Training method of buried pipeline yield displacement prediction model

Abstract

The invention discloses a buried pipeline yield displacement prediction model training method comprising the following steps: obtaining pipe soil information, and carrying out finite element modeling according to the pipe soil information to obtain a pipe soil finite element model; extracting in a calculation result file of the pipe-soil finite element model through a Python language to obtain pipeline yield displacement information corresponding to multiple groups of different pipe-soil information; and training the initial regression model according to the multiple groups of pipeline yield displacement information to obtain a trained buried pipeline yield displacement prediction model. On this basis, multiple groups of pipe-soil yield displacement data under different parameters are generated through pipe-soil finite element simulation, and then a universal buried pipeline yield displacement prediction model is trained based on the data, so that a prediction result can be instantaneously output only by calling the trained model and the input parameters during practical application, repeated finite element calculation is thoroughly avoided, and the prediction efficiency is improved. And the prediction efficiency of the yield displacement of the buried pipeline is improved.

Description

Technical Field

[0001] This application belongs to the field of pipeline application technology, and in particular relates to a training method for a buried pipeline yield displacement prediction model. Background Technology

[0002] In areas with frequent seismic activity, buried pipelines often inevitably cross active faults. Among these, strike-slip faults can cause significant lateral shear displacement, which has a substantial impact on the pipeline structure. When the pipeline is constrained by the surrounding soil and subjected to a large horizontal displacement input, if its deformation exceeds the yield strength of the material or the allowable strain range of the structure, the pipeline will experience local or overall buckling, further leading to material yielding or even structural failure, seriously affecting its service safety and service life.

[0003] The patent with publication number CN107563014B discloses a method for calculating the critical buckling strain and critical fault displacement of a pipeline under fault action. Through finite element simulation and data analysis, the calculation results are extracted. Combining the influence law of each parameter on the critical axial strain of the pipeline, the parameters such as wall thickness, burial depth, pipeline-fault plane angle, pipeline internal pressure, soil compression modulus, soil cohesion, and critical fault displacement are dimensionless. Through multivariate nonlinear regression analysis, the regression formulas for critical compressive strain and critical fault displacement are obtained. Thus, the critical buckling strain and critical fault displacement of the pipeline under the action of soil can be analyzed to ensure the safety and service life of the pipeline.

[0004] Although this method can calculate the critical buckling strain and critical fault displacement of the pipeline under the action of soil through finite element analysis, the calculation process is complex and the calculation cycle is long. It is difficult to meet the rapid response requirements under large-scale parameter changes in soil or pipeline, which reduces the analysis and calculation efficiency of the critical buckling strain and critical fault displacement of the pipeline under the action of soil. Summary of the Invention

[0005] This application provides a training method for a buried pipeline yield displacement prediction model, which improves the prediction efficiency of buried pipeline yield displacement.

[0006] According to a first aspect of this application, embodiments of this application provide a training method for a buried pipeline yield displacement prediction model, which may include:

[0007] Obtain pipe and soil information, which includes: pipe parameters, soil parameters, pipe and soil boundary conditions, and pipe and soil load conditions;

[0008] Finite element modeling is performed based on the pipe and soil information to obtain the pipe and soil finite element model;

[0009] By using Python to extract the calculation results file of the pipe-soil finite element model, we can obtain the pipe yield displacement information corresponding to multiple sets of different pipe-soil information.

[0010] The initial regression model is trained based on multiple sets of pipeline yield displacement information to obtain a trained buried pipeline yield displacement prediction model.

[0011] Optionally, the pipeline parameters include: pipeline geometry parameters and pipeline material parameters; the soil parameters include: soil geometry parameters and soil material parameters;

[0012] Finite element modeling is performed based on the pipe and soil information to obtain the pipe and soil finite element model, including:

[0013] Based on the pipeline geometric parameters and soil geometric parameters, a pipe-soil coupling model is constructed;

[0014] Input the pipe material parameters and soil material parameters into the pipe-soil coupling model, set the parameters of the pipe-soil contact model, and obtain the pipe-soil contact model.

[0015] The parameters of the pipe-soil contact model are set according to the pipe-soil boundary conditions and pipe-soil load conditions to obtain the pipe-soil finite element model.

[0016] Optionally, a pipe-soil coupling model is constructed based on the pipe geometry parameters and soil geometry parameters, including:

[0017] Based on the pipe's geometric parameters, shell element modeling is performed on the pipe to obtain a three-dimensional model of the pipe.

[0018] Based on the soil geometric parameters, solid element modeling of the soil is performed to obtain a three-dimensional soil model;

[0019] The pipe 3D model and the pipe 3D model are coupled to obtain the pipe-soil coupled model.

[0020] Optionally, the initial regression model is trained based on multiple sets of pipeline yield displacement information to obtain a trained buried pipeline yield displacement prediction model, including:

[0021] Multiple sets of pipeline yield displacement information were divided into training and testing sets;

[0022] The training set is input into the initial regression model, and the initial regression model is trained to obtain the target regression model;

[0023] The test set is input into the target regression model to calculate the prediction accuracy, thus obtaining the model's prediction accuracy.

[0024] If the model prediction accuracy meets the model accuracy requirements, the target regression model is determined to be the buried pipeline yield displacement prediction model.

