Intelligent cable force optimization method in long and large irregular pipeline sinking process
Through machine learning technology, the pipeline sinking attitude and duration prediction model is constructed, and the tension distribution of hanging cables is optimized, which solves the problems of low construction accuracy and efficiency in sinking and placement of long and large irregular pipelines, and achieves multi-objective balance and safety improvement.
Patent Information
- Application Number
- CN202411039477.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-31
- Publication Date
- 2024-12-03
AI Technical Summary
During the sinking and laying of large and large irregular pipelines, it is difficult for the existing technology to optimize the cable force, resulting in low construction accuracy, low efficiency and safety risks, and it is difficult to quickly adjust the tension and stress distribution of the suspended cable when manually operated.
Using machine learning technology, by constructing high-precision pipeline sinking attitude and sinking time prediction models, establishing a cable force optimization method that comprehensively considers multiple constraints, and using multiple machine learning models (such as LSTM, GBT, SVR) to analyze real-time data , dynamically adjust the cable release speed and height to optimize the tension distribution of suspended cables.
It effectively improves the accuracy and efficiency of pipeline sinking, reduces safety risks, achieves the optimal balance between multiple goals, avoids performance reduction caused by single target optimization, and ensures the stability and safety of the sinking process.
Smart Images

Figure CN119066951A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of pipeline laying construction, and in particular to a method for intelligently optimizing cable force during the laying of a long and irregular pipeline. Background Art
[0002] The immersed tube method is a construction method in which several prefabricated sections are floated to the sea (river) site and sunk one by one in sequence into a dredged foundation trench to complete the construction of underwater pipelines. The immersed tube method is widely used in the underwater sinking and installation construction of engineering structures such as cross-river pipelines.
[0003] In order to adapt to the complex underwater terrain, the pipeline needs to be designed to fit the contour of the foundation trench, be long and irregular in shape, that is, a long large irregular pipeline. The sinking operation of the long large irregular pipeline is the most critical link in the entire underwater construction process of the structure, involving the coordination of multiple factors, and its core lies in the accurate regulation of the lowering of the suspension cable.
[0004] Currently, pipeline laying mainly relies on manual cable adjustment, which makes it difficult to ensure that each operation can achieve the best adjustment effect. At the same time, it is accompanied by fluctuations in construction precision and accuracy, low efficiency and safety risks. In addition, manual operation is difficult to quickly make the best adjustment in a complex and changeable construction environment. Improper operation can easily cause problems such as loose cable tension, excessive cable tension, uneven cable force distribution, local stress concentration on the pipe body, and inadequate laying.
[0005] It should be pointed out that the current optimization strategies for the laying of long and irregular pipelines mostly focus on the innovative design of laying devices, the improvement of the accuracy of laying docking positioning technology, and the control of system modules. Insufficient consideration is given to the force distribution of the suspension cable and the cable release control strategy, resulting in obvious limitations in the optimization of cable force. In addition, the existing cable force optimization methods are often limited to single factor analysis, and the influence of the coupling of multiple factors on the cable force state is not comprehensively considered, making it difficult to fully reflect the cable force state. Summary of the invention
[0006] The purpose of the present invention is to provide a method for intelligent optimization of cable force in the process of laying long irregular pipelines in view of the deficiencies of the prior art. The method introduces machine learning technology to deeply explore the complex interaction between the coordinated lowering of the suspension cable, the movement posture of the submerged pipe and the laying time period, and constructs a high-precision pipeline laying posture and laying time prediction model through algorithm model training; establishes a cable force optimization method that comprehensively considers multiple constraints such as laying accuracy requirements, suspension cable force balance, environmental factor disturbance, and suspension cable tension limit, and can analyze real-time feedback data to respond quickly according to environmental changes, obtain the optimal cable laying speed and height, and dynamically adjust the laying strategy. It effectively solves the problems of traditional manual cable adjustment, such as high experience dependence, low operating efficiency, limited laying accuracy, and difficulty in quickly and dynamically coordinating multiple suspension cables.
[0007] To achieve the above object, the present invention is implemented through the following technical solutions.
