Petrochemical pipe network fault prediction method and system based on multi-physics coupling digital twinning

Abstract

The method and system for petrochemical pipeline network fault prediction based on multi-physical field coupling digital twinning belong to the technical field of safety monitoring and fault diagnosis of petrochemical equipment. The method comprises the following steps: collecting and preprocessing multi-source data of the petrochemical pipeline network, coupling and integrating multi-physical field sub-models, and obtaining a digital twinning model matched with the physical pipeline network; based on the digital twinning model, carrying out flow-solid-corrosion two-way coupling simulation with real-time operation data as the boundary to obtain multi-physical field state data of the pipeline network; correcting the digital twinning model according to the error between the measured data and the simulation data; based on the corrected digital twinning model and the multi-physical field state data, extracting fault features and matching fault judgment rules to obtain fault diagnosis and trend prediction results. The present application can realize the technical effects of precise prediction, early warning, long-term precise model, diagnosis explanation and whole-process closed-loop operation and maintenance of petrochemical pipeline network faults.

Description

Technical Field

[0001] This invention relates to the field of safety monitoring and fault diagnosis technology for petrochemical equipment, specifically to a method and system for predicting faults in petrochemical pipeline networks using multi-physics field coupled digital twins. Background Technology

[0002] Petrochemical pipelines are subjected to complex conditions of high temperature, high pressure, multiphase flow erosion, and chemical corrosion for a long time. The media they transport are mostly flammable, explosive, toxic and corrosive fluids, which can easily lead to defects such as corrosion, cracks and thinning, and cause safety accidents such as leaks and ruptures.

[0003] Current petrochemical pipeline network inspections primarily rely on periodic shutdowns for manual inspections, which suffers from long cycles, high costs, and an inability to cover hidden defects, making real-time health assessments and fault early warnings difficult. While digital twin technology can construct virtual models of physical entities, existing technologies have significant drawbacks: first, the model fidelity is insufficient, relying on static data and lacking multi-source information fusion and dynamic correction mechanisms; second, the multi-physics coupling capability is weak, failing to accurately depict the bidirectional coupling effects of flow, stress, and corrosion fields, resulting in poor simulation accuracy and real-time performance; and third, purely data-driven predictions lack physical mechanism support, have poor interpretability, and cannot trace the root cause of faults.

[0004] In summary, existing technologies cannot meet the requirements for accurate fault prediction in the long-term safe operation of petrochemical pipeline networks. There is an urgent need for a high-fidelity, multi-field coupled, mechanism-driven fault prediction method and system. Summary of the Invention

[0005] In view of this, the main objective of the present invention is to provide a method and system for predicting petrochemical pipeline network faults using multi-physics field coupled digital twins, in order to at least partially solve the above-mentioned technical problems.

[0006] To achieve the above objectives, as a first aspect of the present invention, a method for predicting faults in petrochemical pipeline networks based on multi-physics coupled digital twins is proposed, comprising the following steps: Collect and preprocess multi-source data from petrochemical pipeline networks, couple and integrate multi-physics sub-models, and obtain a digital twin model that matches the physical pipeline network; Based on the digital twin model, a two-way coupled simulation of fluid-solid-corrosion is performed with real-time running data as the boundary to obtain multi-physics state data of the pipeline network; measured data of the physical entity of the pipeline network are obtained through online non-destructive testing and sensor monitoring, and the measured data are compared with the multi-physics state data obtained by simulation. The digital twin model is then corrected based on the comparison error. Based on the corrected digital twin model and the multiphysics state data obtained from simulation, fault features are extracted and matched with fault judgment rules to obtain fault diagnosis and trend prediction results. The multiphysics sub-model includes a fluid dynamics sub-model, a structural mechanics sub-model, and a corrosion evolution sub-model; The fluid-solid-corrosion bidirectional coupled simulation includes: The fluid dynamics sub-model is solved unidirectionally to obtain the flow field distribution and transfer the wall shear stress and pressure load to the structural field and corrosion field. The structural mechanics sub-model is solved to obtain stress, strain and structural deformation, and the stress and deformation are fed back to the corrosion field; The corrosion evolution sub-model is solved to calculate the corrosion rate and wall thickness damage, and the wall thickness reduction and modulus reduction are fed back to the structural field and flow field. The coupled variables are iterated bidirectionally to synchronize the updates of parameters in the flow field, structural field, and corrosion field. Convergence assessment: Repeat the iteration until all changes in the coupling variables satisfy the convergence threshold.

[0007] In one possible implementation, the unidirectional solution of the fluid dynamics sub-model includes: Solving the continuity and momentum equations using the finite volume method: , , In the formula, u is the fluid velocity vector, ρ is the fluid density, and t is time. For fluid pressure, ρ is the buoyancy of the particulate phase, g is the gravitational acceleration, and μ is the fluid dynamic viscosity; The wall shear stress, wall pressure load, and flow field velocity distribution are obtained by solving the problem.

[0008] In one possible implementation, solving the structural mechanics sub-model includes: Using the finite element method, substituting the wall shear stress and pressure load transmitted by the fluid dynamics sub-model, and combining the thermal strain generated by the temperature field, the structural equilibrium equations are solved: , In the formula, Here is the structural stiffness matrix. It is a displacement vector. For the flow-induced load vector, For thermal load vectors, This is the equivalent load for stress concentration caused by corrosion defects; The stress field, structural deformation, and stress concentration factor are obtained by solving the problem.

[0009] In one possible implementation, solving the corrosion evolution sub-model includes: Based on the synergistic mechanism of electrochemistry and erosion corrosion, a stress-flow field coupled corrosion kinetic equation is constructed: , In the formula, v corr (t) represents the corrosion rate. The baseline corrosion rate is the one without stress or erosion, and χ is the stress acceleration factor. To flush out the sensitivity coefficient, For wall shear stress; The corrosion rate and wall thickness damage are obtained by solving the equation.

[0010] In one possible implementation, the bidirectional iteration of the coupled variables includes: Based on the reduction in corrosion wall thickness and structural deformation, the fluid computational mesh is reconstructed, and the pipe inner diameter, flow cross-sectional area, and wall roughness are corrected. Based on the reduction of wall thickness and modulus after corrosion, the structural stiffness matrix is ​​reconstructed, and the stress calculation benchmark is corrected. The corrosion rate calculation benchmark was revised based on the updated flow field shear stress and structural stress.

[0011] In one possible implementation, the convergence determination includes: Set convergence threshold When the coupling variables satisfy , If the iteration converges, the simulation is terminated; otherwise, the iteration is repeated. Let k be the coupling variable and k be the iteration number. This is the convergence threshold.

[0012] In one possible implementation, the fault features include stress, flow field, corrosion, pressure, and operational trend-related features, and the fault features are combined with pipeline topology coordinates and timestamps to achieve spatiotemporal correlation. Based on the corrected digital twin model and multiphysics state data obtained from simulation, fault features are extracted and matched with fault determination rules to obtain fault diagnosis and trend prediction results, including: Quantitative prediction of crack propagation using the Paris formula: , in, Let be the crack depth at time t+ΔT. Let t be the crack depth at time t, and C be the material fatigue constant. The predicted number of stress cycles is given by ΔK, where ΔK is the stress intensity factor amplitude. The residual strength factor is used to assess the structural load-bearing safety margin. , Where RSF is the residual intensity factor, σ s σ is the yield strength of the pipe. maxThis represents the maximum working stress that the pipeline actually withstands. Type I stress intensity factor is used to determine the critical crack state: , Among them, K I Here, Y is the Type I (open type) stress intensity factor, Y is the geometric correction factor related to pipe geometry and crack location, and σ is the long-range working stress at the crack tip. This refers to the crack depth. By combining the above formulas, quantitative prediction and remaining service assessment of crack propagation, corrosion damage, and stress evolution in petrochemical pipeline networks can be achieved.