[0025] Optionally, after inputting the test set into the target regression model to calculate the prediction accuracy and obtaining the model's prediction accuracy, the method further includes:

[0026] If the model prediction accuracy does not meet the model accuracy requirements, repeat step AC below until the model prediction accuracy meets the model accuracy requirements:

[0027] Step A: Adjust the model parameters of the target regression model to obtain a new target regression model;

[0028] Step B: Input the training set into the new target regression model, train the new target regression model, and obtain the target regression model;

[0029] Step C: Input the test set into the target regression model to calculate the prediction accuracy and obtain the model prediction accuracy.

[0030] Optionally, model parameters include: distance function and number of nearest neighbors.

[0031] Optionally, the training set is input into the initial regression model, and the initial regression model is trained to obtain the target regression model, including:

[0032] The training set is input into the initial regression model to train the initial regression model;

[0033] The initial regression model after training is evaluated based on statistical indicators to obtain the evaluation results;

[0034] If the evaluation results meet the stopping criteria, the initial regression model is determined to be the target regression model.

[0035] Optionally, statistical indicators include: coefficient of determination, root mean square error, mean absolute error, mean absolute percentage error, and maximum relative error.

[0036] Optionally, the test set includes multiple test samples, each of which includes characteristic parameters and a target yield displacement;

[0037] The test set is input into the target regression model to calculate the prediction accuracy. The model prediction accuracy includes:

[0038] For each test sample, perform the following steps D-E:

[0039] Step D: Input the feature parameters into the target regression model for simulation to obtain the predicted yield displacement;

[0040] Step E: Calculate the absolute values ​​of the target yield displacement and the predicted yield to obtain the displacement difference value of the test sample;

[0041] If the displacement difference is less than or equal to the difference threshold, the prediction status of the target regression model corresponding to the test sample is determined to be accurate.

[0042] If the displacement difference value is greater than the difference threshold, the prediction state of the target regression model corresponding to the test sample is determined to be inaccurate.

[0043] The prediction accuracy is calculated by analyzing the predicted states corresponding to each test sample in the statistical test set.

[0044] Optionally, after training the initial regression model based on multiple sets of pipeline yield displacement information to obtain a trained buried pipeline yield displacement prediction model, the method further includes:

[0045] Obtain new land management information;

[0046] The new pipe-soil information is input into the buried pipeline yield displacement prediction model to obtain the buried pipeline yield displacement.

[0047] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects:

[0048] This application provides a training method for a buried pipeline yield displacement prediction model. By acquiring pipeline-soil information and then performing finite element modeling based on this information, a pipeline-soil finite element model is obtained. The calculation results file of the pipeline-soil finite element model is extracted using Python to obtain multiple sets of pipeline yield displacement information corresponding to different pipeline-soil information. Based on these multiple sets of pipeline yield displacement information, an initial regression model is trained to obtain a trained buried pipeline yield displacement prediction model. Based on this, multiple sets of pipeline-soil yield displacement data under different parameters are generated in advance through pipeline-soil finite element simulation. Then, a general buried pipeline yield displacement prediction model is trained based on these data. Therefore, in practical applications, only the trained model needs to be called, and the prediction results can be output instantly by inputting parameters, thus completely avoiding repetitive finite element calculations and improving the prediction efficiency of buried pipeline yield displacement. Furthermore, the simplicity and fast response speed of Python language for file extraction can meet the rapid response requirements under large-scale parameter changes, improving the efficiency of file extraction, which in turn improves the efficiency of model training and subsequent data processing.

[0049] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0050] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application, and do not constitute an undue limitation of this application.

[0051] Figure 1 This is a flowchart illustrating a training method for a buried pipeline yield displacement prediction model according to an exemplary embodiment;

[0052] Figure 2 This is a schematic diagram of the structure of a training device for a buried pipeline yield displacement prediction model, according to an exemplary embodiment.

[0053] Figure 3 This is a structural block diagram of a training device for a buried pipeline yield displacement prediction model, according to an exemplary embodiment. Detailed Implementation

[0054] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0055] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0056] As described in the background section, the calculation process for the yield displacement inside conventional pipelines is complex and time-consuming, making it difficult to meet the rapid response requirements under large-scale parameter changes in soil or pipelines. This reduces the efficiency of analysis and calculation of critical buckling strain and critical fault displacement of pipelines under soil action.

[0057] Based on this, this application provides a training method for a buried pipeline yield displacement prediction model. The training method for the buried pipeline yield displacement prediction model provided in the embodiments of this application will be described below first.

[0058] like Figure 1 - Figure 3 As shown;

[0059] Example 1;

[0060] It may include the following steps:

[0061] S101, Obtain pipe and soil information, which includes: pipe parameters, soil parameters, pipe and soil boundary conditions, and pipe and soil load conditions;

[0062] S102, perform finite element modeling based on the pipe and soil information to obtain the pipe and soil finite element model;

[0063] S103, using Python language to extract from the calculation result file of the pipe-soil finite element model, obtains multiple sets of pipe yield displacement information corresponding to different pipe-soil information;

[0064] S104. The initial regression model is trained based on multiple sets of pipeline yield displacement information to obtain a trained buried pipeline yield displacement prediction model.