[0008] A method for intelligently optimizing cable force during the laying process of a long and irregular pipeline includes the following steps, specifically:
[0009] Step S1, collecting data and constructing an original data set of pipeline laying information;
[0010] Step S2: Establish multiple machine learning prediction models:
[0011] Based on the data set obtained in step S1, training is performed, and the incoming flow velocity, cable pretension, cable release height, and cable release speed are selected as input quantities of each machine learning prediction model, and the pipeline movement posture and total sinking time are selected as output quantities of each machine learning prediction model;
[0012] Determine the key feature set based on the RFE strategy and use the learning rate adjustment strategy to determine the optimal parameter set;
[0013] Comprehensively compare and analyze the prediction performance of each machine learning model, and select the machine learning prediction model with the best accuracy for subsequent constraints;
[0014] Step S3: Based on the machine learning prediction model selected in step S2, a multi-objective function is constructed to optimize the cable tension, and then an objective function for optimizing the cable force of a long irregular pipeline is established. The objective function is:
[0015] f(x)=w1F+w2Var;
[0016] F=max[F1,F2,...,F N ];
[0017]
[0018]
[0019] Where f(x) represents the objective function, w1 and w2 are the weights corresponding to each target value, and the weights are determined according to the importance of the target; F is the maximum tension value of each cable, Var is the variance of the cable tension; F i is the tension of the cable, L is the length of the cable at a certain moment, L0 is the initial length of the cable at a certain moment; c is the damping coefficient of the cable, k is the stiffness of the cable;
[0020]
[0021] F f is the fluid force on the structure, Δ is the displacement, a f is the fluid acceleration relative to the earth, Ca is the inertia coefficient of the structure, a r is the fluid acceleration relative to the structure; ρ w is the density of water, V r is the flow velocity relative to the structure, C d is the drag coefficient, A is the resistance area, and the value is determined according to the actual project;
[0022] Step S4: Set constraints:
[0023] The constraints are the pipe section posture constraint, the sinking period constraint and the objective function value constraint; among them, the pipe section posture constraint and the sinking period constraint are equality constraints, and the objective function value constraint is an inequality constraint. i (x) = 0 is the equality constraint function, g i (x) = 0 is the inequality constraint function;
[0024] The equality constraints are:
[0025] Z t =Z;
[0026] θ t =θ;
[0027] ψ t =ψ;
[0028] T t =T;
[0029] Among them, Z t ,θ t , t is the target value of the pipeline attitude, Z, θ, ψ represent the pipeline attitude prediction model that has been trained and can be called in step S2, and Z, θ, ψ are the heave, pitch, and yaw values of the pipeline at different positions respectively; T t is the target value of the time required to lower the cable to the specified position, and T represents the trained prediction model for the lowering time;
[0030] A weighted square sum function is constructed to quantitatively evaluate the motion posture of the pipe segment. The function is made more sensitive to larger motion amplitudes by processing the motion posture square. Specifically:
[0031] J=c1Z 2 +c2θ 2 +c3ψ 2 ;
[0032] c i is the weight coefficient, which is used to adjust the importance of different motion postures;
[0033] The inequality constraints are:
[0034] F l≤F i ≤F u ;
[0035] L l ≤L i ≤L u ;
[0036] J l ≤J i ≤J u ;
[0037] 0≤T i ≤T u ;
[0038] Among them, F l , L l , J l It is expressed as the lower limit of cable tension, cable length and pipe section posture respectively; F u , L u , J u 、T u It is expressed as the upper limit of the cable tension, cable length, pipe section posture and sinking time;
[0039] Step S5: adopt the penalty function method and introduce the step function to transform the constrained optimization problem into an unconstrained optimization problem, and use the optimization algorithm to perform iterative calculations to find the optimal solution to the problem space, specifically:
[0040]
[0041]
[0042] Among them, u j (g i (X)) is a step function, M is a penalty factor;
[0043] Step S6: Based on the model information database and the optimization mathematical model, a method for optimizing the cable force for laying long and irregular pipelines under multi-objective requirements is proposed;
[0044] Among them, the specific process of the cable force optimization method for long and large irregular pipelines under multi-objective requirements includes the following steps:
[0045] Step a: According to the project's requirements for pipeline laying accuracy and construction period, determine the design variables and constraints, and establish the objective function for optimizing the cable force for laying long and irregular pipelines;
[0046] Step b: Use the training set to train the pipeline laying posture and laying time prediction model, and use the test set to perform error analysis and performance evaluation on the established model; if the accuracy does not meet the preset standard, additional data needs to be added for retraining; if it has met the standard, it can be used directly in the future;
