Vulnerability analysis method of pipeline joints under fault action based on machine learning

Abstract

The invention provides a method for analyzing the vulnerability of a pipe joint under a fault based on machine learning, and belongs to the technical field of machine learning. The method solves the problems of poor accuracy, long time consumption and low efficiency in predicting the damage of a pipe joint under a fault by using the prior art. A nonlinear pipe-soil model is established by using a pipe-soil interaction unit (PSI), and factors such as pipe diameter, pipe wall thickness, pipe burial depth, pipe-soil interaction stiffness, and fault offset are used as model input variables, and the pipe joint rotation angle is used as an output variable. Secondly, a lasso regression model and a logistic regression model are trained by using a large amount of simulation data in combination with machine learning technology, so as to accurately predict the damage of a pipe joint under a fault. The invention is applied to the damage analysis of a pipe joint under a fault.

Description

Technical Field

[0001] The invention provides a method for analyzing the vulnerability of a pipeline joint under fault action based on machine learning, and belongs to the technical field of machine learning. Background Art

[0002] Buried pipelines play a vital role in modern society, not only providing necessary materials for people's lives and economic activities, but also promoting economic development, environmental protection and social stability. However, in recent years, buried pipeline accidents caused by foundation displacement and deformation such as faults have occurred frequently, posing a serious threat to people's lives and property safety. Studies have shown that pipeline joint damage is one of the main causes of accidents.

[0003] At present, the technologies for predicting the damage of pipeline joints under faults mainly include empirical formulas and finite element numerical simulation analysis. The former is based on experimental data and empirical laws. By considering factors such as the geometry of the joint, material properties and fault parameters, a simple mathematical formula is derived to predict the damage of the pipeline. This method has the advantages of being simple and intuitive, but it is highly subjective and has certain limitations in accuracy. The latter usually uses a three-dimensional numerical model to simulate the structure and mechanical behavior of the pipeline joint. This method can consider the nonlinearity of the material and the influence of the pipe-soil contact surface. However, due to the complex mechanical response of the pipeline joint under the action of faults, the three-dimensional numerical simulation analysis is very time-consuming. For a given combination of input parameters, it usually takes days or even weeks to complete the calculation.

[0004] In view of the above problems, the present invention proposes a method for analyzing the vulnerability of pipeline joints under faults based on machine learning. This method can provide calculation results in a shorter time without affecting the accuracy of predicting pipeline joint damage, and the selection of input parameters through machine learning algorithms is more scientific, avoiding the problem of strong subjectivity of human selection. Summary of the invention

[0005] In order to solve the problems of poor accuracy, long time consumption and low efficiency in predicting the damage of pipeline joints under faults by using the existing technology, the present invention proposes a vulnerability analysis method of pipeline joints under faults based on machine learning.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is: a vulnerability analysis method of pipeline joints under fault action based on machine learning, comprising the following steps:

[0007] Step 1, obtaining pipeline geometric size parameters, pipeline material parameters, pipeline laying parameters, earthquake fault parameters and pipe-soil interaction parameters;

[0008] Step 2: Based on the pipeline geometric size parameters, pipeline material parameters, pipeline laying parameters, seismic fault parameters and pipe-soil interaction parameters, the pipe-soil interaction unit PSI is used to write an INP file that can be calculated and run by the finite element software ABAQUS, and the rotation angle of the pipeline joint under the action of the fault is calculated, and the calculation results of the INP file are verified using the known pipe joint rotation angle data;

[0009] Step 3, among the pipeline geometric size parameters, pipeline material parameters, pipeline laying parameters, seismic fault parameters and pipe-soil interaction parameters, one parameter is selected as an independent variable to perform vulnerability analysis on the pipeline joint under the action of the fault, and Latin hypercube sampling is used for the remaining parameters under the assumed probability function distribution to randomly generate multiple sets of parameter sampling values, and then the multiple sets of parameter sampling values ​​are written into the INP file obtained in step 2 and calculated to obtain the corresponding multiple pipeline joint rotation angle values;

[0010] Step 4: Use the multiple groups of parameter sampling values ​​obtained in step 3 and the multiple pipe joint angle values ​​calculated by the INP file as training sample data sets to train the lasso regression model;

[0011] Step 5, repeat step 3 to randomly generate multiple sets of parameter sampling values, and input them into the lasso regression model obtained in step 4 to calculate the corresponding multiple pipe joint angle values;