[0065] Based on the above embodiments, by acquiring pipe-soil information and then performing finite element modeling based on that information, a pipe-soil finite element model is obtained. The calculation results file of this model is then extracted using Python to obtain multiple sets of pipeline yield displacement information corresponding to different pipe-soil conditions. Based on these multiple sets of pipeline yield displacement information, an initial regression model is trained to obtain a trained buried pipeline yield displacement prediction model. Therefore, multiple sets of pipe-soil yield displacement data under different parameters are pre-generated through pipe-soil finite element simulation. A general buried pipeline yield displacement prediction model is then trained based on this data. In practical applications, only the trained model needs to be called, and the input parameters are used to instantly output the prediction results, thus completely avoiding repetitive finite element calculations and improving the prediction efficiency of buried pipeline yield displacement. Furthermore, the simplicity and fast response speed of Python for file extraction meet the need for rapid response under large-scale parameter changes, improving the efficiency of file extraction and consequently improving the efficiency of model training and subsequent data processing.

[0066] In the above S101, the pipeline is buried in the soil. The soil referred to in this application is the soil covering the pipeline within a certain size range around the pipeline. In order to better study the displacement of the pipeline under the action of the soil, it is necessary to obtain pipeline-soil information.

[0067] Pipe-soil information can include: pipe parameters, soil parameters, pipe-soil boundary conditions, and pipe-soil load conditions.

[0068] Among them, pipeline parameters can refer to parameters that reflect the physical characteristics of the pipeline, such as pipeline geometric parameters and pipeline material parameters, which can be obtained directly by calling the design drawings.

[0069] Soil parameters can refer to soil geometric parameters and soil material parameters. Soil geometric parameters refer to the thickness, length, and width of the soil within a certain range around the pipeline. Soil material parameters refer to the parameters corresponding to the material of the soil within that range. The material of the soil can be determined by setting up sensors or soil sampling. The soil material parameters corresponding to different materials can be determined by consulting relevant data.

[0070] Pipe-soil boundary conditions refer to the special boundary settings at both ends of the pipeline, the soil side constraints, and the bottom fixing conditions.

[0071] The soil-pipeline load case involves strike-slip fault displacement input and gravity load application.

[0072] In S102 above, the pipe and soil information is parametrically modeled using a Python script in Abaqus software to obtain the pipe and soil finite element model.

[0073] In S103 above, the finite element calculation files generated by the finite element model of the pipe and soil using Python language are processed in batches to extract the pipe yield displacement information corresponding to multiple different pipe and soil information. The pipe yield displacement information can include key response parameters such as pipe lateral deformation, yield displacement and strain distribution.

[0074] Python is written in C, and many standard and third-party libraries are also written in C. Its language is simple and fast, making it ideal for file extraction. This allows for rapid response under large-scale parameter changes, improving efficiency.

[0075] In S104 above, multiple sets of pipeline yield displacement information are used to train the initial regression model so that the initial regression model meets the accuracy requirements, thereby obtaining a well-trained initial regression model, namely the buried pipeline yield displacement prediction model.

[0076] Specifically, if it is necessary to train a buried pipeline yield displacement prediction model, the yield displacement in the pipeline yield displacement information can be used to train the regression model.

[0077] Optionally, in one example, the pipe parameters include: pipe geometry parameters and pipe material parameters; the soil parameters include: soil geometry parameters and soil material parameters;

[0078] Alternatively, in one example, the above S102 may include:

[0079] S1021, Construct a pipe-soil coupling model based on pipeline geometric parameters and soil geometric parameters;

[0080] S1022, Input the pipe material parameters and soil material parameters into the pipe-soil coupling model, set the parameters of the pipe-soil contact model, and obtain the pipe-soil contact model;

[0081] S1023, Based on the pipe-soil boundary conditions and pipe-soil load conditions, the parameters of the pipe-soil contact model are set to obtain the pipe-soil finite element model.

[0082] In S1031 above, a three-dimensional model of the pipeline can be constructed based on the pipeline's geometric parameters, and a three-dimensional model of the soil can be constructed based on the soil's geometric parameters. Since the pipeline is buried in the soil, in order to accurately reflect the geometric relationship between the soil and the pipeline buried in the soil, the three-dimensional model of the pipeline and the three-dimensional model of the soil need to be coupled together to obtain the pipe-soil coupling model. The pipe-soil coupling model can reflect the three-dimensional geometric relationship between the soil and the pipeline.

[0083] Specifically, S1031 may include:

[0084] S10311, Based on the pipe's geometric parameters, shell element modeling is performed on the pipe to obtain a three-dimensional model of the pipe;

[0085] S10312, Based on the soil geometric parameters, solid element modeling of the soil is performed to obtain a three-dimensional soil model;

[0086] S10313, couple the pipeline 3D model with the pipeline 3D model to obtain the pipeline-soil coupled model.

[0087] In S10311 above, since the pipe is a circular structure with a hollow interior, and the part that comes into contact with the soil is mainly at its outer ring, shell element modeling can be used when performing three-dimensional modeling of the pipe's geometric parameters.

