A numerical method for simulating wear of ultra-long-distance mud pipelines

Abstract

The present invention discloses a numerical method for simulating the wear of ultra-long-distance mud pipelines, and belongs to the technical field of pipeline wear numerical simulation methods. It includes: S1, establishing a fluid control equation; S2, constructing a simplified turbulence model, calculating the transport equation for controlling the turbulent kinetic energy k in the pipeline and the turbulent dissipation rate ε; S3, establishing the collision equations between fluid and particles, and between particles and walls; S4, constructing an erosion wear model to simulate the pipeline wear mechanism, and the erosion wear model at least includes: impact wear rate, scouring wear rate and fatigue wear rate; S5, performing erosion wear calculation to obtain the erosion wear rate; S6, introducing the pipeline operation data actually detected; S7, performing PINNs physical information neural network learning and fluid dynamics analysis, combining the pipeline material properties, obtaining the erosion wear rate law, and realizing pipeline wear prediction. The present invention combines numerical simulation and machine learning to realize intelligent pipeline wear management.

Description

Technical Field

[0001] The invention belongs to the technical field of numerical simulation methods for wear of mud transport pipelines of slurry balance shield machines, and in particular relates to a numerical method for simulating wear of ultra-long distance mud pipelines. Background Art

[0002] The slurry circulation system of the slurry shield is mainly composed of slurry inlet and outlet pumps, slurry inlet and outlet pipelines, control valves, quarrying boxes and pipeline extension mechanisms. It is one of the core systems of the slurry shield system and plays an important role in maintaining the stability of the excavation face, transporting slag and draining sewage from tunnels. During the operation of the ultra-long-distance slurry pipeline, when the slurry shield is excavating in the gravel stratum, due to the continuous collision and friction between the solid particles carried in the fluid and the pipeline wall, a large number of stones cause serious wear on the inner wall of the pipeline, especially at the elbows and joints during the transportation of the slurry pipeline, resulting in a shortened service life of the pipeline and even causing safety accidents. Therefore, accurately predicting and evaluating the wear of the pipeline is crucial to ensure the safe operation of the pipeline.

[0003] However, in the existing technology, the current experimental methods are costly, long-term, and difficult to cover all working conditions, and cannot monitor and predict the wear of pipelines in real time. Although the empirical formula method is simple and easy to use, it has low accuracy, is difficult to adapt to complex working conditions, and cannot provide accurate wear prediction. Most of the existing technologies rely on a single experimental or simulation method, lack intelligent management means, and cannot achieve real-time monitoring and intelligent prediction of pipeline wear. Summary of the invention

[0004] In order to solve the technical problems existing in the above-mentioned background technology, the present invention provides a numerical method for simulating the wear of ultra-long-distance mud pipelines.

[0005] The present invention adopts the following technical solution: a numerical method for simulating wear of ultra-long distance mud pipelines, comprising the following steps:

[0006] S1. Constructing fluid control equations to provide a basic fluid dynamics description for the simplified turbulence model and erosion wear model in steps S2 and S4; including:

[0007] Mass conservation equation: describes the conservation relationship of fluid mass in the control volume, ensuring that the mass of the fluid does not change during the flow process;

[0008] Momentum conservation equation: describes the conservation relationship of fluid momentum in the control volume, taking into account the velocity change of the fluid and the external force;

[0009] Energy conservation equation: describes the conservation relationship of fluid energy in the control volume, taking into account the temperature change and energy transfer of the fluid;

[0010] S2. Construct a simplified turbulence model to calculate the transport equations controlling the turbulent kinetic energy k and the turbulent dissipation rate ε in the pipeline, including:

[0011] The transport equation of turbulent kinetic energy k: describes the transport process of turbulent kinetic energy in the fluid;

[0012] Turbulence dissipation rate ε: describes the rate at which turbulent kinetic energy is converted into heat energy;

[0013] S3: Construct collision equations between fluid and particles, and between particles and wall to simulate the movement and collision behavior of particles in the fluid, and provide data on particle movement and collision for the erosion wear model in S4, including:

