Gas stagnation water hammer analysis method and system coupling actual gas characteristics and CLSVOF
By coupling gas properties with CLSVOF, random distribution images of bubble swarms are acquired and processed, a CLSVOF model is constructed, and the analysis of gas stagnation and water hammer is optimized. This solves the problem that the bubble distribution was not fully considered in the study of gas stagnation and water hammer, and realizes real-time and accurate prediction of gas stagnation and water hammer phenomena.
Patent Information
- Application Number
- CN202410977969.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-22
- Publication Date
- 2024-11-22
AI Technical Summary
Existing technologies have failed to adequately consider the random distribution of bubbles in the analysis of gas-liquid water hammer, resulting in insufficient research on gas-liquid water hammer, especially in the case of gas-liquid two-phase flow, where there is a lack of calculation methods based on the random distribution of bubbles.
By coupling actual gas characteristics with CLSVOF, images of the random distribution of bubble swarms in the pipeline are collected, preprocessed, and a CLSVOF model is constructed to calculate the random distribution of the bubble swarms. The gas stagnation and water hammer analysis model is optimized using a time compensation coefficient.
It improves the accuracy and stability of water hammer analysis, reduces time complexity and local non-uniformity error, and enables real-time and accurate prediction of water hammer phenomena.
Smart Images

Figure CN119004167A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of water hammer analysis, and more particularly to a method and system for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF. Background Art
[0002] The construction of long-distance pressurized water transmission projects across river basins and regions is an effective way to alleviate the contradiction between social water resources supply and demand. In order to solve the water hammer hazards caused by the operation of long-distance pressurized water transmission pipeline systems, the main research aspects of long-distance pressurized water transmission pipeline systems, such as aeration water hammer, gas-liquid two-phase flow, and aeration pipe burst, were reviewed in theoretical analysis, numerical simulation, and physical model tests.
[0003] According to the basic idea of CLSVOF method, VOF and LevelSet are combined to track the calculation of the phase interface of incompressible two-phase flow, which overcomes the shortcomings of false velocity potential and difficulty in accurately calculating the normal and curvature of the interface when solving the momentum transport equation using VOF method. However, in terms of stagnant water hammer, the research on stagnant water hammer caused by small volume bubbles is not sufficient; in terms of gas-liquid two-phase flow, there are few studies on stagnant water hammer based on the random distribution theory of bubbles, and the calculation method of stagnant water hammer based on random distribution theory has not yet been established.
[0004] Therefore, how to propose a stagnant water hammer analysis method and system that couples actual gas characteristics with CLSVOF and optimizes the stagnant water hammer analysis effect through the random distribution of bubbles is an urgent problem that technicians in this field need to solve. Summary of the invention
[0005] In view of this, the present invention provides a method and system for analyzing stagnant water hammer by coupling actual gas characteristics with CLSVOF, and optimizes the stagnant water hammer analysis effect by random distribution of bubbles. In order to achieve the above purpose, the present invention adopts the following technical solutions:
[0006] A method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF, comprising:
[0007] Collect random distribution images of bubble groups in the pipeline, perform preprocessing, and calculate the random distribution of the real-time bubble groups based on the preprocessed images;
[0008] The CLSVOF model for solving the phase interface of incompressible two-phase flow is constructed by coupling VOF and Level Set.
[0009] The interface position and volume fraction information of different phases in each spatial unit under random distribution are calculated by CLSVOF model;
[0010] The stagnant water hammer analysis model is constructed based on the random distribution of bubble groups, gas characteristics, and the interface positions and volume fractions of different phases in each spatial unit under random distribution.
[0011] Optimize the stagnant air water hammer analysis model through time compensation coefficient;
[0012] The optimized aeration water hammer analysis model is used to predict and analyze the aeration water hammer phenomenon in real time.
[0013] Optionally, the randomly distributed image of the bubble group in the pipeline is collected and preprocessed, including: using a high-speed camera to shoot a video of the distribution of the irregular real-time bubble group in the pipeline through a visual window; performing real-time image frame extraction and noise reduction processing on the shot video according to a time series, selecting a morphological stretch area, and obtaining a real-time bubble group image.
[0014] Optionally, calculating the random distribution of the real-time bubble group according to the preprocessed image includes:
[0015] Taking pictures of the real-time bubble group to obtain a real-time bubble group image;
[0016] Extract the centroid coordinates of each irregular bubble in the real-time bubble group and normalize the coordinates;
[0017] Based on the spatial distribution of the bubble group and the normalized centroid coordinates, the average distance method is used to determine the first uniformity of the bubble mixing of the real-time bubble group.
