Pipe network life prediction system oriented to corrosion state

Abstract

The application relates to the field of digital twinning, in particular to a pipe network life prediction system facing an erosion state. The system comprises the following steps: a physical sensing component collects flow velocity signals, pressure signals and pipe wall potential signals of a pipe network medium; a twinning mapping component establishes a flow field-stress field-electrochemical field coupling simulation model, extracts the above signals as dynamic boundary conditions, calculates boundary difference values of current and last time boundary conditions and inputs the coupling simulation model, corrects material transmission flux and charge transfer rate of a multi-field interaction interface, and drives digital twin state updating; an evolution prediction component solves corrosion kinetics equations based on the corrected interface parameters, outputs a pipe network wall thickness space-time distribution state, and calculates residual life according to a thinning trend. The application overcomes state deviation caused by a static boundary, makes the digital twin respond to transient changes of working conditions, eliminates cumulative errors of a steady state assumption, and guarantees corrosion state mapping fidelity and life prediction reliability.

Description

Technical Field

[0001] This invention relates to the field of digital twins, specifically to a pipeline life prediction system for corrosion conditions. Background Technology

[0002] Pipeline transportation systems play a crucial role in material transport in industrial sectors. The internal media and external environment often lead to pipe wall corrosion, potentially causing leaks. To assess the remaining service life of pipeline networks, digital twin technology has been introduced into corrosion monitoring. Existing pipeline digital twin life prediction systems typically deploy sensors at the physical sensing layer to collect pipeline operation data and establish a geometric and physical field simulation model of the pipeline network in virtual space. In this conventional architecture, the boundary conditions of the simulation model are usually set offline based on historical average data or pipeline design parameters. During system operation, the data from the physical sensing layer is primarily used to trigger simulation calculations or as display input. The flow field, stress field, and electrochemical field boundaries in the virtual model remain constant throughout the calculation cycle. The corrosion rate of the pipe wall is obtained through steady-state solution methods, and the remaining service life of the pipeline network is estimated based on linear cumulative damage rules.

[0003] In actual pipeline network operation, fluctuations in flow velocity and pressure pulsations are normal phenomena. These dynamic changes alter the local stress distribution on the pipe wall and the interfacial electrochemical reaction environment in real time. Existing digital twin lifetime prediction systems for pipeline networks still rely on offline-calibrated static boundary conditions to drive multiphysics simulation models when handling these transient conditions. Due to the lack of a response mechanism to real-time changes in operating conditions, the interfaces between the flow field, stress field, and electrochemical field in the simulation model are decoupled or weakly coupled. The calculated mass transfer flux and charge transfer rate cannot be dynamically rebalanced with boundary disturbances. When drastic changes occur in pipeline network operating conditions, the simulation results under static boundary conditions deviate from the actual corrosion evolution state of the physical entity. This deviation accumulates as the simulation cycle lengthens.

[0004] Existing pipeline network digital twin life prediction systems rely on offline, static boundary conditions to drive multiphysics simulation models, resulting in the digital twin failing to keep pace with the transient changes in the physical entity's operating conditions. In the complex pipeline transportation environment, the mass transfer flux and charge transfer rate at multiple interaction interfaces cannot dynamically rebalance with boundary disturbances, causing the corrosion evolution trajectory obtained from the simulation to deviate from the actual state of the physical entity. This coupling distortion caused by boundary solidification leads to a distortion of the corrosion state evolution process. As the operating time increases, the deviation accumulates in the simulation calculations, resulting in cumulative errors in long-term life prediction. This is a core technical problem that urgently needs to be solved in the field of pipeline network digital twins. Summary of the Invention

[0005] The purpose of this invention is to provide a pipeline life prediction system for corrosion conditions, which can effectively solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A pipeline life prediction system for corrosion conditions includes: a physical sensing component that collects the flow velocity, pressure, and pipe wall potential signals of the pipeline operating medium; The twin mapping component establishes a coupled simulation model of the flow field, stress field, and electrochemical field of the pipeline network. It extracts the flow velocity signal, the pressure signal, and the pipe wall potential signal as dynamic boundary conditions, calculates the boundary difference between the dynamic boundary conditions at the current moment and the steady-state boundary conditions at the previous moment, and inputs the boundary difference into the coupled simulation model of the flow field, stress field, and electrochemical field to correct the mass transport flux and charge transfer rate at the multi-field interaction interface, and drives the synchronous update of the digital twin state. The evolution prediction component, based on the modified multi-field interaction interface parameters, solves the corrosion kinetic equation, outputs the spatiotemporal distribution of the pipeline wall thickness, and calculates the remaining life of the pipeline based on the wall thickness reduction trend, thus forming a pipeline life prediction system oriented towards corrosion state.

[0007] Preferably, when the physical sensing component extracts the flow velocity signal, the pressure signal, and the pipe wall potential signal as the dynamic boundary conditions, a time delay compensation mechanism is introduced. Based on the transmission distance of the flow velocity signal, the pressure signal, and the pipe wall potential signal in the pipeline medium and the sound speed of the medium, the time offset is calculated. According to the time offset, the dynamic boundary conditions at different spatial locations are aligned to a unified time reference to eliminate boundary asynchrony caused by signal transmission delay.

[0008] Preferably, when the twin mapping component establishes the coupled simulation model of the flow field-stress field-electrochemical field, it adopts an interleaved iterative solution architecture, sets the flow field and the stress field as external loop variables, sets the electrochemical field as internal loop variables, sets the convergence criterion for the multi-field interaction interface, and alternately updates the internal loop variables after the external loop variables are updated until the convergence criterion is met, and outputs the multi-field coupling equilibrium state.

[0009] Preferably, when the twin mapping component corrects the mass transport flux and the charge transfer rate of the multi-field interaction interface, an adaptive weight allocation mechanism is introduced to calculate the gradient change rate of the boundary difference, and dynamically allocate the weight coefficients of the convection and diffusion terms in the mass transport flux calculation, as well as the weight coefficients of the anodic and cathodic reactions in the charge transfer rate calculation, based on the gradient change rate.