[0014] Fluid-particle collision equations: construct the motion and collision process of particles in the fluid, describe the inertia of particles and the drag force of the fluid;

[0015] Particle-wall collision equation: construct the collision process between particles and pipe walls, and describe the rebound and friction of particles;

[0016] S4. Construct an erosion wear model to simulate the pipeline wear mechanism, wherein the erosion wear model at least includes: impact wear rate, scouring wear rate and fatigue wear rate, wherein:

[0017] Impact wear rate: describes the wear rate caused by particles impacting the pipe wall, taking into account the impact angle and impact force;

[0018] Erosion wear rate: describes the wear rate caused by fluid erosion of the pipe wall, taking into account the velocity and particle concentration of the fluid;

[0019] Fatigue wear rate: describes the fatigue wear rate caused by repeated impact of particles on the pipe wall, taking into account the impact frequency and material fatigue properties;

[0020] S5. Calculate the erosion wear rate of the pipeline wall according to the erosion wear model to obtain the erosion wear rate, which is used to quantify the wear degree of the pipeline wall and provide data support for subsequent wear prediction;

[0021] S6, introducing the pipeline operation data actually detected, including at least the fluid velocity, pressure, and wear thickness of the pipeline in the pipeline, into the model to achieve data correction;

[0022] S7. Conduct PINNs physical information neural network learning and fluid dynamics analysis, combine pipeline material properties, derive the law of erosion wear rate, and realize pipeline wear prediction.

[0023] Preferably, the mass conservation equation, momentum conservation equation and energy conservation equation in step S1 are expressed as follows:

[0024] In three-dimensional space, the mass conservation equation for an incompressible fluid is written as: ▽·u=0;

[0025] Serving size:

[0026] Among them, u=(u x ,u y ,u z ) is the velocity vector of the fluid;

[0027] For an incompressible fluid, the equation for conservation of momentum in each direction is written as:

[0028] Serving size:

[0029] Among them, u i and u j represents the components of velocity in the i and j directions respectively; p represents pressure; ρ represents the density of the fluid; ν represents the dynamic viscosity of the fluid divided by its density; f i Represents the component of the external force acting on the unit volume of the fluid in the i direction;

[0030] For an incompressible fluid, the energy conservation equation is written as:

[0031]

[0032] Where e represents the total energy per unit mass; q represents the heat conduction flux; ϕ ​​represents the dissipation rate due to viscosity, Among them, τ ij is the component of the shear stress tensor; f represents the direction of the external force per unit volume.

[0033] Preferably, the equation for calculating the transport equation for controlling the turbulent kinetic energy k and the turbulent dissipation rate ε in the pipeline in step S2 is as follows:

[0034] Among them, the transport equation of turbulent kinetic energy k is:

[0035]

[0036] Where ρ is the fluid density; u i is the average velocity component; ν is the molecular viscosity; ν t is the eddy viscosity; σ k is the Prandtl turbulence number of the turbulent kinetic energy k; represents the rate of turbulent kinetic energy generation due to Reynolds stress; ε is the turbulent dissipation rate; S k is the user-defined source term one;

[0037] Turbulence dissipation rate equation:

[0038]

[0039] Among them, σ ε is the Prandtl turbulence number of the turbulent dissipation rate ε; C1 and C2 are empirical constants; S ε is the user-defined source term two.

[0040] Preferably, the interaction force between the fluid and the particles includes at least: buoyancy F B , drag force F D , and the pressure gradient force F P ;

[0041] Among them, the buoyancy F B The formula is: F B =V p (ρ f -ρ p )g;

[0042] Among them, V p is the volume of the particle; ρ f is the density of the fluid; ρ p is the density of the particle; g is the acceleration due to gravity;

[0043] Drag force F D The formula is:

[0044]

[0045] Among them, C D is the drag coefficient; ρ f is the density of the fluid; A is the cross-sectional area of ​​the particle; U g is the velocity of the particle; U f is the velocity of the fluid;

[0046] Drag coefficient C D According to the Reynolds number p To determine:

[0047] Where: d p is the particle diameter; ν f is the dynamic viscosity of the fluid;

[0048] Pressure gradient force F P The formula is: F P =-V p ▽p;

[0049] Among them, V p is the volume of the particle; ▽p is the pressure gradient;

[0050] The force on particles in the fluid is comprehensively expressed as: Ftotal =F B +F D +F P ;

[0051] Here: F total is the total force on the particle in the fluid;

[0052] The collision model equations between the fluid and the particles, and between the particles and the wall in step S3 are as follows:

[0053]

[0054] Where: e n represents the normal component; e t represents the tangential component; θ represents the impact angle when the particle collides with the wall.