[0013] As a second aspect of the present invention, a petrochemical pipeline network fault prediction system based on multi-physics field coupled digital twins is also proposed, comprising: a digital twin model construction module, a multi-physics field coupled simulation module, a fault prediction and diagnosis module, and a visualization management module; The output of the digital twin model construction module is connected to the input of the multiphysics coupling simulation module; the output of the multiphysics coupling simulation module is connected to the input of the fault prediction and diagnosis module; and the output of the fault prediction and diagnosis module is connected to the input of the visualization management module. The digital twin model construction module is used to collect and preprocess multi-source data of petrochemical pipeline networks, couple and integrate multi-physics field sub-models, and obtain a digital twin model that matches the physical pipeline network. The multiphysics coupling simulation module is used to perform two-way fluid-solid-corrosion coupling simulation based on a digital twin model and with real-time running data as the boundary to obtain multiphysics state data of the pipeline network; it also obtains measured data of the physical entity of the pipeline network through online non-destructive testing and sensor monitoring, compares the measured data with the multiphysics state data obtained from the simulation, and corrects the digital twin model based on the comparison error; The fault prediction and diagnosis module is used to extract fault features and match fault judgment rules based on the corrected digital twin model and multiphysics state data obtained from simulation, so as to obtain fault diagnosis and trend prediction results. The visualization management module is used to build a platform based on predictive diagnostic results, enabling visualization of pipeline network status, early warning, and output of operation and maintenance decisions. The multiphysics sub-model includes a fluid dynamics sub-model, a structural mechanics sub-model, and a corrosion evolution sub-model; The fluid-solid-corrosion bidirectional coupled simulation includes: The fluid dynamics sub-model is solved unidirectionally to obtain the flow field distribution and transfer the wall shear stress and pressure load to the structural field and corrosion field. The structural mechanics sub-model is solved to obtain stress, strain and structural deformation, and the stress and deformation are fed back to the corrosion field; The corrosion evolution sub-model is solved to calculate the corrosion rate and wall thickness damage, and the wall thickness reduction and modulus reduction are fed back to the structural field and flow field. The coupled variables are iterated bidirectionally to synchronize the updates of parameters in the flow field, structural field, and corrosion field. Convergence assessment: Repeat the iteration until all changes in the coupling variables satisfy the convergence threshold.

[0014] In one possible implementation, the unidirectional solution of the fluid dynamics sub-model includes: Solving the continuity and momentum equations using the finite volume method: , , In the formula: u is the fluid velocity vector, ρ is the fluid density, and t is time. For fluid pressure, ρ is the buoyancy of the particulate phase, g is the gravitational acceleration, and μ is the fluid dynamic viscosity; The wall shear stress, wall pressure load, and flow field velocity distribution are obtained by solving the problem. The solution to the structural mechanics sub-model includes: Using the finite element method, substituting the wall shear stress and pressure load transmitted by the fluid dynamics sub-model, and combining the thermal strain generated by the temperature field, the structural equilibrium equations are solved: , In the formula: Here is the structural stiffness matrix. It is a displacement vector. For the flow-induced load vector, For thermal load vectors, This is the equivalent load for stress concentration caused by corrosion defects; The stress field, structural deformation, and stress concentration factor are obtained. The solution to the corrosion evolution sub-model includes: Based on the synergistic mechanism of electrochemistry and erosion corrosion, a stress-flow field coupled corrosion kinetic equation is constructed: , In the formula: v corr (t) represents the corrosion rate. The baseline corrosion rate is the one without stress or erosion, and χ is the stress acceleration factor. To flush out the sensitivity coefficient, For wall shear stress; The bidirectional iteration of the coupling variables includes: Based on the reduction in corrosion wall thickness and structural deformation, the fluid computational mesh is reconstructed, and the pipe inner diameter, flow cross-sectional area, and wall roughness are corrected. Based on the reduction of wall thickness and modulus after corrosion, the structural stiffness matrix is ​​reconstructed, and the stress calculation benchmark is corrected. The corrosion rate calculation benchmark was revised based on the updated flow field shear stress and structural stress.

[0015] In one possible implementation, the convergence determination includes: Set convergence threshold When the coupling variables satisfy , If the iteration converges, the simulation is terminated; otherwise, the iteration is repeated. Let k be the coupling variable and k be the iteration number. This is the convergence threshold; The fault characteristics include stress, flow field, corrosion, pressure, and operation trend-related features. The fault characteristics are combined with pipeline topology coordinates and timestamps to achieve spatiotemporal correlation. Based on the corrected digital twin model and multiphysics state data obtained from simulation, fault features are extracted and matched with fault determination rules to obtain fault diagnosis and trend prediction results, including: Quantitative prediction of crack propagation using the Paris formula: , in, Let be the crack depth at time t+ΔT. Let t be the crack depth at time t, and C be the material fatigue constant. The predicted number of stress cycles is given by ΔK, where ΔK is the stress intensity factor amplitude. The residual strength factor is used to assess the structural load-bearing safety margin. , Where RSF is the residual intensity factor, σ s σ is the yield strength of the pipe. max This represents the maximum working stress that the pipeline actually withstands. Type I stress intensity factor is used to determine the critical crack state: , Among them, K I Here, Y is the Type I (open type) stress intensity factor, Y is the geometric correction factor related to pipe geometry and crack location, and σ is the long-range working stress at the crack tip. This refers to the crack depth. By combining the above formulas, quantitative prediction and remaining service assessment of crack propagation, corrosion damage, and stress evolution in petrochemical pipeline networks can be achieved.

[0016] Based on the above technical solution, it can be seen that the multi-physics field coupled digital twin petrochemical pipeline network fault prediction method and system of the present invention has at least one of the following beneficial effects compared with the prior art: 1. By collecting multi-source data on the geometry, material, and historical operation and maintenance of petrochemical pipelines and performing outlier removal preprocessing, a digital twin model is constructed by coupling and integrating multiple physical field sub-models. This enables the virtual model to achieve accurate matching with the physical pipeline network in all dimensions, fundamentally solving the problems of single data and large deviation between the model and the entity in traditional modeling, and providing a reliable foundation for pipeline network status simulation and fault prediction. 2. Using real-time running data as the dynamic boundary, a two-way coupled simulation of fluid-solid-corrosion is carried out to realize the iterative transfer of physical quantities of multi-physics field and the real-time update of the computational domain. The simulation calculation is completed according to the convergence threshold. It can obtain the real multi-physics field state data inside the pipeline network in the whole domain, and fully reproduce the collaborative evolution law of fluid scouring, stress concentration and corrosion damage, breaking through the limitations of insufficient accuracy and one-sided state perception of traditional single-field and one-way simulation. 3. Based on the corrected model and multi-physics state data, fault features are extracted, and diagnosis and trend prediction are completed by combining preset fault judgment rules. The pure data-driven black box prediction mode is abandoned. Based on physical mechanisms, the fault can be accurately located, quantitatively graded and its development trend can be deduced. The root cause of the fault can be traced and the false alarm rate of the warning is low, which greatly improves the credibility and practicality of fault prediction. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the multi-physics field coupled digital twin method for predicting petrochemical pipeline network faults according to the present invention. Figure 2 This is a schematic diagram illustrating the construction of the digital twin model of the present invention; Figure 3 This is the multiphysics coupling mapping diagram of the present invention; Figure 4 This is a schematic diagram of fault prediction according to the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to specific embodiments and accompanying drawings.

[0020] The terminology used in this invention is for the purpose of describing particular embodiments only and is not intended to limit the embodiments of the invention. The singular forms “a,” “the,” and “the” as used in the embodiments of the invention and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0021] Existing petrochemical pipeline fault prediction technologies suffer from technical problems such as insufficient model fidelity, low accuracy of multi-physics coupled simulation, lack of physical mechanism support for fault prediction, inability of models to dynamically and adaptively correct themselves, lack of data collaboration among system modules, and lack of a closed-loop operation and maintenance process. Through in-depth research, it has been found that technical means such as constructing a high-fidelity digital twin model through multi-source data fusion, fluid-solid-corrosion bidirectional coupled simulation, iterative correction model for measured-simulation errors, data-mechanism fusion fault diagnosis, modular data collaborative interaction, and edge-cloud collaborative scheduling can achieve the technical effects of accurate fault prediction, early warning, long-term accurate model, interpretable diagnosis, and a closed-loop operation and maintenance process for petrochemical pipelines.

[0022] Therefore, as Figure 1 As shown, this invention proposes a multi-physics coupled digital twin-based method for predicting faults in petrochemical pipeline networks, including: Multi-source data from petrochemical pipeline networks are collected and preprocessed, and then coupled and integrated with multi-physics sub-models to obtain a digital twin model that matches the physical pipeline network.