[0010] Preferably, when the evolution prediction component solves the corrosion kinetic equation, it introduces a Monte Carlo simulation mechanism, generates a random sample set of parameters that conforms to a normal distribution based on the modified multi-field interaction interface parameters, inputs the random sample set of parameters into the corrosion kinetic equation for multiple iterative solutions, statistically analyzes the time probability distribution of the pipe wall thickness reaching the critical failure threshold, and outputs the remaining life confidence interval of the pipeline network.

[0011] Preferably, when calculating the time offset using the time delay compensation mechanism, a pipeline topology model is constructed, a shortest path search algorithm is used to determine the signal transmission path from the physical sensing component to the twin mapping component, the equivalent medium sound velocity is calculated based on the pipe diameter change ratio on the signal transmission path, and the accurate signal propagation delay is calculated as the time offset based on the equivalent medium sound velocity and the length of the signal transmission path.

[0012] Preferably, during the alternating update process of the inner loop variables in the staggered iterative solution architecture, a stress-related polarization correction factor is introduced to extract the equivalent stress parameters of the stress field updated by the outer loop variables. Based on the equivalent stress parameters, the anode Tafel slope and cathode Tafel slope of the electrochemical kinetic equation in the electrochemical field are dynamically corrected, and the corrected electrochemical kinetic equation is applied to the iterative solution process of the inner loop variables.

[0013] Preferably, when dynamically allocating the weight coefficients according to the gradient change rate, a damping smoothing mechanism is introduced. The difference between the real-time weight coefficient of the current iteration cycle and the historical weight coefficient of the previous iteration cycle is calculated. A nonlinear damping factor is applied to the coefficient difference to generate a smoothing compensation amount. The smoothing compensation amount is superimposed on the real-time weight coefficient to output the final weight coefficient, thereby suppressing the weight coefficient oscillation phenomenon caused by the sudden change in the boundary difference gradient.

[0014] Preferably, before the twin mapping component runs, a model reduction operation based on intrinsic orthogonal decomposition is performed on the flow field-stress field-electrochemical field coupled simulation model to extract the global orthogonal basis vectors of the multi-field coupled simulation results. The high-dimensional partial differential equation system is projected onto the low-dimensional subspace spanned by the global orthogonal basis vectors to generate a reduced-order coupled computation model. The reduced-order coupled computation model replaces the original high-dimensional model for the input of the boundary difference and the update of the state of the digital twin.

[0015] Preferably, after the physical sensing component acquires the flow velocity signal, the pressure signal, and the pipe wall potential signal, it introduces an abnormal data reconstruction mechanism based on spatiotemporal correlation, calculates the signal correlation matrix between adjacent spatial nodes and adjacent time nodes in the pipeline network, identifies the abnormal signal positions that deviate from the signal correlation matrix, uses the signal correlation matrix to perform interpolation filling and smoothing filtering on the abnormal signal positions, and transmits the processed signal to the twin mapping component.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention extracts the flow velocity, pressure, and pipe wall potential signals of the pipeline operating medium as dynamic boundary conditions. It calculates the boundary difference between the current dynamic boundary condition and the previous steady-state boundary condition, inputting this difference into a coupled simulation model of the flow field, stress field, and electrochemical field to correct the mass transport flux and charge transfer rate at the multi-field interaction interface. This mechanism enables the digital twin to respond to transient changes in pipeline operating conditions, overcoming the twin state shift defects caused by static boundary conditions. When the medium flow velocity and pressure fluctuate, the boundary difference drives the mass transport and charge transfer processes in the coupled simulation model for real-time rebalancing calculations, eliminating the cumulative errors introduced by steady-state assumptions and ensuring the fidelity of corrosion state mapping and the reliability of remaining life prediction results under long-term pipeline operation. Through dynamic boundary compensation, the system does not need to restart the full simulation calculation every time operating conditions change, maintaining the continuity of model calculations while tracking the corrosion evolution trend of the physical entity, thus solving the cumulative drift problem in long-term life prediction.

[0017] 2. This invention introduces a time delay compensation mechanism, calculating the time offset based on the transmission distance of flow velocity, pressure, and pipe wall potential signals and the medium sound velocity. This aligns dynamic boundary conditions at different spatial locations, eliminating the asynchronous phenomenon of multi-field calculation boundaries caused by signal transmission delay. By calculating the signal transmission path and combining it with the pipe diameter change ratio to convert the equivalent medium sound velocity, the accuracy of time offset calculation is improved. An interleaved iterative solution architecture and stress-related polarization correction factor are adopted, using the flow field and stress field as external circulation variables and the electrochemical field as internal circulation variables. The anode and cathode Tafel slopes are corrected based on equivalent stress parameters, realizing the coupled solution of stress and electrochemistry in the multi-physics interaction process. An adaptive weight allocation mechanism and a damping smoothing mechanism are introduced, allocating weight coefficients for convection and diffusion terms and anode and cathode reactions based on the gradient rate of change of boundary differences. A nonlinear damping factor suppresses weight coefficient oscillations, ensuring the numerical stability of the dynamic correction process. Based on intrinsic orthogonal decomposition, a reduction operation is performed on the high-dimensional model, replacing the original model in the update, reducing the computational resource overhead of the dynamic boundary compensation process. Spatiotemporal correlation is used to interpolate and smooth anomalous signals, avoiding the impact of anomalous data on the simulation model. By combining a random sample set of normally distributed parameters with a Monte Carlo simulation mechanism, the remaining lifetime confidence interval is output, quantifying the uncertainty of lifetime prediction under dynamic boundary perturbations. Attached Figure Description

