An intelligent management platform and method for pipeline high-consequence areas integrated with digital twins

Abstract

The present invention belongs to the technical field of pipeline management, and discloses an intelligent management platform and method for pipeline high consequence areas with integrated digital twins; the platform and method include: acquiring monitoring data of pipeline high consequence areas, constructing a digital twin model of the pipeline high consequence areas, applying the monitoring data to the digital twin model, calculating the risk value of failure in the pipeline high consequence areas, presetting a risk threshold, and triggering a maintenance strategy for the pipeline high consequence areas when the risk value is greater than or equal to the risk threshold; inputting the maintenance strategy into the digital twin model, and optimizing the maintenance strategy using an optimization algorithm; obtaining an optimal maintenance strategy, and performing maintenance operations on the pipeline high consequence areas according to the optimal maintenance strategy, so as to dynamically evaluate the health status of the pipeline high consequence areas, timely discover potential risks, and improve the safety and reliability of pipeline operation.

Description

Technical Field

[0001] The present invention relates to the technical field of pipeline management, and more specifically, to an intelligent management platform and method for pipeline high-consequence areas integrating digital twins. Background Art

[0002] The patent application with publication number CN116642136A discloses a pipeline risk monitoring method, system and storage medium based on digital twins, including: obtaining multi-source attribute data of the target pipe section at the current moment and inputting it into a preset risk assessment formula group to obtain the current comprehensive risk value of the target pipe section; using the multi-source attribute data at the current moment to update the original digital twin model corresponding to the target pipe section to obtain the target digital twin model at the current moment; when the current comprehensive risk value exceeds the preset comprehensive risk value, the risk existing in the target pipe section at the current moment is marked and mapped on the target digital twin model to realize risk monitoring of the target pipe section; the pipeline risk monitoring cycle is greatly shortened, and the digital twin model can be used to evaluate and monitor pipeline risks in real time, thereby improving the accuracy of risk monitoring.

[0003] However, there is still a lack of a platform for comprehensive assessment and intelligent management of pipeline high consequence areas; pipeline high consequence areas refer to pipeline areas where, once a failure occurs, serious consequences (such as casualties, environmental pollution, economic losses, etc.) will occur; due to the importance of these areas, special attention and management are required; but the current management methods mostly rely on manual experience and lack scientific risk assessment models, making it difficult to timely discover potential risks and hidden dangers in pipeline high consequence areas, and it is also impossible to formulate effective maintenance strategies; secondly, the existing methods cannot accurately simulate the characteristics of pipeline high consequence areas; pipeline high consequence areas usually have complex structures and operating conditions (such as multiphase flow, corrosion, etc.), which have a significant impact on the health status and failure risk of the pipeline; but the existing simulation methods are too simple to truly reflect the actual situation of pipeline high consequence areas, resulting in large deviations in risk assessment results; furthermore, the existing methods lack methods for optimizing maintenance strategies; maintenance strategies include multiple methods, and there are complex mutual constraints between these methods; simple empirical maintenance strategies make it difficult to balance various goals and cannot obtain the optimal comprehensive maintenance plan.

[0004] In view of this, the present invention proposes an intelligent management platform and method for pipeline high-consequence areas integrating digital twins to solve the above problems. Summary of the invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above-mentioned purpose, the present invention provides the following technical solutions: an intelligent management platform for pipeline high-consequence areas with integrated digital twins, comprising: a data acquisition module for acquiring monitoring data of pipeline high-consequence areas;

[0006] The comprehensive strategy maintenance module is used to build a digital twin model of the pipeline high-consequence area, apply the monitoring data to the digital twin model, calculate the risk value of failure in the pipeline high-consequence area, preset the risk threshold, and trigger the maintenance strategy of the pipeline high-consequence area when the risk value is greater than or equal to the risk threshold;

[0007] The strategy optimization module is used to input the maintenance strategy into the digital twin model and optimize the maintenance strategy using an optimization algorithm; the optimal maintenance strategy is obtained, and maintenance operations are performed on the high-consequence area of ​​the pipeline according to the optimal maintenance strategy. The modules are connected by wired and / or wireless means.

[0008] Furthermore, the monitoring data includes pipeline internal environment data, pipeline external environment data, pipeline structure health data and pipeline operation data.

[0009] Furthermore, the digital twin model is constructed by:

[0010] Collect basic parameters of high consequence areas of pipelines, including pipeline materials, geometric data and pipeline connection methods; geometric data include pipe diameter, wall thickness and length; pipeline connection method is welding;

[0011] Based on the geometric structure data of the high-consequence area of ​​the pipeline, a physical model of the high-consequence area of ​​the pipeline is established; the physical model is refined based on the pipeline connection method to obtain a perfect physical model; the perfect physical model is adaptively divided to obtain a preliminary twin model; the preliminary twin model is assigned to the pipeline material to obtain a digital twin model of the high-consequence area of ​​the pipeline.

[0012] Furthermore, the method of refining the physical model includes:

[0013] Get the weld geometry at the weld, which includes weld width, weld height and weld angle; for the welding position, define the division interval [-k, k], where k is a positive integer greater than or equal to 1, place the welding position at the zero point of the division interval, use the physical model within the division interval as the pipe segment model, divide the pipe segment model into several simple circular rings, and calculate the inner radius of the simple circular ring and outer radius ;

[0014] inner radius ;

[0015] Outer Radius ;

[0016] in, is the pipe radius, is the wall thickness, is the weld height, is the length of the transition region, is the radius of curvature of the transition area; is the weld angle, is the inner radius shape correction factor, is the outer radius shape correction factor;

[0017] Based on the calculated inner radius of a simple torus and outer radius , a number of simple circular rings are subjected to a union Boolean operation to obtain a weld model;

[0018] The weld model and the physical model within the divided interval are imported into the same coordinate system. According to the position and direction of the weld, the weld model is translated and rotated, and then assembled with the pipe segment model to obtain a complete physical model.

