A method and system for predicting the remaining life of a digitally twinned main steam isolation valve

Abstract

The application discloses a kind of digital twin's main steam isolation valve residual life prediction method and system, belong to nuclear power key equipment state monitoring and life prediction technical field, method includes: real-time acquisition main steam isolation valve part's multidimensional operating state data, pre-processing;The data after pre-processing is input into the response rapid prediction model trained, obtains predicted vibration response and predicted stress time history;The deviation of predicted vibration response and sensor measured vibration response is calculated;If deviation exceeds set threshold, then the model is updated online;If not, then keep the model after training;Based on the predicted stress time history output by online update or the model after training, cyclic load statistics and fatigue damage calculation are carried out, and the residual life of main steam isolation valve is obtained.The application realizes the dynamic, online, high-precision prediction of the health status and fatigue life of main steam isolation valve by constructing digital twin, deeply fusing physical mechanism and real-time data.

Description

Technical Field

[0001] This invention belongs to the field of nuclear power key equipment condition monitoring and life prediction technology, specifically involving a digital twin method and system for predicting the remaining life of a main steam isolation valve. Background Technology

[0002] The main steam isolation valve is a critical safety barrier in the primary loop of a nuclear power plant. Its core function is to quickly cut off the main steam flow in the event of an accident, preventing the leakage of radioactive materials or high-temperature, high-pressure steam. It is considered the "last line of defense" for protecting the nuclear island. During long-term service, the main steam isolation valve continuously withstands high temperatures, high pressures, and severe transient loads. Its critical components are prone to fatigue damage and even cracking due to the multi-field coupling of heat, fluid, and solids, seriously threatening the safe and stable operation of the unit. Therefore, accurate prediction and management of the remaining life of the main steam isolation valve is of great significance for ensuring the safe operation of the power plant, avoiding unplanned shutdowns, and optimizing operation and maintenance costs.

[0003] Currently, the following traditional methods are generally used to predict the lifespan of main steam isolation valves and similar high-temperature and high-pressure valves, all of which have significant limitations: First, there is the conservative calculation method based on design standards. This method calculates a fixed design life based on standards such as ASME BPVC and assumed constant load spectra and material data during the design phase. This method cannot reflect the actual damage and lifespan consumption of the valve under real and variable operating conditions, often leading to "over-maintenance" or "under-maintenance." Second, there is the periodic disassembly inspection and evaluation method. This method requires shutdown and disassembly, consuming a lot of manpower, resources, and time. Although it can provide intuitive information, it is a retrospective, static, and destructive assessment, unable to achieve predictive maintenance, and the disassembly itself may introduce foreign objects and new damage. Third, there is the simple condition monitoring and threshold alarm method. Some power plants monitor the operating data of key valves, such as temperature parameters, and issue early warnings through threshold alarms. This method lacks in-depth insight into the internal physical state of the equipment, cannot quantify the degree of damage and remaining lifespan, and is difficult to predict long-term trends. In addition, although offline finite element method can be used for analysis, it cannot meet the needs of online, real-time life prediction and condition assessment due to its high computational cost and long time consumption.

[0004] In summary, current life prediction methods for main steam isolation valves and similar high-temperature and high-pressure valves are insufficient to reflect the actual damage and life consumption of valves under real and varied operating conditions. They are also difficult to quantify the degree of damage and remaining life, and the degree of predictive maintenance is inadequate. Online real-time life prediction and condition assessment need further optimization. Summary of the Invention

[0005] This invention provides a digital twin method and system for predicting the remaining life of a main steam isolation valve. The purpose is to solve the problems that current life prediction methods for main steam isolation valves and similar high-temperature and high-pressure valves are unable to reflect the actual damage and life consumption of the valve under real and varied operating conditions, are difficult to quantify the degree of damage and remaining life, have insufficient predictive maintenance capabilities, and require further optimization of online real-time life prediction and condition assessment.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: This invention provides a digital twin method for predicting the remaining life of a main steam isolation valve, comprising the following steps: S1. Real-time acquisition of multi-dimensional operating status data of key parts of the main steam isolation valve, and preprocessing of the multi-dimensional operating status data; S2. Input the preprocessed multidimensional operating state data into the trained fast response prediction model to obtain the predicted vibration response and predicted stress time history; calculate the deviation between the predicted vibration response and the sensor-measured vibration response. If the deviation exceeds the set threshold, the fast response prediction model will be updated online; if it does not exceed the threshold, the trained fast response prediction model will be retained. The fast response prediction model after training is obtained in the following way: Using the preprocessed data as boundary conditions and inputs, a high-fidelity digital twin of the main steam isolation valve is constructed. The high-fidelity digital twin is driven to perform multi-physics coupled simulation, generating a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve, and constructing a fast response prediction model. The fast response prediction model is trained, validated, and evaluated to obtain the trained fast response prediction model. S3. Based on the predicted stress time history output by the online updated or trained response fast prediction model, perform cyclic load statistics and fatigue damage calculation to obtain the remaining life of the main steam isolation valve, and complete the prediction of the remaining life of the main steam isolation valve using a digital twin.

