Gas-liquid two-phase flow field real-time synchronous measurement method and system

Abstract

The invention discloses a gas-liquid two-phase flow field real-time synchronous measurement method and system, and relates to the technical field of flow field measurement. The method comprises the following steps: acquiring a high-speed image acquired by a high-speed camera; setting parameters are adopted by the high-speed camera; performing post-processing on the high-speed image to obtain an optimized image; the post-processing comprises image denoising, gray scale segmentation, contour segmentation and roundness denoising; a PWC-Net network is constructed; and performing feature extraction on the optimized image by using the PWC-Net network, and performing gas-liquid two-phase flow field prediction according to the extracted features to obtain a final fluid velocity field. According to the invention, rapid, accurate and stable gas-liquid two-phase flow field velocity measurement can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of flow field measurement, and in particular to a real-time synchronous measurement method and system for a gas-liquid two-phase flow field. Background Art

[0002] Particle Image Velocimetry (PIV) involves adding tracer particles to a flow field, illuminating the flow field with a laser sheet, capturing a sequence of particle images with a high-speed camera, and using image analysis algorithms to calculate particle displacements and thereby determine the velocity distribution. This technique offers the advantages of non-contact, full-field measurement and is widely used in various fluid dynamics experiments and engineering fields.

[0003] Traditional PIV algorithms are primarily categorized into two types: cross-correlation and optical flow. Each method has its own unique principles and applications. The cross-correlation algorithm calculates the displacement of tracer particles between two image frames to determine the fluid's velocity field. The optical flow algorithm in PIV is a velocity field measurement method based on estimating pixel motion within an image sequence. Unlike the cross-correlation algorithm, the optical flow algorithm directly estimates fluid motion by analyzing the grayscale changes of each pixel in the image. However, traditional PIV technology suffers from low computational efficiency, poor real-time performance, and noise sensitivity when processing complex flow fields. Summary of the Invention

[0004] The object of the present invention is to provide a method and system for real-time synchronous measurement of a gas-liquid two-phase flow field, aiming to solve or improve at least one of the above-mentioned technical problems.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for real-time synchronous measurement of a gas-liquid two-phase flow field, comprising:

[0007] Acquiring a high-speed image captured by a high-speed camera; the high-speed camera uses set parameters;

[0008] Post-processing the high-speed image to obtain an optimized image; the post-processing includes image denoising, grayscale segmentation, contour segmentation, and roundness denoising;

[0009] Build the PWC-Net network;

[0010] The PWC-Net network is used to extract features from the optimized image, and the gas-liquid two-phase flow field is predicted based on the extracted features to obtain the final fluid velocity field.

[0011] Optionally, the setting parameters include at least resolution, exposure time, acquisition frame rate and number of saved frames.

[0012] Optionally, the PWC-Net network is constructed as follows:

[0013] Obtaining a training data set; the training data set includes training data and corresponding data labels;

[0014] Constructing a pre-trained network based on a CNN network; the CNN network includes a convolutional layer, an activation function, a pooling layer, and a fully connected layer;

[0015] The training data set is input into the pre-trained network, and training is performed with the goal of minimizing the loss between the network output and the data label. When the network meets the preset accuracy requirements, the trained network is determined as the PWC-Net network.

[0016] Optionally, the utilizing the PWC-Net network to extract features from the optimized image and predicting the gas-liquid two-phase flow field based on the extracted features to obtain a final fluid velocity field specifically includes:

[0017] Input the optimized image into the PWC-Net network, perform multi-scale feature extraction on two adjacent frames of images, and obtain a feature map containing semantic and detail information at different levels;

[0018] Performing image deformation based on the feature map, aligning two frames of image to obtain a deformation feature map;

[0019] Obtaining a similarity metric based on the similarity between each position on the deformation feature map and different optical flow offsets, and constructing a cost volume based on the similarity metric;

[0020] The gas-liquid two-phase flow field is predicted based on the cost volume to obtain the final fluid velocity field.

[0021] Optionally, performing image deformation based on the feature map to align two frames of image to obtain a deformation feature map specifically includes:

[0022] The deformation operation is represented as W(F1,w), and the feature map F1 of the previous frame is deformed by bilinear interpolation so that the feature map F1 of the previous frame is aligned with the feature map F2 of the next frame. In the two-dimensional space, for the coordinate (x, y) on the feature map F1, the coordinate after deformation is (x+w x ,y+w y ), where w x and w y are the components of the optical flow w in the x and y directions, respectively, and the mathematical expression is:

[0023] W(F1,w)(x,y)=∑ i,j F1(x+i,y+i)·max(0,1-|iw x |)·max(0,1-|jw y |)

[0024] Where i and j represent the interpolation offset.