[0055] Preferably, the erosion wear model in step S4 is:

[0056] Where: ER is the erosion wear rate, kg / (m 2 ·s); Np is the number of solid particles; m p is the mass flow rate of particles, kg / s; C(d p ) is the particle diameter function; f (θ) is the impact angle function; v is the particle impact velocity, m / s; b(v) is the velocity index; A face It is the unit impact area caused by a single particle on the wall.

[0057] Preferably, the erosion wear calculation in step S5 includes: the fluid control equation in step S1, extracting the calculation result in the turbulence simplified model in step S2 and the particle velocity Ug data in the collision equation in step S3, and substituting them into the erosion wear model in step S4 for calculation.

[0058] Preferably, the fluid mechanics analysis includes at least: flow state, velocity distribution, and pressure distribution;

[0059] The PINNs physical information neural network at least includes: a neural network model, a loss function L, and a residual;

[0060] The neural network model at least includes: input layer, hidden layer and output layer;

[0061] The loss function L at least includes: data loss L data and physical loss L pde + L bc +L ic ; Data loss L data Used to ensure that the network prediction data is consistent with the actual observation data, the physical loss Lpde + L bc +L ic It is used to ensure that the network prediction data meets the physical laws, boundary conditions bc and initial conditions ic of the fluid control equations.

[0062] Preferably, the PINNs physical information neural network learning comprises the following steps:

[0063] S6.1. Operation data collection; the operation data at least includes: fluid velocity and pressure in the pipeline, and wear thickness of the pipeline;

[0064] S6.2, run data preprocessing;

[0065] S6.3, dividing the training set and the test set; dividing the preprocessed running data into the training set data and the test set data;

[0066] S6.4, Multilayer Perceptron Model Construction;

[0067] S6.5, multi-layer perceptron model training: using the training set data to train the model, including: initializing the network, calculating the loss, optimizing, and iteratively updating;

[0068] S6.6. Multilayer perceptron model testing: Use the test set data to test the trained model and evaluate the performance of the model.

[0069] Beneficial effects of the present invention:

[0070] A numerical method for simulating the wear of ultra-long-distance mud pipelines is provided. By establishing the fluid control equation, the motion state of the fluid in the pipeline can be accurately described, providing basic data for the subsequent simplified turbulence model and erosion wear model. The simplified turbulence model can accurately simulate the complexity of turbulent flow in the pipeline and improve the accuracy of wear prediction. The collision equation can quantify the interaction between particles and fluids, particles and walls, and provide key parameters for the construction of erosion wear models. Through the erosion wear model and its calculation, the erosion wear rate of the pipeline wall can be quantitatively calculated, the wear of the pipeline can be evaluated in real time, potential problems can be discovered in time, and a scientific basis can be provided for pipeline maintenance and optimization. Through the PINNs physical information neural network, numerical simulation and machine learning can be combined, the actual pipeline operation data can be introduced, the erosion wear rate law can be obtained, the accuracy and efficiency of wear prediction can be improved, and intelligent pipeline wear management can be realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0071] Figure 1 The process of the present invention is shown in FIG. Figure 1 .

[0072] Figure 2 The process of the present invention is shown in FIG. Figure 2 .

[0073] Figure 3 Schematic diagram of the PINNs physical information neural network structure.

[0074] Figure 4 Flowchart for learning physical information neural networks (PINNs). DETAILED DESCRIPTION

[0075] The present invention will be further described below in conjunction with the embodiments.