[0007] In some implementations, in S1, the multidimensional operating status data includes inlet valve pressure, outlet valve pressure, temperature, valve opening degree, and vibration acceleration.

[0008] In some implementations, in S1, the preprocessing includes noise reduction and normalization; the noise reduction includes using wavelet thresholding, and the normalization includes using a normalization method.

[0009] In some implementations, in S2, constructing a high-fidelity digital twin of the main steam isolation valve specifically includes: establishing a three-dimensional solid model of the main steam isolation valve, defining the material properties of the three-dimensional solid model and performing mesh generation, refining the mesh in the stress concentration areas, and obtaining a digital twin simulation model; the multiphysics coupling simulation is a thermal-fluid-solid multi-field coupling simulation, and a multi-process optimization platform is used to automatically call finite element analysis software to perform fluid-structure interaction numerical simulation.

[0010] In some implementations, in S2, the high-fidelity simulation sample database includes input variables and corresponding output responses; the input variables are the pressure before the valve, the pressure after the valve, the temperature, and the valve opening degree, and the output responses are the vibration acceleration and stress time history of preset measurement points.

[0011] In some implementations, in S2, the fast response prediction model includes a radial basis function neural network model, which uses a Gaussian kernel function to realize the nonlinear mapping from the input of the operating parameters to the response output; The training, validation, and evaluation process includes: dividing the sample data in the high-fidelity simulation sample database into training, validation, and test sets; training the model using the training set; monitoring the training process using the validation set; and evaluating the model performance using the test set.

[0012] In some implementations, in S2, the deviation includes RMS relative error and spectral similarity; The online update specifically includes: removing abnormal data from the high-fidelity simulation sample database, collecting real-time operating condition data and corresponding measured vibration data for the most recent preset time period, re-inputting the digital twin simulation model to generate new high-fidelity simulation samples, merging them with the sample database after removing abnormal data to form new training samples, retraining the fast response prediction model and replacing the original model.

[0013] In some implementations, in S3, fatigue damage calculation is based on the Miner-Palmgren linear cumulative damage theory, and the calculation formula is as follows: ; in, This represents the total fatigue damage. This represents the actual number of cycles at the current stress level. This represents the number of cycles in the SN curve of the material at the corresponding stress amplitude. The formula for calculating remaining lifetime is as follows: ; in, The current damage rate, For historical running time, The remaining lifespan.

[0014] In some implementations, in S3, the multi-process optimization platform includes the Simulia Isight platform, and the finite element analysis software includes ANSYS Fluent and ANSYS Mechanical; the three-dimensional solid model is created using SolidWorks software.

[0015] This invention also provides a digital twin-based system for predicting the remaining life of a main steam isolation valve, used to implement the aforementioned digital twin-based method for predicting the remaining life of a main steam isolation valve. The system includes a data acquisition and preprocessing module, a response deviation calculation module, and a remaining life output module, wherein: Data acquisition and preprocessing module: used to acquire multi-dimensional operating status data of the main steam isolation valve in real time and preprocess the multi-dimensional operating status data; Response deviation calculation module: used to input preprocessed multidimensional operating status data into the trained fast response prediction model to obtain the predicted vibration response and predicted stress time history; calculate the deviation between the predicted vibration response and the sensor measured vibration response; If the deviation exceeds the set threshold, the fast response prediction model will be updated online; if it does not exceed the threshold, the trained fast response prediction model will be retained. The fast response prediction model after training is obtained in the following way: Using the preprocessed data as boundary conditions and inputs, a high-fidelity digital twin of the main steam isolation valve is constructed. The high-fidelity digital twin is driven to perform multi-physics coupled simulation, generating a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve, and constructing a fast response prediction model. The fast response prediction model is trained, validated, and evaluated to obtain the trained fast response prediction model. Remaining life output module: Based on the predicted stress time history output by the online updated or trained response fast prediction model, perform cyclic load statistics and fatigue damage calculation to obtain the remaining life of the main steam isolation valve, and complete the prediction of the remaining life of the main steam isolation valve by digital twin.