[0018] Figure 1 This is a flowchart of the overall operation of the pipeline life prediction system for corrosion conditions according to the present invention. Figure 2 This is a flowchart of the pipeline sensing signal delay compensation and multi-measuring point dynamic boundary timing unified alignment processing of the present invention; Figure 3 This is a flowchart of the entire process of simulation and solution of the flow field-stress field-electrochemical field alternating iterative coupling in this invention; Figure 4 This is a flowchart of the multi-field interface parameter adaptive weight allocation and damping smoothing anti-oscillation correction process of the present invention; Figure 5 This is a flowchart of the calculation and processing of the confidence interval of pipeline corrosion remaining life based on Monte Carlo sampling in this invention. Figure 6 This is a flowchart of the linkage process for abnormal sensor signal data preprocessing and POD model order reduction optimization in this invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Please refer to Figure 1 This embodiment provides a pipeline life prediction system for corrosion conditions, including a physical sensing component, a twin mapping component, and an evolution prediction component. The physical sensing component is deployed at key nodes in the pipeline network to collect flow velocity, pressure, and pipe wall potential signals of the operating medium. The twin mapping component pre-establishes a coupled simulation model of the pipeline network's flow field, stress field, and electrochemical field. This model is constructed based on the pipeline network's geometric parameters, material properties, and medium characteristics, and includes flow field control equations, stress field control equations, and electrochemical field control equations, as well as the coupling relationships of multiple field interfaces. The twin mapping component extracts the flow velocity, pressure, and pipe wall potential signals collected by the physical sensing component as dynamic boundary conditions, calculates the boundary difference between the current dynamic boundary conditions and the previous steady-state boundary conditions, and inputs this boundary difference into the coupled simulation model to correct the mass transfer flux and charge transfer rate at the multiple field interfaces, driving the synchronous update of the digital twin's state. The evolution prediction component solves the corrosion kinetic equation based on the modified multi-field interaction interface parameters, outputs the spatiotemporal distribution of the pipe network wall thickness, and calculates the remaining life of the pipe network based on the wall thickness reduction trend.

[0021] In this embodiment, the physical sensing component includes a flow velocity sensor, a pressure sensor, and a potential sensor. The flow velocity sensor measures the medium flow velocity using the Doppler principle, outputting an electrical signal with a frequency proportional to the flow velocity. The pressure sensor measures the medium pressure using the piezoresistive principle, outputting an electrical signal with a voltage proportional to the pressure. The potential sensor measures the pipe wall potential using the reference electrode method, outputting an electrical signal with a voltage proportional to the potential difference between the pipe wall and the reference electrode. The physical sensing component acquires these signals at a fixed sampling frequency, determined based on the dynamic characteristics of the pipeline network's operating conditions, ensuring the capture of transient changes in the operating conditions. The acquired raw signals are converted from analog to digital and then transmitted to the twin mapping component.

[0022] When establishing a coupled simulation model of flow field, stress field, and electrochemical field using the twin mapping component, the geometric model of the pipeline network is first constructed. This geometric model is generated based on the pipeline network's design drawings and field survey data, and includes geometric features such as pipe diameter, wall thickness, length, elbows, tees, and valves. Then, the geometric model is meshed, using either structured or unstructured meshes, with the mesh density determined based on computational accuracy requirements and available resources. After meshing, the governing equations and boundary conditions for each physical field are defined. The Navier-Stokes equations are used for the flow field to describe the flow characteristics of the medium; the elastic equilibrium equations are used for the stress field to describe the stress distribution on the pipe wall; and the Nernst and Tafel equations are used for the electrochemical field to describe the electrochemical reaction process on the pipe wall. The coupling relationships at the multi-field interface include: the pressure distribution in the flow field serves as the load boundary condition for the stress field; the velocity distribution in the flow field affects the mass transport process in the electrochemical field; the stress distribution in the stress field affects the charge transfer process in the electrochemical field; and the distribution of corrosion products in the electrochemical field affects the flow resistance and load distribution in the stress field.

[0023] In this embodiment, when the twin mapping component calculates the boundary difference between the current dynamic boundary condition and the previous steady-state boundary condition, it first stores the steady-state boundary conditions obtained after the simulation calculation at the previous moment. The steady-state boundary conditions include the flow velocity, pressure, and potential values ​​at each boundary node. Then, it compares the dynamic boundary conditions collected by the physical sensing component at the current moment with the steady-state boundary conditions at the previous moment node by node, calculating the boundary difference for each boundary node. The formula for calculating the boundary difference is:

[0024] in, Indicates the first Each boundary node at time... The boundary difference, Indicates the first Each boundary node at time... The dynamic boundary condition values, Indicates the first Each boundary node at time... The steady-state boundary condition values, This indicates the simulation time step.

[0025] The twin mapping component inputs boundary difference values ​​into the coupled simulation model of the flow field-stress field-electrochemical field to correct the mass transport flux and charge transfer rate at the multi-field interaction interface. Mass transport flux includes the transport flux from the corrosive medium to the pipe wall surface and the transport flux of corrosion products from the pipe wall surface to the bulk medium. Charge transfer rate includes the charge transfer rate of the anodic reaction and the charge transfer rate of the cathodic reaction. The correction process adjusts the transport coefficient and reaction rate constant at the multi-field interaction interface based on the magnitude and direction of the boundary difference value. When the boundary difference is positive, it indicates that the boundary condition value at the current moment is greater than the steady-state boundary condition value at the previous moment, correspondingly increasing the mass transport flux and charge transfer rate; when the boundary difference is negative, it indicates that the boundary condition value at the current moment is less than the steady-state boundary condition value at the previous moment, correspondingly decreasing the mass transport flux and charge transfer rate.

[0026] The evolution prediction component solves the corrosion kinetic equation based on the modified multi-field interaction parameters. The corrosion kinetic equation describes the relationship between the pipe wall corrosion rate and the multi-field interaction parameters, and its general form is:

[0027] in, Indicates the corrosion rate of the pipe wall. Indicates the concentration of the corrosive medium. This represents the equivalent stress of the pipe wall. Indicates the tube wall potential. This represents the temperature of the medium. In this embodiment, the corrosion kinetic equation is derived based on Faraday's law and the Tafel equation, and its specific form is:

[0028] in, Indicates the molar mass of a metal. Indicates the number of electrons transferred in an electrochemical reaction. Denotes Faraday's constant. The density of a metal, This represents the corrosion current density. The corrosion current density is determined by the current density at which the anodic and cathodic reaction current densities are equal, and its calculation formula is:

[0029] in, and These represent the exchange current densities for the anodic and cathodic reactions, respectively. and These represent the anode Tafel slope and the cathode Tafel slope, respectively. and These represent the equilibrium potentials for the anodic and cathodic reactions, respectively. Represents the gas constant. It represents thermodynamic temperature.