[0018] Optionally, also include:
[0019] Collect the number or density data of bubbles in each area of the real-time bubble group;
[0020] Fit the data to a Poisson distribution model and calculate the mean incidence of the Poisson distribution;
[0021] The fitted Poisson distribution was tested using the chi-square test to verify whether the data met the assumptions of the Poisson distribution;
[0022] The degree of agreement between the fitted Poisson distribution and the observed data was determined by model comparison, and the secondary uniformity and Poisson property of the bubble distribution were evaluated;
[0023] The random distribution of the real-time bubble group is obtained by coupling the first uniformity, the second uniformity and the Poisson property.
[0024] Optionally, the coupling VOF and Level Set to construct a CLSVOF model for solving the phase interface of an incompressible two-phase flow includes:
[0025] Initialize the Level Set function and VOF function respectively;
[0026] After initialization, the function velocity field and pressure field are updated by solving the pressure correction equation and momentum equation;
[0027] After the update, the governing equations for the flow transport are solved by the characterization function;
[0028] Reconstruct the phase interface according to the solution results and reinitialize the Level Set function;
[0029] The CLSVOF model for solving the phase interface of incompressible two-phase flow is obtained through cyclic iteration.
[0030] Optionally, reconstructing the phase interface according to the solution result includes:
[0031] According to the solution results, the discrete VOF function and the continuous Level Set function are obtained;
[0032] The interface reconstruction is performed using the piecewise linear interface reconstruction method. The interface is moved in the normal direction to match the unit liquid area ratio with the VOF function value. The VOF function flux between two adjacent units is calculated, and the vertical distance from the unit center to the phase interface is iteratively solved using the secant method.
[0033] The Level Set function is initialized again as a distance function using the reconstructed phase interface.
[0034] Optionally, the re-initialization of the Level Set function includes: determining the function sign and the Level Set function value of the Level Set, and controlling the Level Set function value of each computing unit to be the minimum value of the distance from the unit center to the edge, the distance from the unit center to the unit vertex, the distance from the unit center to the intersection of the phase interface and the grid edge, and the distance from the unit center to the projection point of the phase interface.
[0035] Optionally, constructing an air stagnation water hammer analysis model includes:
[0036] The collected data in the pipeline is preprocessed to obtain the random distribution of the bubble group, gas characteristics, and the interface position and volume fraction information of different phases in each spatial unit under random distribution;
[0037] Normalize the preprocessed data and divide the data set;
[0038] Based on the divided data set, a stagnant air water hammer analysis model based on a recurrent neural network is constructed and trained;
[0039] The real-time data in the pipeline is collected as the sample to be tested, and the sample to be tested is input into the stagnant water hammer analysis model to obtain the predicted values of the interface position and volume fraction information of different phases in the pipeline at the next moment.
[0040] Optionally, the optimization of the stagnant air water hammer analysis model by the time compensation coefficient includes:
[0041] Obtain time information in time sequence according to the captured video;
[0042] Acquiring and normalizing time information, temperature compensation coefficients, and machine compensation coefficients to obtain a characteristic value group;
[0043] The time compensation coefficient is obtained by calculating the fuzzy neural network trained in advance based on the characteristic value group;
[0044] The time compensation coefficient is used to optimize the memory unit and weight of the stagnant water hammer analysis model.
[0045] Optionally, a stagnant gas water hammer analysis system that couples actual gas properties with CLSVOF, including:
[0046] Acquisition module: used to acquire randomly distributed images of bubble groups in the pipeline;
[0047] Preprocessing module: used for preprocessing and calculating the random distribution of real-time bubble groups based on the preprocessed images;
[0048] CLSVOF model building module: used to couple VOF and Level Set to build a CLSVOF model for solving the phase interface of incompressible two-phase flow;
[0049] Calculation module: used to calculate the interface position and volume fraction information of different phases in each spatial unit under random distribution through the CLSVOF model;
[0050] Analysis model building module: used to build a stagnant water hammer analysis model based on the random distribution of bubble groups, gas characteristics, and the interface positions and volume fractions of different phases in each spatial unit under random distribution;
[0051] Optimization module: used to optimize the stagnant water hammer analysis model through the time compensation coefficient;
[0052] Prediction and analysis module: used to predict and analyze the aeration water hammer phenomenon in real time through the optimized aeration water hammer analysis model.