[0016] Compared with the prior art, the present invention, a digital twin method and system for predicting the remaining life of a main steam isolation valve, has the following advantages: This invention discloses a digital twin method for predicting the remaining life of a main steam isolation valve. It applies digital twin technology to the prediction of the valve's lifespan, integrating physical mechanisms with real-time data and utilizing a high-fidelity digital twin to simulate real multi-physics coupling effects. This enables online, dynamic monitoring and accurate prediction of the valve's health status and remaining lifespan. This provides crucial technical support for predictive maintenance, avoiding unplanned outages, and optimizing operation and maintenance costs, effectively ensuring the safe and economical operation of nuclear power plants. The invention proposes a hybrid modeling approach combining mechanistic models and data-driven methods, balancing accuracy and efficiency. This approach integrates high-fidelity physical simulation with machine learning models. First, a high-fidelity simulation sample library covering all operating conditions is constructed using finite element analysis. Then, this sample library is used to train a surrogate model capable of rapidly predicting stress and vibration responses. This approach retains the accuracy of physical mechanisms while achieving the computational efficiency of data-driven models, thus achieving a leap from high-precision but time-consuming methods to a fast and accurate one. This invention constructs a digital twin model by integrating real-time data acquisition with an efficient proxy model. While maintaining accuracy, it reduces prediction time from hours or days to milliseconds, solving the problems of high cost, long processing time, and poor real-time performance of traditional offline simulation calculations. This enables online, high-fidelity, real-time prediction of valve stress state and fatigue life. Attached Figure Description

[0017] The accompanying drawings are provided to further understand the invention and constitute a part of this invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0018] Figure 1 This is a flowchart illustrating a digital twin method for predicting the remaining life of a main steam isolation valve according to the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] Therefore, the following detailed description of the embodiments of the invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the invention without inventive effort are within the scope of protection of the invention.

[0021] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0022] It should be noted that the apparatus and methods disclosed in the embodiments herein can also be implemented in other ways. The apparatus embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments herein. In this regard, each block in a flowchart or block diagram may represent a module, program, or part of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system to perform the specified function or action, or can be implemented using a combination of dedicated hardware and computer instructions.

[0023] In addition, the functional modules in the various embodiments of this article can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0024] How can we provide a method for predicting the remaining life of a main steam isolation valve that can accurately map the actual damage state of the valve and achieve accurate and timely prediction of the health status and fatigue life of the main steam isolation valve?

[0025] like Figure 1 As shown, the present invention provides a digital twin method for predicting the remaining life of a main steam isolation valve, comprising the following steps: S1. Real-time acquisition of multi-dimensional operating status data of the main steam isolation valve, and preprocessing of the multi-dimensional operating status data; S2. Input the preprocessed multidimensional operating state data into the trained fast response prediction model to obtain the predicted vibration response and predicted stress time history; calculate the deviation between the predicted vibration response and the sensor-measured vibration response. If the deviation exceeds the set threshold, the fast response prediction model will be updated online; if it does not exceed the threshold, the trained fast response prediction model will be retained. The fast response prediction model after training is obtained in the following way: Using the preprocessed data as boundary conditions and inputs, a high-fidelity digital twin of the main steam isolation valve is constructed. The high-fidelity digital twin is driven to perform multi-physics coupled simulation, generating a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve, and constructing a fast response prediction model. The fast response prediction model is trained, validated, and evaluated to obtain the trained fast response prediction model. S3. Based on the predicted stress time history output by the online updated or trained response fast prediction model, perform cyclic load statistics and fatigue damage calculation to obtain the remaining life of the main steam isolation valve, and complete the prediction of the remaining life of the main steam isolation valve using a digital twin.