[0030] After solving the corrosion kinetic equations, the evolution prediction component outputs the spatiotemporal distribution of the pipeline wall thickness. This distribution represents the wall thickness at different locations and times within the pipeline. The initial wall thickness is determined based on the design wall thickness or the thickness measured in the field. Within each simulation time step, the wall thickness reduction is calculated based on the corrosion rate. This reduction is then subtracted from the current wall thickness to obtain the wall thickness for the next time step. The formula for calculating the wall thickness reduction is:

[0031] in, Indicates the amount of wall thickness reduction. Indicates the corrosion rate, This indicates the simulation time step.

[0032] The evolution prediction component calculates the remaining life of the pipeline network based on the wall thickness reduction trend. The remaining life is defined as the time from the current moment until the pipe wall thickness reaches the critical failure threshold. The critical failure threshold is determined according to the pipeline network's design standards and safety requirements. The evolution prediction component fits the wall thickness reduction trend of each grid node to obtain a function relating the wall thickness to time, and then solves for the time when this function equals the critical failure threshold; this is the remaining life of that node. The overall remaining life of the pipeline network is taken as the minimum of the remaining lives of all nodes.

[0033] At the end of this embodiment, the system collects real-time operational data through physical sensing components as dynamic boundary conditions, calculates boundary differences, and corrects multi-field interaction interface parameters to drive synchronous updates of the digital twin's state. Then, based on the corrected parameters, it solves the corrosion kinetic equation to calculate the spatiotemporal distribution of the pipeline wall thickness and its remaining life. This mechanism enables the digital twin to respond to transient changes in pipeline operating conditions, overcoming the twin state offset defects caused by static boundary conditions.

[0034] In a preferred embodiment, reference Figure 2 When the physical sensing component extracts flow velocity, pressure, and pipe wall potential signals as dynamic boundary conditions, a time delay compensation mechanism is introduced. This mechanism calculates the time offset based on the transmission distance of the flow velocity, pressure, and pipe wall potential signals within the pipe network medium and the speed of sound in the medium. It then aligns the dynamic boundary conditions at different spatial locations to a unified time reference based on this time offset, eliminating boundary asynchrony caused by signal transmission delays.

[0035] In this embodiment, the signal transmission speed in the pipeline medium is equal to the speed of sound in the medium. Because different physical sensing components are deployed at different spatial locations within the pipeline, the distance the signal travels from the physical sensing components to the twin mapping components varies, resulting in differences in the time it takes for the signal to reach the twin mapping components. Without time delay compensation, the dynamic boundary conditions at different spatial locations will be at different time bases, leading to asynchronous boundary conditions in multi-field coupled simulation calculations, thus affecting the accuracy of the simulation results.

[0036] The working principle of the time delay compensation mechanism is as follows: First, determine the spatial coordinates of each physical sensing component and the twin mapping component. Then, calculate the signal transmission distance from each physical sensing component to the twin mapping component. The signal transmission distance is the distance along the pipeline axis between the physical sensing component and the twin mapping component. Next, calculate the signal propagation delay based on the medium's sound speed; the signal propagation delay is equal to the signal transmission distance divided by the medium's sound speed. Finally, calculate the time offset based on the signal propagation delay, aligning the signals collected by different physical sensing components to a unified time reference. The formula for calculating the time offset is:

[0037] in, Indicates the first Time offset of each physical sensing component Indicates the first The signal transmission distance from the physical sensing component to the twin mapping component Indicates the speed of sound in the medium. This represents the signal propagation delay of the reference physical sensing component. Any physical sensing component can be selected as the reference; typically, the one with the shortest signal transmission distance is chosen.

[0038] In this embodiment, a pipeline network topology model is constructed when calculating the time offset using a time delay compensation mechanism. The pipeline network topology model consists of nodes and edges. Nodes represent connection points such as endpoints, elbows, tees, and valves in the pipeline network, while edges represent pipe segments connecting two nodes. Each edge has attributes such as length, diameter, and wall thickness. A shortest path search algorithm is used to determine the signal transmission path from the physical sensing component to the twin mapping component. The shortest path search algorithm can be either Dijkstra's algorithm or Floyd's algorithm. After determining the signal transmission path, the equivalent medium sound velocity is calculated based on the pipe diameter variation ratio along the transmission path. The medium sound velocity is related to the pipe diameter; when the pipe diameter changes, the medium sound velocity also changes. The formula for calculating the equivalent medium sound velocity is:

[0039] in, Represents the sound velocity in the equivalent medium. Indicates the number of pipe segments in the signal transmission path. Indicates the first The length of each pipe section Indicates the first The velocity of sound of the medium in each pipe section. The sound velocity of the medium in a pipe section is determined based on the diameter of that pipe section, and the calculation formula is as follows:

[0040] in, This represents the speed of sound in an infinite space. Indicates the bulk modulus of the medium. Indicates the density of the medium. Indicates the first The diameter of each pipe section, Indicates the first The wall thickness of each pipe section.

[0041] The accurate signal propagation delay is calculated as the time offset based on the equivalent medium sound velocity and the length of the signal transmission path. The accurate signal propagation delay is equal to the total length of the signal transmission path divided by the equivalent medium sound velocity. By constructing a pipeline network topology model and using a shortest path search algorithm to determine the signal transmission path, and by combining the equivalent medium sound velocity with the pipe diameter variation ratio, the calculation accuracy of the time offset is improved, thereby improving the accuracy of dynamic boundary condition alignment.

[0042] At the end of this embodiment, by introducing a time delay compensation mechanism, the time offset of the signals collected by the physical sensing components at different spatial locations is calculated, and the dynamic boundary conditions are aligned to a unified time reference. This eliminates the asynchronous phenomenon of multi-field calculation boundaries caused by signal transmission delay and improves the accuracy of multi-field coupled simulation calculation.

[0043] In a preferred embodiment, reference Figure 3 When establishing a coupled simulation model of flow field, stress field, and electrochemical field using the twin mapping component, an interleaved iterative solution architecture is adopted. This architecture sets the flow field and stress field as external loop variables and the electrochemical field as internal loop variables. A convergence criterion is set for the multi-field interaction interface. After updating the external loop variables, the internal loop variables are alternately updated until the convergence criterion is met, outputting the multi-field coupled equilibrium state.