[0025] Optionally, obtaining a similarity metric based on the similarity between each position on the deformation feature map and different optical flow offsets, and constructing a cost volume based on the similarity metric, specifically includes:

[0026] For each position (x, y) on the deformation feature map and different optical flow offsets (δx, δy), calculate the similarity metric:

[0027] C(x,y,δx,δy)=W(F1,w)(x,y)·F2(x+δx,y+δy)

[0028] Among them, W(F1,w) represents the deformation operation, w represents the optical flow, F1 represents the feature map of the previous frame, and F2 represents the feature map of the next frame;

[0029] The similarity measures of all positions and optical flow offsets are combined to form a cost volume C; the cost volume C is a four-dimensional tensor used to record the similarity under different positions and optical flow assumptions.

[0030] Optionally, the gas-liquid two-phase flow field prediction based on the cost volume to obtain a final fluid velocity field specifically includes:

[0031] Based on the cost volume, a coarse-to-fine strategy is adopted to upsample the prediction results in order from low to high resolution levels to obtain the final fluid velocity field.

[0032] The present invention also provides a gas-liquid two-phase flow field real-time synchronous measurement system, which applies the above method and includes: a high-power laser light source, a high-speed camera, a synchronous controller, an optical correction box and a host computer;

[0033] A plurality of high-speed cameras are arranged around the optical correction box, and the light emitting end of the high-power laser light source is arranged at the top of the optical correction box. Each of the high-speed cameras and the high-power laser light source is connected to the synchronization controller, and the synchronization controller is also connected to the host computer.

[0034] According to the specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0035] The present invention discloses a method and system for real-time synchronous measurement of a gas-liquid two-phase flow field. The method comprises acquiring a high-speed image captured by a high-speed camera; using set parameters for the high-speed camera; post-processing the high-speed image to obtain an optimized image; the post-processing includes image denoising, grayscale segmentation, contour segmentation, and roundness denoising; constructing a PWC-Net network; utilizing the PWC-Net network to extract features from the optimized image, and predicting the gas-liquid two-phase flow field based on the extracted features to obtain a final fluid velocity field. The present invention enables rapid, accurate, and stable measurement of the velocity of a gas-liquid two-phase flow field. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 is a system structure diagram in this embodiment;

[0038] Figure 2 This is the PWC-Net structure diagram in this embodiment;

[0039] Figure 3 1 is a structural diagram of the convolution module and a structural diagram of the deconvolution module in this embodiment; wherein, part (a) is a structural diagram of the convolution module; part (b) is a structural diagram of the deconvolution module;

[0040] Figure 4 This is a diagram of the high-speed camera configuration procedure in this embodiment;

[0041] Figure 5 This is the particle image velocimetry program diagram in this embodiment. DETAILED DESCRIPTION

[0042] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0043] The object of the present invention is to provide a method and system for real-time synchronous measurement of a gas-liquid two-phase flow field, aiming to solve or improve at least one of the above-mentioned technical problems.

[0044] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0045] The present invention provides a method for real-time synchronous measurement of a gas-liquid two-phase flow field, comprising:

[0046] Step 100: Acquire a high-speed image captured by a high-speed camera; the high-speed camera adopts set parameters.

[0047] Step 200: Post-processing the high-speed image to obtain an optimized image; the post-processing includes image denoising, grayscale segmentation, contour segmentation, and roundness denoising.

[0048] Step 300: Construct a PWC-Net network.

[0049] Step 400: Using the PWC-Net network, extract features from the optimized image, and predict the gas-liquid two-phase flow field based on the extracted features to obtain a final fluid velocity field.

[0050] As a specific implementation method, the specific processing process shown below is provided.

[0051] First, a solution is constructed to address the existing problems. The overall concept is:

[0052] Algorithmically, a deep learning optical flow network is introduced into the field of particle image velocimetry. By constructing a deep learning framework and an optical flow estimation network architecture, features can be extracted directly from particle image pairs and the velocity field can be predicted, thereby improving the accuracy and robustness of the velocity field measurement results. The network structure fully considers the characteristics of the PIV task, such as combining the sparsity of particle images, noise interference, and continuity constraints of the flow field to construct a learning dataset training model, thereby ensuring accuracy while being closer to the field of fluid mechanics. The system introduces optical flow models with different structures, such as traditional optical flow networks that directly use convolutional neural networks and convolutional optical flow networks combined with multi-scale feature fusion mechanisms and attention mechanisms, to provide calculation schemes with different levels of accuracy and computational loss, and to provide different solutions for different task requirements.