[0023] Based on the digital twin model, a two-way coupled simulation of fluid-solid-corrosion is performed with real-time running data as the boundary to obtain multi-physics state data of the pipeline network; measured data of the physical entity of the pipeline network are obtained through online non-destructive testing and sensor monitoring, and the measured data are compared with the multi-physics state data obtained from the simulation. The digital twin model is then corrected based on the comparison error.

[0024] Based on the corrected digital twin model and multiphysics state data obtained from simulation, fault features are extracted and matched with fault judgment rules to obtain fault diagnosis and trend prediction results.

[0025] In one possible implementation, the multi-source data includes geometric data, material physical parameters, and historical operation and maintenance data. The preprocessing employs the 3σ criterion to remove outliers, and the preprocessed data is imported into simulation software to complete the model coupling and construction. By collecting multi-source data on the geometry, material properties, and historical operation and maintenance of the petrochemical pipeline network and performing outlier removal preprocessing, a multi-physics sub-model is coupled and integrated to construct a digital twin model. The multi-physics sub-model includes a fluid dynamics sub-model, a structural mechanics sub-model, and a corrosion evolution sub-model, enabling the virtual model to achieve full-dimensional and accurate matching with the physical pipeline network. This fundamentally solves the problems of single data and large deviations between the model and the entity in traditional modeling, providing a reliable foundation for pipeline network status simulation and fault prediction.

[0026] In this embodiment, as Figure 2 As shown, a laser scanning device was used to perform a full-area 3D scan of the petrochemical pipeline network, simultaneously analyzing pipeline design drawings, as-built data, and construction acceptance documents, and collecting core geometric parameters, including: pipe diameter D, nominal pipe wall thickness w. n Actual wall thickness w act 3D coordinates L of the pipe section laying path j (x,y,z), valve position coordinates S k (x i ,y i ,z i ), Welded joint distribution coordinates T(x) j ,y j ,z j The high-precision geometric dataset G is formed by considering factors such as the pipe bending radius R, etc. G = {D, w} n w act L j (x,y,z), S k (x i ,y i ,z i ), T(x j ,y j ,z j Let i be the valve number, i = 1, 2, ..., n, and n be the total number of valves; j be the weld joint number, j = 1, 2, ..., m, and m be the total number of weld joints. Import the geometric dataset into 3D modeling software such as SolidWorks and UGNX to build a preliminary geometric model of the pipeline network.

[0027] Based on a high-precision geometric model, the mechanical and physical property parameters of the materials of each component of the pipeline network are collected, including: elastic modulus E, Poisson's ratio μ, and tensile strength σ. b Yield strength σ s Parameters such as thermal conductivity λ, coefficient of thermal expansion α, and corrosion resistance coefficient η form a material parameter set P, P={E, μ, σ}. b , σ s ,λ,α,η,…}. Combining fluid mechanics, solid mechanics, and electrochemical corrosion theories, a multi-physics coupled physical model is constructed, including a fluid mechanics sub-model, a structural mechanics sub-model, and a corrosion evolution sub-model.

[0028] Collect historical operation and maintenance and defect data of the pipeline network throughout its entire life cycle, including: historical corrosion thinning amount b. corr (t), historical crack size a(t), high-risk defect location set L, historical fault handling records, routine maintenance cycles, annual temperature / pressure / flow operation data, etc., form a historical dataset H. H={b corr(t), a(t), L, historical fault handling records, routine maintenance cycle, operating data, ...}, where t is the service time.

[0029] The geometric dataset G, material parameter set P, and historical dataset H are all preprocessed using the 3σ criterion to remove outliers. The sample mean is then calculated. Calculate the sample standard deviation ,use Anomaly detection rules. For samples that satisfy the above formula... Values ​​identified as outliers are removed.

[0030] The preprocessed multi-source fusion data is imported into the Comsol finite element simulation software, and the geometric model and physical model are iteratively corrected and coupled together to finally form a digital twin virtual model that is fully matched with the physical pipeline network and supports multi-physics simulation.

[0031] Based on the aforementioned digital twin model, a two-way coupled simulation of fluid-solid-corrosion is performed with real-time operational data as the boundary to obtain multi-physics state data of the pipeline network. Measured physical data of the pipeline network are obtained through online non-destructive testing and sensor monitoring. This measured data is compared with the multi-physics state data obtained from the simulation, and the digital twin model is corrected based on the comparison error. Using real-time operational data as the dynamic boundary, a two-way coupled simulation of fluid-solid-corrosion is carried out to achieve iterative transfer of multi-physics physical quantities and real-time updating of the computational domain. The simulation calculation is completed according to the convergence threshold, enabling the acquisition of real multi-physics state data within the pipeline network across the entire domain. This fully reproduces the collaborative evolution of fluid scouring, stress concentration, and corrosion damage, overcoming the limitations of insufficient accuracy and one-sided state perception in traditional single-field, unidirectional simulations.

[0032] In one possible implementation, the real-time operating data includes temperature, pressure, flow rate, vibration, and electrochemical parameters; the bidirectional coupled simulation completes the iterative transfer of the three physical quantities and the update of the computational domain until the convergence threshold is met, at which point it terminates.

[0033] In one possible implementation, the model correction employs a relative error method to calculate the data matching error, iteratively correcting the simulation parameters when the error exceeds a threshold. The fault characteristics include stress, flow field, corrosion, pressure, and operational trend-related features, with trend predictions encompassing corrosion, crack, and stress development predictions. Iteratively correcting model parameters based on the matching error between measured and simulated data can automatically eliminate model accuracy degradation caused by material aging and operating condition drift, ensuring the digital twin model maintains long-term accuracy throughout the pipeline network's entire lifecycle. This avoids the shortcomings of traditional one-time modeling that leads to later failures, continuously guaranteeing the reliability of simulation and prediction results.

[0034] By iteratively correcting model parameters based on the matching error between measured and simulated data, the model accuracy attenuation caused by material aging and operating condition drift can be automatically eliminated. This allows the digital twin model to maintain long-term accuracy throughout the entire life cycle of the pipeline network, avoiding the defects of traditional one-time modeling that fails later, and continuously ensuring the reliability of simulation and prediction results.

[0035] In this embodiment, a sensor network is deployed at key locations in the pipeline network to collect temperature T in real time. s (t), pressure P s (t), flow rate Q(t), vibration acceleration a v Multivariate time-series data, including T, pH value, and electrochemical potential, constitute the real-time monitoring data vector M(t), where M(t) = [T]. s (t), P s (t), Q(t), a v (t), pH...] T .

[0036] Secondly, the real-time monitoring data M(t) will be compared with the media composition and particulate matter concentration distribution C in the process knowledge base. p By combining such information, a time-varying dynamic boundary condition set is constructed, which serves as the dynamic boundary condition for multiphysics simulation.

[0037] A two-way fluid-solid-corrosion coupling algorithm is used for real-time simulation. Real-time monitoring data serves as the dynamic boundary driving iterative simulation, enabling real-time interactive transfer of the three physical quantities, bidirectional iterative convergence, and global state solution. Compared to unidirectional coupling, which only transfers loads, bidirectional coupling can simultaneously reflect the changes in geometric degradation and structural deformation caused by corrosion on the flow field boundary, as well as the impact of changes in flow field parameters on corrosion dynamics. This fully reproduces the multi-physics co-evolution law of the petrochemical pipeline network. The iterative process is as follows: (a) Unidirectional solution of fluid dynamics sub-model and output of flow field load The Reynolds-averaged Navier-Stokes equations (RANS) are solved using the finite volume method (FVM) to adapt to multiphase and turbulent flow conditions in petrochemical pipeline networks. For gas-liquid / solid-liquid multiphase flows, VOF or Mixture multiphase flow models are introduced. Using dynamic boundary conditions as input, the continuity equation, momentum equation, and energy equation are solved to obtain the transient flow field distribution across the entire pipe network: In the formula: u, i.e., u(x, y, z, t), is the fluid velocity vector (three-dimensional vector), ρ is the fluid density, t is time, P is the fluid pressure (scalar), ∇P is the pressure gradient (three-dimensional vector), μ is the fluid dynamic viscosity, and F... pThe buoyancy / force vector of the particles (three-dimensional vector), where g is the acceleration due to gravity. For Hamiltonian operators (divergence / gradient operators). The tensor cross product (dyadic vector) of the velocity vectors is commonly abbreviated as . , used to characterize the convection term.