[0076] A numerical method for simulating wear of ultra-long-distance mud pipelines, such as Figure 1 and Figure 4 As shown, the following steps are included:

[0077] S1. Constructing fluid control equations to provide a basic fluid dynamics description for the simplified turbulence model and erosion wear model in steps S2 and S4; including:

[0078] Mass conservation equation: describes the conservation relationship of fluid mass in the control volume, ensuring that the mass of the fluid does not change during the flow process;

[0079] Momentum conservation equation: describes the conservation relationship of fluid momentum in the control volume, taking into account the velocity change of the fluid and the external force;

[0080] Energy conservation equation: describes the conservation relationship of fluid energy in the control volume, taking into account the temperature change and energy transfer of the fluid;

[0081] By establishing the fluid control equation, the movement state of the fluid in the pipeline can be accurately described, providing basic data for the subsequent simplified turbulence model and erosion wear model.

[0082] S2. Construct a simplified turbulence model to calculate the transport equations controlling the turbulent kinetic energy k and the turbulent dissipation rate ε in the pipeline, including:

[0083] The transport equation of turbulent kinetic energy k: describes the transport process of turbulent kinetic energy in the fluid;

[0084] Turbulence dissipation rate ε: describes the rate at which turbulent kinetic energy is converted into heat energy;

[0085] The simplified turbulence model can accurately simulate the complexity of turbulent flow in the pipeline and improve the accuracy of wear prediction.

[0086] S3: Establish the collision equations between fluid and particles, and between particles and wall to simulate the movement and collision behavior of particles in the fluid, and provide the data of particle movement and collision for the erosion wear model in S4, including:

[0087] Fluid-particle collision equations: construct the motion and collision process of particles in the fluid, describe the inertia of particles and the drag force of the fluid;

[0088] Particle-wall collision equation: construct the collision process between particles and pipe walls, and describe the rebound and friction of particles;

[0089] Through the collision equation, the interaction between particles and fluid, particles and wall can be quantified, providing key parameters for the construction of erosion wear model.

[0090] S4. Construct an erosion wear model to simulate the pipeline wear mechanism. The erosion wear model at least includes: impact wear rate, scouring wear rate and fatigue wear rate; wherein:

[0091] Impact wear rate: describes the wear rate caused by particles impacting the pipe wall, taking into account the impact angle and impact force;

[0092] Erosion wear rate: describes the wear rate caused by fluid erosion of the pipe wall, taking into account the velocity and particle concentration of the fluid;

[0093] Fatigue wear rate: describes the fatigue wear rate caused by repeated impact of particles on the pipe wall, taking into account the impact frequency and material fatigue properties;

[0094] S5. Perform erosion wear calculation to obtain the erosion wear rate, including impact wear rate calculation, scour wear rate calculation and fatigue wear rate calculation. Through the erosion wear model and its calculation, the erosion wear rate of the pipeline wall can be quantitatively calculated, the wear of the pipeline can be evaluated in real time, potential problems can be discovered in time, and a scientific basis can be provided for pipeline maintenance and optimization.

[0095] S6. Introducing the pipeline operation data actually detected, including at least the fluid velocity, pressure, and wear thickness of the pipeline in the pipeline, into the model to achieve data correction;

[0096] S7. Conduct PINNs physical information neural network learning and fluid dynamics analysis, combine pipeline material properties, derive the law of erosion wear rate, and realize pipeline wear prediction. Through PINNs physical information neural network, it is possible to combine numerical simulation and machine learning, introduce actual pipeline operation data, derive the law of erosion wear rate, improve the accuracy and efficiency of wear prediction, and realize intelligent pipeline wear management.

[0097] In a further embodiment, the mass conservation equation describes that the mass of the fluid remains constant within a closed system. For an incompressible fluid, this means that the density of the fluid is constant, so the conservation of mass is simplified to the conservation of volume flow.