[0088] In S10312 above, similarly, the soil is a solid covering the pipe, so solid element modeling is used when modeling the soil in three dimensions.

[0089] In S10313 above, the three-dimensional soil model and the three-dimensional pipeline model are coupled to obtain a three-dimensional model in which the pipeline is buried in the figure, namely the pipe-soil coupled model.

[0090] In S1032 above, the pipeline is buried in the soil, and there will be an interaction between the two. The magnitude of these forces is related to parameters such as soil material and pipeline material. In order to better reflect the interaction between soil and pipeline, it is necessary to input information such as pipeline material parameters and soil material parameters into the pipe-soil coupling model, set some material parameters in the pipe-soil coupling model, and thus obtain the pipe-soil contact model, which can characterize the actual mechanical behavior under different soil properties, fault displacement and pipeline parameter combinations.

[0091] Among them, soil material parameters are set according to actual engineering conditions, which may include elastic modulus, cohesion, etc.

[0092] In the above S1033, the pipe-soil boundary conditions include special boundary settings at both ends of the pipe, soil lateral constraints, and bottom fixing conditions; the pipe-soil load conditions may include strike-slip fault displacement input and gravity load application.

[0093] By inputting the pipe-soil boundary conditions and pipe-soil load conditions into the pipe-soil contact model, further constraints can be imposed based on engineering requirements on the basis of the interaction between the soil and the pipeline, so as to ensure that the pipe-soil contact model can meet the actual requirements, thus obtaining a pipe-soil finite element model suitable for engineering needs.

[0094] Example 2:

[0095] Alternatively, in one example, S104 may include:

[0096] S1041, divide multiple sets of pipeline yield displacement information into training set and test set;

[0097] S1042, Input the training set into the initial regression model, train the initial regression model, and obtain the target regression model;

[0098] S1043, Input the test set into the target regression model to calculate the prediction accuracy and obtain the model prediction accuracy;

[0099] S1044, if the model prediction accuracy meets the model accuracy requirements, the target regression model is determined to be the buried pipeline yield displacement prediction model.

[0100] In S1041 above, multiple sets of pipeline yield displacement information are combined into a dataset. The dataset is then divided into two sets: a training set and a test set. Both the test set and the training set contain multiple sets of pipeline yield displacement information. The pipeline yield displacement information in the test set is called the test sample, and the pipeline yield displacement information in the training set is called the training sample.

[0101] For training and testing samples, the input features for training and testing the model include: pipe geometric parameters, soil geometric parameters, friction coefficient between pipe and soil, etc., and the output target is the target yield displacement.

[0102] Preferably, 85% of the dataset is the training set and 15% is the test set.

[0103] In S1042 above, the training samples in the training set are input into the initial regression model, and the initial regression model is trained until the termination condition is met, that is, the training is completed and the target regression model is obtained.

[0104] Optionally, the initial regression model can be a KNN (K-Nearest Neighbors) model. The training principle of this model is: in the feature space, if the majority of the K nearest (i.e., the closest) samples in a sample belong to a certain category, then that sample also belongs to that category. The KNN model is simple and intuitive, requires no complex training process, has strong interpretability, and the prediction results can be directly interpreted through neighboring samples.

[0105] In S1043 above, after obtaining the target regression model, the input features in the test sample are input into the target regression model, and the target regression model will output the predicted yield displacement. Then, the difference between the predicted yield displacement and the target yield displacement in the test sample is compared to calculate the prediction accuracy, thereby obtaining the model prediction accuracy. The model prediction accuracy can determine whether the prediction of the target regression model is accurate.

[0106] In S1044 above, if the model prediction accuracy meets the model accuracy requirements, it can be considered that the target regression model obtained after training the training set can accurately predict the yield displacement and can be applied in reality to predict the target yield displacement. The target regression model obtained in this way is the buried pipeline yield displacement prediction model.

[0107] In one example, after S1043 above, the method may further include:

[0108] S1045, If the model prediction accuracy does not meet the model accuracy requirements, repeat step AC below until the model prediction accuracy meets the model accuracy requirements:

[0109] Step A: Adjust the model parameters of the target regression model to obtain a new target regression model;

[0110] Step B: Input the training set into the new target regression model, train the new target regression model, and obtain the target regression model;

[0111] Step C: Input the test set into the target regression model to calculate the prediction accuracy and obtain the model prediction accuracy.

[0112] In S1045 above, if the model prediction accuracy does not meet the model accuracy requirements, the target regression model needs to be retrained using the training set until the model prediction accuracy of the retrained target regression model meets the model accuracy requirements.

[0113] Since the model prediction accuracy of the target regression model tested on the test set does not meet the model accuracy requirements, the model parameters of the target regression model need to be adjusted when training the target regression model again to obtain a new target regression model.

[0114] Specifically, the model parameters include: distance function and number of nearest neighbors (K), where the distance function can include Manhattan distance function and Euclidean distance function, etc.; in this example, the optimal distance function is Euclidean distance function, and the optimal number of nearest neighbors K = 3.

[0115] The new target regression model is then trained, and the prediction progress is calculated to obtain the new model prediction accuracy. The specific implementation methods of these two steps are the same as those of S1042-S1043 above, except that there are differences in the model, which will not be elaborated on here.