[0019] Furthermore, the method of adaptively dividing the improved physical model includes:

[0020] Identify the key areas in the perfect physical model; decompose the perfect physical model into M sub-areas, each of which contains one or more key areas; perform special divisions on the sub-areas whose number of key areas is greater than or equal to the preset division threshold, and perform ordinary divisions on the sub-areas whose number of key areas is less than the preset division threshold, to obtain several grids; for the grids between adjacent sub-areas, perform grid docking, and integrate the grids of each sub-area into a preliminary twin model;

[0021] The identification method of the key area includes:

[0022] Divide the perfect physical model into m preliminary regions, and divide the preliminary regions into n curves, and calculate the curvature of each curve ;

[0023] ;

[0024] in, is the curvature weighting function, is the parametric equation of the curve, is the parameter variable of the curve, for The first derivative of for The second derivative, is the torsion rate of the curve, is the weight coefficient;

[0025] A curvature threshold is preset. If the mean curvature of all curves in the preliminary area is greater than the curvature threshold, the corresponding preliminary area is taken as the key area.

[0026] Special divisions include:

[0027] All key areas in the sub-area are taken as initial frontier units. For each frontier unit, several new frontier units are generated according to the preset size, and the new frontier units are added to the preset active frontier set.

[0028] Check whether the newly added frontier unit intersects with the existing frontier unit. For the two intersecting frontier units, perform unit clipping, that is, remove the intersecting parts from the two frontier units; until the active frontier set is empty and the frontier unit corresponds to the grid.

[0029] Furthermore, the curvature weighting function is obtained by:

[0030] Collecting N groups of curve data, the curve data including geometric parameters of the curve and corresponding key area marks; extracting geometric features from the curve data, the geometric features including curvature, torsion rate and high-order derivatives of the parametric equation of the curve;

[0031] definition The posterior probability ;

[0032] in, is the geometric feature, Mark the key areas. is the likelihood function, is the prior probability, is proportional to the sign of ; the prior probability ;in, is the mean function of the prior probability, is the covariance function of the prior probability; is a multivariate normal distribution;

[0033] Based on the posterior probability, calculate the curvature weighting function The posterior probability distribution of is obtained, and sample points of the curvature weighted function are extracted from the posterior probability distribution; based on the extracted sample points, the curvature weighted function is fitted;

[0034] The sampling point extraction method includes:

[0035] Define a Markov chain, whose state space is defined as the parameter space of the curvature weighted function, and define a proposal distribution; randomly select a sample point from the prior distribution obtained based on the prior probability as the initial state, that is, the starting point of the Markov chain;

[0036] For the initial state, a new sample point is extracted from the proposed distribution, the posterior probability value of the new sample point is calculated, and the acceptance rate is calculated based on the posterior probability values ​​of the initial state and the new sample point;

[0037] The formula for calculating the acceptance rate is:

[0038] ;

[0039] in, is the new sample point, For new sample points The value of the posterior probability is is the initial state, for The value of the posterior probability is For the time step When, from arrive The proposed distribution of For the time step When, from arrive Proposed distribution of; is the adaptively adjusted variance function; is the acceptance rate;

[0040] Adaptive variance function ;in, From the initial state to the time step The covariance matrix of all sample points sampled so far; is a positive constant; is the identity matrix;

[0041] Define a uniform distribution, draw a random number from a uniform distribution, if the random number is less than the acceptance rate, accept the new sample point as the next state; otherwise, keep the current initial state unchanged; repeat until the Markov chain converges to the posterior probability distribution;

[0042] After the Markov chain converges, a fixed number of sample points are extracted from the Markov chain. These sample points are the sample points of the curvature weighted function extracted from the posterior probability distribution.

[0043] Furthermore, the risk value is calculated by:

[0044] The internal environment data of the pipeline is used as the environmental load of the digital twin model, and the external environment data of the pipeline is used as the boundary condition of the digital twin model; the pipeline structure health data is updated to the corresponding position of the digital twin model; the pipeline operation data is used as the working condition load of the digital twin model;

[0045] The finite element numerical calculation method is used to solve the stress of each grid of the digital twin model at this time; the failure criterion function is defined , substitute the stress of each grid into the failure criterion function, and then calculate the corresponding failure index of each grid; calculate the failure probability of the grid according to the distribution of the failure index; perform weighted summation on the failure probabilities of all grids to obtain the probability of failure in the high consequence area of ​​the entire pipeline, that is, the risk value.

[0046] Furthermore, the failure criterion function The formula is:

[0047] ML= [ ( s 1 - s 2) 2 + ( s 2 - s 3) 2 + ( s 3 - s 1) 2 ] (2 × (1+ φ× f( oh , d ))) ;

[0048] in, , and are the three principal stresses of the mesh, is the correction factor; is the stress direction function, is the stress tensor, is the stress direction angle;

[0049] For isotropic materials:

[0050] ;

[0051] For anisotropic materials:

[0052] ;

[0053] in, , , , , , and is a parameter to be determined;

[0054] Failure indicators ,in, is the yield strength of the material, For Grid The value of the failure criterion function;

[0055] The calculation method of failure probability includes:

[0056] For each grid, the failure index is calculated by Monte Carlo simulation. Draw a number of samples from the distribution of and calculate the failure index in the samples The weighted ratio less than 0 is the failure probability;

[0057] The formula for calculating the weighted ratio is:

[0058] Pf= 1 N 3 × ∑ [ e1 ×( 1+ α1× exp( - β1× d) ) +e 2 ×( 1+ c1 × ML(i) Sy ) ] × g_ j ;

[0059] in, is the total number of samples, , and To adjust the parameters, and is the equilibrium parameter, and , is the average distance from the sample to all failure indicators, For the The value of the failure index of the sample, is the weighted ratio.