[0038] Extracting the shear stress field (i.e., wall shear stress) τ of the pipe wall w (x,y,z,t), tangential velocity of the fluid at the wall, u τ (x, y, z, t), near-wall turbulence intensity I t (x, y, z, t), wall mass transfer coefficient k m The load data is transmitted to the structural stress field sub-model and the corrosion evolution sub-model; the flow field pressure load P is output simultaneously. w (x, y, z, t) serves as the surface force boundary condition for the structural mechanics field.

[0039] (b) Solving the structural mechanics sub-model and feedback of structural response The finite element method (FEM) is used to solve the governing equations of structural mechanics, taking into account the coupled effects of thermal stress, flow-induced loads, and corrosion-induced wall thickness reduction, thus constructing a thermo-structural coupled solution system. (1) Flow-induced load coupling: Substitute the wall shear stress τ transferred by the fluid dynamics sub-model w Pressure load P w Construct the flow-induced load vector F flow ; (2) Thermal-structural coupling: combined with temperature field T s The thermal strain ε generated by (t) th =αΔT, where α is the coefficient of thermal expansion and ΔT is the temperature change, constructing the thermal load vector F. thermal ; (3) Corrosion-structure coupling: The wall thickness reduction b output by the corrosion evolution sub-model corr Introducing a structural solution framework: ①Based on wall thickness reduction amount b corr Update the effective wall thickness w of the pipe section eff =w 0- b corr (Where w0 is the initial wall thickness); ②Based on effective wall thickness w eff Corrected structural stiffness matrix K Str K Str =f(E,w eff (E is the elastic modulus, L is the element length), structural stiffness is positively correlated with wall thickness; reducing wall thickness will lead to a decrease in stiffness matrix K. StrThe corresponding element stiffness component decreases, and the structural stress and deformation increase under the same load, realizing the quantitative coupling of corrosion damage to the structural mechanical response; ③ Simultaneously, the local stress concentration effect caused by the morphology of corrosion defects (such as corrosion pits) is mitigated by introducing an equivalent load F for stress concentration due to corrosion defects. corr Perform quantitative characterization; (4) Solve simultaneously: based on the updated stiffness matrix K Str and load vector F flow F thermal F corr Solve the structural equilibrium equations: In the formula: Here is the structural stiffness matrix. It is a displacement vector. For the flow-induced load vector, For thermal load vectors, This is the equivalent load for stress concentration caused by corrosion defects.

[0040] The global stress tensor field (stress field) σ(x, y, z, t), strain tensor field ε(x, y, z, t), structural deformation d(x, y, z, t), and peak stress σ in stress concentration areas such as welds, elbows, and diameter changes were calculated. max（i,j）。

[0041] Based on the finite element solution of the above structural equilibrium equations, the global stress tensor field σ(x,y,z,t) is obtained by element stress interpolation, and its component form is as follows: , Based on Hooke's Law , where E is the elastic modulus, ν is Poisson's ratio, tr(σ) is the stress tensor trace, that is, the matrix trace of the stress tensor σ(x,y,z,t) (the sum of the elements on the main diagonal of the matrix), I is the identity matrix, and the global strain tensor field ε(x,y,z,t) is calculated from the stress tensor field. The displacement vector δ directly corresponds to the structural deformation d(x,y,z,t), that is, d(x,y,z,t)=δ(x,y,z,t); For geometrically discontinuous areas such as welds, elbows, reducers, and tees, extract the Mises equivalent stress of each element: , The peak stress σ in each region was obtained through screening. max (i,j), where (i,j) is the region number.

[0042] The stress field σ, the wall geometric offset Δd caused by structural deformation, and the stress-accelerated corrosion factor χ, which is the coefficient of stress on corrosion electrochemistry, are transferred to the corrosion evolution sub-model. At the same time, the wall profile update data after structural deformation is fed back to the fluid dynamics sub-model to correct the boundary of the flow field calculation domain.

[0043] (1) Parameter transfer to the corrosion evolution sub-model: The following parameters are transferred from the structural mechanics sub-model to the corrosion evolution sub-model through the data interaction interface of the multiphysics coupled simulation platform: ① Stress field σ(x,y,z,t): The global stress tensor field is mapped to the computational mesh of the corrosion sub-model, and the Mises equivalent stress σ at each node is extracted. eq (x,y,z,t) are used as inputs for calculating the corrosion rate; ② Wall geometric offset Δd(x,y,z,t): The normal displacement components of the nodes on the inner wall of the pipe are extracted from the structural deformation d(x,y,z,t) to obtain the wall geometric offset Δd(x,y,z,t), which is used to update the computational domain geometric boundary of the corrosion sub-model. ③ Stress-accelerated corrosion factor χ(σ): Calculated based on the stress field σ(x,y,z,t), its expression is as follows: (where k) σ σ is the stress influence coefficient. s (where the yield strength is the material's yield strength) can be directly substituted into the corrosion evolution equation to quantify the accelerating effect of stress on the corrosion electrochemical process.

[0044] (2) Backpropagation to the boundary of the fluid dynamics sub-model: Updated wall profile data after structural deformation, referring to the updated coordinate values ​​of the pipe inner wall nodes extracted from the structural deformation d(x,y,z,t): The initial wall node coordinates are (x0, y0, z0), and after structural deformation, the node coordinates are updated to (x0 + d). x ,y0+d y ,z0+d z ), where d x d y d z These are the three components of the displacement vector δ; The updated wall node coordinates are fed back to the fluid dynamics sub-model through the coupling interface, the fluid computation mesh is reconstructed, the geometric boundary of the flow field computation domain is corrected, and the flow field calculation is adapted to the actual shape of the pipe after structural deformation, so as to realize the bidirectional coupling iteration of the structural field and the flow field.

[0045] (c) Solving the corrosion evolution sub-model and quantifying corrosion damage Based on the synergistic mechanism of electrochemical corrosion kinetics and erosion corrosion, a stress-flow field coupled corrosion evolution model is constructed to distinguish three damage modes: uniform corrosion, pitting corrosion, and stress corrosion cracking. The corrosion rate and wall thickness damage are then calculated. Substituting the wall mass transfer coefficient k into the fluid dynamics sub-model m Wall shear stress τ w (Including fluid viscous shear stress and particle scouring shear stress. In scenarios where high precision is not required, fluid viscous shear stress can be ignored, and particle scouring shear stress is equivalent to wall shear stress.) The stress σ and stress-accelerated corrosion factor χ transmitted by the structural mechanics sub-model are combined with the medium electrochemical parameters to solve the stress-flow field coupled corrosion kinetic equation. Erosion corrosion synergistic rate: In the formula: v corr (t) is v corr (x, y, z, t) represents the corrosion rate. The baseline corrosion rate is the one without stress or erosion. The scouring sensitivity coefficient.

[0046] Through the data interaction interface of the multiphysics coupled simulation platform, the calculation results of the fluid dynamics sub-model and the structural mechanics sub-model are mapped to the computational mesh of the corrosion evolution sub-model to complete parameter substitution and equation solving. The specific steps are as follows: ① Substitute the flow field parameters: Use the wall mass transfer coefficient k obtained from the fluid dynamics sub-model. m Wall shear stress τ w Mapped to the corresponding wall node of the erosion sub-model, where τ w Substituting directly into the above corrosion kinetic equation, we quantify the accelerating effect of flow field erosion on corrosion; the wall mass transfer coefficient k m Used to correct the baseline corrosion rate v 0corr Characterizes the effect of the mass transfer process of the medium on electrochemical corrosion; ② Substitution of structural field parameters: Extract the Mises equivalent stress σ at each node from the global stress field σ(x,y,z,t) obtained by solving the structural mechanics sub-model. eq Substitute (x,y,z,t) into the formula for calculating the stress-accelerated corrosion factor: In the formula, k σ σ is the stress influence coefficient (calibrated by stress corrosion test). s The yield strength of the pipe is given; the calculated stress-accelerated corrosion factor χ is directly substituted into the corrosion kinetic equation to quantify the accelerating effect of stress on the corrosion electrochemical process. ③ Substitute the electrochemical parameters of the medium: Combine the electrochemical parameters of the medium in the pipeline, such as pH value, temperature, and ion concentration, to calibrate the reference corrosion rate v. 0corrWith the scouring sensitivity coefficient λ, complete the assignment of all parameters for the corrosion kinetic equation; ④ Solving the equations: Based on all the parameters substituted above, the corrosion kinetic equations are solved node by node to obtain the corrosion rate field v across the entire pipeline at all times. corr (x,y,z,t) and distinguish the corrosion rate distribution of three damage modes: uniform corrosion, pitting corrosion, and stress corrosion cracking.