[0116] Through continuous training and verification, it can be ensured that the final trained target regression model meets the model accuracy requirements, thereby ensuring the accuracy of the buried pipeline yield displacement prediction model.

[0117] Alternatively, in one example, the above S1042 may include:

[0118] S10421, Input the training set into the initial regression model to train the initial regression model;

[0119] S10422, Evaluate the initial regression model after training based on statistical indicators to obtain the evaluation results;

[0120] S10423, if the evaluation results meet the stopping condition, determine the initial regression model as the target regression model.

[0121] In S10421 above, the training samples in the training set are input into the initial regression model to train the initial regression model. The training process involves continuously adjusting the model parameters of the initial regression model.

[0122] In S10422 above, each time the model training is completed using training samples, the initial regression model after training is evaluated by statistical indicators to obtain the evaluation results. The evaluation results refer to the determination coefficient, root mean square error, mean absolute error, mean absolute percentage error, and maximum relative error of the target regression model.

[0123] Specifically, statistical indicators may include: coefficient of determination (R²). 2 The parameters are: root mean square error (RMSE), mean absolute error (MAE), mean absolute percentage error (MAE), and maximum relative error (δ_max). The coefficient of determination, as a core indicator of model fit, ranges from [0, 1]. When R... 2When the value approaches 1, it indicates that the model has excellent predictive reliability and explanatory power. Three complementary error metrics (RMSE, MAE, and MAPE) characterize the residual distribution of predicted and actual values ​​from different dimensions: RMSE amplifies larger errors through the squared term and is highly sensitive to outliers; MAE provides a robust measure of mean absolute error; and MAPE achieves cross-dataset comparability through percentage-based relative error. Furthermore, the maximum relative error (δ_max) is included in the evaluation system as a key warning indicator of predictive reliability.

[0124] In S10423 above, when the evaluation result meets the stopping condition, the training can be indicated as complete, that is, the target regression model is obtained.

[0125] The stopping condition refers to model convergence, which is reflected in R. 2 They approach 1; the maximum relative error becomes smaller and smaller; that is to say, these two indicators approach constant or change less and less.

[0126] Optionally, S1043 above may include:

[0127] S10431, For each test sample, perform the following steps D-E respectively:

[0128] Step D: Input the feature parameters into the target regression model for simulation to obtain the predicted yield displacement;

[0129] Step E: Calculate the absolute values ​​of the target yield displacement and the predicted yield to obtain the displacement difference value of the test sample;

[0130] S10432, when the displacement difference value is less than or equal to the difference threshold, the prediction state of the target regression model corresponding to the test sample is determined to be accurate.

[0131] S10433, when the displacement difference value is greater than the difference threshold, the prediction state of the target regression model corresponding to the test sample is determined to be inaccurate.

[0132] S10434, calculates the prediction accuracy by analyzing the predicted states corresponding to each test sample in the statistical test set.

[0133] In S10431 above, for each test sample, the target regression model makes a prediction based on the input features in the test sample to obtain an output feature, namely the predicted yield displacement. Then, the predicted yield displacement is compared with the target yield displacement in the test sample to obtain the displacement difference value between the predicted yield displacement and the target yield displacement.

[0134] Specifically, for each test sample, the following steps D-E are performed:

[0135] In step D, the feature parameters are input into the target regression model for simulation to obtain the predicted yield displacement.

[0136] Step E: Calculate the absolute values ​​of the target yield displacement and the predicted yield to obtain the displacement difference value of the test sample;

[0137] In S10432 above, the displacement difference value is compared with the difference threshold. If the displacement difference value is less than or equal to the difference threshold, it means that the predicted yield displacement is very close to the actual target yield displacement. The displacement difference value between them can be regarded as error. Therefore, the prediction can be considered accurate, that is, the prediction state of the target regression model is accurate.

[0138] The difference threshold is a value set by technicians based on their design experience and requirements.

[0139] In S10433 above, conversely, if the displacement difference value is larger than the difference threshold, then the displacement difference value is considered to be an error, and the prediction of the target regression model is inaccurate, that is, the prediction state of the target regression model is inaccurate.

[0140] In S10434 above, the test set includes multiple test samples. In the above steps, the prediction state corresponding to each test sample is determined, and then the prediction state corresponding to each test sample is statistically analyzed to obtain the prediction accuracy (percentage) of the target regression model for all test samples in the test set.

[0141] In the above embodiment 2, each method in the above embodiment 1 is included and each process in the above embodiment 1 can be implemented, and the same technical effect can be achieved. To avoid repetition, it will not be described again here.

[0142] Example 3:

[0143] Alternatively, in one example, after S104 above, the method may further include:

[0144] S106, Obtain new soil and pipe information;

[0145] S107. Input the new pipe-soil information into the buried pipeline yield displacement prediction model to obtain the buried pipeline yield displacement.

[0146] In S106 above, new pipe-soil information is obtained by setting up sensors or by directly retrieving database information. The new pipe-soil information can refer to the information of soil and pipeline in the area that needs to be predicted using the buried pipeline yield displacement prediction model.