[0053] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method and system for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF, which has the following beneficial effects:
[0054] The present invention proposes a stagnant water hammer analysis method that couples actual gas characteristics with CLSVOF, including: collecting random distribution images of bubble groups in a pipeline, and performing preprocessing, and calculating the random distribution of real-time bubble groups according to the preprocessed images; coupling VOF and Level Set to construct a CLSVOF model for solving the phase interface of incompressible two-phase flow; calculating the interface position and volume fraction information of different phases in each spatial unit under random distribution through the CLSVOF model; constructing a stagnant water hammer analysis model through the random distribution of bubble groups, gas characteristics, and the interface position and volume fraction information of different phases in each spatial unit under random distribution; optimizing the stagnant water hammer analysis model through a time compensation coefficient; and performing real-time prediction and analysis of the stagnant water hammer phenomenon through the optimized stagnant water hammer analysis model. The present invention reduces the time complexity by calculating the random distribution of real-time bubble groups and considering the spatial distribution of bubbles, and effectively reduces the errors of local and overall inhomogeneity. Considering the correlation in time series, the time prior correlation information is learned to improve the stability of the network and the accuracy of prediction. At the same time, in the bubble group detection in the pipeline environment of continuous scenes, the feature information of the previous image in the continuous image is used for the subsequent image, which reduces the parameters and calculation complexity of the network model. The time compensation coefficient is combined to construct and optimize the stagnant water hammer analysis model. The optimized stagnant water hammer analysis model is used to perform real-time and accurate prediction and analysis of the stagnant water hammer phenomenon. BRIEF DESCRIPTION OF THE DRAWINGS
[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.
[0056] Figure 1 A schematic flow chart of a stagnant gas water hammer analysis method for coupling actual gas characteristics with CLSVOF provided by the present invention.
[0057] Figure 2 A structural framework diagram of a gas stagnation water hammer analysis system that couples actual gas characteristics with CLSVOF provided by the present invention.
[0058] Figure 3 This is a schematic diagram of the CLSVOF model provided by the present invention. DETAILED DESCRIPTION
[0059] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. 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 creative work are within the scope of protection of the present invention.
[0060] The embodiment of the present invention discloses a method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF. Figure 1 As shown, including:
[0061] Collect random distribution images of bubble groups in the pipeline, perform preprocessing, and calculate the random distribution of the real-time bubble groups based on the preprocessed images;
[0062] The CLSVOF model for solving the phase interface of incompressible two-phase flow is constructed by coupling VOF and Level Set.
[0063] The interface position and volume fraction information of different phases in each spatial unit under random distribution are calculated by CLSVOF model;
[0064] The stagnant water hammer analysis model is constructed based on the random distribution of bubble groups, gas characteristics, and the interface positions and volume fractions of different phases in each spatial unit under random distribution.
[0065] Optimize the stagnant air water hammer analysis model through time compensation coefficient;
[0066] The optimized aeration water hammer analysis model is used to predict and analyze the aeration water hammer phenomenon in real time.
[0067] Furthermore, the random distribution image of the bubble group in the pipeline is collected and preprocessed, including: using a high-speed camera to shoot a video of the distribution of the irregular real-time bubble group in the pipeline through a visual window; performing real-time image frame extraction and Gaussian noise reduction processing on the shot video according to a time series, selecting a morphological stretch area, and obtaining a real-time bubble group image.
[0068] Furthermore, the random distribution of the real-time bubble group is calculated based on the preprocessed image, including:
[0069] Taking pictures of the real-time bubble group to obtain a real-time bubble group image;
[0070] Extract the centroid coordinates of each irregular bubble in the real-time bubble group and normalize the coordinates;
[0071] Based on the spatial distribution of the bubble group and the normalized centroid coordinates, the average distance method is used to determine the first uniformity of the bubble mixing of the real-time bubble group.
[0072] Specifically, the normalization process includes:
[0073] Performing top hat transformation and filter transformation preprocessing on the real-time bubble group image, and segmenting the processed real-time bubble group image to obtain a series of grayscale images;
[0074] Convert the grayscale image into a black and white image, and use a threshold to convert the binary image;
[0075] Based on the binary image, the centroid coordinates of each irregular bubble in the real-time bubble group are found, and the coordinates are normalized.