[0026] This invention discloses a digital twin method for predicting the remaining life of a main steam isolation valve. By constructing a digital twin that is synchronized and interacts with the physical valve in real time, and deeply integrating the physical mechanism with real-time operating data, it achieves dynamic, online, and high-precision prediction of the health status and fatigue life of the main steam isolation valve, providing a basis for predictive maintenance and intelligent operation and maintenance of power plants.

[0027] Specifically, the present invention includes the following steps: I. Real-time acquisition and processing of multi-source sensor data.

[0028] Multi-source sensors, including acceleration sensors and pressure sensors, are installed at key parts of the main steam isolation valve, and the existing valve monitoring system (such as temperature and valve stem displacement sensors) is integrated to collect multi-dimensional operating status data such as vibration acceleration, pressure before and after the valve, and temperature in real time, and the signals are preprocessed.

[0029] 1) Sensor Placement and Data Acquisition: Multi-source sensors, including acceleration sensors and pressure sensors, are placed in key parts of the main steam isolation valve (such as the valve body, valve stem, and actuator), and integrated with the existing valve monitoring system (such as temperature sensors and valve stem displacement sensors) to collect the upstream pressure in real time. Pressure after valve ,temperature Valve opening Vibration acceleration ( , , And signal synchronization and uploading are completed through the data transmission module.

[0030] 2) Signal Preprocessing: The acquired raw signals contain a large amount of background noise and interference, requiring preprocessing to accurately extract the true signal features. Noise reduction and standardization are performed on the acquired raw signals. Wavelet thresholding is used to filter out background noise; then, normalization is used to eliminate differences in the dimensions and numerical ranges of different physical quantities, providing a high-quality data foundation for subsequent analysis.

[0031] II. Construction of a high-fidelity digital twin of the main steam isolation valve.

[0032] By using real-time acquired operational status data as boundary conditions and inputs to a digital twin model, the model is driven to perform multiphysics coupled simulation calculations to obtain the vibration and stress responses of key valve components (such as valve body, valve seat, and valve stem) under various operating conditions, providing core data inputs for valve condition assessment and life prediction.

[0033] 1) Establish a digital twin simulation model: A three-dimensional solid model of the main steam isolation valve is established, material properties are defined, and meshing is performed. Mesh refinement is applied to stress concentration areas to improve computational accuracy. This model serves as a digital twin model for performing thermo-fluid-solid multi-field coupled simulations of the main steam isolation valve, calculating the valve's flow field, temperature field, stress field, etc.

[0034] 2) Automatic simulation of the digital twin: The Simulia Isight multi-process optimization platform is used to automatically call Ansys finite element analysis software for fluid-structure interaction numerical simulation. Real-time collected valve inlet pressure is used as the basis for this simulation. Pressure after valve ,temperature and valve opening ( Using as input variables, the valve digital twin is driven to perform fluid-structure interaction numerical simulation, generating batch sample data covering various working conditions. The simulation output is the vibration acceleration at key measuring points. , , and stress time history Constructing a simulation sample database: Integrating all input variables and their corresponding output responses, a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve is constructed.

[0035] 3) Characteristic Space Construction: This involves constructing the characteristic space from the input variables in the finite element analysis (inlet pressure of the valve). Pressure after valve ,temperature and valve opening ) as input features for the machine learning model, and the calculated vibration acceleration ( , , and stress time history As the target output features of machine learning models, establish data structures oriented towards machine learning.

[0036] 4) Construction and Training of a Rapid Prediction Model for Response: A radial basis function neural network model is constructed, using the aforementioned input variables as inputs and the target output variable as the output. A Gaussian kernel function is used to achieve a nonlinear mapping from the input operating parameters to the response output. The sample data is divided into training, validation, and test sets to complete the training, validation, and evaluation of the model, ultimately obtaining a rapid prediction model for the fluid-structure interaction response of the main steam isolation valve.

[0037] 5) Drive the prediction model to respond quickly: Use the real-time collected working condition data as input to drive the trained prediction model to respond quickly and output the vibration acceleration and stress time history of key measuring points in milliseconds.

[0038] 6) Deviation calculation: The actual vibration signal measured by the sensor ( , , ) and predicted vibration signals ( , , The two systems are compared to calculate their RMS relative error and spectral similarity. A deviation rate threshold is set. If the deviation rate does not exceed the threshold, online fatigue damage calculation and remaining life prediction are performed. If the deviation rate exceeds the threshold, the model is triggered to update online, enabling the digital twin to learn and track the current state of the physical valve and achieve dynamic evolution.