[0047] Then, wall thickness damage quantification and multi-field feedback updates were performed: Calculate the wall thickness corrosion reduction (i.e., wall thickness damage) within a single iteration step: b(x,y,z,t)= Local real-time wall thickness w is updated based on corrosion reduction. act (t)=w 0- b(x,y,z,t) (where w0 is the initial wall thickness of the pipe), and calculate the elastic modulus reduction factor η. corr (η) corr It is positively correlated with the degree of corrosion damage and is used to characterize the deterioration effect of corrosion on the mechanical properties of materials.

[0048] The wall thickness distribution field after corrosion damage w act (t), elastic modulus reduction factor η corr The data is then fed back to the structural mechanics sub-model based on the updated wall thickness w. act (t) Corrected structural stiffness matrix K Str Based on η corr By modifying the material's elastic modulus E, a two-way coupling iteration of corrosion and structure is achieved.

[0049] Simultaneously, the change in cross-sectional area ΔS of the flow channel caused by wall thickness reduction and the updated wall profile data after structural deformation (i.e., the coordinates of the nodes on the inner wall of the deformed pipe, which are obtained by superimposing the initial wall coordinates with the structural deformation d(x,y,z,t))) are fed back to the fluid dynamics sub-model: the fluid computation mesh is reconstructed, the boundary between the flow field domain and the computation domain is corrected, so that the flow field calculation is adapted to the actual morphology of the pipe after corrosion thinning and structural deformation, and the closed-loop bidirectional coupling iteration of the three fields of flow-solid-corrosion is realized.

[0050] (d) Bidirectional iteration of coupled variables and update of computational domain This stage is the core of the two-way coupling process, enabling closed-loop transfer of the three physical quantities and real-time updates of geometry and parameters: Flow field update: Based on the structural deformation d(x,y,z,t) obtained from the structural mechanics sub-model, the normal displacement components of the nodes on the inner wall of the pipe are extracted to obtain the wall deformation Δd(x,y,z,t). Based on the wall thickness corrosion reduction b(x,y,z,t) obtained from the corrosion evolution sub-model, the wall deformation Δd and the wall thickness corrosion reduction b are geometrically superimposed to obtain the final morphological change of the inner wall of the pipe. Based on the superimposed morphological change, the fluid computation mesh is reconstructed, the inner diameter and flow cross-sectional area of ​​the pipe are corrected, and the wall roughness is updated according to the degree of corrosion damage, completing the real-time update of the flow field computation domain and providing accurate geometric boundaries for the flow field solution of the next iteration.

[0051] Structural field update: The real-time wall thickness w obtained based on the corrosion evolution sub-model act (t), elastic modulus reduction factor η corr (η) corr (This is the degradation coefficient of the material's elastic modulus due to corrosion damage, which is positively correlated with the degree of wall thickness reduction). Substituting the above parameters into the structural mechanics solution system, based on the updated wall thickness w... act (t) and the reduced elastic modulus E′=E0·η corr (E0 is the initial elastic modulus), reconstruct the structural stiffness matrix K Str This corrects the stress field calculation benchmark, providing accurate mechanical parameters for the structural field solution in the next iteration.

[0052] Corrosion field parameters updated: The wall shear stress τ obtained by resolving the problem after updating the flow field domain is based on the updated wall shear stress τ. w (x,y,z,t), the stress field σ(x,y,z,t) obtained after resolving the stress field after updating the structural field; and the updated τ w Substituting σ back into the stress-flow field coupled corrosion kinetics equation, the corrosion rate calculation benchmark for the next iteration is corrected, enabling real-time iterative updates of corrosion field parameters and completing the closed-loop bidirectional coupling of the flow-solid-corrosion three fields.

[0053] (e) Convergence test and iteration termination Repeat steps (a)-(d) until the changes in all coupled variables satisfy the convergence criterion for coupled iteration convergence. The criterion is: ,in, Let k be the coupling variable (e.g., stress, corrosion rate) and k be the iteration number. To achieve a convergence threshold and ensure the stability and accuracy of the simulation results.

[0054] like Figure 3 As shown, the global flow field distribution u(x,y,z,t) and wall shear stress τ inside the pipe are obtained through coupled calculations. wReal-time multiphysics state data (x,y,z,t), stress distribution σ(x,y,z,t), strain distribution ε(x,y,z,t), and corrosion rate distribution v(x,y,z,t) are obtained. These state data are correlated and mapped with a digital twin model, and real-time visualization of the internal flow state, stress and strain distribution, and corrosion damage rate of the pipeline network is achieved through three-dimensional visualization technology.

[0055] To achieve continuous consistency between the digital twin model and the physical entity, and to eliminate the model accuracy degradation caused by material aging, operating condition drift, and cumulative corrosion damage, this invention adopts an iterative correction method for model parameters based on error feedback from "multi-source detection data - simulation data" to achieve adaptive calibration and long-term fidelity of the twin model. The specific implementation process is as follows: (1) Regularly use online non-destructive testing equipment such as pulsed eddy current, AC electromagnetic field, and ultrasonic guided wave to perform full-area scanning of high-risk areas such as pipe elbows, welds, diameter changes, and valve connections to obtain a dataset of measured defects in key areas. NDT ={dact, l, θact}, where d is the measured defect depth, l is the measured defect length, and θ is the measured three-dimensional coordinates of the defect. Simultaneously, real-time operational monitoring data such as temperature, pressure, stress, and electrochemical potential collected by the sensor network are fused to form a unified physical entity measured benchmark dataset D. Real All measured data were subjected to Kalman filtering for noise reduction and the 3σ criterion for outlier removal to complete data standardization preprocessing and ensure the reliability of the detection benchmark.

[0056] (2) Extract the data from the fluid-solid-corrosion multiphysics coupling simulation results related to D. Real Simulation output data D with the same spatial location, timestamp, and physical dimension Sim Including the simulated defect depth d sim Simulated wall thickness w sim Simulated stress value σ sim Simulated corrosion rate v sim etc. Using the three-dimensional coordinates of the pipeline network topology as the spatial reference and a unified timestamp as the time reference, D... Real With D Sim Spatiotemporal registration is performed to establish a one-to-one mapping relationship between physical measurement points and simulation grid nodes, thereby eliminating calculation errors caused by spatiotemporal misalignment.

[0057] (3) The matching error e between the simulation data and the measured data is calculated using the relative error method. The calculation formula is as follows: , Set the allowable error coefficient ε e In this embodiment, the value is taken as 5%, when the error e ≤ ε e When the accuracy of the digital twin model meets the standard, no correction is needed; when the error e > εe When this happens, the model parameter iterative correction process is automatically triggered.

[0058] (4) Based on the multiphysics coupling mechanism and parameter sensitivity analysis, adjustable parameters that have a significant impact on the simulation results are selected as correction variables, and a set of correction parameter vectors is constructed. : Where: k1 is the medium viscosity correction coefficient of the fluid field, k2 is the material elastic modulus correction coefficient of the structural mechanics field, and k3 is the environmental correction coefficient of the corrosion field, etc.

[0059] The corrected parameter vector set ψ is substituted into the Comsol finite element simulation software to update the parameter configurations of the fluid dynamics sub-model, structural mechanics sub-model, and corrosion evolution sub-model. The fluid-solid-corrosion bidirectional coupled simulation is then re-executed to generate the corrected simulation dataset. The error between the corrected simulation data and the measured data is calculated again to complete the accuracy verification until the model accuracy meets the standard.

[0060] To ensure that the iterative correction process does not affect the real-time performance of multiphysics simulation, the computational efficiency of the model is improved by combining model order reduction technology, collaborative scheduling of cloud computing and edge computing, and preset working condition simulation library for rapid real-time simulation.