[0076] Specifically, judging the first uniformity of bubble mixing of the real-time bubble group by using the average distance method includes:
[0077] The hypercube line selection method in computational geometry theory is used to calculate the average distance of random points in a two-dimensional image when the image reaches a uniform state, and the average distance is used as a standard reference value.
[0078] Based on the spatial distribution of the bubble group and the normalized centroid coordinates, an average distance between random irregular bubbles is obtained, and the average distance is compared with a standard reference value to determine the first uniformity of the bubble mixing of the real-time bubble group.
[0079] The average distance method is used to effectively judge the mixing uniformity of random irregular bubble groups, taking into account the spatial distribution of bubbles, reducing the time complexity and effectively reducing the errors of local and overall unevenness.
[0080] Furthermore, it also includes:
[0081] Collect the number or density data of bubbles in each area of the real-time bubble group;
[0082] Fit the data to a Poisson distribution model and calculate the mean incidence of the Poisson distribution;
[0083] The fitted Poisson distribution was tested using the chi-square test to verify whether the data met the assumptions of the Poisson distribution;
[0084] The degree of agreement between the fitted Poisson distribution and the observed data was determined by model comparison, and the secondary uniformity and Poisson property of the bubble distribution were evaluated;
[0085] The random distribution of the real-time bubble group is obtained by coupling the first uniformity, the second uniformity and the Poisson property, and the first uniformity, the second uniformity and the Poisson property are weightedly summed to obtain a parameter value characterizing the random distribution of the real-time bubble group.
[0086] Furthermore, the coupled VOF and Level Set construct a CLSVOF model for solving the phase interface of incompressible two-phase flow, such as Figure 3 As shown, including:
[0087] Initialize the Level Set function and VOF function respectively;
[0088] After initialization, the function velocity field and pressure field are updated by solving the pressure correction equation and momentum equation; the control equations of gas-liquid two-phase flow mainly include the continuity equation and momentum equation, and the control equations of single-phase and mixed-phase two-phase flow are written in a unified form;
[0089] After the update, the governing equations for the flow transport are solved by the characterization function;
[0090] Reconstruct the phase interface according to the solution results and reinitialize the Level Set function;
[0091] The CLSVOF model for solving the phase interface of incompressible two-phase flow is obtained through cyclic iteration.
[0092] Furthermore, the phase interface reconstruction according to the solution result includes:
[0093] According to the solution results, the discrete VOF function and the continuous Level Set function are obtained;
[0094] The interface reconstruction is performed using the piecewise linear interface reconstruction method. The interface is moved in the normal direction to match the unit liquid area ratio with the VOF function value. The VOF function flux between two adjacent units is calculated, and the vertical distance from the unit center to the phase interface is iteratively solved using the secant method.
[0095] The Level Set function is initialized again as a distance function using the reconstructed phase interface.
[0096] Furthermore, the re-initialization of the Level Set function includes: determining the function sign of LevelSet by the phase mark function, and determining the Level Set function value by calculating the minimum distance from the center of the grid unit to the phase interface, and controlling the Level Set function value of each calculation unit to be the minimum value of the distance from the center of the unit to the edge, the distance from the center of the unit to the unit vertex, the distance from the center of the unit to the intersection of the phase interface and the grid edge line, and the distance from the center of the unit to the projection point of the phase interface.
[0097] Furthermore, the construction of the stagnant air water hammer analysis model includes:
[0098] The collected data in the pipeline is preprocessed to obtain the random distribution of the bubble group, gas characteristics, and the interface position and volume fraction information of different phases in each spatial unit under random distribution;
[0099] Normalize the preprocessed data, divide the data set, and set the original data;
[0100] Construct and train a stagnant air water hammer analysis model based on recurrent neural network;
[0101] The real-time data in the pipeline is collected as the sample to be tested, and the sample to be tested is input into the stagnant water hammer analysis model to obtain the predicted values of the interface position and volume fraction information of different phases in the pipeline at the next moment.
[0102] In a specific implementation, it also includes: predicting the random distribution of bubble groups through a deep neural network, the steps are as follows:
[0103] Obtain the image to be identified and N consecutive images before and after it, and form a set of consecutive image sequences in time sequence;
[0104] Inputting the continuous image sequence into a pre-trained deep neural network model; the deep neural network model includes an encoding network module, a recurrent convolutional network module and a decoding network module;
[0105] Extracting features from each image in the continuous image sequence in sequence through the encoding network module to obtain a feature graph sequence containing semantic features of bubble groups;
[0106] The recurrent convolutional network module receives the recurrent input of the feature map sequence from the encoding network module, performs multi-layer recurrent convolution and temporal feature fusion, and outputs a semantic feature map after feature fusion;
[0107] The semantic feature map is decoded by the decoding network module, and a predicted value of the random distribution of the bubble group is output.