[0147] In step S107 above, the new pipe-soil information is input into the buried pipeline yield displacement prediction model. The buried pipeline yield displacement prediction model simulates the yield displacement of the buried pipeline based on the pipe-soil information, thereby obtaining an accurate prediction of the buried pipeline yield displacement. This achieves accurate prediction of the yield displacement of the pipeline buried in the soil based on the acquired pipe-soil information.

[0148] In the above embodiment 3, each method in the above embodiment 1 and / or 2 is included, and each process in the above embodiment 1 and / or 2 can be implemented to achieve the same technical effect. To avoid repetition, it will not be described again here.

[0149] Example 4;

[0150] like Figure 2 As shown, based on the same inventive concept, this embodiment, in addition to Embodiment 1, also provides a training device 200 for predicting the yield displacement of buried pipelines, which includes:

[0151] The first acquisition module 210 is used to acquire pipe and soil information, which includes: pipe parameters, soil parameters, pipe and soil boundary conditions, and pipe and soil load conditions.

[0152] Modeling module 220 is used to perform finite element modeling based on the pipe-soil information to obtain the pipe-soil finite element model.

[0153] The extraction module 230 is used to extract the pipeline yield displacement information corresponding to multiple sets of different pipeline information from the calculation result file of the pipe-soil finite element model using the Python language.

[0154] Training module 240 is used to train the initial regression model based on multiple sets of pipeline yield displacement information to obtain a trained buried pipeline yield displacement prediction model.

[0155] Based on this, the first acquisition module 210 acquires pipe-soil information, the modeling module 220 performs finite element modeling based on the pipe-soil information to obtain a pipe-soil finite element model, and the extraction module 230 extracts the calculation result file of the pipe-soil finite element model using Python to obtain multiple sets of pipe yield displacement information corresponding to different pipe-soil information; the training module 240 trains the initial regression model based on multiple sets of pipe yield displacement information to obtain a trained buried pipeline yield displacement prediction model. Based on this, multiple sets of pipe-soil yield displacement data under different parameters are generated in advance through pipe-soil finite element simulation, and then a general buried pipeline yield displacement prediction model is trained based on these data. Therefore, in practical applications, only the trained model needs to be called and the parameters input to instantly output the prediction result, thereby completely avoiding repeated finite element calculations and improving the prediction efficiency of buried pipeline yield displacement. Furthermore, the simplicity and fast response speed of Python language for file extraction can meet the rapid response requirements under large-scale parameter changes, improving the efficiency of file extraction, which in turn improves the efficiency of model training and subsequent data processing.

[0156] Optionally, the pipeline parameters include: pipeline geometry parameters and pipeline material parameters; the soil parameters include: soil geometry parameters and soil material parameters;

[0157] Modeling module 220 includes:

[0158] The building unit is used to construct a pipe-soil coupling model based on the pipe geometry parameters and soil geometry parameters;

[0159] The first input unit is used to input the pipe material parameters and soil material parameters into the pipe-soil coupling model, set the parameters of the pipe-soil contact model, and obtain the pipe-soil contact model.

[0160] The second input unit is used to set parameters for the pipe-soil contact model based on the pipe-soil boundary conditions and pipe-soil load conditions, thereby obtaining the pipe-soil finite element model.

[0161] Optionally, the building unit may include:

[0162] The first modeling sub-unit is used to perform shell element modeling of the pipeline based on the pipeline's geometric parameters, thereby obtaining a three-dimensional model of the pipeline.

[0163] The second modeling sub-unit is used to model the soil as a solid element based on the soil geometric parameters, thereby obtaining a three-dimensional soil model.

[0164] The coupling sub-unit is used to couple the 3D model of the pipeline with another 3D model of the pipeline to obtain the pipeline-soil coupled model.

[0165] Optionally, the training module 240 may include:

[0166] The partitioning unit is used to divide multiple sets of pipeline yield displacement information into training and test sets;

[0167] The training unit is used to input the training set into the initial regression model, train the initial regression model, and obtain the target regression model.

[0168] The calculation unit is used to input the test set into the target regression model to calculate the prediction accuracy and obtain the model prediction accuracy.

[0169] The determination unit is used to determine the target regression model as the buried pipeline yield displacement prediction model, provided that the model prediction accuracy meets the model accuracy requirements.

[0170] Optionally, the training module 240 may also include:

[0171] The loop unit is used to repeatedly execute the following steps AC until the model prediction accuracy meets the model accuracy requirements, provided that the model prediction accuracy does not meet the requirements:

[0172] Step A: Adjust the model parameters of the target regression model to obtain a new target regression model;

[0173] Step B: Input the training set into the new target regression model, train the new target regression model, and obtain the target regression model;

[0174] Step C: Input the test set into the target regression model to calculate the prediction accuracy and obtain the model prediction accuracy.

[0175] Optionally, model parameters include: distance function and number of nearest neighbors.

[0176] Optionally, the training unit may include:

[0177] The first training subunit is used to input the training set into the initial regression model to train the initial regression model;

[0178] The evaluation sub-unit is used to evaluate the initial regression model after training based on statistical indicators and obtain the evaluation results;

[0179] The first determining sub-unit is used to determine the initial regression model as the target regression model when the evaluation result meets the stopping condition.