[0098] The mass conservation equation, momentum conservation equation, and energy conservation equation in step S1 are expressed as follows:

[0099] The mass conservation equation describes that the mass of a fluid remains constant within a closed system. For an incompressible fluid, this means that the density of the fluid is constant, so the conservation of mass simplifies to the conservation of volume flow. In three dimensions, the mass conservation equation for an incompressible fluid is written as: ▽·u=0;

[0100] Serving size:

[0101] Among them, u=(u x ,u y ,u z ) is the velocity vector of the fluid;

[0102] The momentum conservation equation describes the effects of forces inside and outside the fluid on the fluid. For an incompressible fluid, the momentum conservation equation in each direction is written as:

[0103] Serving size:

[0104] Among them, u i and u j represents the components of velocity in the i and j directions respectively; p represents pressure; ρ represents the density of the fluid; ν represents the dynamic viscosity of the fluid divided by its density; f i Represents the component of the external force acting on the unit volume of the fluid in the i direction;

[0105] The energy conservation equation describes the energy changes within the fluid, including changes in kinetic energy and internal energy. For incompressible fluids, the energy conservation equation is written as:

[0106] Where e represents the total energy per unit mass; q represents the heat conduction flux; ϕ ​​represents the dissipation rate due to viscosity, Among them, τ ij is the component of the shear stress tensor; f represents the direction of the external force per unit volume.

[0107] In a further embodiment,

[0108] The equation for calculating the transport equation for controlling the turbulent kinetic energy k in the pipeline and the turbulent dissipation rate ε in step S2 is as follows:

[0109] Among them, the transport equation of turbulent kinetic energy k is:

[0110] Where ρ is the fluid density; u i is the average velocity component; ν is the molecular viscosity; ν t is the eddy viscosity; σ kis the Prandtl turbulence number of the turbulent kinetic energy k; represents the rate of turbulent kinetic energy generation due to Reynolds stress; ε is the turbulent dissipation rate; S k is the user-defined source term one;

[0111] Turbulence dissipation rate ε equation:

[0112] Among them, σ ε is the Prandtl turbulence number of the turbulent dissipation rate ε; C1 and C2 are empirical constants; S ε is the user-defined source term two.

[0113] In a further embodiment, the interaction force between the fluid and the particles includes at least: buoyancy F B , drag force F D , and the pressure gradient force F P ;

[0114] Among them, the buoyancy F B It is caused by Archimedes' principle. When a particle is immersed in a fluid, it is subject to an upward force equal to the weight of the fluid displaced by the particle. The direction of the buoyancy is always upward. The buoyancy F B The formula is: F B =V p (ρ f -ρ p )g;

[0115] Among them, V p is the volume of the particle; ρ f is the density of the fluid; ρ p is the density of the particle; g is the acceleration due to gravity;

[0116] Drag force F D The drag force F is the resistance of the fluid to the particle, which hinders the movement of the particle. The magnitude of the drag force depends on the velocity of the particle relative to the fluid. D The formula is:

[0117]

[0118] Among them, C D is the drag coefficient; ρ f is the density of the fluid; A is the cross-sectional area of ​​the particle; U g is the velocity of the particle; U f is the velocity of the fluid;

[0119] Drag coefficient C D According to the Reynolds number p To determine:

[0120]

[0121] Where: d p is the particle diameter; ν f is the dynamic viscosity of the fluid;

[0122] When there is a pressure gradient in the fluid, the particle will be subjected to a force along the direction of the pressure gradient, namely, the pressure gradient force F P . Pressure gradient force F P The formula is: F P =-V p ▽p;

[0123] Among them, V p is the volume of the particle; ▽p is the pressure gradient;

[0124] The force on particles in the fluid is comprehensively expressed as: F total =F B +F D +F P ;

[0125] Here: F total is the total force on the particle in the fluid;

[0126] In order to construct the collision model between particles and tube wall, the concept of collision elastic recovery coefficient is introduced, and e n (normal component) and e t (Tangential component) represents the momentum change in the normal and tangential directions before and after the collision. Grant and Tabakoff conducted a large number of collision experiments and summarized a collision model with strong applicability, which is also the collision model used in this paper. The collision model equations between the fluid and the particle, and between the particle and the wall in step S3 are as follows:

[0127] Where: e n represents the normal component; e t represents the tangential component; θ represents the impact angle when the particle collides with the wall.