[0108] Furthermore, the construction of an enhanced aeration water hammer analysis model includes:
[0109] Preprocessing the collected data in the pipeline to obtain the random distribution of bubble groups, the predicted value of the random distribution of bubble groups, gas properties, and the interface positions and volume fraction information of different phases in each spatial unit under random distribution. The gas properties include: bubble size, gas type, gas solubility and gas compressibility;
[0110] Normalize the preprocessed data and divide the data set;
[0111] Based on the divided data set, a stagnant air water hammer analysis model based on a recurrent neural network is constructed and trained;
[0112] The real-time data in the pipeline is collected as the sample to be tested, and the sample to be tested is input into the stagnant water hammer analysis model to obtain the predicted values of the interface position and volume fraction information of different phases in the pipeline at the next moment.
[0113] Specifically, the steps of constructing and training a stagnant air water hammer analysis model based on a recurrent neural network are as follows:
[0114] The preprocessed data is randomly divided into a training set and a test set. The training set is then used to generate the model, and the test set is used to test the accuracy and error of the model to verify the effectiveness of the model.
[0115] Build and train the initial prediction model and initialize the model parameters;
[0116] Collect the preprocessed data within a given window and the true value observed under the given window;
[0117] Calculate the final forecast value within a given window for the heating equipment;
[0118] Based on the error between the final predicted value and the true value, the model parameters are updated to complete the construction and training of the final prediction model.
[0119] Further, the updating model parameters includes:
[0120] Use gradient descent backpropagation to update the parameters of the neural network module, attention mechanism component, and vector autoregression module:
[0121] Using mean square error MSE as the objective function, its calculation formula is as follows:
[0122]
[0123] Among them, O n Represents the true value of the nth sample, O' n Represents the predicted value of the nth sample, where N represents the number of samples. If the current error is less than the set error threshold, it is determined that the iteration reaches the preset maximum number of iterations and a trained prediction model is obtained, otherwise the prediction model continues to be trained.
[0124] Specifically, the gas size includes:
[0125] Draw a bubble size frequency density distribution map based on the real-time bubble group image;
[0126] Construct the Gauss distribution of the Gauss model. The Gauss distribution is:
[0127]
[0128] Where: d-frequency density, characterizing the probability of bubble size appearing;
[0129] j - bubble size, j k - Bubble size expectation, characterizing the average bubble size; z-2 times the standard deviation;
[0130] The Gauss distribution of the Gauss model is used to fit the bubble size density distribution according to the obtained bubble size frequency density distribution diagram to obtain the maximum bubble size and the expected bubble size j. k and 2 times the standard deviation z; according to the maximum corrosion depth and expected corrosion depth j obtained by fitting k and 2 times the standard deviation z result, explaining the bubble size type and the probability of occurrence of different bubble sizes.
[0131] This embodiment fully considers the continuity of the scene, that is, the association in the time series, and learns the time prior association information, so as to improve the stability of the network and the accuracy of the prediction. At the same time, in the bubble group detection in the pipeline environment of the continuous scene, the feature information of the previous image in the continuous image is used for the next image, so as to reduce the parameters and computational complexity of the network model.
[0132] Furthermore, the optimization of the stagnant air water hammer analysis model by the time compensation coefficient includes:
[0133] Obtain time information in time sequence according to the captured video;
[0134] Acquiring and normalizing time information, temperature compensation coefficients, and machine compensation coefficients to obtain a characteristic value group;
[0135] The time compensation coefficient is obtained by calculating the fuzzy neural network trained in advance based on the characteristic value group;
[0136] The time compensation coefficient is used to optimize the memory unit and weight of the stagnant water hammer analysis model.
[0137] Specifically, the fuzzy neural network includes an input layer, a fuzzification layer, a fuzzy reasoning layer and an output layer. The structure of the fuzzy neural network is changed by setting the weights and node points of the fuzzification layer and the fuzzy reasoning layer. The input layer is filled with the normalized eigenvalues of the three parameters of time information, temperature compensation coefficient and machine compensation coefficient, and finally a time compensation coefficient is output through the fuzzification layer, the fuzzification layer and the defuzzification layer to complete the time compensation of the stagnant air water hammer analysis process.