[0012] Step 6: based on the known pipeline joint angle damage value, the multiple pipeline joint angle values ​​obtained in step 5 are converted into a set of binary life and death vectors with multiple elements;

[0013] Step 7: Using the sample values ​​of the multiple groups of parameters in step 5 and a group of binary life and death vectors with multiple elements obtained in step 6 as training sample data sets, training the logistic regression model, and calculating the coefficients of the multi-parameter vulnerability function;

[0014] Step 8: Perform convolution calculation on the multi-parameter vulnerability function and the probability density function corresponding to each parameter to obtain the vulnerability curve of pipeline joint damage.

[0015] The pipeline geometric size parameters include pipeline length, pipeline diameter and pipeline wall thickness, the pipeline material parameters include pipeline elastic modulus and pipe material Poisson's ratio, the pipeline laying parameters include pipeline burial depth, the earthquake fault parameters include fault dip and fault offset, and the pipe-soil interaction parameters include T u The corresponding pipe-soil axial stiffness, Q u The corresponding pipe-soil vertical stiffness, Q d The corresponding pipe-soil vertical stiffness, where T uQ is the soil restraint force on the buried pipeline when it undergoes axial displacement relative to the surrounding soil; u Q is the soil restraint force on the buried pipeline when it moves upward relative to the soil; d It is the soil restraint force on the buried pipeline when it moves downward relative to the soil.

[0016] In step 2, the INP file can be calculated and run by the finite element software ABAQUS, and the command flow of the INP file is used to define the pipe joint and the behavior characteristics of the pipe joint, specifically including: the definition of the element set constituting the pipe joint, the definition of the mechanical behavior of the pipe joint, the definition of whether the pipe joint is a rigid connection, and the definition of the rotational freedom constraint of the pipe joint on the X-axis and the Y-axis, wherein the element set of the pipe joint includes two nodes constituting the pipe joint and a two-node connection unit constituting the pipe joint.

[0017] The step 6 specifically includes the following steps:

[0018] Step 6.1, among the multiple pipe joint angle values ​​obtained in step 5, the data which is greater than or equal to the pipe joint angle damage value is recorded as 1;

[0019] Step 6.2: Among the multiple pipe joint angle values ​​obtained in step 5, the data that is smaller than the pipe joint angle damage value is recorded as 0.

[0020] The beneficial effects of the present invention compared to the prior art are as follows: the vulnerability analysis method of pipeline joints under faults based on machine learning provided by the present invention explores the influence of different parameters on pipeline joints under faults of buried pipelines through powerful data mining of machine learning algorithms, and accurately gives the vulnerability curve of pipeline joints. By performing vulnerability analysis on pipeline joints, it is possible to predict the corner damage of buried pipelines under faults, so as to achieve timely inspection and maintenance, reduce the probability of accidents, and have important significance for improving the management and operation of pipeline systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The present invention will be further described below in conjunction with the accompanying drawings:

[0022] Figure 1 Flow chart of the method of the present invention.

[0023] Figure 2 This is the pipe-soil axial interaction model diagram.

[0024] Figure 3 This is the pipe-soil vertical interaction model diagram.

[0025] Figure 4 Schematic diagram of the pipeline and PSI unit in step 2 of the present invention.

[0026] Figure 5 It is a comparison chart of the INP calculation results of the embodiment of the present invention and the model test data.

[0027] Figure 6 It is a schematic diagram of 200 pipe joint angle values ​​calculated by INP file in an embodiment of the present invention.

[0028] Figure 7 1 is a schematic diagram of one million pipe joint rotation angle values ​​calculated by the lasso regression model in an embodiment of the present invention.

[0029] Figure 8 It is a schematic diagram of converting one million pipe joint angle values ​​into a set of binary life and death vectors with one million elements through the method of the present invention.

[0030] Fig. 9 It is a fragility curve diagram of pipeline joints obtained by simulation using the method of the present invention. DETAILED DESCRIPTION

[0031] like Figures 1 to 9 As shown, the present invention is further described according to the following specific embodiments.