[0053] Secondly, the specific technical solution includes the design of hardware and software parts.

[0054] Hardware:

[0055] In the traditional particle image velocimetry system, the hardware foundation is mainly composed of high-power laser light source, high-speed camera, synchronous controller, optical correction box and host computer. Figure 1As shown in the figure, its core design is to achieve high-precision flow field measurement by integrating a transparent pipe and a correction box into the observation engineering system. Specifically, a laser light source is set up on the top of the pipe to generate a uniform and stable laser sheet light source to illuminate the tracer particles in the flow field. Two cameras are placed on both sides of the transparent pipe to synchronously capture the particle motion data in the pipe. These cameras are connected to the synchronization controller via a data cable and transmit the collected image data to the host computer for real-time processing and analysis. Through this design, the system can efficiently obtain flow field information and combine deep learning algorithms to achieve high-precision, real-time particle image velocimetry, providing reliable technical support for research in complex flow field environments.

[0056] The laser light source is one of the core components of the PIV system, used to illuminate the tracer particles in the flow field. The high-speed camera is used to capture the moving images of the tracer particles in the flow field, and its performance directly affects the temporal resolution and spatial resolution of the PIV system. The synchronization controller is used to coordinate the working timing of the laser light source and the camera to ensure that the laser pulse is precisely synchronized with the camera exposure time. The optical correction box improves the accuracy of the flow field changes in the test tube by reducing the optical distortion caused by factors such as the curvature of the pipe and the difference in refractive index between the fluid in the pipe and the pipe medium. The computer is the control center of the PIV system, used to control the operation of the hardware equipment, collect image data and perform subsequent processing. The data processing system is usually equipped with high-performance computing hardware and professional PIV software to realize image analysis, velocity field calculation and result visualization.

[0057] Software part:

[0058] The software design is divided into three core projects: image post-processing, particle image velocimetry, and high-speed camera configuration. Each project is developed to meet specific functional requirements, and together they form a complete high-speed image acquisition and processing system.

[0059] Image post-processing programs are primarily used to correct and optimize acquired image data to ensure its accuracy and usability. Their core functions include image denoising, grayscale segmentation, contour segmentation, and circularity denoising.

[0060] The particle image velocimetry program is the key computing module of the system, which is used to perform real-time or offline particle image velocimetry analysis on the data collected by the high-speed camera. The program can read and process real-time data signals, and can also perform post-processing analysis on offline data, ensuring that it can be flexibly applied in various scenarios. Figure 5 As shown in Figure 1, the program is divided into two modules: real-time PIV and offline PIV. Real-time PIV measures by reading camera data, while offline PIV supports reading video or image data. This data is fed into the PIV velocity measurement process, which then generates three PIV results: vector field, energy field, and vorticity field, which are then stored.

[0061] The high-speed camera configuration program can set parameters and control the camera to ensure accurate image acquisition during data acquisition. The program has two major functions: camera switch control and real-time image preview, which allows users to quickly adjust the shooting angle and field of view. Through the flexible setting of parameters such as resolution, exposure time, acquisition frame rate and number of saved frames, it can adapt to various experimental needs. During the real-time acquisition process, the program can dynamically adjust the camera settings to ensure the quality of the image data. Figure 4 As shown in the figure, the camera control module includes functions such as shooting preview, high-speed acquisition, dark field calculation, camera formatting, and video export, ensuring that shooting and data processing are carried out as required. The parameter setting module adjusts parameters such as exposure time, shooting frame rate, resolution, and number of saved frames. These settings directly affect the captured image quality and data volume.

[0062] Based on the above two parts, the working principle and overall processing flow of the solution are provided.

[0063] Traditional PIV algorithms work by scattering tracer particles in a flow field, illuminating it with a pulsed laser, and using a camera to record images of the particles at different times. Images are analyzed using either cross-correlation or optical flow methods. The cross-correlation method divides the image into blocks and calculates correlations to determine particle displacements, but this results in an "averaging effect" that reduces spatial resolution. The optical flow method estimates the motion vector field as a global function, achieving pixel-level resolution but being sensitive to noise and imaging conditions. While widely used in flow field measurements, these traditional algorithms suffer from complex parameter settings and low computational efficiency. With technological advancements, deep learning-based PIV velocity measurement algorithms are emerging to overcome these shortcomings.