[0061] Based on the modified digital twin model and multiphysics state data, fault features are extracted and matched with fault judgment rules to obtain fault diagnosis and trend prediction results.

[0062] In this embodiment, as Figure 4 As shown, all multi-source heterogeneous data are interpolated into a spatially continuous field using pipeline topology coordinates and a unified timestamp via Kriging interpolation. This field is then matched and fused with the simulation grid data at the same spatiotemporal resolution. A set of core parameters that directly and sensitively reflect the fault evolution mechanism is extracted and calculated, forming a set including the local stress concentration factor K. t Flow field disturbance intensity factor C f Equivalent wall thickness attenuation rate V eff The fault feature set F(t) is the deviation between the pressure loss anomaly coefficient ξ and the trend of the monitoring parameter e, where F(t) = {Kt, C}. f V eff ,ξ,e}. Construct a complete fault prediction and diagnosis mechanism of "data fusion representation - mechanism-driven judgment - quantitative analysis and source tracing".

[0063] Wherein the local stress concentration factor K t To calculate the theoretical stress concentration factor for geometrically discontinuous regions, the formula can be expressed as follows: , Let K be the defect depth, and ζ be the radius of curvature at the notch root. Simultaneously, simulation results verify that K... t =σmax / σ nom , σ max For the peak stress in this region, σ nom This is the theoretical stress.

[0064] The flow field disturbance intensity factor C f To quantify the intensity of flow separation and secondary flow in areas such as elbows and reducers, which is directly related to the risk of erosion corrosion, the formula is C. f =(u local -u inlet ) / u inlet ×(D / R) 0.5 , where u local For the local average flow velocity, u inlet Where is the average inlet velocity, D is the pipe diameter, and R is the bend radius.

[0065] Where the equivalent wall thickness attenuation rate V eff To comprehensively consider the synergistic damage effects of electrochemical uniform corrosion and localized fluid erosion, V eff (x,y,z,t)=V corr (t)+β·τ w (t), where V corr (t) represents the electrochemical corrosion rate, τ w (t) represents the wall shear stress, and β represents the material erosion sensitivity coefficient.

[0066] Where the pressure loss anomaly coefficient ξ is specific to valves and filters, ξ=(ΔP measured- ΔP design ) / ΔP design It is used for early detection of blockages or internal leaks. ΔP measured For the actual measured voltage drop, ΔP design This is the rated voltage drop under design operating conditions.

[0067] Wherein, the trend deviation of the monitoring parameter e is the degree of deviation between the short-term change rate of key process parameters such as pressure, temperature, and flow rate and their historical normal operating range or process set value, e=|r(t)-r normal ∣ / σ normal Where r(t) is the instantaneous rate of change, r normal With σ normal These represent the historical normal mean and standard deviation.

[0068] Based on the physical mechanisms of typical failure modes in petrochemical pipeline networks, a deterministic mapping rule base is established between characteristic parameters and failure types such as corrosion thinning, scouring corrosion damage, stress corrosion cracking risk, valve internal leakage / blockage and flow-induced vibration, and abnormal operating conditions.

[0069] The corrosion thinning criterion is: at the equivalent wall thickness decay rate V...eff (t) exceeds the allowable limit V calculated based on the remaining life target, and the actual wall thickness w act Less than the theoretical wall thickness w of the current design design The corrosion thinning fault was determined to have occurred.

[0070] The erosion corrosion damage criterion is used to identify localized corrosion caused by accelerated flow. When the flow field disturbance intensity factor C in a certain region... f Greater than threshold C th And the equivalent wall thickness attenuation rate V in this region eff When (t) is significantly higher than the average level of the pipeline network, it is determined that there is synergistic damage from scouring and corrosion at that location.

[0071] The criterion for judging the risk of stress corrosion cracking is: in a corrosive environment, when the local stress concentration factor K... t Exceeding threshold K th If so, it is determined that there is a risk of stress corrosion cracking.

[0072] The criteria for valve internal leakage / blockage and flow-induced vibration mainly rely on fluid mechanics and vibration analysis. When the pressure loss anomaly coefficient ξ continuously exceeds the threshold ξ... th When this is accompanied by abnormal flow or pressure monitoring values, it is determined to be internal leakage or blockage of the valve. When the flow field disturbance intensity factor C... f Exceeding threshold C th And a significant vibration acceleration a was detected. v When (t), it is determined that there is a risk of flow-induced vibration.

[0073] The criterion for determining abnormal operating status is as follows: when the trend deviation of the monitored parameter e exceeds three times the standard deviation of the normal baseline, the system or local operating status is determined to be abnormal.

[0074] The location of the fault is determined by the geometric coordinates G associated with the characteristic parameters that triggered the above criteria. k Direct determination. Severity is then quantified and graded based on physical indicators: for corrosion-related failures, the residual strength factor (RSF) and V-based... eff Predicted remaining life classification; for crack-related failures, classification based on safety margin or crack propagation rate using stress intensity factor; for flow field and valve failures, classification based on ξ and C. f The degree of exceedance and duration are classified.

[0075] Based on the current state and physical model, predictions are made regarding corrosion development, crack propagation, and stress distribution development. The corrosion development prediction includes the wall thickness at future moments. In the formula Calculations are based on predicted operating conditions. Crack propagation prediction uses the Paris formula for iteration. In the formula The predicted number of stress cycles, ΔK, represents the stress intensity factor amplitude. Stress distribution prediction involves: based on known or predicted future load spectra (pressure and temperature cycles), rapid analysis using the parameterized finite element model established in step 2 to calculate the stress field distribution σ(x,y,z,t) at key future time points. future ).

[0076] Specifically, fault prediction can include: Quantitative prediction of crack propagation using the Paris formula: , in, Let be the crack depth at time t+ΔT. Let t be the crack depth at time t, and C be the material fatigue constant. The predicted number of stress cycles is given by ΔK, where ΔK is the stress intensity factor amplitude. The residual strength factor is used to assess the structural load-bearing safety margin. , Where RSF is the residual intensity factor, σ s σ is the yield strength of the pipe. max This represents the maximum working stress that the pipeline actually withstands. Type I stress intensity factor is used to determine the critical crack state: , Among them, K I Here, Y is the Type I (open type) stress intensity factor, Y is the geometric correction factor related to pipe geometry and crack location, and σ is the long-range working stress at the crack tip. This refers to the crack depth. By combining the above formulas, quantitative prediction and remaining service assessment of crack propagation, corrosion damage, and stress evolution in petrochemical pipeline networks can be achieved.

[0077] As a second aspect of the present invention, a petrochemical pipeline network fault prediction system based on multi-physics field coupled digital twins is also proposed, comprising: a digital twin model construction module, a multi-physics field coupled simulation module, a fault prediction and diagnosis module, and a visualization management module; The output of the digital twin model construction module is connected to the input of the multiphysics coupling simulation module; the output of the multiphysics coupling simulation module is connected to the input of the fault prediction and diagnosis module; and the output of the fault prediction and diagnosis module is connected to the input of the visualization management module. The digital twin model construction module is used to collect and preprocess multi-source data of petrochemical pipeline networks, couple and integrate multi-physics field sub-models, and obtain a digital twin model that matches the physical pipeline network. The multiphysics coupling simulation module is used to perform two-way fluid-solid-corrosion coupling simulation based on a digital twin model and with real-time running data as the boundary to obtain multiphysics state data of the pipeline network; it also obtains measured data of the physical entity of the pipeline network through online non-destructive testing and sensor monitoring, compares the measured data with the multiphysics state data obtained from the simulation, and corrects the digital twin model based on the comparison error; The fault prediction and diagnosis module is used to extract fault features and match fault judgment rules based on the corrected digital twin model and multiphysics state data obtained from simulation, so as to obtain fault diagnosis and trend prediction results. The visualization management module is used to build a platform based on predictive diagnostic results, enabling visualization of pipeline network status, early warning, and output of operation and maintenance decisions.