[0128] In a further embodiment, there are various erosion wear models for calculating the numerical value of wear, which can accurately predict the wear of the pipe wall. The erosion theory model provided by the fluid dynamics analysis software Fluen is used for research. The erosion wear model in step S4 is:

[0129] Where: ER is the erosion wear rate, kg / (m 2 ·s); Np is the number of solid particles; m p is the mass flow rate of particles, kg / s; C(d p) is the particle diameter function; f (θ) is the impact angle function; v is the particle impact velocity, m / s; b(v) is the velocity index; A face is the unit impact area caused by a single particle on the wall; where f(θ) is a piecewise function, and the specific values ​​are shown in the table:

[0130] In a further embodiment, the calculation of the erosion and wear model in step S5 includes: extracting the fluid motion control equation of step S1, the calculation results (fluid velocity field, pressure field) in the simplified turbulence model in step S2, and the particle velocity Ug data in the collision equation of step S3, and substituting them into the erosion and wear model in step S4 for calculation.

[0131] In a further embodiment, in step S6, the operation data of the measured pipeline at the current moment is obtained; the operation data is divided into a training set and a test set, and the operation data of the training set and the test set at least include: the fluid velocity and pressure in the pipeline, and the thickness of the pipeline, which are input into a multi-layer perceptron to train the erosion wear rate; the test set is used to predict the accuracy of the judgment model to obtain the thickness of the measured pipeline at the current moment; according to the thickness of the measured pipeline at the current moment, the erosion wear rate of the measured pipeline at the current moment is determined.

[0132] In a further embodiment, the fluid mechanics analysis includes at least: flow state, velocity distribution, and pressure distribution;

[0133] PINNs physical information neural network at least includes: neural network model, loss function L, residual;

[0134] The neural network model includes at least: This is the basic component, usually composed of multiple layers, including input layer, hidden layer and output layer. The input can be time and space coordinates, and the output is the physical quantity to be solved (such as temperature, pressure, etc.).

[0135] The loss function L at least includes: data loss L data and physical loss L pde + L bc +L ic ; Data loss L data Used to ensure that the network prediction data is consistent with the actual observation data, the physical loss L pde + L bc +L ic It is used to ensure that the network prediction data meets the physical laws, boundary conditions bc and initial conditions ic of the fluid control equations.

[0136] Residual: In PINNs, the residual is the result of subtracting the expression on the right from the expression on the left side of the physical equation. Ideally, when the network training is complete, the residual should be close to zero, which means that the network's predictions are consistent with the laws of physics.

[0137] In a further embodiment, PINNs physical information neural network learning comprises the following steps:

[0138] S6.1. Operation data collection: obtain the operation data of the tested pipeline at the current moment. The operation data at least includes: fluid velocity, pressure, and thickness of the pipeline. These data are the basis for training PINNs and are used for subsequent model training and verification.

[0139] S6.2. Run data preprocessing; perform preprocessing operations such as cleaning, normalization, and interpolation on the collected data to ensure the quality and consistency of the data and facilitate subsequent model training.

[0140] S6.3. Divide the preprocessed running data into training set data and test set data. The training set is used for model training, and the test set is used for model verification and evaluation to ensure that the model has good generalization ability.

[0141] S6.4. Multilayer perceptron model construction: Select a suitable neural network architecture, usually a multilayer perceptron, and define the number of layers, the number of neurons in each layer, the activation function, etc. This step determines the basic structure of the model.

[0142] S6.5. Multilayer perceptron model training: Use the training set data to train the model, including: initializing the network (randomly initializing the parameters of the neural network), calculating the loss (using the current network parameters to calculate the predicted value, and based on this, calculating the data loss and physical loss), optimizing (adjusting the network parameters through the back propagation algorithm to minimize the loss function), iterative updating (repeatedly calculating the loss and optimizing steps until a predetermined stopping criterion is reached, such as the number of iterations or the loss threshold);

[0143] S6.6, Multilayer Perceptron Model Testing: Use the test set data to test the trained model, evaluate the model performance, and determine whether the model has good generalization ability by calculating the loss value, accuracy and other indicators on the test set. Through training and testing, the model can eventually output the law of erosion wear rate consistent with the fluid mechanics state (such as velocity distribution and pressure distribution). This law can be used to predict the wear condition in the pipeline, guide engineering design and maintenance, and improve the safety and reliability of the system.