[0032] like Figure 1 As shown, the present invention proposes a method for analyzing the vulnerability of a pipeline joint under fault action based on machine learning, comprising the following steps:

[0033] Step 1, obtaining pipeline geometric size parameters, pipeline material parameters, pipeline laying parameters, earthquake fault parameters and pipe-soil interaction parameters;

[0034] Specifically, in this embodiment, the pipeline geometric size parameters include: the HDPE pipeline length is 7100mm, the pipeline diameter is 600mm, and the pipeline wall thickness is 3.5mm; the pipeline material parameters include: the pipeline elastic modulus is 600MPa, and the Poisson's ratio is 0.3; the pipeline burial depth in the pipeline laying parameters is 1.2m; the earthquake fault parameters include: the fault dip is 90°, and the fault offset is 0-100mm; the pipe-soil interaction parameters include: T u The corresponding pipe-soil axial stiffness, Q u The corresponding pipe-soil vertical stiffness, Q d The corresponding pipe-soil vertical stiffness, where T u Q is the soil restraint force on the buried pipeline when it undergoes axial displacement relative to the surrounding soil; u Q is the soil restraint force on the buried pipeline when it moves upward relative to the soil; d It is the soil restraint force on the buried pipeline when it moves downward relative to the soil.

[0035] For the pipe-soil axial interaction model, Figure 2 As shown in the figure, according to the ALA specification, when the buried pipeline undergoes axial displacement relative to the surrounding soil, the soil restraint force it is subjected to can be expressed by the formula Calculation, where: γ is the density of the soil (kN / m 3 );H s is the buried depth of the pipeline (m); D is the diameter of the pipeline (m); K0 is the static earth pressure coefficient, which is Calculated; is the friction angle of the soil body)°); f′ is the surface correction coefficient of the pipeline material. In this embodiment, γ, The values ​​of f′ are 16.3 kN / m 3 , 42°, 0.6, calculated T u =11.5kN / m, the corresponding relative displacement of pipe and soil is taken as t u =3mm;

[0036] For the pipe-soil vertical interaction model, Figure 3 As shown in the figure, the soil restraint force on the buried pipeline when it moves upward relative to the soil can be expressed by the formula Q u =N qv γH s D calculation, where: N qv is the correction factor, which can be expressed as Calculate, after calculating Q u =22.4kN / m, the corresponding relative displacement of pipe and soil is taken as q u =0.01H s =12mm; When the buried pipeline moves downward relative to the soil, the soil restraint force it is subjected to can be expressed by the formula Q d =N q γH s D+N γ γD 2 / 2 calculation, where N q 、N γ Are correction coefficients, which can be respectively expressed by the formula and Calculate, after calculating Q d =1457.6kN / m, the corresponding relative displacement of pipe and soil is taken as q d =0.1D=60mm. In this embodiment, the pipe-soil interaction stiffness is used to characterize the interaction force between the pipe and the soil. According to the relationship formula between stiffness and force: We can find T u The corresponding pipe-soil axial interaction Same reason Q u The corresponding pipe-soil vertical interaction Q d The corresponding pipe-soil vertical interaction

[0037] Step 2: Write the INP file of finite element software ABAQUS according to pipeline geometric size parameters, material parameters, earthquake fault parameters, pipeline laying parameters and pipe-soil interaction parameters. Specifically, it includes the following steps:

[0038] Step 2.1. In the INP file, establish a buried pipeline model based on the geometric and physical and mechanical parameters of the pipeline (pipeline length, diameter, wall thickness, elastic modulus, Poisson's ratio), divide the units, and assign pipeline material properties;

[0039] Step 2.2: Establish the pipe-soil interaction element (PSI), such as Figure 4 As shown, one side of the PSI unit shares nodes (node ​​1 and node 2) with the pipeline unit, and the nodes on the other side (node ​​2 and node 3) represent the far-field surface. ABAQUS connects the far-field side of the PSI unit and the side of the pipeline unit sharing the nodes by defining the unit connection properties.

[0040] Step 2.3: Figure 2 and Figure 3 The pipe-soil interaction model parameters are shown. The nonlinear anisotropic constitutive model of the PSI element in the axial and vertical directions is defined using the following command:

[0041] *PIPE-SOIL STIFFNESS, TYPE=NONLINEAR, DIRECTION=direction;

[0042] The direction parameter can be axial or vertical. Repeat the DIRECTION parameter of this command multiple times to define the behavior characteristics of the constitutive model in each direction.