[0180] Optionally, statistical indicators include: coefficient of determination, root mean square error, mean absolute error, mean absolute percentage error, and maximum relative error.

[0181] Optionally, the test set includes multiple test samples, each of which includes characteristic parameters and a target yield displacement;

[0182] The computing unit may include:

[0183] The execution subunit is used to perform steps D-E for each test sample:

[0184] Step D: Input the feature parameters into the target regression model for simulation to obtain the predicted yield displacement;

[0185] Step E: Calculate the absolute values ​​of the target yield displacement and the predicted yield to obtain the displacement difference value of the test sample;

[0186] The second determining subunit is used to determine the prediction state of the target regression model corresponding to the test sample as accurate when the displacement difference value is less than or equal to the difference threshold.

[0187] The third determining subunit is used to determine the prediction state of the target regression model corresponding to the test sample as inaccurate when the displacement difference value is greater than the difference threshold.

[0188] The statistical subunit is used to statistically analyze the predicted state corresponding to each test sample in the test set and calculate the prediction accuracy.

[0189] Optionally, the training device 200 for the buried pipeline yield displacement prediction model may further include:

[0190] The second acquisition module is used to acquire new soil and pipe information;

[0191] The input module is used to input new pipe and soil information into the buried pipeline yield displacement prediction model to obtain the buried pipeline yield displacement.

[0192] In Example 4, the processes of at least one of the above Examples 1, 2 and 3 can be implemented and the same technical effects can be achieved. To avoid repetition, they will not be described again here.

[0193] Example 5:

[0194] Figure 3 The diagram shows a hardware structure schematic of a training device for a buried pipeline yield displacement prediction model provided in an embodiment of the present invention.

[0195] The training device for the buried pipeline yield displacement prediction model may include a processor 301 and a memory 302 storing computer program instructions.

[0196] Specifically, the processor 301 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of the present invention.

[0197] Memory 302 may include mass storage for data or instructions. For example, and not limitingly, memory 302 may include a hard disk drive (HDD), floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 302 may include removable or non-removable (or fixed) media. Where appropriate, memory 302 may be internal or external to the integrated gateway disaster recovery device. In a particular embodiment, memory 32 is non-volatile solid-state memory.

[0198] In certain embodiments, memory 302 may include read-only memory (ROM), random access memory (RAM), disk storage media device, optical storage media device, flash memory device, electrical, optical, or other physical / tangible memory storage device. Thus, generally, memory 302 includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors 301), it is operable to perform the operations described with reference to the method according to one aspect of this application.

[0199] The processor 301 reads and executes computer program instructions stored in the memory 302 to implement any of the training methods for the buried pipeline yield displacement prediction model in the above embodiments.

[0200] In one example, the training device for the buried pipeline yield displacement prediction model may also include a communication interface 303 and a bus 304. As shown in the figure, the processor 301, memory 302, and communication interface 303 are connected through the bus 304 and communicate with each other.

[0201] The communication interface 303 is mainly used to realize communication between various modules, devices, units and / or equipment in the embodiments of the present invention.

[0202] Bus 304 may include hardware, software, or both. For example, and not as a limitation, bus 304 may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, a Wireless Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a Microchannel Architecture (MCA) bus, a Peripheral Control Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses, or combinations of two or more of these. Where appropriate, bus 304 may include one or more buses 304. Although specific bus 304s are described and illustrated in embodiments of this application, this application contemplates any suitable bus 304 or interconnect.

[0203] The training equipment for this buried pipeline yield displacement prediction model can be based on the current training method for buried pipeline yield displacement prediction models, thereby achieving a combination of... Figure 1 , 2 The training method and training device 200 for the buried pipeline yield displacement prediction model are described.

[0204] In addition, this application also provides a computer program product, including computer program instructions, which, when executed by processor 301, can implement the steps and corresponding content of the aforementioned method embodiments.

[0205] It should be clarified that this application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of this application is not limited to the specific steps described and shown. Those skilled in the art can make various changes, modifications, and additions, or change the order of steps, after understanding the spirit of this application.

[0206] The functional blocks shown in the above-described structural diagram can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, they can be, for example, electronic circuits, application-specific integrated circuits (ASICs), appropriate firmware, plug-ins, function cards, etc. When implemented in software, the elements of this application are programs or code segments used to perform the required tasks. Programs or code segments can be stored on a machine-readable medium or transmitted over a transmission medium or communication link via data signals carried on a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, fiber optic media, radio frequency (RF) links, etc. Code segments can be downloaded via computer networks such as the Internet, intranets, etc.

[0207] It should also be noted that the exemplary embodiments mentioned in this application describe methods or systems based on a series of steps or apparatus. However, this application is not limited to the order of the above steps; that is, the steps can be performed in the order mentioned in the embodiments, or in a different order, or several steps can be performed simultaneously.

[0208] The aspects of this disclosure have been described above with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block in the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, or other programmable buried pipeline yield displacement prediction model training method, apparatus, or device to produce a machine such that these instructions, executed via the processor of the computer or other programmable buried pipeline yield displacement prediction model training method, apparatus, or device, enable the implementation of the functions / actions specified in one or more blocks of the flowchart illustrations and / or block diagrams. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor, or a field-programmable logic circuit. It is also understood that each block in the block diagrams and / or flowchart illustrations, and combinations of blocks in the block diagrams and / or flowchart illustrations, can also be implemented by special-purpose hardware performing the specified functions or actions, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0209] The above description is merely a specific implementation of this application. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, modules, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the protection scope of this application.