[0064] The core idea of the deep learning PIV algorithm is to use a convolutional neural network (CNN) to learn the velocity field mapping relationship directly from particle image pairs. Compared with traditional PIV methods, this algorithm avoids manual feature design and complex optimization processes, and can automatically extract motion information from images and generate a highly accurate velocity field.

[0065] The core components of CNN include convolutional layers, activation functions, pooling layers, and fully connected layers.

[0066] The convolutional layer extracts local features of an image using a convolution kernel (Filter) to generate a feature map. The convolutional layer is a core component of a convolutional neural network (CNN), used to extract features from input data (such as an image). Its core operation is the convolution operation, which slides the convolution kernel (Filter) over the input data and calculates the weighted sum of the local area. The following is the formula for the convolutional layer:

[0067]

[0068] Among them, O is the output feature, K is the convolution kernel, and I is the input feature.

[0069] Activation functions are key components in neural networks, introducing nonlinearity that enables them to learn and represent complex patterns. Pooling layers reduce the dimensionality of feature maps, reducing computational effort and enhancing feature robustness.

[0070] The fully connected layer maps the extracted features to the final output through a fully connected output. The fully connected layer fully connects each neuron of the input data to each neuron of the output data, implements linear transformation through the weight matrix and bias vector, and is usually combined with an activation function to introduce nonlinearity. The following is the formula for the fully connected layer:

[0071] y=W·x+b

[0072] Where x is the input vector, W is the weight matrix, b is the matrix bias, and y is the output vector.

[0073] The CNN workflow typically involves inputting an image, extracting features, and generating a classification output. The image first passes through multiple convolutional and pooling layers to extract hierarchical features, which are then mapped to the classification result through fully connected layers. CNN's advantages lie in its local receptive field, parameter sharing, and hierarchical feature extraction, enabling it to efficiently capture local image features while reducing the number of model parameters.

[0074] PWC-Net is essentially based on the CNN architecture, but unlike traditional CNN, PWC-Net is designed specifically for optical flow estimation tasks, and its structure and functions are optimized around optical flow calculation. The working principle of this model is as follows:

[0075] 1. Feature extraction

[0076] For the two input frames I1 and I2, multi-scale features are extracted through convolutional neural networks. Let the feature extraction function of the first layer be f l (·), then the extracted feature map Where l represents the pyramid level. These feature maps contain semantic and detail information at different levels, providing a basis for subsequent optical flow estimation.

[0077] 2. Image deformation

[0078] According to the currently estimated optical flow w, the feature map F1 of the previous frame is deformed so that it is more aligned with the feature map F2 of the next frame. The deformation operation can be expressed as W(F1,w), which is implemented by bilinear interpolation. In two-dimensional space, for the coordinate (x,y) on the feature map F1, the deformed coordinate is (x+w x ,y+w y ), where w x and w yare the components of the optical flow w in the x and y directions respectively. The mathematical expression is:

[0079] W(F1,w)(x,y)=∑ i,j F1(x+i,y+i)·max(0,1-|iw x |)·max(0,1-|jw y |)

[0080] 3. Cost volume

[0081] Construct a cost volume to measure the similarity between the deformed feature map W(F1,w) and F2. For each position (x,y) on the feature map and different optical flow offsets (δx,δy), calculate the similarity metric. Common similarity metrics include dot product:

[0082] C(x,y,δx,δy)=W(F1,w)(x,y)·F2(x+δx,y+δy)

[0083] Combining the similarity measures of all positions and optical flow offsets forms the cost volume C, which is a four-dimensional tensor that records the degree of similarity under different position and optical flow assumptions.

[0084] 4. Optical flow estimation and refinement

[0085] The cost volume C is processed by the convolutional neural network to predict the optical flow offset of each position and obtain the initial optical flow estimate. In order to improve the accuracy, a coarse-to-fine strategy is adopted, where a preliminary estimate is made at the low-resolution pyramid level, and then the result is upsampled and refined with the information from the high-resolution level. At each scale, a loss function L (such as L1 or L2 loss) is used to supervise the training and optimize the model parameters so that the predicted optical flow As close as possible to the real optical flow w gt . Take L1 loss as an example:

[0086] Where N is the number of samples, and w gt are the predicted optical flow and the true optical flow of the i-th sample, respectively. Finally, the optical flow estimation can be further refined through the context network to improve its accuracy and smoothness.