[0060] Furthermore, the method for obtaining the optimal maintenance strategy includes:

[0061] The maintenance strategy is encoded into a high-dimensional vector, where each dimension corresponds to a decision variable. The decision variables include maintenance time point, maintenance method, maintenance location, maintenance material and maintenance process.

[0062] Randomly generate a fixed number of initial plants in the solution space to form a seed population; define the growth function ;in, is the maintenance cost objective function, Cost weight, is the maintenance time objective function, is the time weight, is the objective function of risk reduction, Reduce weight for risk;

[0063] The maintenance cost objective function is a discretized function of the sum of maintenance material cost, maintenance process cost and maintenance labor cost; the maintenance time objective function is a discretized function of the sum of preparation maintenance time, actual maintenance time and acceptance time;

[0064] Risk reduction objective function ;in, is the risk value after maintenance, is the risk value before maintenance;

[0065] For each plant, calculate its value on the growth function, record it as the growth value, and calculate its growth probability based on the growth value; the higher the growth probability, the greater the chance that the plant will survive and reproduce in the next generation;

[0066] The growth probability is calculated as:

[0067] G_v= (exp( φ1 × (f_v - f_avg)) × (1+ e1 × (f_v - f_med) ϑ1 )) ∑ [ (exp( φ1 × (f_v - f_avg)) × (1+ e1 × (f_v - f_med) ϑ1 )) ] ;

[0068] in, For the The growth probability of each plant, For the The growth value of each plant, is the average growth value of all plants in the current seed population, is the median of all plant growth values ​​in the current seed population; To adjust the parameters; is the balance coefficient, is the index parameter;

[0069] Preset the growth threshold, select the plants with growth probability greater than the growth threshold from the current seed population as the parent plants according to the growth probability, generate new daughter plants through crossover and mutation operations; merge the parent plants and daughter plants to form a temporary population; sort the plants in the temporary population in descending order according to the growth value, and select the first Q plants as the next generation population;

[0070] Until the maximum number of iterations is reached or the population converges, and the plant with the highest current growth value is output as the optimal solution, which is the optimal maintenance strategy.

[0071] A pipeline high-consequence area intelligent management method integrated with digital twins, which is implemented based on the pipeline high-consequence area intelligent management platform integrated with digital twins, comprising: S1, obtaining monitoring data of the pipeline high-consequence area;

[0072] S2. Build a digital twin model of the pipeline high consequence area, apply the monitoring data to the digital twin model, calculate the risk value of failure in the pipeline high consequence area, preset the risk threshold, and trigger the maintenance strategy of the pipeline high consequence area when the risk value is greater than or equal to the risk threshold;

[0073] S3. Input the maintenance strategy into the digital twin model and use the optimization algorithm to optimize the maintenance strategy; obtain the optimal maintenance strategy, and implement maintenance operations on the high consequence area of ​​the pipeline according to the optimal maintenance strategy.

[0074] The technical effects and advantages of the pipeline high-consequence area intelligent management platform and method integrated with digital twins of the present invention are as follows:

[0075] The present invention constructs a digital twin model of the pipeline high consequence area, which can accurately simulate the characteristics of the pipeline high consequence area, so as to accurately calculate the risk value of failure in the pipeline high consequence area; the risk assessment method based on digital twins is more scientific and reliable than the traditional method; secondly, the maintenance strategy is triggered according to the risk value, and the maintenance strategy is optimized by the optimization algorithm, so as to obtain the optimal maintenance strategy, thereby maximizing the maintenance efficiency and reducing the maintenance cost; the optimized maintenance strategy not only considers the degree of risk reduction, but also weighs the maintenance cost and maintenance time, realizes multi-objective optimization, and ensures the economy and timeliness of maintenance; furthermore, the real-time monitoring data is applied to the digital twin model, which can dynamically evaluate the health status of the pipeline high consequence area, timely discover risk hazards, and improve the safety and reliability of pipeline operation. Active prevention is achieved, major accidents are avoided, and the safety of personnel and the environment is guaranteed. BRIEF DESCRIPTION OF THE DRAWINGS

[0076] Figure 1 It is a schematic diagram of an intelligent management platform for pipeline high-consequence areas integrating digital twins of the present invention;

[0077] Figure 2 This is a schematic diagram of an intelligent management method for high-consequence areas in pipelines integrating digital twins according to the present invention. DETAILED DESCRIPTION

[0078] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0079] Example 1

[0080] See also Figure 1 As shown, this embodiment provides an intelligent management platform for pipeline high-consequence areas integrating digital twins, including:

[0081] A data acquisition module, used to obtain monitoring data of high consequence areas of pipelines;

[0082] The comprehensive strategy maintenance module is used to build a digital twin model of the pipeline high-consequence area, apply the monitoring data to the digital twin model, calculate the risk value of failure in the pipeline high-consequence area, preset the risk threshold, and trigger the maintenance strategy of the pipeline high-consequence area when the risk value is greater than or equal to the risk threshold;

[0083] The strategy optimization module is used to input the maintenance strategy into the digital twin model and optimize the maintenance strategy using an optimization algorithm; the optimal maintenance strategy is obtained, and maintenance operations are performed on the high-consequence areas of the pipeline according to the optimal maintenance strategy. The modules are connected by wired and / or wireless means to achieve data transmission between modules.

[0084] Furthermore, the monitoring data includes pipeline internal environment data, pipeline external environment data, pipeline structural health data and pipeline operation data.

[0085] Specifically, the internal environment data of the pipeline include medium temperature, medium pressure, medium flow velocity, and medium flow rate; the external environment data of the pipeline include soil temperature, soil moisture, soil pH, and climate environment (temperature, humidity, wind speed, etc.); the pipeline structural health data include the location and size of pipe body defects (cracks, local expansion of corrosion, etc.); the pipeline operation data include working pressure and working temperature.