[0044] In this embodiment, the workflow of the staggered iterative solution architecture is as follows: First, the variable values ​​of the flow field, stress field, and electrochemical field are initialized. The initial values ​​can be determined based on the design parameters of the pipeline network or historical operating data. Then, the outer loop is entered to update the flow field variables. The update of the flow field variables is based on the Navier-Stokes equations and solved using the finite volume method or the finite element method. After the flow field variables are updated, the stress field variables are updated. The update of the stress field variables is based on the elasticity equilibrium equations, using the pressure distribution of the flow field as the load boundary condition, and solved using the finite element method. After the stress field variables are updated, the inner loop is entered to update the electrochemical field variables. The update of the electrochemical field variables is based on the Nernst and Tafel equations, using the velocity distribution of the flow field and the stress distribution of the stress field as boundary conditions, and solved using the finite difference method or the finite element method. During the inner loop iteration, the convergence criterion of the multi-field interaction interface is checked. If the convergence criterion is not met, the electrochemical field variables continue to be updated; if the convergence criterion is met, the inner loop is exited, and the next outer loop iteration begins. When the number of iterations in the outer loop reaches the set maximum number of iterations or the multi-field coupling state reaches stability, the iteration stops and the multi-field coupling equilibrium state is output.

[0045] The convergence criterion for multi-interaction interfaces is determined based on the changes in the parameters of the multi-interaction interfaces. In this embodiment, the convergence criterion is:

[0046] in, This represents the change in mass transport flux at multiple interaction interfaces. This represents the maximum mass transport flux at multiple interaction interfaces. This represents the change in charge transfer rate at a multi-field interaction interface. This represents the maximum charge transfer rate at the multi-field interaction interface. This represents the convergence threshold. The convergence threshold is determined based on the required computational accuracy and is typically set to a value of [value missing]. to .

[0047] In this embodiment, a stress-related polarization correction factor is introduced during the alternating update process of the inner loop variables in the staggered iterative solution architecture. The stress-related polarization correction factor extracts the equivalent stress parameters of the stress field updated by the outer loop variables. Based on the equivalent stress parameters, the anode Tafel slope and cathode Tafel slope of the electrochemical kinetic equation in the electrochemical field are dynamically corrected. The corrected electrochemical kinetic equation is then applied to the iterative solution process of the inner loop variables.

[0048] The equivalent stress parameter adopts the von Mises equivalent stress, and its calculation formula is as follows:

[0049] in, This indicates the von Mises equivalent effect. , , These represent the three principal stresses.

[0050] The formula for calculating the stress-related polarization correction factor is:

[0051] in, This represents the stress-related polarization correction factor. Indicates the stress sensitivity coefficient. This represents the von Mises equivalent stress. The stress sensitivity coefficient is determined based on the properties of the metallic material and is obtained through experimental measurement.

[0052] The corrected formulas for calculating the anode Tafel slope and the cathode Tafel slope are as follows:

[0053]

[0054] in, and These represent the corrected anode Tafel slope and cathode Tafel slope, respectively. and These represent the original anode Tafel slope and cathode Tafel slope, respectively. This represents the stress-related polarization correction factor.

[0055] By introducing a stress-dependent polarization correction factor, the stress distribution of the stress field is coupled with the electrochemical reaction process of the electrochemical field, realizing the stress-electrochemical coupling solution in the multi-physics interaction process and improving the accuracy of electrochemical field calculation. The stress-dependent polarization correction factor and Tafel slope under different equivalent stresses are shown in Table 1.

[0056] Table 1. Stress-related polarization correction factor and Tafel slope under different equivalent stresses

[0057] Table 1 shows the stress-related polarization correction factor and Tafel slope under different equivalent stresses. As can be seen from Table 1, the stress-related polarization correction factor increases linearly with increasing equivalent stress, and the corrected anodic and cathodic Tafel slopes also increase accordingly. This indicates that the presence of stress accelerates the electrochemical reaction process, leading to an increased corrosion rate.

[0058] At the end of this embodiment, by employing an interleaved iterative solution architecture, the flow field and stress field are used as external loop variables, and the electrochemical field as an internal loop variable, achieving a coupled solution for multiple physics fields. A stress-related polarization correction factor is introduced to dynamically correct the anode and cathode Tafel slopes based on equivalent stress parameters, improving the accuracy of the electrochemical field calculation and thus enhancing the reliability of the multi-field coupled simulation results.

[0059] In a preferred embodiment, reference Figure 4 When the twin mapping component corrects the mass transport flux and charge transfer rate at multi-field interaction interfaces, an adaptive weight allocation mechanism is introduced. The adaptive weight allocation mechanism calculates the gradient rate of change of the boundary difference and dynamically allocates the weight coefficients of the convection and diffusion terms in the mass transport flux calculation, as well as the weight coefficients of the anodic and cathodic reactions in the charge transfer rate calculation, based on the gradient rate of change.

[0060] In this embodiment, the mass transport flux consists of a convection term and a diffusion term, and its calculation formula is as follows:

[0061] in, Represents the total mass transport flux. Represents the convective mass transport flux. This represents the mass transport flux during diffusion. This represents the weighting coefficient of the convection term. This represents the weighting coefficient of the diffusion term. The sum of the weighting coefficients of the convection and diffusion terms is 1, i.e. .

[0062] The formula for calculating the convective mass transport flux is:

[0063] in, Indicates the flow rate of the medium. This indicates the concentration of the corrosive medium.

[0064] The formula for calculating the diffusion term mass transport flux is:

[0065] in, This represents the diffusion coefficient of the corrosive medium. This represents the concentration gradient of the corrosive medium.

[0066] The charge transfer rate consists of the charge transfer rate of the anodic reaction and the charge transfer rate of the cathodic reaction, and its calculation formula is as follows:

[0067] in, Indicates the total charge transfer rate. Indicates the charge transfer rate of the anodic reaction. Indicates the charge transfer rate of the cathode reaction. The weighting coefficients represent the anodic reaction. This represents the weighting coefficient for the cathode reaction. The sum of the weighting coefficients for the anode reaction and the cathode reaction is 1, i.e. .

[0068] The formula for calculating the charge transfer rate of the anodic reaction is:

[0069] The formula for calculating the charge transfer rate of the cathode reaction is:

[0070] The adaptive weight allocation mechanism calculates the gradient rate of change of the boundary difference. The gradient rate of change of the boundary difference represents the rate of change of the boundary difference over time, and its calculation formula is:

[0071] in, Indicates time The rate of change of the boundary difference gradient, Indicates time The boundary difference, Indicates time The boundary difference, This indicates the simulation time step.