[0043] Step 2.4: Based on the connector definition, use the following command to define the pipe joint and its behavior characteristics:

[0044] *Connector Section, elset=Wire-1-Set-1, behavior=ConnSect-1

[0045] *Nset, nset = Wire-1-Set-1

[0046] 201,

[0047] *Nset,nset=Wire-1-Set-1

[0048] 202,

[0049] *Elset,elset=Wire-1-Set-1 2222

[0051] *Connector Behavior,name=ConnSect-1

[0052] *Connector Elasticity, rigidity

[0053] 4,5

[0054] In this embodiment, "Wire-1-Set-1" represents the element set constituting the pipe joint, which is composed of node 201, node 202 and unit 2222, where "201, 202" are two nodes constituting the pipe joint, and "2222" is a two-node connection unit constituting the pipe joint; "ConnSect-1" refers to the mechanical behavior of the pipe joint, where "Connector Elasticity, rigid" means that the pipe joint is regarded as a rigid connection, and "4" and "5" represent that the rotational freedom of the pipe joint on the X-axis and Y-axis is constrained.

[0055] Step 2.5. Add the *Nodefile statement after the *Restart statement to output the calculation results of the pipeline joint node displacement to a file with the file name suffix .fil. Figure 5 This is a comparison chart of the pipe joint rotation angle calculated by the INP file in this embodiment and the angle measured by the model test published by Zhou Min in "Research on Mechanical Response Characteristics of Buried HDPE Pipeline Induced by Uneven Foundation Settlement". Figure 5 It can be seen that the error between the result of INP calculation and the actual measured value of the model is small, within 5%, which shows that the pipe joint angle calculated by the INP file in this embodiment is effective.

[0056] Step 3: Among the pipeline geometric dimension parameters, pipeline material parameters, pipeline laying parameters, earthquake fault parameters and pipe-soil interaction parameters, this embodiment selects the pipeline diameter in the geometric dimension parameters as the independent variable to perform vulnerability analysis on the pipeline joint under the action of the fault. For the remaining parameters, Latin hypercube sampling is used under the assumed probability function distribution (Table 1) to randomly generate 200 sets of sampling values. Then, the sampling values ​​of these 200 sets of parameters are written into the INP file obtained in step 2 and calculated to obtain the corresponding 200 pipeline joint rotation angle values ​​as shown in FIG. Figure 6 The probability function distribution of each parameter in this embodiment is shown in Table 1:

[0057] parameter unit Probability function distribution average value Standard Deviation Pipe length m Constant 7.1 - Pipe diameter mm Constant 600 - Pipe wall thickness mm Constant 3.5 - Tube elastic modulus MPa Constant 600 - Poisson's ratio of pipe - Constant 0.3 - Pipeline burial depth m Constant 1.2 - Fault dip ° Constant 90 - Fault offset mm Average distribution [0，100] - <![CDATA[T u Corresponding pipe-soil axial stiffness]]> kN / m / m Gaussian distribution 3833.3 1149.9 <![CDATA[Q u Corresponding pipe-soil vertical stiffness]]> kN / m / m Gaussian distribution 1866.7 560.0 <![CDATA[Q d The corresponding pipe-soil stiffness]]> kN / m / m Gaussian distribution 24293.3 7287.9

[0058] Table 1 Probability functions of various parameters.

[0059] Step 4: The two hundred sets of parameter sampling values ​​in step 3 and the two hundred pipe joint angle values ​​calculated by the INP file are used as training sample data sets. The lasso regression model is trained using the training sample data sets. The formula is as follows:

[0060]

[0061] In the above formula: N is the number of samples, β0 is the intercept, x j is the input parameter, β j is the coefficient of the input parameter, p is the number of input parameters, λ is the scaling factor, d i is the required parameter.

[0062] Specifically, firstly, the two hundred groups of parameter sampling values ​​in step 3 are normalized, and then these 200 groups of normalized values ​​and the two hundred pipe joint angle values ​​calculated by the INP file are used as training sample data sets to train the lasso regression model to obtain a trained lasso regression model. The number of samples N in this embodiment is 200; the number of input parameters p is 11; the input parameter x j They are pipeline length, pipeline diameter, pipeline wall thickness, pipe elastic modulus, pipe Poisson's ratio, pipeline burial depth, fault inclination, T u The corresponding pipe-soil axial stiffness, Q u The corresponding pipe-soil vertical stiffness, Q d The corresponding pipe-soil vertical stiffness, fault offset, and the coefficient β corresponding to the above input parameters i They are 0, 0, 0, 0, 0, 0, 0, 0, 0.022, 0.199, and 1.308 respectively; the intercept β0 is 1.31; and the scaling factor λ is 0.01.