[0086] It comes from multiple channels including on-site detection (such as intelligent detection robots, infrared thermal imagers, ultrasound, etc.), online monitoring systems (such as pressure, temperature, flow, etc. sensors), historical operation and maintenance records, etc.

[0087] Furthermore, the digital twin model is constructed in the following ways:

[0088] Collect basic parameters of high consequence areas of pipelines, including pipeline materials, geometric data and pipeline connection methods; geometric data include pipe diameter, wall thickness and length; pipeline connection method is welding.

[0089] Based on the geometric structure data of the high-consequence area of ​​the pipeline, a physical model of the high-consequence area of ​​the pipeline is established; the physical model is refined based on the pipeline connection method to obtain a perfect physical model; the perfect physical model is adaptively divided to obtain a preliminary twin model; the preliminary twin model is assigned to the pipeline material to obtain a digital twin model of the high-consequence area of ​​the pipeline.

[0090] Use computer-aided design (CAD) software or geometric modeling software to create a physical model of the high-consequence area of ​​the pipeline.

[0091] Ways to refine the physical model include:

[0092] Get the weld geometry at the weld, which includes weld width, weld height and weld angle; for the welding position, define the division interval [-k, k], where k is a positive integer greater than or equal to 1, place the welding position at the zero point of the division interval, use the physical model within the division interval as the pipe segment model, divide the pipe segment model into several simple circular rings, and calculate the inner radius of the simple circular ring and outer radius .

[0093] inner radius ;

[0094] Outer Radius ;in, is the pipe section radius (i.e. the diameter of a pipe section), is the wall thickness, is the weld height, is the length of the transition region, is the radius of curvature of the transition area; is the weld angle, is the inner radius shape correction factor, is the outer radius shape correction factor.

[0095] The weld area is usually not a simple annular surface, but has a certain transition area; the length of the transition area refers to the length of the transition area from the pipe section diameter to the weld height; the transition area is usually not a straight line segment, but a curve, and the curvature radius of this curve is the curvature radius of the transition area.

[0096] It should be noted that a large amount of measured data of weld cross sections (weld geometry) is collected as a training data set; a machine learning algorithm (such as neural network, decision tree, etc.) is used to train the model, taking the measured data as input. and As output, the model obtained through training predicts the corresponding weld geometry size and value.

[0097] Based on the calculated inner radius of a simple torus and outer radius , a number of simple circular rings are unioned by Boolean operation to obtain the weld model.

[0098] The weld model and the physical model within the divided interval are imported into the same coordinate system. According to the position and direction of the weld, the weld model is translated and rotated, and then assembled with the pipe segment model to obtain a complete physical model.

[0099] Ways to adaptively partition the improved physical model include:

[0100] Identify the key areas in the perfect physical model; specifically, divide the perfect physical model into m preliminary areas, and divide the preliminary areas into n curves, and calculate the curvature of each curve ;

[0101] ;in, is the curvature weighting function, is the parametric equation of the curve (obtained by fitting), is the parameter variable of the curve, for The first derivative of for The second derivative, is the torsion rate of the curve, is the weight coefficient, which is used to adjust the influence of the torsion rate; the geometric meaning of the torsion rate is to take a small arc length on the curve , the corresponding change in the torsion angle , then the bending torsion rate is .

[0102] The curvature weighting function can be obtained by:

[0103] N groups of curve data are collected, where the curve data include geometric parameters of the curve and corresponding key area marks; the geometric parameters include coordinate points and tangent vectors.

[0104] Geometric features are extracted from curve data, including curvature, torsion and higher-order derivatives of the curve's parametric equations (obtained by fitting geometric parameters).

[0105] definition The posterior probability ;in, is the geometric feature, Mark the key areas. is the likelihood function, is the prior probability, Is proportional to the sign of .

[0106] Likelihood function Describes the given geometric features and the curvature weighted function Under the condition of The probability of; using machine learning models (such as neural networks, decision trees, etc.) to fit the likelihood function, and As input, It is obtained as the output of training.

[0107] Prior probability ;in, is the mean function of the prior probability, is the covariance function of the prior probability, both of which are about the parameter variables Function of is a multivariate normal distribution; the prior probability Reflects the prior knowledge or assumptions; specifically, the prior probability Modeled as a parameter variable For the changing Gaussian process, the mean function adopts a constant function or a polynomial function; the covariance function adopts a generalized exponential kernel function.

[0108] Based on the posterior probability, calculate the curvature weighting function The posterior probability distribution is obtained, and sample points of the curvature weighted function are extracted from the posterior probability distribution; based on the extracted sample points, the curvature weighted function is fitted.

[0109] Specifically, a Markov chain is defined, whose state space is defined as the parameter space of the curvature weighted function. A proposed distribution is defined to generate new sample points, and the proposed distribution selects Gaussian distribution or other symmetric distribution. A sample point is randomly selected from the prior distribution obtained based on the prior probability as the initial state, that is, the starting point of the Markov chain.

[0110] For the initial state, a new sample point is extracted from the proposed distribution, the posterior probability value of the new sample point is calculated, and the acceptance rate is calculated based on the posterior probability values ​​of the initial state and the new sample point.

[0111] The formula for calculating the acceptance rate is:

[0112] ;in, is the new sample point, For new sample points The value of the posterior probability is is the initial state, for The value of the posterior probability is For the time step When, from arrive The proposed distribution of For the time step When, from arrive Proposed distribution of; is the adaptively adjusted variance function; For the acceptance rate.

[0113] Adaptive variance function ;in, From the initial state to the time step The covariance matrix of all sample points sampled so far; is a small positive constant to ensure It is positive; is the identity matrix.