[0072] The weighting coefficients are dynamically allocated based on the gradient rate of change. When the absolute value of the gradient rate of change is large, it indicates that the boundary conditions are changing rapidly. In this case, the convection term and the anodic reaction dominate the mass transport and charge transfer processes, and the weighting coefficients for the convection term and the anodic reaction are correspondingly increased. When the absolute value of the gradient rate of change is small, it indicates that the boundary conditions are changing slowly. In this case, the diffusion term and the cathodic reaction dominate the mass transport and charge transfer processes, and the weighting coefficients for the diffusion term and the cathodic reaction are correspondingly increased. The formula for allocating the weighting coefficients is:

[0073]

[0074]

[0075]

[0076] in, This represents the gradient sensitivity coefficient, which is determined based on the dynamic response characteristics of the system.

[0077] In this embodiment, a damping smoothing mechanism is introduced when dynamically allocating weight coefficients based on the gradient change rate. The damping smoothing mechanism calculates the difference between the real-time weight coefficient of the current iteration cycle and the historical weight coefficient of the previous iteration cycle, applies a nonlinear damping factor to the coefficient difference to generate a smoothing compensation amount, and superimposes the smoothing compensation amount onto the real-time weight coefficient to output the final weight coefficient, thereby suppressing the weight coefficient oscillation phenomenon caused by abrupt changes in the boundary difference gradient.

[0078] The damping smoothing mechanism works as follows: First, it calculates the difference between the real-time weight coefficients of the current iteration and the historical weight coefficients of the previous iteration. Then, it applies a nonlinear damping factor to this difference. The formula for calculating the nonlinear damping factor is:

[0079] in, Represents the nonlinear damping factor. Indicates the damping coefficient. This represents the difference in coefficients. The damping coefficient is determined based on the system's stability requirements.

[0080] Next, calculate the smoothing compensation amount. The formula for calculating the smoothing compensation amount is:

[0081] Finally, the smoothing compensation is added to the real-time weight coefficients to obtain the final weight coefficients. The formula for calculating the final weight coefficients is:

[0082] in, This represents the final weighting coefficient. This represents the historical weighting coefficients from the previous iteration period. This represents the smoothing compensation amount. The weight coefficient allocation under different boundary difference gradient change rates is shown in Table 2.

[0083] Table 2 Weight coefficient allocation under different gradient rates of boundary difference.

[0084] Table 2 shows the weighting coefficients under different boundary difference gradient change rates. As can be seen from Table 2, with the increase of the boundary difference gradient change rate, the weighting coefficients of the convection term and the anodic reaction gradually increase, while the weighting coefficients of the diffusion term and the cathodic reaction gradually decrease. This indicates that when the boundary conditions change rapidly, the convection term and the anodic reaction dominate the mass transport and charge transfer processes.

[0085] At the end of this embodiment, an adaptive weight allocation mechanism is introduced to dynamically allocate the weight coefficients of the convection and diffusion terms in the mass transport flux calculation, and the weight coefficients of the anodic and cathodic reactions in the charge transfer rate calculation, based on the gradient rate of change of the boundary difference. This allows the mass transport flux and charge transfer rate at the multi-field interface to better adapt to the dynamic changes in boundary conditions. A damped smoothing mechanism is introduced to smooth the changes in weight coefficients, suppressing the oscillations in weight coefficients caused by abrupt changes in the boundary difference gradient and ensuring the numerical stability of the dynamic correction process.

[0086] In a preferred embodiment, reference Figure 5 When solving the corrosion kinetic equations, the evolution prediction component introduces a Monte Carlo simulation mechanism. The Monte Carlo simulation mechanism generates a normally distributed random sample set of parameters based on the modified multi-field interactive interface parameters. This random sample set is then input into the corrosion kinetic equations for multiple iterative solutions. The probability distribution of the time when the pipe wall thickness reaches the critical failure threshold is statistically analyzed, and the remaining life confidence interval of the pipeline network is output.

[0087] In this embodiment, the multi-field interaction interface parameters include corrosive medium concentration, pipe wall equivalent stress, pipe wall potential, and medium temperature. These parameters exhibit a certain degree of randomness and uncertainty during actual operation, and their probability distribution can be approximated as a normal distribution. The Monte Carlo simulation mechanism first determines the mean and standard deviation of each multi-field interaction interface parameter. The mean is determined based on the corrected multi-field interaction interface parameters, and the standard deviation is determined based on statistical analysis of historical operating data. Then, a random sample set of parameters conforming to a normal distribution is generated. The size of the random sample set of parameters is determined according to the computational accuracy requirements and computational resource conditions, typically ranging from 1000 to 10000.

[0088] For each random sample of parameters, the corrosion rate is solved by inputting it into the corrosion kinetic equation. Then, based on the corrosion rate, the change in pipe wall thickness over time is calculated until the pipe wall thickness reaches the critical failure threshold, and the remaining lifetime corresponding to that sample is recorded. After all random samples of parameters have been solved, the probability distribution of the remaining lifetime is statistically analyzed. The probability distribution of the remaining lifetime can be represented by a histogram or a probability density function.

[0089] Based on the probability distribution of remaining lifetime, the confidence interval for remaining lifetime is calculated. A confidence interval represents the range of remaining lifetime at a given confidence level. The confidence level is determined according to the system's security requirements, typically set to 95% or 99%. The formula for calculating the confidence interval is:

[0090] in, Indicates the confidence interval. This represents the mean of remaining lifespan. The standard deviation of remaining life. The upper part of the standard normal distribution Quantiles Indicates the significance level. Confidence levels. The confidence intervals for remaining lifetime at different confidence levels are shown in Table 3.

[0091] Table 3. Confidence intervals for remaining life at different confidence levels

[0092] Table 3 shows the remaining lifetime confidence intervals at different confidence levels. As can be seen from Table 3, the range of the confidence interval gradually expands with increasing confidence level. This indicates that a larger range of remaining lifetime uncertainty needs to be accepted in order to obtain a higher confidence level.