[0047] Step c: Filter out the cable adjustment control parameters that meet the constraints from the current test database, denoted as
[0048] Step d: Use the optimization algorithm to solve the multi-objective long large irregular pipeline laying cable optimization mathematical model, and obtain the candidate solution, which is the optimal cable adjustment control parameter, denoted as
[0049] Step S7: and Merge into {X i} M+N , according to the importance of each optimization goal, determine the weight vector pair {X i} M+N Compare and select the decision-making schemes in the paper to obtain the theoretical optimal cable adjustment control parameter x # ;
[0050] Step S8: Combine the finite element software to process the theoretical optimal cable adjustment control parameters selected above, and finally obtain the optimal cable adjustment control parameters x ## , and update the database;
[0051] Step S9: After the model training is completed, data processing and analysis can be quickly performed based on the data feedback during the pipeline laying process to dynamically adjust the cable laying strategy.
[0052] Among them, in step d, the step of obtaining the candidate solution as the optimal cable adjustment control parameter includes:
[0053] Step d1: Initialize the population space range S, population size N, iteration number I, and allocate population positions and individual velocities in the problem space. Evaluate the fitness values of all particles in the current population P;
[0054] Step d2, adopting the differential evolution strategy and utilizing the individual differences of particles to update the position, using global information to avoid falling into the local optimal solution, and updating the position of the population particles to be close to the optimal position P;
[0055] Step d3, introduce the flight strategy and add nonlinear perturbation factors to speed up the individual approaching the optimal position to update the solution space and recalculate the target values of each particle;
[0056] Step d4: Whether the obtained particles are close to the global optimal solution is used as the convergence criterion. If convergence is not achieved, return to step d2 and continue to update the population; if convergence is achieved, output the population P as the optimal solution.
[0057] Wherein, in said step S1, the collected data comes from finite element software data, specifically: using Orcaflex finite element software, a finite element model is established according to the flow field environment, pipeline structure, and laying scheme information, and the accuracy of the finite element model is enhanced by reasonably setting the construction conditions, so as to obtain relevant data of the pipeline laying model;
[0058] Wherein, in said step S1, a data preprocessing step is also included;
[0059] The data preprocessing step is to preprocess the data through data cleaning and normalization, eliminate abnormal data and eliminate the dimensional influence between data.
[0060] Among them, in step S2, the multiple machine learning prediction models are LSTM machine learning prediction model, GBT machine learning prediction model, and SVR machine learning prediction model respectively.
[0061] Compared with the prior art, the present invention has the following beneficial effects, specifically:
[0062] 1. The intelligent optimization method of cable force in the laying process of long and irregular pipelines can effectively solve the problems of large experience dependence, low operation efficiency, limited laying accuracy and difficulty in fast dynamic coordination of multiple suspension cables in manual cable adjustment. The present invention introduces machine learning algorithms to extract, select and construct key features from multi-dimensional and highly complex raw data, train pipeline motion posture and laying time prediction models, and effectively capture multi-factor coupling dynamic interactions for accurate prediction and effective control. It can achieve the optimal balance between multiple objectives and effectively avoid performance reduction caused by single-objective optimization.
[0063] 2. The method of the present invention is used to regulate and control the laying of cables in pipelines, which can simultaneously meet the requirements of pipeline laying accuracy, optimization of cable tension at multiple locations, and the specified construction period. By optimizing the laying speed and height, problems such as pipeline collision and cable overload caused by excessive speed or improper height can be avoided, reducing potential safety hazards during the construction process; at the same time, the influence of multiple factors can be comprehensively considered to ensure the laying accuracy of pipelines; and according to the real-time feedback data during the laying of pipelines, the laying speed and height can be quickly adjusted to ensure the stability and safety of the laying process, reduce the time of pauses and rework, and improve construction efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] The present invention is further described below with the aid of the accompanying drawings. However, the embodiments in the accompanying drawings do not constitute any limitation to the present invention.