[0078] The multiphysics sub-model includes a fluid dynamics sub-model, a structural mechanics sub-model, and a corrosion evolution sub-model; The fluid-solid-corrosion bidirectional coupled simulation includes: The fluid dynamics sub-model is solved unidirectionally to obtain the flow field distribution and transfer the wall shear stress and pressure load to the structural field and corrosion field. The structural mechanics sub-model is solved to obtain stress, strain and structural deformation, and the stress and deformation are fed back to the corrosion field; The corrosion evolution sub-model is solved to calculate the corrosion rate and wall thickness damage, and the wall thickness reduction and modulus reduction are fed back to the structural field and flow field. The coupled variables are iterated bidirectionally to synchronize the updates of parameters in the flow field, structural field, and corrosion field. Convergence assessment: Repeat the iteration until all changes in the coupling variables satisfy the convergence threshold.

[0079] In this embodiment, a comprehensive software platform is constructed that integrates a pipeline network model visualization module, a status assessment module, a prediction and early warning module, and a decision support module to achieve comprehensive visualization monitoring and intelligent management of the health status of the pipeline network throughout its entire life cycle.

[0080] The pipeline network model visualization module serves as the system's main interface, displaying a 1:1 mapping of the 3D digital twin pipeline network model to the physical entity. This module supports dynamically loading and overlaying multi-dimensional cloud maps from multiphysics simulations and real-time monitoring, including: real-time temperature / pressure distribution thermal cloud maps, stress-strain cloud maps, flow field vector cloud maps, corrosion rate cloud maps, and risk point cloud maps based on risk index coloring, enabling the visualization of invisible states within the pipeline network.

[0081] The status assessment module receives and processes fault diagnosis results in real time. This module visually displays the overall health status of the pipeline network, various fault modes and their severity levels, and key risk indicator values ​​in the form of dashboards, lists, and 3D labels, providing support for rapid status awareness.

[0082] The prediction and early warning module provides a forward-looking view based on trend prediction results. This module can dynamically display corrosion development prediction cloud maps, crack propagation prediction cloud maps, and stress distribution prediction cloud maps, and plot the predicted evolution trend curves of risk indicators at key locations. When the prediction results reach the warning line, the system automatically pops up visual and audible alarms.

[0083] The decision support module is geared towards maintenance decision-making, integrating all conclusions from diagnostics and predictions. This module automatically generates structured maintenance recommendation reports, including details of fault location, ranking of primary causes, assessment of the scope of impact, and mechanism-based targeted maintenance decision recommendations.

[0084] In one possible implementation, the digital twin model construction module includes a data acquisition unit, a data preprocessing unit, and a model coupling unit; the data acquisition unit, data preprocessing unit, and model coupling unit are sequentially signal-connected; the data acquisition unit is used to acquire geometric data, material physical parameters, and historical operation and maintenance data of the petrochemical pipeline network; the data preprocessing unit is used to remove outliers from multi-source data using the 3σ criterion, and simultaneously complete the spatiotemporal unification and format standardization of multi-source data; the model coupling unit is used to couple multi-physics sub-models to construct a digital twin model, and its output is connected to the multi-physics coupling simulation module.

[0085] In one possible implementation, the multiphysics coupled simulation module includes a sensor acquisition unit, a dynamic boundary construction unit, a simulation calculation unit, and a model correction unit. The sensor acquisition unit, dynamic boundary construction unit, and simulation calculation unit are sequentially signal-connected, and the model correction unit is bidirectionally data-connected to both the simulation calculation unit and the digital twin model construction module. The sensor acquisition unit collects real-time operating data of the pipeline network, including temperature, pressure, flow rate, vibration, and electrochemical parameters. This real-time operating data is collected in a time-series format, and the dynamic boundary conditions are updated in real-time according to the operating conditions. The dynamic boundary construction unit constructs simulation dynamic boundaries based on the real-time operating data. The simulation calculation unit performs a fluid-solid-corrosion bidirectional coupled simulation and outputs multiphysics state data. The model correction unit iteratively corrects the digital twin model based on the error between measured and simulated data. The corrected digital twin model uses the relative error method to calculate the data matching error. When the error exceeds a threshold, the simulation parameters are iteratively corrected. A 5% error coefficient is used as the judgment threshold. Based on parameter sensitivity analysis, adjustable simulation parameters are selected to construct a correction vector. After iterative correction, the coupled simulation is re-executed until the error reaches the target.

[0086] In one possible implementation, the fault prediction and diagnosis module includes a feature extraction unit, a rule base unit, a fault determination unit, and a trend prediction unit. The feature extraction unit, fault determination unit, and trend prediction unit are sequentially signal-connected, and the output of the rule base unit is signal-connected to the fault determination unit. The feature extraction unit is used to extract fault features based on models and state data. These fault features include stress, flow field, corrosion, pressure, and operational trend-related features. The trend prediction includes corrosion, crack, and stress development prediction. The fault features are combined with pipeline topology coordinates and timestamps to achieve spatiotemporal correlation. The trend prediction uses the Paris formula to predict crack propagation, the residual strength factor to assess corrosion faults, and the stress intensity factor to determine crack safety margins. The rule base unit is used to store judgment rules for typical faults in petrochemical pipelines. The fault determination unit is used to match features and rules to complete fault diagnosis. The trend prediction unit is used to predict fault development trends based on the diagnostic results.

[0087] In one possible implementation, the visualization management module includes a model display unit, a status assessment unit, an early warning and alarm unit, and a decision generation unit; the status assessment unit, early warning and alarm unit, and decision generation unit are all signal-connected to the model display unit; the model display unit is used to visualize the pipeline twin model and multiphysics cloud map; the status assessment unit is used to assess the health status of the pipeline based on diagnostic results; the early warning and alarm unit is used to issue early warning prompts when risks exceed standards; and the decision generation unit is used to generate pipeline operation and maintenance decisions and maintenance suggestions.

[0088] In one possible implementation, this system employs a collaborative scheduling of edge computing and cloud computing. The edge computing is responsible for real-time data acquisition, preprocessing, and local rapid simulation, while the cloud computing is responsible for large-scale coupled computing, model training, and long-term data storage.

[0089] The foregoing has described specific embodiments of the present invention. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired results. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0090] In the description of the embodiments of the present invention, the terms "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In the embodiments of the present invention, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described may be combined in a suitable manner in any one or more embodiments or examples. Furthermore, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in the embodiments of the present invention, as well as the features of the different embodiments or examples.

[0091] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of embodiments of the present invention, "multiple" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0092] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as should be understood by those skilled in the art to which embodiments of the invention pertain.

[0093] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for predicting faults in petrochemical pipeline networks using multi-physics coupled digital twins, characterized in that, Includes the following steps: Collect and preprocess multi-source data from petrochemical pipeline networks, couple and integrate multi-physics sub-models, and obtain a digital twin model that matches the physical pipeline network; Based on the digital twin model, a two-way coupled simulation of fluid-solid-corrosion is performed with real-time running data as the boundary to obtain multi-physics state data of the pipeline network; measured data of the physical entity of the pipeline network are obtained through online non-destructive testing and sensor monitoring, and the measured data are compared with the multi-physics state data obtained by simulation. The digital twin model is then corrected based on the comparison error. Based on the corrected digital twin model and the multiphysics state data obtained from simulation, fault features are extracted and matched with fault judgment rules to obtain fault diagnosis and trend prediction results. The multiphysics sub-model includes a fluid dynamics sub-model, a structural mechanics sub-model, and a corrosion evolution sub-model; The fluid-solid-corrosion bidirectional coupled simulation includes: The fluid dynamics sub-model is solved unidirectionally to obtain the flow field distribution and transfer the wall shear stress and pressure load to the structural field and corrosion field. The structural mechanics sub-model is solved to obtain stress, strain and structural deformation, and the stress and deformation are fed back to the corrosion field; The corrosion evolution sub-model is solved to calculate the corrosion rate and wall thickness damage, and the wall thickness reduction and modulus reduction are fed back to the structural field and flow field. The coupled variables are iterated bidirectionally to synchronize the updates of parameters in the flow field, structural field, and corrosion field. Convergence assessment: Repeat the iteration until all changes in the coupling variables satisfy the convergence threshold.

2. The fault prediction method according to claim 1, characterized in that, The unidirectional solution of the fluid dynamics sub-model includes: Solving the continuity and momentum equations using the finite volume method: ， ， In the formula, u is the fluid velocity vector, ρ is the fluid density, t is time, and μ is the fluid dynamic viscosity. For fluid pressure, ρ is the buoyancy of the particles, and g is the acceleration due to gravity. The wall shear stress, wall pressure load, and flow field velocity distribution are obtained by solving the problem.