[0093] At the end of this embodiment, a Monte Carlo simulation mechanism is introduced to consider the randomness and uncertainty of multi-field interactive interface parameters. A random sample set of parameters is generated and iteratively solved multiple times to statistically analyze the probability distribution of remaining lifetime and output the confidence interval of remaining lifetime. This mechanism quantifies the uncertainty of lifetime prediction under dynamic boundary disturbances, providing a more comprehensive basis for pipeline maintenance decisions.

[0094] In a preferred embodiment, before the twin mapping component runs, a model reduction operation based on intrinsic orthogonal decomposition is performed on the coupled simulation model of the flow field-stress field-electrochemical field. The model reduction operation based on intrinsic orthogonal decomposition extracts global orthogonal basis vectors from the multi-field coupled simulation results, projects the high-dimensional partial differential equation system onto a low-dimensional subspace spanned by the global orthogonal basis vectors, and generates a reduced-order coupled computational model. This reduced-order coupled computational model replaces the original high-dimensional model for boundary difference input and digital twin state updates.

[0095] In this embodiment, the workflow of the model order reduction operation based on intrinsic orthogonal decomposition is as follows: First, a snapshot matrix of multi-field coupled simulation is generated. The snapshot matrix consists of multi-field coupled simulation results at different times and under different boundary conditions. Each snapshot vector contains all nodal variable values ​​of the flow field, stress field, and electrochemical field. The number of columns in the snapshot matrix is ​​the number of snapshots, and the number of rows is the dimension of each snapshot vector, i.e., the degrees of freedom of the high-dimensional model.

[0096] Then, calculate the covariance matrix of the snapshot matrix. The formula for calculating the covariance matrix is:

[0097] in, Represents the covariance matrix. Represents a snapshot matrix. Indicates the number of snapshots.

[0098] Next, the covariance matrix is ​​decomposed into eigenvalues ​​to obtain eigenvalues ​​and eigenvectors. The eigenvalues ​​are arranged in descending order, and the corresponding eigenvectors are the global orthogonal basis vectors. The global orthogonal basis vectors have orthogonality and completeness, and can represent the main characteristics of the multi-field coupled simulation results.

[0099] Then, select before The eigenvector corresponding to the largest eigenvalue is used as the reduced-order basis vector. The value of is determined according to the energy criterion, that is, the preceding... The ratio of the sum of individual eigenvalues ​​to the sum of all eigenvalues ​​reaches a set energy threshold. This energy threshold is typically set to 99% or 99.9%.

[0100] Finally, the high-dimensional partial differential equations are projected onto a low-dimensional subspace spanned by the reduced-order basis vectors, generating a reduced-order coupled computational model. The degrees of freedom of the reduced-order coupled computational model are... The number of reduced basis vectors is much smaller than the original high-dimensional model's degrees of freedom. The solution time of the reduced-order coupled computational model is much shorter than that of the original high-dimensional model, which can significantly reduce computational resource consumption. The number of reduced-order basis vectors and computation time under different energy thresholds are shown in Table 4.

[0101] Table 4. Number of reduced basis vectors and computation time ratio under different energy thresholds.

[0102] Table 4 shows the ratio of the number of reduced basis vectors to the computation time under different energy thresholds. As can be seen from Table 4, the number of reduced basis vectors gradually increases with the increase of the energy threshold, and the computation time ratio also gradually increases. When the energy threshold is 99%, the number of reduced basis vectors is 50, and the computation time ratio is 0.15, meaning that the computation time of the reduced model is only 15% of the original model, while maintaining high computational accuracy.

[0103] At the end of this embodiment, by performing a reduction operation on the high-dimensional model based on intrinsic orthogonal decomposition, a reduced-order coupled computational model is generated and replaces the original model in the update, which reduces the computational resource overhead of the dynamic boundary compensation process and improves the real-time performance of the system.

[0104] In a preferred embodiment, reference Figure 6 After the physical sensing component collects flow velocity, pressure, and pipe wall potential signals, an anomaly data reconstruction mechanism based on spatiotemporal correlation is introduced. This mechanism calculates the signal correlation matrix between adjacent spatial nodes and adjacent time nodes in the pipeline network, identifies the locations of anomaly signals deviating from the correlation matrix, and uses the correlation matrix to perform interpolation, filling, and smoothing filtering on the anomaly signal locations. The processed signals are then transmitted to the twin mapping component.

[0105] In this embodiment, the working principle of the spatiotemporal correlation-based anomaly data reconstruction mechanism is as follows: First, a spatiotemporal matrix of the signal is constructed. The rows of the spatiotemporal matrix represent different spatial nodes, the columns represent different time nodes, and the matrix elements represent the signal values ​​of the corresponding spatial and time nodes.

[0106] Then, the spatial correlation matrix and temporal correlation matrix of the signals are calculated. The spatial correlation matrix represents the signal correlation between different spatial nodes, and its elements are the Pearson correlation coefficients of the signals at two spatial nodes. The temporal correlation matrix represents the signal correlation between different time nodes, and its elements are the Pearson correlation coefficients of the signals at two time nodes. The formula for calculating the Pearson correlation coefficient is:

[0107] in, Representing variables and variables Pearson correlation coefficient, and Representing variables respectively and variables The Each sample value and Representing variables respectively and variables The sample mean, Indicates the number of samples.

[0108] Next, the locations of anomalous signals are identified. For each signal value, its residuals with those of adjacent spatial and temporal nodes are calculated. The formula for calculating the residuals is:

[0109] in, Indicates the first The spatial node, the first The residual of the signal value at each time point Indicates the first The spatial node, the first The original signal values ​​at each time point, This represents the predicted value based on the signal values ​​of adjacent spatial nodes and adjacent temporal nodes. The predicted value is calculated using the spatial correlation matrix and the temporal correlation matrix.

[0110] If the absolute value of the residual is greater than the set anomaly threshold, the signal is considered an anomaly. The anomaly threshold is determined based on the signal's noise level and the system's fault tolerance requirements.

[0111] Then, the location of the anomalous signal is filled in using the correlation matrix. The interpolation fill uses Kriging interpolation or inverse distance weighted interpolation, and calculates the estimated location of the anomalous signal based on the normal signal values ​​around the anomalous signal location and the spatial correlation matrix.