[0114] Define a uniform distribution (such as N(0, 1), a uniform distribution from 0 to 1), draw a random number from a uniform distribution, if the random number is less than the acceptance rate, accept the new sample point as the next state; otherwise, keep the current initial state unchanged; repeat until the Markov chain converges to the posterior probability distribution.

[0115] After the Markov chain converges, a fixed number of sample points are extracted from the Markov chain. These sample points are the sample points of the curvature weighted function extracted from the posterior probability distribution. It should be noted that it is usually necessary to discard some of the initial samples to reduce the impact of the initial value.

[0116] A curvature threshold is preset. If the mean curvature of all curves in the preliminary area is greater than the curvature threshold, the corresponding preliminary area is taken as the key area.

[0117] The perfect physical model is decomposed into M sub-areas, each of which contains one or more key areas; the decomposition method is manual division or automatic partitioning, and it is necessary to ensure that the interfaces between the sub-areas are clear and have no overlap.

[0118] The sub-regions whose number of key areas is greater than or equal to the preset division threshold are specially divided, and the sub-regions whose number of key areas is less than the preset division threshold are generally divided (such as the mapping grid division algorithm) to obtain several grids; for the grids between adjacent sub-regions, the grids are reasonably connected to avoid gaps or overlaps; the grids of each sub-region are integrated into a preliminary twin model.

[0119] Special divisions include:

[0120] All key areas in the sub-area are taken as initial frontier units. For each frontier unit, several new frontier units are generated according to a preset size, and the new frontier units are added to the preset active frontier set.

[0121] Check whether the newly added frontier unit intersects with the existing frontier unit. For the two intersecting frontier units, perform unit clipping, that is, remove the intersecting parts from the two frontier units; until the active frontier set is empty, it means that the entire sub-area has been covered; then the frontier unit corresponds to the grid.

[0122] Assign pipe materials to each grid and apply preliminary boundary conditions, that is, assign the preliminary twin model to the pipe material; the preliminary boundary conditions include displacement boundaries and force boundaries; the displacement boundary defines the displacement component value of the constrained grid, and the force boundary defines the magnitude and direction of the concentrated force or grid load.

[0123] Furthermore, the risk value calculation method includes:

[0124] The internal environmental data of the pipeline is used as the environmental load of the digital twin model, and the external environmental data of the pipeline is used as the boundary condition of the digital twin model; the pipeline structural health data is updated to the corresponding position of the digital twin model; and the pipeline operation data is used as the operating load of the digital twin model.

[0125] The finite element numerical calculation method is used to solve the stress (principal stress, equivalent stress and stress tensor) of each grid of the digital twin model at this time; the failure criterion function is defined , substitute the stress of each grid into the failure criterion function, and then calculate the corresponding failure index of each grid; calculate the failure probability of the grid according to the distribution of the failure index; perform weighted summation on the failure probabilities of all grids to obtain the probability of failure in the high consequence area of ​​the entire pipeline, that is, the risk value.

[0126] Failure Criteria Function The formula is:

[0127] ML= [ ( s 1 - s 2) 2 + ( s 2 - s 3) 2 + ( s 3 - s 1) 2 ] (2 × (1+ φ× f( oh , d ))) ;

[0128] in, , and are the three principal stresses of the mesh, is a correction factor used to consider the effect of stress multiaxiality on the failure criterion. For the uniaxial stress state, ; For multiaxial stress states, Take a value greater than 0, obtained by fitting the experimental data; is the stress direction function, is the stress tensor, is the stress direction angle.

[0129] For isotropic materials (materials whose mechanical properties are the same in all directions and are not affected by direction), ; For anisotropic materials (such as weld areas in pipelines),

[0130] ;

[0131] in, , , , , , and is an undetermined parameter obtained by fitting the test data; the sum of the trigonometric functions is used to describe the sudden change in the performance of the material in certain specific directions; for example, for the pipeline weld area, the performance along the weld direction may suddenly change, which is represented by the sum.

[0132] Failure indicators ,in, is the yield strength of the material, For Grid The value of the failure criterion function.

[0133] The calculation method of failure probability includes:

[0134] For each grid, the failure index is calculated by Monte Carlo simulation. Draw a number of samples from the distribution of and calculate the failure index in the samples The weighted proportion that is less than 0; that is, the failure probability.

[0135] The formula for calculating the weighted ratio is:

[0136] Pf= 1 N 3 × ∑ [ e1 ×( 1+ α1× exp( - β1× d) ) +e 2 ×( 1+ c1 × ML(i) Sy ) ] × g_ j ;

[0137] in, is the total number of samples, , and To adjust the parameters (determined by fitting the experimental data), and is the equilibrium parameter, and , used to regulate the weights of stress state and sample position, is the average distance from the sample to all failure indicators, For the The value of the failure index of the samples; is the weighted ratio.

[0138] The method of presetting risk thresholds includes: constructing a risk matrix, in which one dimension represents the failure probability (i.e., the calculated risk value) and the other dimension represents the severity of the consequences of failure (such as economic losses, environmental impact, etc.). According to the risk level of different areas in the matrix, the corresponding risk thresholds are preset; for example, when both the failure probability and the severity of the consequences are high, the corresponding risk threshold should be set lower to adopt a more conservative maintenance strategy.

[0139] The maintenance strategy includes maintenance time (periodic maintenance or maintenance triggered by risk value), maintenance method (overall replacement, local repair, reinforcement), maintenance location (maintenance of the entire pipeline or a specific area), maintenance material (what material is used for repair or replacement), and maintenance process (what process method is used, such as hot welding or implantation).