[0063] Step 5: Repeat step 3 to randomly generate one million sets of parameter sampling values, perform normalization, and then input them into the lasso regression model obtained in step 4 to calculate the corresponding one million pipe joint angle values, such as Figure 7 shown.

[0064] Step 6: Based on the pipe joint angle damage values ​​in the published "Study on Mechanical Response Characteristics of Buried HDPE Pipelines Induced by Uneven Foundation Settlement", the one million pipe joint angle values ​​obtained in step 5 are converted into a set of binary life and death vectors with one million elements, such as Figure 8 As shown. Specifically including the following steps:

[0065] Step 6.1. Among the one million pipe joint angle values ​​obtained in step 5, the data that is greater than or equal to the pipe joint angle damage value in the published document "Study on Mechanical Response Characteristics of Buried HDPE Pipelines Induced by Uneven Foundation Settlement" is recorded as 1.

[0066] Step 6.2: Among the one million pipe joint angle values ​​obtained in step 5, the data that is smaller than the pipe joint angle damage value in the published document "Study on Mechanical Response Characteristics of Buried HDPE Pipelines Induced by Uneven Foundation Settlement" is recorded as 0.

[0067] According to the model test data in the published "Study on Mechanical Response Characteristics of Buried HDPE Pipes Induced by Uneven Foundation Settlement", when the rotation angle of the HDPE pipe joint reaches 0.58°, the joint is damaged. In this embodiment, when the rotation angle of the pipe joint is greater than or equal to 0.58°, it is recorded as 1; when the rotation angle of the pipe joint is less than 0.58°, it is recorded as 0.

[0068] Step 7: Use the one million sets of parameter sampling values ​​normalized in step 5 and a set of binary life and death vectors with one million elements obtained in step 6 as training sample data sets to train the logistic regression model and calculate the coefficients of the multi-parameter vulnerability function. The calculation formula of the logistic regression model is as follows:

[0069]

[0070] In the above formula, p l is the number of important input parameters, x j are important input parameters, θ0 is the intercept, θ δ and θ j are the coefficients of important input parameters.

[0071] Specifically, the important input parameter p in this embodiment is l The number of is 4; the important input parameter x i T u The corresponding pipe-soil axial stiffness, Q u The corresponding pipe-soil vertical stiffness, Q d The corresponding pipe-soil vertical stiffness and fault offset, the coefficients θ1, θ2, θ3, θ δ They are 0.162, 2.615, 22.281, and 146.553 respectively; the intercept θ0 is 194.194.

[0072] Step 8: Using formula (3), perform convolution calculation on the multi-parameter vulnerability function and the probability density function corresponding to each parameter to obtain the vulnerability curve of pipeline joint damage.

[0073]

[0074] Where f(x1)...f(x pl ) is the probability density function corresponding to each important parameter.

[0075] in, T u The corresponding probability density function of the pipe-soil axial interaction stiffness; Q u The corresponding probability density function of the pipe-soil vertical interaction stiffness; Q d The probability density function of the corresponding pipe-soil vertical interaction stiffness.

[0076] In this embodiment, f(x1), f(x2), f(x3) are convolved with the multi-parameter vulnerability function obtained in step 7 to obtain the pipeline joint vulnerability curve as shown in FIG. Fig. 9 As shown. Fig. 9 It can be seen that as the fault offset increases, the failure probability of the pipe joint also increases. For example, when the fault offset is 10 mm, the failure probability of the pipe joint is 39.5%; and when the fault offset increases to 20 mm, the failure probability of the pipe joint reaches 98.1%.