[0112] Finally, the interpolated signal is smoothed using a smoothing filter. Moving average filtering or median filtering is employed to eliminate high-frequency noise and improve signal quality.

[0113] At the end of this embodiment, an abnormal data reconstruction mechanism based on spatiotemporal correlation is introduced to identify and reconstruct abnormal signals, thereby avoiding the impact of abnormal data on the simulation model and improving the robustness and reliability of the system.

Claims

1. A pipeline life prediction system for corrosion conditions, characterized in that, include: Physical sensing components collect flow velocity, pressure, and pipe wall potential signals of the operating medium in the pipeline network; The twin mapping component establishes a coupled simulation model of the flow field, stress field, and electrochemical field of the pipeline network. It extracts the flow velocity signal, the pressure signal, and the pipe wall potential signal as dynamic boundary conditions, calculates the boundary difference between the dynamic boundary conditions at the current moment and the steady-state boundary conditions at the previous moment, and inputs the boundary difference into the coupled simulation model of the flow field, stress field, and electrochemical field to correct the mass transport flux and charge transfer rate at the multi-field interaction interface, and drives the synchronous update of the digital twin state. The evolution prediction component, based on the modified multi-field interaction interface parameters, solves the corrosion kinetic equation, outputs the spatiotemporal distribution of the pipeline wall thickness, and calculates the remaining life of the pipeline based on the wall thickness reduction trend, thus forming a pipeline life prediction system oriented towards corrosion state.

2. The pipeline life prediction system for corrosion conditions according to claim 1, characterized in that, When the physical sensing component extracts the flow velocity signal, the pressure signal, and the pipe wall potential signal as the dynamic boundary condition, it introduces a time delay compensation mechanism. Based on the transmission distance of the flow velocity signal, the pressure signal, and the pipe wall potential signal in the pipe network medium and the sound speed of the medium, the time offset is calculated, and the dynamic boundary conditions at different spatial locations are aligned to a unified time reference according to the time offset.

3. The pipeline life prediction system for corrosion conditions according to claim 1, characterized in that, When the twin mapping component establishes the coupled simulation model of the flow field-stress field-electrochemical field, it adopts an interleaved iterative solution architecture, sets the flow field and stress field as external loop variables, sets the electrochemical field as internal loop variables, sets the convergence criterion for the multi-field interaction interface, and alternately updates the internal loop variables after the external loop variables are updated until the convergence criterion is met, and outputs the multi-field coupling equilibrium state.

4. The pipeline life prediction system for corrosion conditions according to claim 1, characterized in that, When the twin mapping component corrects the mass transport flux and charge transfer rate of the multi-field interaction interface, it introduces an adaptive weight allocation mechanism to calculate the gradient change rate of the boundary difference and dynamically allocate the weight coefficients of the convection and diffusion terms in the mass transport flux calculation, as well as the weight coefficients of the anodic and cathodic reactions in the charge transfer rate calculation, based on the gradient change rate.

5. The pipeline life prediction system for corrosion conditions according to claim 1, characterized in that, When the evolution prediction component solves the corrosion kinetic equation, it introduces a Monte Carlo simulation mechanism. Based on the modified multi-field interaction interface parameters, it generates a random sample set of parameters that conforms to a normal distribution. The random sample set of parameters is input into the corrosion kinetic equation for multiple iterations. The probability distribution of the time when the pipe wall thickness reaches the critical failure threshold is statistically analyzed, and the remaining life confidence interval of the pipeline is output.

6. The pipeline life prediction system for corrosion conditions according to claim 2, characterized in that, When calculating the time offset by introducing the time delay compensation mechanism, a pipeline topology model is constructed, and the shortest path search algorithm is used to determine the signal transmission path from the physical sensing component to the twin mapping component. The equivalent medium sound velocity is calculated by combining the pipe diameter change ratio on the signal transmission path, and the accurate signal propagation delay is calculated as the time offset based on the equivalent medium sound velocity and the length of the signal transmission path.

7. The pipeline life prediction system for corrosion conditions according to claim 3, characterized in that, In the alternating update process of the inner loop variables in the staggered iterative solution architecture, a stress-related polarization correction factor is introduced to extract the equivalent stress parameters of the stress field updated by the outer loop variables. Based on the equivalent stress parameters, the anode Tafel slope and cathode Tafel slope of the electrochemical kinetic equation in the electrochemical field are dynamically corrected, and the corrected electrochemical kinetic equation is applied to the iterative solution process of the inner loop variables.

8. The pipeline life prediction system for corrosion conditions according to claim 4, characterized in that, When dynamically allocating the weight coefficients based on the gradient change rate, a damping smoothing mechanism is introduced. The difference between the real-time weight coefficients of the current iteration cycle and the historical weight coefficients of the previous iteration cycle is calculated. A nonlinear damping factor is applied to the coefficient difference to generate a smoothing compensation amount. The smoothing compensation amount is superimposed on the real-time weight coefficients to output the final weight coefficients, thereby suppressing the weight coefficient oscillation phenomenon caused by the sudden change in the boundary difference gradient.

9. The pipeline life prediction system for corrosion conditions according to claim 1, characterized in that, Before the twin mapping component runs, a model reduction operation based on intrinsic orthogonal decomposition is performed on the flow field-stress field-electrochemical field coupled simulation model. The global orthogonal basis vectors of the multi-field coupled simulation results are extracted, and the high-dimensional partial differential equation system is projected onto the low-dimensional subspace spanned by the global orthogonal basis vectors to generate a reduced-order coupled computation model. The reduced-order coupled computation model replaces the original high-dimensional model for the input of the boundary difference and the update of the state of the digital twin.

10. The pipeline life prediction system for corrosion conditions according to claim 1, characterized in that, After the physical sensing component acquires the flow velocity signal, the pressure signal, and the pipe wall potential signal, it introduces an abnormal data reconstruction mechanism based on spatiotemporal correlation to calculate the signal correlation matrix between adjacent spatial nodes and adjacent time nodes in the pipeline network, identify the abnormal signal positions that deviate from the signal correlation matrix, and use the signal correlation matrix to perform interpolation filling and smoothing filtering on the abnormal signal positions. The processed signals are then transmitted to the twin mapping component.

Images