[0140] Ways to input maintenance strategies into the digital twin model include:

[0141] According to the maintenance time point, the service life of the digital twin model is updated; according to the maintenance method and maintenance location, repair or replacement operations are performed on the corresponding area of ​​the digital twin model; according to the maintenance materials and maintenance processes, the material properties and geometric parameters of the corresponding area in the digital twin model are updated.

[0142] The optimal maintenance strategy can be obtained by:

[0143] The maintenance strategy is encoded into a high-dimensional vector, where each dimension corresponds to a decision variable. The decision variables include maintenance time point, maintenance method, maintenance location, maintenance material and maintenance process.

[0144] A fixed number of initial plants (candidate solutions) are randomly generated in the solution space (the combination of all possible values ​​of the decision variables constitutes a solution space) to form a seed population.

[0145] Define the growth function , used to evaluate the quality of each plant (maintenance strategy) in the seed population;

[0146] ;

[0147] in, is the maintenance cost objective function, Cost weight, is the maintenance time objective function, is the time weight, is the objective function of risk reduction, Reduce the weight for risk.

[0148] The maintenance cost objective function is a discretized function consisting of the sum of maintenance material cost, maintenance process cost and maintenance labor cost; the maintenance time objective function is a discretized function consisting of the sum of preparation maintenance time, actual maintenance time and acceptance time.

[0149] Risk reduction objective function ;in, is the risk value after maintenance, is the risk value before maintenance.

[0150] For each plant, its value on the growth function is calculated, recorded as the growth value, and its growth probability is calculated based on the growth value; the higher the growth probability, the greater the chance that the plant will survive and reproduce in the next generation.

[0151] The growth probability is calculated as:

[0152] G_v= (exp( φ1 × (f_v - f_avg)) × (1+ e1 × (f_v - f_med) ϑ1 )) ∑ [ (exp( φ1 × (f_v - f_avg)) × (1+ e1 × (f_v - f_med) ϑ1 )) ] ;

[0153] in, For the The growth probability of each plant, For the The growth value of each plant, is the average growth value of all plants in the current seed population, is the median of all plant growth values ​​in the current seed population; It is an adjustment parameter used to control the influence of the difference between the fitness value and the average fitness value on the growth probability; is the balance coefficient, is an index parameter used to control the influence of the difference between the fitness value and the median on the growth probability; it takes into account the superiority of each plant relative to the average level of the population, and also takes into account the degree of deviation from the center position of the population, thereby more accurately reflecting the relative superiority of each plant.

[0154] A growth threshold is preset, and according to the growth probability, plants with a growth probability greater than the growth threshold are selected from the current seed population as parent plants, and new daughter plants are generated through crossover and mutation operations; the crossover operation adopts single-point crossover, multi-point crossover or uniform crossover; the mutation operation adopts Gaussian mutation or uniform mutation.

[0155] The parent plants and the daughter plants are merged to form a temporary population; the plants are sorted in descending order according to the growth value from the temporary population, and the first Q plants are selected as the next generation population.

[0156] Until the maximum number of iterations is reached or the population converges, and the plant with the highest current growth value is output as the optimal solution, which is the optimal maintenance strategy.

[0157] In this embodiment, a digital twin model of the pipeline high consequence area is constructed, which can accurately simulate the characteristics of the pipeline high consequence area, so as to accurately calculate the risk value of failure in the pipeline high consequence area; the risk assessment method based on digital twins is more scientific and reliable than the traditional method; secondly, the maintenance strategy is triggered according to the risk value, and the maintenance strategy is optimized by the optimization algorithm, so as to obtain the optimal maintenance strategy, thereby maximizing the maintenance efficiency and reducing the maintenance cost; the optimized maintenance strategy not only considers the degree of risk reduction, but also weighs the maintenance cost and maintenance time, realizes multi-objective optimization, and ensures the economy and timeliness of maintenance; furthermore, applying real-time monitoring data to the digital twin model can dynamically evaluate the health status of the pipeline high consequence area, timely discover risk hazards, and improve the safety and reliability of pipeline operation. Active prevention is achieved, major accidents are avoided, and the safety of personnel and the environment is guaranteed.

[0158] Example 2

[0159] See also Figure 2 As shown, the part not described in detail in this embodiment is described in Example 1, which provides an intelligent management method for pipeline high consequence areas integrating digital twins, including:

[0160] S1. Obtain monitoring data of pipeline high consequence areas;

[0161] S2. Build a digital twin model of the pipeline high consequence area, apply the monitoring data to the digital twin model, calculate the risk value of failure in the pipeline high consequence area, preset the risk threshold, and trigger the maintenance strategy of the pipeline high consequence area when the risk value is greater than or equal to the risk threshold;

[0162] S3. Input the maintenance strategy into the digital twin model and use the optimization algorithm to optimize the maintenance strategy; obtain the optimal maintenance strategy, and implement maintenance operations on the high consequence area of ​​the pipeline according to the optimal maintenance strategy.

[0163] Example 3

[0164] This embodiment discloses an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the operation mode of the above-mentioned intelligent management method for high-consequence areas of pipelines integrating digital twins is implemented.

[0165] Since the electronic device introduced in this embodiment is an electronic device used to implement a pipeline high-consequence area intelligent management method integrating digital twins in the embodiment of this application, based on the pipeline high-consequence area intelligent management method integrating digital twins introduced in the embodiment of this application, the technical personnel of this field can understand the specific implementation of the electronic device of this embodiment and its various variations, so how the electronic device implements the method in the embodiment of this application is not described in detail here. As long as the technical personnel of this field implement the electronic device used in the pipeline high-consequence area intelligent management method integrating digital twins in the embodiment of this application, it belongs to the scope of protection of this application.