Claims

1. A training method for a buried pipeline yield displacement prediction model, characterized in that, The method comprises the following steps: obtaining pipe-soil information, wherein the pipe-soil information comprises pipe parameters, soil parameters, pipe-soil boundary conditions and pipe-soil load conditions; performing finite element modeling according to the pipe-soil information to obtain a pipe-soil finite element model; extracting, through a Python language, from a calculation result file of the pipe-soil finite element model, pipe yield displacement information corresponding to multiple groups of different pipe-soil information; training an initial regression model according to the multiple groups of pipe yield displacement information to obtain a trained buried pipe yield displacement prediction model.

2. The method of claim 1, wherein, The pipe parameters comprise pipe geometric parameters and pipe material parameters; the soil parameters comprise soil geometric parameters and soil material parameters; The finite element modeling according to the pipe-soil information to obtain a pipe-soil finite element model comprises: constructing a pipe-soil coupling model according to the pipe geometric parameters and the soil geometric parameters; inputting the pipe material parameters and the soil material parameters into the pipe-soil coupling model to perform parameter setting on the pipe-soil contact model to obtain a pipe-soil contact model; performing parameter setting on the pipe-soil contact model according to the pipe-soil boundary conditions and the pipe-soil load conditions to obtain the pipe-soil finite element model.

3. The method of claim 2, wherein the buried pipeline yield displacement prediction model is trained using a machine learning algorithm. The construction of the pipe-soil coupling model according to the pipe geometric parameters and the soil geometric parameters comprises: performing shell element modeling on a pipe according to the pipe geometric parameters to obtain a pipe three-dimensional model; performing solid element modeling on soil according to the soil geometric parameters to obtain a soil three-dimensional model; coupling the pipe three-dimensional model and the soil three-dimensional model to obtain a pipe-soil coupling model.

4. The method of claim 1, wherein, The training of the initial regression model according to the multiple groups of pipe yield displacement information to obtain a trained buried pipe yield displacement prediction model comprises: dividing the multiple groups of pipe yield displacement information into a training set and a test set; inputting the training set into an initial regression model to train the initial regression model to obtain a target regression model; inputting the test set into the target regression model to calculate prediction accuracy to obtain model prediction accuracy; in a case where the model prediction accuracy meets model accuracy requirements, determining that the target regression model is the buried pipe yield displacement prediction model.

5. The training method for the buried pipeline yield displacement prediction model as described in claim 4, characterized in that, After the inputting of the test set into the target regression model to calculate prediction accuracy to obtain model prediction accuracy, the method further comprises: in a case where the model prediction accuracy does not meet model accuracy requirements, cyclically performing steps A-C until the model prediction accuracy meets model accuracy requirements: Step A, adjusting model parameters of the target regression model to obtain a new target regression model; Step B, inputting the training set into the new target regression model to train the new target regression model to obtain a target regression model; Step C, inputting the test set into the target regression model to calculate prediction accuracy to obtain model prediction accuracy.

6. The training method for the buried pipeline yield displacement prediction model as described in claim 5, characterized in that, The model parameters comprise a distance function and a number of nearest neighbors.

7. The training method for the buried pipeline yield displacement prediction model as described in claim 4, characterized in that, The inputting of the training set into an initial regression model to train the initial regression model to obtain a target regression model comprises: inputting the training set into the initial regression model to train the initial regression model; judging the trained initial regression model according to a statistical index to obtain a judgment result; in a case where the judgment result meets a stop condition, determining the initial regression model as a target regression model.

8. The training method for the buried pipeline yield displacement prediction model as described in claim 7, characterized in that, The statistical index includes a coefficient of determination, a root mean square error, a mean absolute error, a mean absolute percentage error, and a maximum relative error.

9. The method of claim 4, wherein, The test set includes a plurality of test samples, and each test sample includes a characteristic parameter and a target yield displacement; The test set is input into the target regression model to calculate the prediction accuracy, and a model prediction accuracy is obtained, including: For each test sample, the following steps D-E are performed: Step D, inputting the characteristic parameter into the target regression model to simulate and obtain a predicted yield displacement; Step E, calculating the absolute value of the target yield displacement and the predicted yield displacement to obtain a displacement difference value of the test sample; in a case where the displacement difference value is less than or equal to a difference threshold, determining that the prediction state of the target regression model corresponding to the test sample is prediction accurate; in a case where the displacement difference value is greater than the difference threshold, determining that the prediction state of the target regression model corresponding to the test sample is prediction inaccurate; statistically counting the prediction state corresponding to each test sample in the test set to calculate the prediction accuracy.

10. The method of claim 1-9, wherein, After the initial regression model is trained according to a plurality of groups of pipe yield displacement information to obtain a trained buried pipe yield displacement prediction model, the method further includes: obtaining new pipe-soil information; inputting the new pipe-soil information into the buried pipe yield displacement prediction model to obtain a buried pipe yield displacement.

Images