[0144] The above is a detailed introduction to a numerical method for simulating ultra-long distance mud pipeline wear provided by the present invention. It is worth noting that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical details described in the aforementioned embodiments, or to make equivalent replacements for some of the technical features therein. Any equivalent replacements, modifications, etc. made within the core ideas and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A numerical method for simulating wear of ultra-long distance mud pipelines, characterized in that: The following steps are involved: S1. Constructing fluid control equations to provide a basic fluid dynamics description for the simplified turbulence model and erosion wear model in steps S2 and S4; including: Mass conservation equation: describes the conservation relationship of fluid mass in the control volume, ensuring that the mass of the fluid does not change during the flow process; Momentum conservation equation: describes the conservation relationship of fluid momentum in the control volume, taking into account the velocity change of the fluid and the external force; Energy conservation equation: describes the conservation relationship of fluid energy in the control volume, taking into account the temperature change and energy transfer of the fluid; S2. Construct a simplified turbulence model to calculate the transport equations controlling the turbulent kinetic energy k and the turbulent dissipation rate ε in the pipeline, including: The transport equation of turbulent kinetic energy k: describes the transport process of turbulent kinetic energy in the fluid; Turbulence dissipation rate ε: describes the rate at which turbulent kinetic energy is converted into heat energy; S3: Establish the collision equations between fluid and particles, and between particles and wall to simulate the movement and collision behavior of particles in the fluid, and provide the data of particle movement and collision for the erosion wear model in S4, including: Fluid-particle collision equations: construct the motion and collision process of particles in the fluid, describe the inertia of particles and the drag force of the fluid; Particle-wall collision equation: construct the collision process between particles and pipe walls, and describe the rebound and friction of particles; S4. Construct an erosion wear model to simulate the pipeline wear mechanism, wherein the erosion wear model at least includes: impact wear rate, scouring wear rate and fatigue wear rate, wherein: Impact wear rate: describes the wear rate caused by particles impacting the pipe wall, taking into account the impact angle and impact force; Erosion wear rate: describes the wear rate caused by fluid erosion of the pipe wall, taking into account the velocity and particle concentration of the fluid; Fatigue wear rate: describes the fatigue wear rate caused by repeated impact of particles on the pipe wall, taking into account the impact frequency and material fatigue properties; S5. Calculate the erosion wear rate of the pipeline wall according to the erosion wear model to obtain the erosion wear rate, which is used to quantify the wear degree of the pipeline wall and provide data support for subsequent wear prediction; S6, introducing the pipeline operation data actually detected, including at least the fluid velocity, pressure, and wear thickness of the pipeline in the pipeline, into the model to achieve data correction; S7. Conduct PINNs physical information neural network learning and fluid dynamics analysis, combine pipeline material properties, derive the law of erosion wear rate, and realize pipeline wear prediction.

2. A numerical method for simulating wear of ultra-long distance mud pipelines according to claim 1, characterized in that: The mass conservation equation, momentum conservation equation and energy conservation equation in step S1 are expressed as follows: In three-dimensional space, the mass conservation equation for an incompressible fluid is written as: ▽·u=0; Serving size: Among them, u=(u x ,u y ,u z ) is the velocity vector of the fluid; For an incompressible fluid, the equation for conservation of momentum in each direction is written as: Serving size: Among them, u i and u j represents the components of velocity in the i and j directions respectively; p represents pressure; ρ represents the density of the fluid; ν represents the dynamic viscosity of the fluid divided by its density; f i Represents the component of the external force acting on the unit volume of the fluid in the i direction; For an incompressible fluid, the energy conservation equation is written as: Where e represents the total energy per unit mass; q represents the heat conduction flux; ϕ ​​represents the dissipation rate due to viscosity, Among them, τ ij are the components of the shear stress tensor; f represents the direction of the external force per unit volume.