[0166] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0167] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technical users in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. An intelligent management platform for pipeline high-consequence areas with integrated digital twins, characterized in that: include: A data acquisition module, used to obtain monitoring data of high consequence areas of pipelines; The comprehensive strategy maintenance module is used to build a digital twin model of the pipeline high-consequence area, apply the monitoring data to the digital twin model, calculate the risk value of failure in the pipeline high-consequence area, preset the risk threshold, and trigger the maintenance strategy of the pipeline high-consequence area when the risk value is greater than or equal to the risk threshold; The strategy optimization module is used to input the maintenance strategy into the digital twin model and optimize the maintenance strategy using an optimization algorithm; the optimal maintenance strategy is obtained, and maintenance operations are performed on the high-consequence area of ​​the pipeline according to the optimal maintenance strategy. The modules are connected by wired and / or wireless means; The method for obtaining the optimal maintenance strategy includes: The maintenance strategy is encoded into a high-dimensional vector, where each dimension corresponds to a decision variable. The decision variables include maintenance time point, maintenance method, maintenance location, maintenance material and maintenance process. A fixed number of initial plants are randomly generated in the solution space to form a seed population; the growth function SZ = w1×f1+w2×f2+w3×f3 is defined; where f1 is the maintenance cost objective function, w1 is the cost weight, f2 is the maintenance time objective function, w2 is the time weight, f3 is the risk reduction objective function, and w3 is the risk reduction weight; The maintenance cost objective function is a discretized function of the sum of maintenance material cost, maintenance process cost and maintenance labor cost; the maintenance time objective function is a discretized function of the sum of preparation maintenance time, actual maintenance time and acceptance time; Risk reduction objective function Among them, Fh is the risk value after maintenance, and Fq is the risk value before maintenance; For each plant, calculate its value on the growth function, record it as the growth value, and calculate its growth probability based on the growth value; the higher the growth probability, the greater the chance that the plant will survive and reproduce in the next generation; The growth probability is calculated as: Among them, G_v is the growth probability of the vth plant, f_v is the growth value of the vth plant, f_avg is the average growth value of all plants in the current seed population, and f_med is the median growth value of all plants in the current seed population; is the adjustment parameter; ε1 is the balance coefficient, is the index parameter; Preset the growth threshold, select the plants with growth probability greater than the growth threshold from the current seed population as the parent plants according to the growth probability, generate new daughter plants through crossover and mutation operations; merge the parent plants and daughter plants to form a temporary population; sort the plants in the temporary population in descending order according to the growth value, and select the first Q plants as the next generation population; Until the maximum number of iterations is reached or the population converges, and the plant with the highest current growth value is output as the optimal solution, which is the optimal maintenance strategy.

2. The pipeline high-consequence area intelligent management platform integrated with digital twin according to claim 1 is characterized in that: The monitoring data includes pipeline internal environment data, pipeline external environment data, pipeline structure health data and pipeline operation data.

3. The pipeline high-consequence area intelligent management platform integrated with digital twin according to claim 2 is characterized in that: The digital twin model is constructed in the following ways: Collect basic parameters of high consequence areas of pipelines, including pipeline materials, geometric data and pipeline connection methods; geometric data include pipe diameter, wall thickness and length; pipeline connection method is welding; Based on the geometric structure data of the high-consequence area of ​​the pipeline, a physical model of the high-consequence area of ​​the pipeline is established; the physical model is refined based on the pipeline connection method to obtain a perfect physical model; the perfect physical model is adaptively divided to obtain a preliminary twin model; the preliminary twin model is assigned to the pipeline material to obtain a digital twin model of the high-consequence area of ​​the pipeline.

4. The pipeline high-consequence area intelligent management platform integrated with digital twin according to claim 3 is characterized in that: The method of refining the physical model includes: The weld geometry dimensions at the weld are obtained, including weld width, weld height and weld angle; for the welding position, a division interval [-k, k] is defined, where k is a positive integer greater than or equal to 1, the welding position is placed at the zero point of the division interval, the physical model within the division interval is used as the pipe segment model, the pipe segment model is divided into several simple circular rings, and the inner radius r_in and outer radius r_out of the simple circular rings are calculated; Inner radius r_in=(R-t_w-l_trans)×cos(α)×f_in; Outer radius r_out = (R + h + l_trans + r_curve) × cos (α) × f_out; where R is the radius of the pipe section, t_w is the wall thickness, h is the weld height, l_trans is the transition area length, r_curve is the transition area curvature radius; α is the weld angle, f_in is the inner radius shape correction coefficient, and f_out is the outer radius shape correction coefficient; Based on the calculated inner radius r_in and outer radius r_out of the simple circular ring, a Boolean operation is performed on the union of several simple circular rings to obtain a weld model; The weld model and the physical model within the divided interval are imported into the same coordinate system. According to the position and direction of the weld, the weld model is translated and rotated, and then assembled with the pipe segment model to obtain a complete physical model.

5. The pipeline high-consequence area intelligent management platform integrated with digital twin according to claim 4 is characterized in that: The method of adaptively dividing the improved physical model includes: Identify the key areas in the perfect physical model; decompose the perfect physical model into M sub-areas, each of which contains one or more key areas; perform special divisions on the sub-areas whose number of key areas is greater than or equal to the preset division threshold, and perform ordinary divisions on the sub-areas whose number of key areas is less than the preset division threshold, to obtain several grids; for the grids between adjacent sub-areas, perform grid docking, and integrate the grids of each sub-area into a preliminary twin model; The identification method of the key area includes: Divide the perfected physical model into m preliminary regions, and divide the preliminary regions into n curves, and calculate the curvature K of each curve; Where w(t) is the curvature weighting function, r(t) is the parametric equation of the curve, t is the parameter variable of the curve, r'(t) is the first-order derivative of r(t), r"(t) is the second-order derivative of r(t), τ(t) is the torsion rate of the curve, and γ is the weight coefficient; A curvature threshold is preset. If the mean curvature of all curves in the preliminary area is greater than the curvature threshold, the corresponding preliminary area is taken as the key area. Special divisions include: All key areas in the sub-area are taken as initial frontier units. For each frontier unit, several new frontier units are generated according to the preset size, and the new frontier units are added to the preset active frontier set. Check whether the newly added frontier unit intersects with the existing frontier unit. For the two intersecting frontier units, perform unit clipping, that is, remove the intersecting parts from the two frontier units; until the active frontier set is empty and the frontier unit corresponds to the grid.