[0039] 7) Online Model Update: When a model update is triggered, abnormal data is removed from the sample library. Simultaneously, the system automatically collects real-time data from the most recent period (e.g., the past hour), including operating condition data and corresponding measured vibration data. A high-fidelity simulation sample is generated by re-inputting the digital twin simulation model and merged with the sample library after removing abnormal data to form a new training sample. The prediction model is then retrained. After training is complete, the original prediction model is replaced.

[0040] 8) Model Validation: The real-time collected working condition data is used as input to drive the trained surrogate model to make rapid predictions, calculate deviations and update the model online until the model meets the deviation requirements. The vibration acceleration and stress time history of key measuring points are output in milliseconds.

[0041] III. Online fatigue damage calculation and remaining life prediction.

[0042] Obtain the predicted stress time history of key components of the main steam isolation valve from the digital twin model. The rainflow counting method was used to statistically analyze cyclic loads and obtain the stress amplitude of each stress cycle at the valve measuring point. Mean stress and number of stress cycles .

[0043] Based on the SN curves of the main steam isolation valve component materials, determine the corresponding stress amplitude. The number of cycles required for a material to undergo fatigue failure. .

[0044] Based on the Miner-Palmgren linear cumulative damage theory, the total fatigue damage of the main steam isolation valve during its service life is calculated. The calculation formula is as follows: ; in, This represents the total fatigue damage. This represents the actual number of cycles at the current stress level. This represents the number of cycles of the SN curve of the material at the corresponding stress amplitude.

[0045] Calculate the fatigue life of the main steam isolation valve: ; in, The current damage rate, For historical running time, The remaining lifespan.

[0046] Visualization and Decision Support: The simulation results, real-time cumulative damage, remaining life, and other key information of the digital twin model are dynamically displayed through a visual interface, and early warning information and maintenance suggestions are provided to maintenance personnel.

[0047] This invention also provides a digital twin-based system for predicting the remaining life of a main steam isolation valve, used to implement a digital twin-based method for predicting the remaining life of a main steam isolation valve. The system includes a data acquisition and preprocessing module, a response deviation calculation module, and a remaining life output module, wherein: Data acquisition and preprocessing module: used to acquire multi-dimensional operating status data of the main steam isolation valve in real time and preprocess the multi-dimensional operating status data; Response deviation calculation module: used to input preprocessed multidimensional operating status data into the trained fast response prediction model to obtain the predicted vibration response and predicted stress time history; calculate the deviation between the predicted vibration response and the sensor measured vibration response; If the deviation exceeds the set threshold, the fast response prediction model will be updated online; if it does not exceed the threshold, the trained fast response prediction model will be retained. The fast response prediction model after training is obtained in the following way: Using the preprocessed data as boundary conditions and inputs, a high-fidelity digital twin of the main steam isolation valve is constructed. The high-fidelity digital twin is driven to perform multi-physics coupled simulation, generating a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve, and constructing a fast response prediction model. The fast response prediction model is trained, validated, and evaluated to obtain the trained fast response prediction model. Remaining life output module: Based on the predicted stress time history output by the online updated or trained response fast prediction model, perform cyclic load statistics and fatigue damage calculation to obtain the remaining life of the main steam isolation valve, and complete the prediction of the remaining life of the main steam isolation valve by digital twin.

[0048] The following detailed description of the digital twin method and system for predicting the remaining life of a main steam isolation valve according to the present invention will be provided through specific embodiments.

[0049] This embodiment takes the main steam isolation valve as an example. The material is ASTM A182 F316H, which has been in continuous operation for 20 years and its remaining life is predicted.

[0050] I. Real-time acquisition and processing of multi-source sensor data.

[0051] 1) Sensor Arrangement and Data Acquisition: A triaxial acceleration sensor is installed on the valve body (measuring point 1) and valve stem coupling (measuring point 2) of the main steam isolation valve. Pressure sensors are installed before and after the valve. The existing valve monitoring system (such as temperature sensor and valve stem displacement sensor LVDT) is integrated to collect the pressure before the valve in real time. Pressure after valve ,temperature Valve opening Vibration acceleration ( , , , , , And signal synchronization and uploading are completed through the data transmission module.