[0065] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION
[0066] The present invention is described below in conjunction with specific implementation methods.
[0067] An intelligent optimization method for cable force during the laying of long and irregular pipelines. Figure 1 As shown, the following steps are included, specifically:
[0068] Step S1, collecting data and constructing an original data set of pipeline laying information;
[0069] Step S2: Establish multiple machine learning prediction models:
[0070] Based on the data set obtained in step S1, training is performed, and the incoming flow velocity, cable pretension, cable release height, and cable release speed are selected as input quantities of each machine learning prediction model, and the pipeline movement posture and total sinking time are selected as output quantities of each machine learning prediction model;
[0071] The motion posture and total sinking time of long and irregular pipelines are affected by multiple factors, and there is a complex nonlinear relationship between each factor and the target variable. The key feature set is determined based on the RFE strategy and the optimal parameter set is determined using the learning rate adjustment strategy.
[0072] Comprehensively compare and analyze the prediction performance of each machine learning model, and select the machine learning prediction model with the best accuracy for subsequent constraints;
[0073] Step S3: Based on the machine learning prediction model selected in step S2, a multi-objective function is constructed to optimize the cable tension, and then an objective function for optimizing the cable force of a long irregular pipeline is established. The objective function is:
[0074] f(x)=w1F+w2Var;
[0075] F=max[F1,F2,...,F N ];
[0076]
[0077]
[0078] Where f(x) represents the objective function, w1 and w2 are the weights corresponding to each target value, and the weights are determined according to the importance of the target; F is the maximum tension value of each cable, Var is the variance of the cable tension; F i is the tension of the cable, L is the length of the cable at a certain moment, L0 is the initial length of the cable at a certain moment; c is the damping coefficient of the cable, k is the stiffness of the cable;
[0079]
[0080] F fis the fluid force on the structure, Δ is the displacement, a f is the fluid acceleration relative to the earth, C a is the inertia coefficient of the structure, a r is the fluid acceleration relative to the structure; ρ w is the density of water, V r is the flow velocity relative to the structure, C d is the drag coefficient, A is the resistance area, and the value is determined according to the actual project;
[0081] Step S4: Set constraints:
[0082] The constraints are the pipe section posture constraint, the sinking period constraint and the objective function value constraint; among them, the pipe section posture constraint and the sinking period constraint are equality constraints, and the objective function value constraint is an inequality constraint. i (x) = 0 is the equality constraint function, g i (x) = 0 is the inequality constraint function;
[0083] The equality constraints are:
[0084] Z t =Z;
[0085] θ t =θ;
[0086] ψ t =ψ;
[0087] T t =T;
[0088] Among them, Z t ,θ t , t is the target value of the pipeline attitude, Z, θ, ψ represent the pipeline attitude prediction model that has been trained and can be called in step S2, and Z, θ, ψ are the heave, pitch, and yaw values of the pipeline at different positions respectively; T t is the target value of the time required to lower the cable to the specified position, and T represents the trained prediction model for the lowering time;
[0089] A weighted square sum function is constructed to quantitatively evaluate the motion posture of the pipe segment. The function is made more sensitive to larger motion amplitudes by processing the motion posture square. Specifically:
[0090] J=c1Z 2 +c2θ 2 +c3ψ 2 ;
[0091] c i is the weight coefficient, which is used to adjust the importance of different motion postures;
[0092] The inequality constraints are:
[0093] F l ≤F i ≤F u ;
[0094] L l ≤L i ≤L u ;
[0095] J l ≤J i ≤J u ;
[0096] 0≤T i ≤T u ;
[0097] Among them, F l , L l , J l It is expressed as the lower limit of cable tension, cable length and pipe section posture respectively; F u , L u , J u 、T u It is expressed as the upper limit of the cable tension, cable length, pipe section posture and sinking time;
[0098] Step S5: adopt the penalty function method and introduce the step function to transform the constrained optimization problem into an unconstrained optimization problem, and use the optimization algorithm to perform iterative calculations to find the optimal solution to the problem space, specifically:
