deep learning algorithm and the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process, a loss function is constructed, and supervised regression learning of the output matching data is realized after machine learning training to obtain the rectangular channel two-phase parameter intelligent calculation model; Step 4: Through the rectangular channel two-phase parameter intelligent calculation model, calculate the input parameters related to the rectangular channel two-phase flow and heat transfer process to obtain the corresponding output parameters, and realize the rapid calculation of parameters in the rectangular channel two-phase flow and heat transfer process.

2. A rectangular flow channel multi-physics field intelligent calculation method as claimed in claim 1, characterized in that: The step 1 specifically includes the following: Step 10: By analyzing the key influencing factors of multi-physics field characteristics in the two-phase flow and heat transfer process in the rectangular channel, the input parameters related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm are selected; Step 11: By analyzing the multi-physics field characteristics of the two-phase flow and heat transfer process in the rectangular channel, the output parameters related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm are selected; Step 12: Based on the multi-physics field computing requirements, determine the input-output matching data pairs related to the two-phase flow and heat transfer process in the rectangular channel required by the deep learning algorithm.

3. A rectangular flow channel multi-physical field intelligent calculation method as claimed in claim 2, characterized in that: The deep learning algorithm described in step 10 includes a convolutional neural network CNN, a recurrent neural network RNN, and a long short-term memory neural network LSTM.

4. A rectangular flow channel multi-physics field intelligent calculation method as claimed in claim 2, characterized in that: The input parameters related to the two-phase flow and heat transfer process in the rectangular channel described in step 10 include the gas phase reduced velocity j g , liquid phase conversion velocity j f , channel cross-sectional dimensions including radial height H and width G, channel inlet and outlet fluid pressure P, channel inlet and outlet fluid temperature T, channel wall roughness ε, channel inlet fluid flow rate W, channel heating power Q or heat flux density q", fluid physical properties parameters, and one or more combinations of motion freedom parameters.

5. A rectangular flow channel multi-physics field intelligent calculation method as claimed in claim 2, characterized in that: The output parameters related to the two-phase flow and heat transfer process in the rectangular channel described in step 11 include the position z of the measuring points in different length directions. i , Width direction measuring point position x i and the measuring point position L in the axial direction i One or more combinations of two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soter average diameter, two-phase temperature, pressure, flow rate, and velocity.

6. A rectangular flow channel multi-physics field intelligent calculation method as claimed in claim 2, characterized in that: The data pairs described in step 12 include input parameters related to the two-phase flow and heat transfer process in the rectangular channel, including the gas phase reduced velocity j g , liquid phase conversion velocity j f , channel cross-sectional dimensions including radial height H and width G, channel inlet and outlet fluid pressure P, channel inlet and outlet fluid temperature T, channel wall roughness ε, channel inlet fluid flow rate W, channel heating power Q or heat flux density q″, fluid physical properties, one or more combinations of motion freedom parameters, and corresponding output parameters including two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soth average diameter parameter, one or more combinations of two-phase temperature, pressure, flow rate, and velocity parameters.

7. A rectangular flow channel multi-physics field intelligent calculation method as claimed in claim 1, characterized in that: The step 2 comprises the following steps: Step 20: Divide the rectangular channel into multiple spatial grids in radial and axial directions; Step 21: Calculate the initial multi-physical field state parameters of the two-phase flow and heat transfer process of each spatial grid by high-precision three-dimensional CFD fluid software or core multi-channel thermal hydraulic software, or measure the corresponding parameters by experimental means to obtain the two-phase parameters that characterize the multi-physical field characteristics of the two-phase flow and heat transfer process, including interface parameters such as two-phase interface concentration, cavitation fraction, gas phase velocity, gas phase frequency, gas phase chord length distribution, cross-sectional gas phase distribution, bubble Soth average diameter, and phase parameters such as two-phase temperature, pressure, flow rate, and velocity; Step 22: Select the phase parameters of the rectangular channel two-phase flow and heat transfer process in the spatial grid where the test measurement points are located, establish the input-output matching data pairs related to the rectangular channel two-phase flow and heat transfer process required by the deep learning algorithm, and form the data pool required for the multi-physical field.

8. The multi-physics intelligent calculation method for rectangular flow channel according to claim 1, characterized in that: One expression of the loss function in step 3 is expressed in the form of mean square error, as shown below:

9. A rectangular flow channel multi-physics field intelligent calculation method as claimed in claim 1, characterized in that: The step 3 includes building a deep learning neural network structure based on a deep learning algorithm and input-output matching data pairs related to the two-phase flow and heat transfer process in a rectangular channel, constructing loss functions corresponding to different physical fields, and realizing supervised regression learning of the output matching data after machine learning training. Based on the K-fold cross-validation method, when the calculation result error is less than the set value, the learning stops, otherwise it continues to relearn, thereby obtaining an intelligent calculation model of the two-phase parameters of the rectangular channel.

10. A rectangular flow channel multi-physical field intelligent calculation method according to claim 9, characterized in that: The deep learning neural network described in step 3 includes a combination of one or more structures of a fully connected neural network, a long short-term memory neural network or a recurrent neural network, and the combination is selected according to the multi-physical field computing requirements.

11. The multi-physical field intelligent calculation method for a rectangular flow channel according to claim 9, characterized in that: In step 3, the input-output matching data related to the two-phase flow and heat transfer process in the rectangular channel is processed by dimensionless parameter processing to eliminate the influence of different magnitudes of phase parameters.

12. A rectangular flow channel multi-physics field intelligent calculation method as claimed in claim 9, characterized in that: The neuron connection weights and biases of the neural network described in step 3 are used as adjustable optimization parameters for deep machine learning to adjust the loss function of partial spatial sample data one by one until the loss function of all spatial sample data is adjusted.

Images