[0052] 2) Signal Preprocessing: The acquired raw signals contain a large amount of background noise and interference, requiring preprocessing to accurately extract the true signal features. Noise reduction and standardization are performed on the acquired raw signals. Wavelet thresholding is used to filter out background noise; then, normalization is used to eliminate differences in the dimensions and numerical ranges of different physical quantities, providing a high-quality data foundation for subsequent analysis.

[0053] II. Construction of a high-fidelity digital twin of the main steam isolation valve.

[0054] 1) Establish a digital twin simulation model: Based on the valve drawings, create a 3D geometric model of the valve in SolidWorks. Import the completed 3D model into ANSYS software, setting the valve body material to ASTM A182 F316H special steel, and setting the material's elastic modulus, Poisson's ratio, density, coefficient of thermal expansion, etc. Perform mesh generation, especially refining the mesh in stress concentration areas such as the sealing contact surface between the valve seat and valve disc, and the connection between the valve stem and valve disc, to ensure that the mesh quality meets the calculation accuracy requirements.

[0055] 2) Automatic simulation of the digital twin: The Simulia Isight multi-process optimization platform is used to automatically call Ansys finite element analysis software for fluid-structure interaction numerical simulation. Real-time collected valve inlet pressure is used as the basis for this simulation. p 1. Pressure after valve p 2. Temperature T and valve opening ( l Using L as the input variable, the valve digital twin is driven to perform fluid-structure interaction numerical simulation, generating 1000 sets of sample data covering various working conditions. The simulation output is the vibration acceleration at key measuring points. , , , , , ) and stress time history ( , ).

[0056] 3) Automated Simulation Workflow Setup: An automated simulation workflow was built using the Simulia Isight multi-process optimization platform, integrating ANSYS Fluent and ANSYS Mechanical for fluid-structure interaction numerical simulations. The automated workflow was established within Isight to achieve parameter transfer, automatic software invocation, and data extraction.

[0057] 4) Set input variables and boundary conditions: Use the valve inlet pressure collected from the real-time database. Pressure after valve ,temperature and valve opening The input variables are: In Fluent, set the corresponding pressure inlet, pressure outlet, and temperature boundary conditions, perform transient flow field simulation, and calculate the unsteady fluid forces acting on the valve body and valve disc.

[0058] 5) Batch Simulation and Data Generation: Fluid dynamic data is transferred as loads to the structural model, and transient dynamic analysis is performed in Mechanical. Using Isight's design-of-experiments (BOE) function, 1000 different working condition combinations are generated according to an orthogonal array within the actual variation range of the input variables, driving an automated simulation process to complete the fluid-structure interaction calculations for all working conditions in batches. Simulation outputs include vibration acceleration time histories and stress time histories at key measuring points. The vibration measuring point locations are consistent with the actual sensor layout, including the valve body and valve stem coupling; stress outputs focus on fatigue-prone areas such as the valve seat sealing surface and valve stem connection.

[0059] 6) Construct a simulation sample database: This involves storing the input variables of 1000 sets of simulation conditions ( , , , ) and their corresponding output response ( , , , We will integrate these data to build a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve.

[0060] 7) Feature Space Construction: Extract features from the simulation database. Input features are 4-dimensional vectors: , , , ; The target output features are the time history curves of vibration acceleration and stress at key measuring points, denoted as: ; 8) Neural Network Model Construction and Training: A radial basis function neural network model is constructed, using the aforementioned input variables as inputs and the target output variable as the output. A Gaussian kernel function is used to achieve a nonlinear mapping from the input operating parameters to the response output. The sample data is divided into training, validation, and test sets to complete the training, validation, and evaluation of the model, ultimately obtaining a fast prediction model for the fluid-structure interaction response of the main steam isolation valve.

[0061] The 1000 sets of sample data were randomly divided into a training set (800 sets), a validation set (100 sets), and a test set (100 sets) in a ratio of 8:1:1. The model was trained using the training set, the training process was monitored using the validation set to prevent overfitting, and the model performance was finally evaluated using the test set.

[0062] Construct a radial basis function neural network model, with the input layer consisting of 4-dimensional working condition parameters: , , , ; The output layer provides an 8-dimensional output response. ; The RBF neural network model can be constructed and trained using the newrb function in MATLAB.