[0099]
[0100]
[0101] Among them, u j (g i (X)) is a step function, M is a penalty factor;
[0102] Step S6: Based on the model information database and the optimization mathematical model, a method for optimizing the cable force for laying long and irregular pipelines under multi-objective requirements is proposed;
[0103] Among them, the specific process of the cable force optimization method for long and large irregular pipelines under multi-objective requirements includes the following steps:
[0104] Step a: According to the project's requirements for pipeline laying accuracy and construction period, determine the design variables and constraints, and establish the objective function for optimizing the cable force for laying long and irregular pipelines;
[0105] Step b: Use the training set to train the pipeline laying posture and laying time prediction model, and use the test set to perform error analysis and performance evaluation on the established model; if the accuracy does not meet the preset standard, additional data needs to be added for retraining; if it has met the standard, it can be used directly in the future;
[0106] Step c: Filter out the cable adjustment control parameters that meet the constraints from the current test database, denoted as
[0107] Step d: Use the optimization algorithm to solve the multi-objective long large irregular pipeline laying cable optimization mathematical model, and obtain the candidate solution, which is the optimal cable adjustment control parameter, denoted as
[0108] Step S7: and Merge into {X i} M+N , according to the importance of each optimization goal, determine the weight vector pair {X i} M+N Compare and select the decision-making schemes in the paper to obtain the theoretical optimal cable adjustment control parameter x # ;
[0109] Step S8: Combine the finite element software to process the theoretical optimal cable adjustment control parameters selected above, and finally obtain the optimal cable adjustment control parameters x ## , and update the database;
[0110] Step S9: After the model training is completed, data processing and analysis can be quickly performed based on the data feedback during the pipeline laying process to dynamically adjust the cable laying strategy.
[0111] It should be noted that in step d, the step of obtaining the candidate solution, i.e., the optimal cable adjustment control parameter, includes:
[0112] Step d1: Initialize the population space range S, population size N, iteration number I, and allocate population positions and individual velocities in the problem space. Evaluate the fitness values of all particles in the current population P;
[0113] Step d2, adopting the differential evolution strategy and utilizing the individual differences of particles to update the position, using global information to avoid falling into the local optimal solution, and updating the position of the population particles to be close to the optimal position P;
[0114] Step d3, introduce the flight strategy and add nonlinear perturbation factors to speed up the individual approaching the optimal position to update the solution space and recalculate the target values of each particle;
[0115] Step d4: Whether the obtained particles are close to the global optimal solution is used as the convergence criterion. If convergence is not achieved, return to step d2 and continue to update the population; if convergence is achieved, output the population P as the optimal solution.
[0116] Wherein, in said step S1, the collected data comes from finite element software data, specifically: using Orcaflex finite element software, a finite element model is established according to the flow field environment, pipeline structure, and laying scheme information, and the accuracy of the finite element model is enhanced by reasonably setting the construction conditions, so as to obtain relevant data of the pipeline laying model;
[0117] In addition, in the step S1, a data preprocessing step is also included;
[0118] The data preprocessing step is to preprocess the data through data cleaning and normalization, eliminate abnormal data and eliminate the dimensional influence between data.
[0119] In addition, in step S2, the multiple machine learning prediction models are, respectively, LSTM machine learning prediction model (long short-term memory network machine learning prediction model), GBT machine learning prediction model (gradient boosting tree machine learning prediction model), and SVR machine learning prediction model (support vector regression machine learning prediction model).