[0087] In the above description, the PWC-Net structure diagram is as follows Figure 2 As shown in FIG3 , the structure diagram of the convolution module and the deconvolution module in the algorithm is shown in FIG3 .

[0088] exist Figure 2The main structure of PWC-Net is described in

[15] . It uses convolutional layers to achieve feature separation. Each convolutional feature layer then undergoes a deconvolution layer to upsample the optical flow estimation to produce the optical flow estimated features. The upper layer's optical flow features are then warped and cost-matched, and then concatenated and deconvolved to obtain the optical flow estimated features. This entire process is repeated, with a final convolution layer used to generate a detailed output prediction of the optical flow.

[0089] exist Figure 3 The convolutional module extracts features through two Conv 3×3 convolutional layers, introduces nonlinearity through the ReLU activation function, and finally uses a Pool 2×2 pooling layer for downsampling, reducing the size of the feature map, reducing the data dimensionality, and extracting more abstract feature representations. The deconvolutional module also uses two Conv 3×3 convolutional layers and ReLU activation, and then uses an UpConv 2×2 deconvolution layer for upsampling, expanding the feature map size and restoring data resolution.

[0090] As a specific example, to measure experimental parameters of a horizontal tube gas-liquid two-phase flow field, 7-μm melamine-formaldehyde resin particles coated with rhodamine dye, which have a higher hydrophilicity, were used as liquid-phase tracer particles (also known as fluorescent tracer particles). Dioctyl sebacate (DEHS) droplets with a particle size of 5-6 μm, treated with a high-pressure atomizer, were used as a gas-phase tracer.

[0091] Image Acquisition: Mount a high-speed camera near the gas-liquid two-phase flow experimental apparatus. Adjust the camera's shooting angle and focal length to ensure a complete and clear capture of the tracer particle's motion within the flow field. During acquisition, adjust camera parameters such as exposure time and gain based on the dynamics of the flow field to obtain high-quality particle images.

[0092] Model training: Collect a large number of particle image sequences from gas-liquid two-phase flows under various operating conditions and accurately annotate them, including information such as the particles' true velocity vectors. Divide the annotated image sequences into training, validation, and test sets. Use the training set data to train the model, employing a stochastic gradient descent optimization algorithm. After a certain number of training iterations, validate the model using the validation set data, observing metrics such as the model's loss function and accuracy. Adjust model parameters based on the validation results to prevent overfitting or underfitting. When the model achieves satisfactory performance on the test set, stop training and save the trained model parameters.

[0093] Image post-processing operations: The particle image undergoes image denoising, grayscale segmentation, contour segmentation, and roundness denoising. The processed image data will be saved and used for subsequent analysis.

[0094] Velocity Field Calculation: A sequence of collected gas-liquid two-phase flow particle images is fed into a trained deep learning model. The model then outputs the corresponding velocity field information based on the learned features and patterns. The calculation results are converted into vector fields, energy fields, and vorticity gradient fields, and displayed directly through a visual interface, allowing users to fully understand the flow field characteristics.

[0095] Therefore, this embodiment has the following beneficial effects:

[0096] High Precision: By using deep learning models to learn and extract features from a large number of complex gas-liquid two-phase flow field images, particle motion trajectories can be identified more accurately. Compared with traditional PIV technology, this significantly improves velocity field measurement accuracy and can accurately capture subtle velocity changes and complex flow pattern characteristics in the flow field.

[0097] High efficiency: The parallel computing capability and optimized algorithm structure of the deep learning model enable velocity field calculation to be completed in a short time, meeting real-time measurement requirements and greatly improving measurement efficiency.

[0098] Strong robustness: Because the model has been trained with a large amount of diverse data, it has strong adaptability to interference factors such as noise, bubble deformation, and liquid phase turbulence in the gas-liquid two-phase flow field. It can stably obtain accurate velocity field information under complex working conditions, thereby improving the robustness of the system.

[0099] High versatility: This system is not only suitable for gas-liquid two-phase flow field measurement, but after appropriate adjustment and training, it can also be widely used to measure other types of flow fields, such as pure liquid flow fields and gas-solid two-phase flow fields. It has good versatility and can meet the flow field measurement needs of different fields.