[0077] The present invention establishes a nonlinear pipe-soil model by using a pipe-soil interaction unit (PSI), taking factors such as pipe diameter, pipe wall thickness, pipe burial depth, pipe-soil interaction stiffness, fault offset and the like as model input variables, and the pipe joint angle as the output variable. Secondly, in combination with machine learning technology, a large amount of simulation data is used to train a lasso regression model and a logistic regression model to accurately predict the damage of pipe joints under the action of faults. Compared with the three-dimensional finite element numerical simulation method, the method proposed in the present invention can save a lot of calculation time and improve the prediction efficiency without affecting the accuracy of predicting the damage of pipe joints, and has strong practicality and economic benefits.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A vulnerability analysis method for pipeline joints under fault action based on machine learning, characterized in that: The steps include: Step 1, obtaining pipeline geometric size parameters, pipeline material parameters, pipeline laying parameters, earthquake fault parameters and pipe-soil interaction parameters; Step 2: Based on the pipeline geometric size parameters, pipeline material parameters, pipeline laying parameters, seismic fault parameters and pipe-soil interaction parameters, the pipe-soil interaction unit PSI is used to write an INP file that can be calculated and run by the finite element software ABAQUS, and the rotation angle of the pipeline joint under the action of the fault is calculated, and the calculation results of the INP file are verified using the known pipe joint rotation angle data; Step 3, among the pipeline geometric size parameters, pipeline material parameters, pipeline laying parameters, seismic fault parameters and pipe-soil interaction parameters, one parameter is selected as an independent variable to perform vulnerability analysis on the pipeline joint under the action of the fault, and Latin hypercube sampling is used for the remaining parameters under the assumed probability function distribution to randomly generate multiple sets of parameter sampling values, and then the multiple sets of parameter sampling values ​​are written into the INP file obtained in step 2 and calculated to obtain the corresponding multiple pipeline joint rotation angle values; Step 4: Use the multiple groups of parameter sampling values ​​obtained in step 3 and the multiple pipe joint angle values ​​calculated by the INP file as training sample data sets to train the lasso regression model; Step 5, repeat step 3 to randomly generate multiple sets of parameter sampling values, and input them into the lasso regression model obtained in step 4 to calculate the corresponding multiple pipe joint angle values; Step 6: based on the known pipeline joint angle damage value, the multiple pipeline joint angle values ​​obtained in step 5 are converted into a set of binary life and death vectors with multiple elements; Step 7: Using the sample values ​​of the multiple groups of parameters in step 5 and a group of binary life and death vectors with multiple elements obtained in step 6 as training sample data sets, training the logistic regression model, and calculating the coefficients of the multi-parameter vulnerability function; Step 8: Perform convolution calculation on the multi-parameter vulnerability function and the probability density function corresponding to each parameter to obtain the vulnerability curve of pipeline joint damage.

2. The vulnerability analysis method of pipeline joints under fault action based on machine learning according to claim 1 is characterized in that: The pipeline geometric size parameters include pipeline length, pipeline diameter and pipeline wall thickness, the pipeline material parameters include pipeline elastic modulus and pipe material Poisson's ratio, the pipeline laying parameters include pipeline burial depth, the earthquake fault parameters include fault dip and fault offset, and the pipe-soil interaction parameters include The corresponding pipe-soil axial stiffness, The corresponding pipe-soil vertical stiffness, The corresponding pipe-soil vertical stiffness is It is the soil restraint force on the buried pipeline when it undergoes axial displacement relative to the surrounding soil; It is the soil restraint force on the buried pipeline when it moves upward relative to the soil; It is the soil restraint force on the buried pipeline when it moves downward relative to the soil.

3. The vulnerability analysis method of pipeline joints under fault action based on machine learning according to claim 2 is characterized in that: In step 2, the INP file can be calculated and run by the finite element software ABAQUS, and the command flow of the INP file is used to define the pipe joint and the behavior characteristics of the pipe joint, specifically including: the definition of the element set constituting the pipe joint, the definition of the mechanical behavior of the pipe joint, the definition of whether the pipe joint is a rigid connection, and the definition of the rotational freedom constraint of the pipe joint on the X-axis and the Y-axis, wherein the element set of the pipe joint includes two nodes constituting the pipe joint and a two-node connection unit constituting the pipe joint.

4. The vulnerability analysis method of pipeline joints under fault action based on machine learning according to claim 3 is characterized by: The step 6 specifically includes the following steps: Step 6.1, the data of the multiple pipe joint angle values ​​obtained in step 5 that is greater than or equal to the pipe joint angle damage value is recorded as 1; Step 6.2: Among the multiple pipe joint angle values ​​obtained in step 5, the data that is smaller than the pipe joint angle damage value is recorded as 0.

Images