[0138] In a specific embodiment, the training process of the fuzzy neural network includes:
[0139] Initialize the collected three characteristic values of temperature compensation coefficient, time information and machine compensation coefficient, wherein the temperature compensation coefficient is determined according to the temperature history data, the time information is determined according to the collected time series, and the machine compensation coefficient is determined according to the running state of the collection device;
[0140] Using the first set of eigenvalues collected by initialization, calculate the corresponding output values, the deviation between the output values and the expected output values when the three eigenvalues are used as inputs;
[0141] Using a fuzzy neural network algorithm to calculate weights and result values based on the output values and deviations;
[0142] Compare the result value of the time compensation coefficient with the expected value. If the result value is less than the expected value, it means that the function variables and result values calculated by the algorithm have reached the expected target. If not, the weights and function variables that do not meet the performance calculation will be used as the required conditions for the next calculation and re-iterate until they meet the threshold range.
[0143] Specifically, the time compensation coefficient is multiplied by the memory unit of the stagnant water hammer analysis model, and the weight of the memory unit is adjusted to make the model pay more attention to the memory of the time step, thereby improving the performance of the model when dealing with time dependency and time offset problems; or the time compensation coefficient is multiplied by the connection weight or output weight in the recurrent neural network, and the importance of the model to the input of different time steps is adjusted, so that the model can more flexibly adapt to the input data of different time steps.
[0144] In a specific embodiment, a gas stagnation water hammer analysis system that couples actual gas properties with CLSVOF, such as Figure 2 As shown, including:
[0145] Acquisition module: used to acquire randomly distributed images of bubble groups in the pipeline;
[0146] Preprocessing module: used for preprocessing and calculating the random distribution of real-time bubble groups based on the preprocessed images;
[0147] CLSVOF model building module: used to couple VOF and Level Set to build a CLSVOF model for solving the phase interface of incompressible two-phase flow;
[0148] Calculation module: used to calculate the interface position and volume fraction information of different phases in each spatial unit under random distribution through the CLSVOF model;
[0149] Analysis model building module: used to build a stagnant water hammer analysis model based on the random distribution of bubble groups, gas characteristics, and the interface positions and volume fractions of different phases in each spatial unit under random distribution;
[0150] Optimization module: used to optimize the stagnant water hammer analysis model through the time compensation coefficient;
[0151] Prediction and analysis module: used to predict and analyze the aeration water hammer phenomenon in real time through the optimized aeration water hammer analysis model.
[0152] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.
[0153] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF. It is characterized in that include: Collect random distribution images of bubble groups in the pipeline, perform preprocessing, and calculate the random distribution of the real-time bubble groups based on the preprocessed images; The CLSVOF model for solving the phase interface of incompressible two-phase flow is constructed by coupling VOF and Level Set. The interface position and volume fraction information of different phases in each spatial unit under random distribution are calculated by CLSVOF model; The stagnant water hammer analysis model is constructed based on the random distribution of bubble groups, gas characteristics, and the interface positions and volume fractions of different phases in each spatial unit under random distribution. Optimize the stagnant air water hammer analysis model through time compensation coefficient; The optimized aeration water hammer analysis model is used to predict and analyze the aeration water hammer phenomenon in real time.
2. A method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 1, It is characterized in that The method of collecting the randomly distributed image of the bubble group in the pipeline and performing preprocessing includes: using a high-speed camera to shoot a video of the distribution of the irregular real-time bubble group in the pipeline through a visual window; performing real-time image frame extraction and noise reduction processing on the shot video according to a time series, selecting a morphological stretch area, and obtaining a real-time bubble group image.
3. A method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 1, It is characterized in that The random distribution of the real-time bubble group is calculated based on the preprocessed image, including: Taking pictures of the real-time bubble group to obtain a real-time bubble group image; Extract the centroid coordinates of each irregular bubble in the real-time bubble group and normalize the coordinates; Based on the spatial distribution of the bubble group and the normalized centroid coordinates, the average distance method is used to determine the first uniformity of the bubble mixing of the real-time bubble group.