6. The pipeline high-consequence area intelligent management platform integrated with digital twin according to claim 5 is characterized in that: The method for obtaining the curvature weighting function includes: Collecting N groups of curve data, the curve data including geometric parameters of the curve and corresponding key area marks; extracting geometric features from the curve data, the geometric features including curvature, torsion rate and high-order derivatives of the parametric equation of the curve; Define the posterior probability of w(t) as p(w(t)|X,Y)∝p(Y|X,w(t))×p(w(t)); where X is the geometric feature, Y is the key area marker, p(Y|X,w(t)) is the likelihood function, p(w(t)) is the prior probability, and ∝ is the sign proportional to ; prior probability p(w(t))=NJ(w(t)|μ(t),E(t)); where μ(t) is the mean function of the prior probability, and E(t) is the covariance function of the prior probability; NJ() is a multivariate normal distribution; Based on the posterior probability, the posterior probability distribution of the curvature weighted function w(t) is calculated, and sample points of the curvature weighted function are extracted from the posterior probability distribution; based on the extracted sample points, the curvature weighted function is fitted; The sampling point extraction method includes: Define a Markov chain, whose state space is defined as the parameter space of the curvature weighted function, and define a proposal distribution; randomly select a sample point from the prior distribution obtained based on the prior probability as the initial state, that is, the starting point of the Markov chain; For the initial state, a new sample point is extracted from the proposed distribution, the posterior probability value of the new sample point is calculated, and the acceptance rate is calculated based on the posterior probability values ​​of the initial state and the new sample point; The formula for calculating the acceptance rate is: Where θ' is a new sample point, p(θ') is the value of the posterior probability of the new sample point θ', θ is the initial state, p(θ) is the value of the posterior probability of θ, q(θ|θ',Y(t')) is the proposed distribution from θ to θ' at time step t', q(θ'|θ,Y(t')) is the proposed distribution from θ' to θ at time step t'; Y(t') is the variance function of adaptive adjustment; β is the acceptance rate; Adaptively adjusted variance function Y(t') = cov + ε × I; where cov is the covariance matrix of all sample points sampled from the initial state to time step t'; ε is a positive constant; I is the unit matrix; Define a uniform distribution, draw a random number from a uniform distribution, if the random number is less than the acceptance rate, accept the new sample point as the next state; otherwise, keep the current initial state unchanged; repeat until the Markov chain converges to the posterior probability distribution; After the Markov chain converges, a fixed number of sample points are extracted from the Markov chain. These sample points are the sample points of the curvature weighted function extracted from the posterior probability distribution.

7. The pipeline high-consequence area intelligent management platform integrated with digital twin according to claim 6 is characterized in that: The risk value is calculated by: The internal environment data of the pipeline is used as the environmental load of the digital twin model, and the external environment data of the pipeline is used as the boundary condition of the digital twin model; the pipeline structure health data is updated to the corresponding position of the digital twin model; the pipeline operation data is used as the working condition load of the digital twin model; The finite element numerical calculation method is used to solve the stress of each grid in the digital twin model at this time; the failure criterion function ML is defined, the stress of each grid is substituted into the failure criterion function, and then the corresponding failure index of each grid is calculated; according to the distribution of the failure index, the failure probability of the grid is calculated; the failure probabilities of all grids are weighted and summed to obtain the probability of failure in the high consequence area of ​​the entire pipeline, that is, the risk value.

8. The pipeline high-consequence area intelligent management platform integrated with digital twin according to claim 7 is characterized in that: The formula of the failure criterion function ML is: Among them, σ1, σ2 and σ3 are the three principal stresses of the grid, is the correction coefficient; f(ω,δ) is the stress direction function, ω is the stress tensor, and δ is the stress direction angle; For isotropic materials, f(ω,δ)=1; for anisotropic materials, Among them, c1, c2, c3, m1, n1, m2 and n2 are parameters to be determined; Failure index g = Sy - ML(i), where Sy is the yield strength of the material and ML(i) is the value of the failure criterion function of grid i; The calculation method of failure probability includes: For each grid, a number of samples are extracted from the distribution of the failure index g through Monte Carlo simulation, and the weighted proportion of the failure index g in the samples that is less than 0 is calculated; this is the failure probability; The formula for calculating the weighted ratio is: Wherein, N3 is the total number of samples, α1, β1 and γ1 are adjustment parameters, e1 and e2 are balance parameters, and e1+e2=1, d is the average distance from the sample to all failure indicators, g_j is the value of the failure indicator of the jth sample, and Pf is the weighted ratio.

9. A pipeline high-consequence area intelligent management method integrated with digital twins, which is implemented based on the pipeline high-consequence area intelligent management platform integrated with digital twins according to any one of claims 1 to 8, characterized in that: include: S1. Obtain monitoring data of pipeline high consequence areas; S2. Build a digital twin model of the pipeline high consequence area, apply the monitoring data to the digital twin model, calculate the risk value of failure in the pipeline high consequence area, preset the risk threshold, and trigger the maintenance strategy of the pipeline high consequence area when the risk value is greater than or equal to the risk threshold; S3. Input the maintenance strategy into the digital twin model and use the optimization algorithm to optimize the maintenance strategy; obtain the optimal maintenance strategy, and implement maintenance operations on the high consequence area of ​​the pipeline according to the optimal maintenance strategy.

Images