[0120] Through the above steps, the intelligent optimization method for cable force in the long irregular pipeline laying process of the present invention has the following advantages, specifically:
[0121] 1. The intelligent optimization method of cable force in the laying process of long and irregular pipelines can effectively solve the problems of large experience dependence, low operation efficiency, limited laying accuracy and difficulty in fast dynamic coordination of multiple suspension cables in manual cable adjustment. The present invention introduces machine learning algorithms to extract, select and construct key features from multi-dimensional and highly complex raw data, train pipeline motion posture and laying time prediction models, and effectively capture multi-factor coupling dynamic interactions for accurate prediction and effective control. It can achieve the optimal balance between multiple objectives and effectively avoid performance reduction caused by single-objective optimization.
[0122] 2. The method of the present invention is used to regulate and control the laying of cables in pipelines, which can simultaneously meet the requirements of pipeline laying accuracy, optimization of cable tension at multiple locations, and the specified construction period. By optimizing the laying speed and height, problems such as pipeline collision and cable overload caused by excessive speed or improper height can be avoided, reducing potential safety hazards during the construction process; at the same time, the influence of multiple factors can be comprehensively considered to ensure the laying accuracy of pipelines; and according to the real-time feedback data during the laying of pipelines, the laying speed and height can be quickly adjusted to ensure the stability and safety of the laying process, reduce the time of pauses and rework, and improve construction efficiency.
[0123] The above contents are only preferred embodiments of the present invention. For ordinary technicians in this field, according to the concept of the present invention, there will be changes in the specific implementation methods and application scopes. The content of this specification should not be understood as limiting the present invention.
Claims
1. A method for intelligent optimization of cable force during the laying of long and irregular pipelines, characterized in that: The following steps are included, specifically: Step S1, collecting data and constructing an original data set of pipeline laying information; Step S2: Establish multiple machine learning prediction models: Based on the data set obtained in step S1, training is performed, and the incoming flow velocity, cable pretension, cable release height, and cable release speed are selected as input quantities of each machine learning prediction model, and the pipeline movement posture and total sinking time are selected as output quantities of each machine learning prediction model; Determine the key feature set based on the RFE strategy and use the learning rate adjustment strategy to determine the optimal parameter set; Comprehensively compare and analyze the prediction performance of each machine learning model, and select the machine learning prediction model with the best accuracy for subsequent constraints; Step S3: Based on the machine learning prediction model selected in step S2, a multi-objective function is constructed to optimize the cable tension, and then an objective function for optimizing the cable force of a long irregular pipeline is established. The objective function is: f(x)=w1F+w2Var; F=max[F1,F2,...,F N ]; Where f(x) represents the objective function, w1 and w2 are the weights corresponding to each target value, and the weights are determined according to the importance of the target; F is the maximum tension value of each cable, Var is the variance of the cable tension; F i is the tension of the cable, L is the length of the cable at a certain moment, L0 is the initial length of the cable at a certain moment; c is the damping coefficient of the cable, k is the stiffness of the cable; F f is the fluid force on the structure, Δ is the displacement, a f is the fluid acceleration relative to the earth, C a is the inertia coefficient of the structure, a r is the fluid acceleration relative to the structure; ρ w is the density of water, V r is the flow velocity relative to the structure, C d is the drag coefficient, A is the resistance area, and the value is determined according to the actual project; Step S4: Set constraints: The constraints are the pipe section posture constraint, the sinking period constraint and the objective function value constraint; among them, the pipe section posture constraint and the sinking period constraint are equality constraints, and the objective function value constraint is an inequality constraint. i (x) = 0 is the equality constraint function, g i (x) = 0 is the inequality constraint function; The equality constraints are: WITH t =Z; i t =θ; ψ t =ψ; T t =T; Among them, Z t ,θ t , t is the target value of the pipeline attitude, Z, θ, ψ represent the pipeline attitude prediction model that has been trained and can be called in step S2, and Z, θ, ψ are the heave, pitch, and yaw values of the pipeline at different positions respectively; T t is the target value of the time required to lower the cable to the specified position, and T represents the trained prediction model for the lowering time; A weighted square sum function is constructed to quantitatively evaluate the motion posture of the pipe segment. The function is made more sensitive to larger