[0100] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0101] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The above examples are only intended to help understand the core concept of the present invention. At the same time, those skilled in the art will find that the specific implementation methods and application scopes may vary based on the concept of the present invention. In summary, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A method for real-time synchronous measurement of gas-liquid two-phase flow field, characterized in that: include: Acquiring a high-speed image captured by a high-speed camera; the high-speed camera uses set parameters; performing post-processing on the high-speed image to obtain an optimized image; The post-processing includes image denoising, grayscale segmentation, contour segmentation and roundness denoising; Build the PWC-Net network; The PWC-Net network is used to extract features from the optimized image, and the gas-liquid two-phase flow field is predicted based on the extracted features to obtain the final fluid velocity field.

2. The method for real-time synchronous measurement of gas-liquid two-phase flow field according to claim 1, characterized in that: The setting parameters include at least resolution, exposure time, acquisition frame rate and number of saved frames.

3. The method for real-time synchronous measurement of gas-liquid two-phase flow field according to claim 1, characterized in that: The construction process of the PWC-Net network is as follows: Obtaining a training data set; the training data set includes training data and corresponding data labels; Constructing a pre-trained network based on a CNN network; the CNN network includes a convolutional layer, an activation function, a pooling layer, and a fully connected layer; The training data set is input into the pre-trained network, and training is performed with the goal of minimizing the loss between the network output and the data label. When the network meets the preset accuracy requirements, the trained network is determined as the PWC-Net network.

4. The method for real-time synchronous measurement of gas-liquid two-phase flow field according to claim 1, characterized in that: The method of extracting features from the optimized image using the PWC-Net network and predicting the gas-liquid two-phase flow field based on the extracted features to obtain the final fluid velocity field specifically includes: Input the optimized image into the PWC-Net network, perform multi-scale feature extraction on two adjacent frames of images, and obtain a feature map containing semantic and detail information at different levels; Performing image deformation based on the feature map, aligning two frames of image to obtain a deformation feature map; Obtaining a similarity metric based on the similarity between each position on the deformation feature map and different optical flow offsets, and constructing a cost volume based on the similarity metric; The gas-liquid two-phase flow field is predicted based on the cost volume to obtain the final fluid velocity field.

5. The method for real-time synchronous measurement of gas-liquid two-phase flow field according to claim 4, characterized in that: The image deformation is performed based on the feature map, and two frames of images are aligned to obtain a deformation feature map, specifically including: The deformation operation is represented as W(F1,w), and the feature map F1 of the previous frame is deformed by bilinear interpolation so that the feature map F1 of the previous frame is aligned with the feature map F2 of the next frame. In the two-dimensional space, for the coordinate (x, y) on the feature map F1, the coordinate after deformation is (x+w x ,y+w y ), where w x and w y are the components of the optical flow w in the x and y directions, respectively, and the mathematical expression is: W(F1,w)(x,y)=∑ i,j F1(x+i,w+i)·max(0,1-|i-w x |)·max(0,1-|j-w y |) Where i and j represent the interpolation offset.

6. The method for real-time synchronous measurement of gas-liquid two-phase flow field according to claim 4, characterized in that: The similarity between each position on the deformation feature map and different optical flow offsets is obtained to obtain a similarity metric, and a cost volume is constructed according to the similarity metric, specifically including: For each position (x, y) on the deformation feature map and different optical flow offsets (δx, δy), calculate the similarity metric: C(x,y,δx,δy)=W(F1,w)(x,y)·F2(x+δx,y+δy) Among them, W(F1,w) represents the deformation operation, w represents the optical flow, F1 represents the feature map of the previous frame, and F2 represents the feature map of the next frame; The similarity measures of all positions and optical flow offsets are combined to form a cost volume C; the cost volume C is a four-dimensional tensor used to record the similarity under different positions and optical flow assumptions.

7. The method for real-time synchronous measurement of gas-liquid two-phase flow field according to claim 4, characterized in that: The gas-liquid two-phase flow field prediction based on the cost volume to obtain the final fluid velocity field specifically includes: Based on the cost volume, a coarse-to-fine strategy is adopted to upsample the prediction results in order from low to high resolution levels to obtain the final fluid velocity field.

8. A gas-liquid two-phase flow field real-time synchronous measurement system, using the method according to any one of claims 1 to 7, characterized in that: include: High-power laser light source, high-speed camera, synchronization controller, optical correction box and host computer; A plurality of high-speed cameras are arranged around the optical correction box, and the light emitting end of the high-power laser light source is arranged at the top of the optical correction box. Each of the high-speed cameras and the high-power laser light source is connected to the synchronization controller, and the synchronization controller is also connected to the host computer.

Images