4. A method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 3, It is characterized in that Also includes: Collect the number or density data of bubbles in each area of the real-time bubble group; Fit the data to a Poisson distribution model and calculate the mean incidence of the Poisson distribution; The fitted Poisson distribution was tested using the chi-square test to verify whether the data met the assumptions of the Poisson distribution; The degree of agreement between the fitted Poisson distribution and the observed data was determined by model comparison, and the secondary uniformity and Poisson property of the bubble distribution were evaluated; The random distribution of the real-time bubble group is obtained by coupling the first uniformity, the second uniformity and the Poisson property.
5. The method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 1, It is characterized in that The CLSVOF model for solving the phase interface of incompressible two-phase flow by coupling VOF and Level Set includes: Initialize the Level Set function and VOF function respectively; After initialization, the function velocity field and pressure field are updated by solving the pressure correction equation and momentum equation; After the update, the governing equations for the flow transport are solved by the characterization function; Reconstruct the phase interface according to the solution results and reinitialize the Level Set function; The CLSVOF model for solving the phase interface of incompressible two-phase flow is obtained through cyclic iteration.
6. A method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 5, It is characterized in that The phase interface reconstruction according to the solution result comprises: According to the solution results, the discrete VOF function and the continuous Level Set function are obtained; The interface reconstruction is performed using the piecewise linear interface reconstruction method. The interface is moved in the normal direction to match the unit liquid area ratio with the VOF function value. The VOF function flux between two adjacent units is calculated, and the vertical distance from the unit center to the phase interface is iteratively solved using the secant method. The Level Set function is initialized again as a distance function using the reconstructed phase interface.
7. The method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 5, It is characterized in that The re-initialization of the Level Set function includes: determining the function sign and the Level Set function value of the Level Set, and controlling the Level Set function value of each computing unit to be the minimum value of the distance from the unit center to the edge, the distance from the unit center to the unit vertex, the distance from the unit center to the intersection of the phase interface and the grid edge line, and the distance from the unit center to the projection point of the phase interface.
8. The method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 1, It is characterized in that The construction of the stagnant air water hammer analysis model comprises: The collected data in the pipeline is preprocessed to obtain the random distribution of the bubble group, gas characteristics, and the interface position and volume fraction information of different phases in each spatial unit under random distribution; Normalize the preprocessed data and divide the data set; Based on the divided data set, a stagnant air water hammer analysis model based on a recurrent neural network is constructed and trained; The real-time data in the pipeline is collected as the sample to be tested, and the sample to be tested is input into the stagnant water hammer analysis model to obtain the predicted values of the interface position and volume fraction information of different phases in the pipeline at the next moment.
9. The method for analyzing stagnant gas water hammer by coupling actual gas characteristics with CLSVOF according to claim 1, It is characterized in that The time compensation coefficient optimization of the stagnant air water hammer analysis model includes: Obtain time information in time sequence according to the captured video; Acquiring and normalizing time information, temperature compensation coefficients, and machine compensation coefficients to obtain a characteristic value group; The time compensation coefficient is obtained by calculating the fuzzy neural network trained in advance based on the characteristic value group; The time compensation coefficient is used to optimize the memory unit and weight of the stagnant water hammer analysis model.
10. A gas stagnation water hammer analysis system that couples actual gas characteristics with CLSVOF. It is characterized in that include: Acquisition module: used to acquire randomly distributed images of bubble groups in the pipeline; Preprocessing module: used for preprocessing and calculating the random distribution of real-time bubble groups based on the preprocessed images; CLSVOF model building module: used to couple VOF and Level Set to build a CLSVOF model for solving the phase interface of incompressible two-phase flow; Calculation module: used to calculate the interface position and volume fraction information of different phases in each spatial unit under random distribution through the CLSVOF model; Analysis model building module: used to build a stagnant water hammer analysis model based on the random distribution of bubble groups, gas characteristics, and the interface positions and volume fractions of different phases in each spatial unit under random distribution; Optimization module: used to optimize the stagnant water hammer analysis model through the time compensation coefficient; Prediction and analysis module: used to predict and analyze the aeration water hammer phenomenon in real time through the optimized aeration water hammer analysis model.
Citation Information
Cited By
High-pressure bubble pulsating load prediction and initial parameter inversion method and system based on artificial intelligence
CN120542323A
Electricity-heat-flow-gas multi-physics field collaborative simulation model construction method
CN121052172A
A Method for Constructing a Multiphysics Collaborative Simulation Model of Electricity, Heat, Fluidity, and Gas
CN121052172B