motion amplitudes by processing the motion posture square. Specifically: J=c1Z 2 +c2θ 2 +c3ψ 2 ; c i is the weight coefficient, which is used to adjust the importance of different motion postures; The inequality constraints are: F l ≤F i ≤F u ; L l ≤L i ≤L u ; J l ≤J i ≤J u ; 0≤T i ≤T u ; Among them, F l , L l , J l It is expressed as the lower limit of cable tension, cable length and pipe section posture respectively; F u , L u , J u , T u It is expressed as the upper limit of the cable tension, cable length, pipe section posture and sinking time; Step S5: adopt the penalty function method and introduce the step function to transform the constrained optimization problem into an unconstrained optimization problem, and use the optimization algorithm to perform iterative calculations to find the optimal solution to the problem space, specifically: Among them, u j (g i (X)) is a step function, M is a penalty factor; Step S6: Based on the model information database and the optimization mathematical model, a method for optimizing the cable force for laying long and irregular pipelines under multi-objective requirements is proposed; Among them, the specific process of the cable force optimization method for long and large irregular pipelines under multi-objective requirements includes the following steps: Step a: According to the project's requirements for pipeline laying accuracy and construction period, determine the design variables and constraints, and establish the objective function for optimizing the cable force for laying long and irregular pipelines; Step b: Use the training set to train the pipeline laying posture and laying time prediction model, and use the test set to perform error analysis and performance evaluation on the established model; if the accuracy does not meet the preset standard, additional data needs to be added for retraining; if it has met the standard, it can be used directly in the future; Step c: Filter out the cable adjustment control parameters that meet the constraints from the current test database, denoted as Step d: Use the optimization algorithm to solve the multi-objective long large irregular pipeline laying cable optimization mathematical model, and obtain the candidate solution, which is the optimal cable adjustment control parameter, denoted as Step S7: and Merge into {X i } M+N , according to the importance of each optimization goal, determine the weight vector pair {X i } M+N Compare and select the decision-making schemes in the paper to obtain the theoretical optimal cable adjustment control parameter x # ; Step S8: Combine the finite element software to process the theoretical optimal cable adjustment control parameters selected above, and finally obtain the optimal cable adjustment control parameters x ## , and update the database; Step S9: After the model training is completed, data processing and analysis can be quickly performed based on the data feedback during the pipeline laying process to dynamically adjust the cable laying strategy.
2. According to claim 1, a method for intelligent optimization of cable force during the laying of a long irregular pipeline, characterized in that: In step d, the step of obtaining a candidate solution, i.e., the optimal cable adjustment control parameter, includes: Step d1: Initialize the population space range S, population size N, iteration number I, and allocate population positions and individual velocities in the problem space. Evaluate the fitness values of all particles in the current population P; Step d2, adopting the differential evolution strategy and utilizing the individual differences of particles to update the position, using global information to avoid falling into the local optimal solution, and updating the position of the population particles to be close to the optimal position P; Step d3, introduce the flight strategy and add nonlinear perturbation factors to speed up the individual approaching the optimal position to update the solution space and recalculate the target values of each particle; Step d4: Whether the obtained particles are close to the global optimal solution is used as the convergence criterion. If convergence is not reached, return to step d2 and continue to update the population; If converged, the population P is output as the optimal solution.
3. The intelligent optimization method for cable force in the laying process of a long irregular pipeline according to claim 1 is characterized in that: In step S1, the collected data comes from finite element software data. Specifically, Orcaflex finite element software is used to establish a finite element model based on the flow field environment, pipeline structure, and laying plan information, and the accuracy of the finite element model is enhanced by reasonably setting the construction conditions to obtain relevant data of the pipeline laying model.
4. The intelligent optimization method for cable force in the laying process of a long irregular pipeline according to claim 1 is characterized in that: In the step S1, a data preprocessing step is also included; The data preprocessing step is to preprocess the data through data cleaning and normalization, eliminate abnormal data and eliminate the dimensional influence between data.
5. The intelligent optimization method for cable force in the laying process of a long irregular pipeline according to claim 1 is characterized in that: In step S2, the multiple machine learning prediction models are, in sequence, an LSTM machine learning prediction model, a GBT machine learning prediction model, and an SVR machine learning prediction model.