3. The fault prediction method according to claim 1, characterized in that, The solution to the structural mechanics sub-model includes: Using the finite element method, substituting the wall shear stress and pressure load transmitted by the fluid dynamics sub-model, and combining the thermal strain generated by the temperature field, the structural equilibrium equations are solved: ， In the formula, Here is the structural stiffness matrix. It is a displacement vector. For the flow-induced load vector, For thermal load vectors, This is the equivalent load for stress concentration caused by corrosion defects; The stress field, structural deformation, and stress concentration factor are obtained by solving the problem.

4. The fault prediction method according to claim 1, characterized in that, The solution to the corrosion evolution sub-model includes: Based on the synergistic mechanism of electrochemistry and erosion corrosion, a stress-flow field coupled corrosion kinetic equation is constructed: ， In the formula, v corr (t) represents the corrosion rate. The baseline corrosion rate is the one without stress or erosion, and χ is the stress acceleration factor. To flush out the sensitivity coefficient, For wall shear stress; The corrosion rate and wall thickness damage are obtained by solving the equation.

5. The fault prediction method according to claim 1, characterized in that, The bidirectional iteration of the coupling variables includes: Based on the reduction in corrosion wall thickness and structural deformation, the fluid computational mesh is reconstructed, and the pipe inner diameter, flow cross-sectional area, and wall roughness are corrected. Based on the reduction of wall thickness and modulus after corrosion, the structural stiffness matrix is ​​reconstructed, and the stress calculation benchmark is corrected. The corrosion rate calculation benchmark was revised based on the updated flow field shear stress and structural stress.

6. The fault prediction method according to claim 1, characterized in that, The convergence criterion includes: Set convergence threshold When the coupling variables satisfy ， If the iteration converges, the simulation is terminated; otherwise, the iteration is repeated. Let k be the coupling variable and k be the iteration number. This is the convergence threshold.

7. The fault prediction method according to claim 1, characterized in that, The fault characteristics include stress, flow field, corrosion, pressure, and operating trend. The fault characteristics are combined with pipeline topology coordinates and timestamps to achieve spatiotemporal correlation. Based on the corrected digital twin model and multiphysics state data obtained from simulation, fault features are extracted and matched with fault determination rules to obtain fault diagnosis and trend prediction results, including: Quantitative prediction of crack propagation using the Paris formula: ， in, Let be the crack depth at time t+ΔT. Let t be the crack depth at time t, and C be the material fatigue constant. The predicted number of stress cycles is given by ΔK, where ΔK is the stress intensity factor amplitude. The residual strength factor is used to assess the structural load-bearing safety margin. ， Where RSF is the residual intensity factor, σ s σ is the yield strength of the pipe. max This represents the maximum working stress that the pipeline actually withstands. Type I stress intensity factor is used to determine the critical crack state: ， Among them, K I Here, Y is the Type I stress intensity factor, which is a geometric correction factor related to pipe geometry and crack location, and σ is the long-range working stress at the crack tip. This refers to the crack depth. By combining calculations, we can achieve quantitative prediction and remaining life assessment of crack propagation, corrosion damage, and stress evolution in petrochemical pipeline networks.

8. A petrochemical pipeline network fault prediction system based on multi-physics coupled digital twins, characterized in that, include: The digital twin model construction module is used to collect and preprocess multi-source data of petrochemical pipeline networks, couple and integrate multi-physics field sub-models, and obtain a digital twin model that matches the physical pipeline network. The multiphysics coupling simulation module is used to perform two-way fluid-solid-corrosion coupling simulation based on a digital twin model and with real-time running data as the boundary to obtain multiphysics state data of the pipeline network; it also obtains measured data of the physical entity of the pipeline network through online non-destructive testing and sensor monitoring, compares the measured data with the multiphysics state data obtained from the simulation, and corrects the digital twin model based on the comparison error; The fault prediction and diagnosis module is used to extract fault features and match fault judgment rules based on the corrected digital twin model and multiphysics state data obtained from simulation, so as to obtain fault diagnosis and trend prediction results. The visualization management module is used to build a platform based on predictive diagnostic results, enabling visualization of pipeline network status, early warning, and output of operation and maintenance decisions. The multiphysics sub-model includes a fluid dynamics sub-model, a structural mechanics sub-model, and a corrosion evolution sub-model; The fluid-solid-corrosion bidirectional coupled simulation includes: The fluid dynamics sub-model is solved unidirectionally to obtain the flow field distribution and transfer the wall shear stress and pressure load to the structural field and corrosion field. The structural mechanics sub-model is solved to obtain stress, strain and structural deformation, and the stress and deformation are fed back to the corrosion field; The corrosion evolution sub-model is solved to calculate the corrosion rate and wall thickness damage, and the wall thickness reduction and modulus reduction are fed back to the structural field and flow field. The coupled variables are iterated bidirectionally to synchronize the updates of parameters in the flow field, structural field, and corrosion field. Convergence assessment: Repeat the iteration until all changes in the coupling variables satisfy the convergence threshold.

9. The fault prediction system according to claim 8, characterized in that, The unidirectional solution of the fluid dynamics sub-model includes: Solving the continuity and momentum equations using the finite volume method: ， ， In the formula: u is the fluid velocity vector, ρ is the fluid density, and t is time. For fluid pressure, ρ is the buoyancy of the particulate phase, g is the gravitational acceleration, and μ is the fluid dynamic viscosity; The wall shear stress, wall pressure load, and flow field velocity distribution are obtained by solving the problem. The solution to the structural mechanics sub-model includes: Using the finite element method, substituting the wall shear stress and pressure load transmitted by the fluid dynamics sub-model, and combining the thermal strain generated by the temperature field, the structural equilibrium equations are solved: ， In the formula: Here is the structural stiffness matrix. It is a displacement vector. For the flow-induced load vector, For thermal load vectors, This is the equivalent load for stress concentration caused by corrosion defects; The stress field, structural deformation, and stress concentration factor are obtained. The solution to the corrosion evolution sub-model includes: Based on the synergistic mechanism of electrochemistry and erosion corrosion, a stress-flow field coupled corrosion kinetic equation is constructed: ， In the formula: v corr (t) represents the corrosion rate. The baseline corrosion rate is the one without stress or erosion, and χ is the stress acceleration factor. To flush out the sensitivity coefficient, For wall shear stress; The bidirectional iteration of the coupling variables includes: Based on the reduction in corrosion wall thickness and structural deformation, the fluid computational mesh is reconstructed, and the pipe inner diameter, flow cross-sectional area, and wall roughness are corrected. Based on the reduction of wall thickness and modulus after corrosion, the structural stiffness matrix is ​​reconstructed, and the stress calculation benchmark is corrected. The corrosion rate calculation benchmark was revised based on the updated flow field shear stress and structural stress.

10. The fault prediction system according to claim 9, characterized in that, The convergence criterion includes: Set convergence threshold When the coupling variables satisfy ， If the iteration converges, the simulation is terminated; otherwise, the iteration is repeated. Let k be the coupling variable and k be the iteration number. This is the convergence threshold; The fault characteristics include stress, flow field, corrosion, pressure, and operating trend. The fault characteristics are combined with pipeline topology coordinates and timestamps to achieve spatiotemporal correlation. Based on the corrected digital twin model and multiphysics state data obtained from simulation, fault features are extracted and matched with fault determination rules to obtain fault diagnosis and trend prediction results, including: Quantitative prediction of crack propagation using the Paris formula: ， in, Let be the crack depth at time t+ΔT. Let t be the crack depth at time t, and C be the material fatigue constant. The predicted number of stress cycles is given by ΔK, where ΔK is the stress intensity factor amplitude. The residual strength factor is used to assess the structural load-bearing safety margin. ， Where RSF is the residual intensity factor, σ s σ is the yield strength of the pipe. max This represents the maximum working stress that the pipeline actually withstands. Type I stress intensity factor is used to determine the critical crack state: ， Among them, K I Here, Y is the Type I stress intensity factor, which is a geometric correction factor related to pipe geometry and crack location, and σ is the long-range working stress at the crack tip. This refers to the crack depth. By combining calculations, we can achieve quantitative prediction and remaining life assessment of crack propagation, corrosion damage, and stress evolution in petrochemical pipeline networks.

Images