3. A numerical method for simulating wear of ultra-long distance mud pipelines according to claim 2, characterized in that: The equation for calculating the transport equation for controlling the turbulent kinetic energy k in the pipeline and the turbulent dissipation rate ε in step S2 is as follows: Among them, the transport equation of turbulent kinetic energy k is: Where ρ is the fluid density; u i is the average velocity component; ν is the molecular viscosity; ν t is the eddy viscosity; σ k is the Prandtl turbulence number of the turbulent kinetic energy k; represents the rate of turbulent kinetic energy generation due to Reynolds stress; ε is the turbulent dissipation rate; S k is the user-defined source term one; Turbulence dissipation rate equation: Among them, σ ε is the Prandtl turbulence number of the turbulent dissipation rate ε; C1 and C2 are empirical constants; S ε is the user-defined source term two.

4. A numerical method for simulating wear of ultra-long distance mud pipelines according to claim 3, characterized in that: The interaction forces between fluid and particles include at least: buoyancy F B , drag force F D , and the pressure gradient force F P ; Among them, the buoyancy F B The formula is: F B =V p (ρ f -ρ p )g; Among them, V p is the volume of the particle; ρ f is the density of the fluid; ρ p is the density of the particle; g is the acceleration due to gravity; Drag force F D The formula is: Among them, C D is the drag coefficient; ρ f is the density of the fluid; A is the cross-sectional area of ​​the particle; U g is the velocity of the particle; U f is the velocity of the fluid; Drag coefficient C D According to the Reynolds number p To determine: Where: d p is the particle diameter; ν f is the dynamic viscosity of the fluid; Pressure gradient force F P The formula is: F P =-V p ▽p; Among them, V p is the volume of the particle; ▽p is the pressure gradient; The force on particles in the fluid is comprehensively expressed as: F total =F B +F D +F P ; Here: F total is the total force on the particle in the fluid; The collision model equations between the fluid and the particles, and between the particles and the wall in step S3 are as follows: Where: e n represents the normal component; e t represents the tangential component; θ represents the impact angle when the particle collides with the wall.

5. A numerical method for simulating wear of ultra-long distance mud pipelines according to claim 4, characterized in that: The erosion wear model in step S4 is: Where: ER is the erosion wear rate, kg / (m 2 ·s); Np is the number of solid particles; m p is the mass flow rate of particles, kg / s; C(d p ) is the particle diameter function; f (θ) is the impact angle function; v is the particle impact velocity, m / s; b(v) is the velocity index; A face It is the unit impact area caused by a single particle on the wall.

6. A numerical method for simulating wear of ultra-long distance mud pipelines according to claim 5, characterized in that: The erosion wear calculation in step S5 includes: the fluid control equation in step S1, extracting the calculation result in the turbulence simplified model in step S2 and the particle velocity Ug data in the collision equation in step S3, and substituting them into the erosion wear model in step S4 for calculation.

7. The numerical method for simulating wear of ultra-long distance mud pipeline according to claim 1 is characterized in that: The fluid dynamics analysis includes at least: flow state, velocity distribution, and pressure distribution; The PINNs physical information neural network at least includes: a neural network model, a loss function L, and a residual; The neural network model at least includes: input layer, hidden layer and output layer; The loss function L at least includes: data loss L data and physical loss L pde + L bc +L ic ; Data loss L data Used to ensure that the network prediction data is consistent with the actual observation data, the physical loss L pde + L bc +L ic It is used to ensure that the network prediction data meets the physical laws, boundary conditions bc and initial conditions ic of the fluid control equations.

8. A numerical method for simulating wear of ultra-long distance mud pipelines according to claim 7, characterized in that: The PINNs physical information neural network learning includes the following steps: S6.

1. Operation data collection; the operation data at least includes: fluid velocity and pressure in the pipeline, and wear thickness of the pipeline; S6.2, run data preprocessing; S6.3, dividing the training set and the test set; dividing the preprocessed running data into the training set data and the test set data; S6.4, Multilayer Perceptron Model Construction; S6.5, multi-layer perceptron model training: using the training set data to train the model, including: initializing the network, calculating the loss, optimizing, and iteratively updating; S6.

6. Multilayer perceptron model testing: Use the test set data to test the trained model and evaluate the performance of the model.

Images