[0063] net=newrb (X, Y, goal, spread, MN, DF); The variables in the function will be described below: X represents the input data, a 4×1000 matrix with 4-dimensional operating parameters: p1, p2, T, l / L, arranged in columns with 1000 samples. Y represents the output data, an 8×1000 matrix with 8-dimensional output response, also arranged in columns with 1000 samples. `goal` is the target error; the model automatically terminates training when the error is less than or equal to the `goal` value. `spread` is the spread constant, controlling the width of the RBF radial basis function. `MN` is the minimum number of training iterations, set to 1. `DF` is the maximum number of training iterations; typically, `DF` is set to 1000 or greater to ensure the neural network is sufficiently trained.

[0064] Tests showed that the model's average relative error in predicting vibration acceleration RMS values ​​on the test set was less than 3%, meeting engineering accuracy requirements.

[0065] The system drives the neural network model to respond quickly: it collects real-world operating data once per second. , , , The data is then input into a neural network model, which drives the trained neural network model to make rapid predictions. The model outputs the predicted vibration response Y within milliseconds.

[0066] Deviation calculation: The system synchronously reads the measured signals from the acceleration sensors installed at the corresponding measuring points on the physical valves. Vibration response predicted by neural network Calculate the deviation between the predicted and measured values, including RMS relative error and spectral similarity. Set a threshold: if the RMS relative error is <5% and the spectral similarity is >0.95, the model is considered to be in good condition, and proceed to step 8) to calculate fatigue damage. Otherwise, trigger a model update alarm.

[0067] 8) Data collection and retraining: Remove sample data that does not meet the deviation requirements. The system automatically collects real-time operating condition data and corresponding measured vibration data from the past hour, re-inputs them into the digital twin model to generate high-fidelity sample data, merges them with the removed sample database, and retrains the RBF neural network model.

[0068] 9) Model Update and Validation: After training, the old model is replaced with the new model. New real-time data is then collected and input into the new model to calculate the bias. Validation shows that the updated model's predictive RMS relative error has decreased to 2.5%, and the spectral similarity has increased to 0.97, meeting the requirements. The updated digital twin accurately learns and tracks the state of the physical valve under new operating conditions, achieving dynamic evolution and outputting the vibration acceleration and stress time history of key measuring points within milliseconds.

[0069] III. Online fatigue damage calculation and remaining life prediction are performed using the aforementioned methods, including historical operating time. Use a 20-year timeframe. Then, apply the aforementioned visualization and decision support.

[0070] In summary, this invention provides a digital twin method and system for predicting the remaining life of a main steam isolation valve. It involves deploying multi-source sensors at key locations on the main steam isolation valve to collect operational data in real time; constructing a digital twin that includes a high-fidelity simulation model and a rapid machine learning prediction model to achieve second-level response calculations of the main steam isolation valve's vibration response and stress state; calculating the fatigue damage of the main steam isolation valve based on the stress time history using the rainflow counting method and Miner's linear cumulative damage theory; and ultimately achieving dynamic prediction of the remaining life. This enables dynamic, online, and high-precision prediction of the health status and fatigue life of the main steam isolation valve, providing a scientific basis for predictive maintenance and intelligent operation and maintenance in power plants.

[0071] Finally, it should be noted that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Anyone skilled in the art can readily implement the present invention according to the description and above. Any modifications, alterations, or equivalent variations made using the technical content disclosed above are equivalent embodiments of the present invention. Furthermore, any modifications, alterations, or variations made to the above embodiments based on the essential technology of the present invention are still within the protection scope of the present invention.

Claims

1. A method for predicting the remaining life of a main steam isolation valve using digital twins, characterized in that, Includes the following steps: S1. Real-time acquisition of multi-dimensional operating status data of the main steam isolation valve, and preprocessing of the multi-dimensional operating status data; S2. Input the preprocessed multidimensional operating state data into the trained fast response prediction model to obtain the predicted vibration response and the predicted stress time history. Calculate the deviation between the predicted vibration response and the actual vibration response measured by the sensor; If the deviation exceeds the set threshold, the fast response prediction model will be updated online; if it does not exceed the threshold, the trained fast response prediction model will be retained. The fast response prediction model after training is obtained in the following way: Using the preprocessed data as boundary conditions and inputs, a high-fidelity digital twin of the main steam isolation valve is constructed. The high-fidelity digital twin is driven to perform multi-physics coupled simulation, generating a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve, and constructing a fast response prediction model. The fast response prediction model is trained, validated, and evaluated to obtain the trained fast response prediction model. S3. Based on the predicted stress time history output by the online updated or trained response fast prediction model, perform cyclic load statistics and fatigue damage calculation to obtain the remaining life of the main steam isolation valve, and complete the prediction of the remaining life of the main steam isolation valve using a digital twin.

2. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S1, the multidimensional operating status data includes inlet valve pressure, outlet valve pressure, temperature, valve opening degree, and vibration acceleration.

3. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S1, the preprocessing includes noise reduction and normalization; the noise reduction includes wavelet thresholding and the normalization includes normalization.

4. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S2, constructing a high-fidelity digital twin of the main steam isolation valve specifically includes: establishing a three-dimensional solid model of the main steam isolation valve, defining the material properties of the three-dimensional solid model and performing mesh generation, refining the mesh in the stress concentration area, and obtaining a digital twin simulation model; the multiphysics coupling simulation is a thermal-fluid-solid multi-field coupling simulation, and a multi-process optimization platform is used to automatically call finite element analysis software to perform fluid-structure interaction numerical simulation.

5. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S2, the high-fidelity simulation sample database includes input variables and corresponding output responses; the input variables are the pressure before the valve, the pressure after the valve, the temperature, and the valve opening degree, and the output responses are the vibration acceleration and stress time history of preset measurement points.

6. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S2, the fast response prediction model includes a radial basis function neural network model, which uses a Gaussian kernel function to realize the nonlinear mapping from the input of the operating parameters to the response output; The training, validation, and evaluation process includes: dividing the sample data in the high-fidelity simulation sample database into training, validation, and test sets; training the model using the training set; monitoring the training process using the validation set; and evaluating the model performance using the test set.

7. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S2, the deviation includes RMS relative error and spectral similarity; The online update specifically includes: removing abnormal data from the high-fidelity simulation sample database, collecting real-time operating condition data and corresponding measured vibration data for the most recent preset time period, re-inputting the digital twin simulation model to generate new high-fidelity simulation samples, merging them with the sample database after removing abnormal data to form new training samples, retraining the fast response prediction model and replacing the original model.

8. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S3, fatigue damage calculation is based on the Miner-Palmgren linear cumulative damage theory, and the calculation formula is as follows: ； in, This represents the total fatigue damage. This represents the actual number of cycles at the current stress level. This represents the number of cycles in the SN curve of the material at the corresponding stress amplitude. The formula for calculating remaining lifetime is as follows: ； in, The current damage rate, For historical running time, The remaining lifespan.

9. The method for predicting the remaining life of a main steam isolation valve using digital twins according to claim 1, characterized in that, In S3, the multi-process optimization platform includes the Simulia Isight platform, and the finite element analysis software includes ANSYS Fluent and ANSYS Mechanical. The 3D solid model was created using SolidWorks software.

10. A digital twin-based system for predicting the remaining life of a main steam isolation valve, used to implement the digital twin-based method for predicting the remaining life of a main steam isolation valve according to any one of claims 1-9, characterized in that, It includes a data acquisition and preprocessing module, a response deviation calculation module, and a remaining lifetime output module, wherein: Data acquisition and preprocessing module: used to acquire multi-dimensional operating status data of the main steam isolation valve in real time and preprocess the multi-dimensional operating status data; Response deviation calculation module: used to input preprocessed multidimensional operating status data into the trained fast response prediction model to obtain the predicted vibration response and predicted stress time history; calculate the deviation between the predicted vibration response and the sensor measured vibration response; If the deviation exceeds the set threshold, the fast response prediction model will be updated online; if it does not exceed the threshold, the trained fast response prediction model will be retained. The fast response prediction model after training is obtained in the following way: Using the preprocessed data as boundary conditions and inputs, a high-fidelity digital twin of the main steam isolation valve is constructed. The high-fidelity digital twin is driven to perform multi-physics coupled simulation, generating a high-fidelity simulation sample database covering all operating conditions of the main steam isolation valve, and constructing a fast response prediction model. The fast response prediction model is trained, validated, and evaluated to obtain the trained fast response prediction model. Remaining life output module: Based on the predicted stress time history output by the online updated or trained response fast prediction model, perform cyclic load statistics and fatigue damage calculation to obtain the remaining life of the main steam isolation valve, and complete the prediction of the remaining life of the main steam isolation valve by digital twin.

Images