Hydroelectric generating unit axis adjustment method and system based on digital twinning

Abstract

The present application relates to the field of installation and operation and maintenance of hydroelectric generating set, and discloses a method and system for adjusting the axis of hydroelectric generating set based on digital twinning. The method comprises: during the installation stage of the generating set, sensors are arranged at key parts of the generating set and connected to construction period environmental monitoring data, and a full-factor perception network covering the generating set body and the construction environment is constructed; a geometric model is established based on the design data of the generating set and the sensor arrangement information, a coupled dynamics model of the foundation-rack-rotor system is constructed by fusing material properties and mechanical characteristics, and a digital twin of the axis of the generating set is formed; during the construction and operation stage of the generating set, real-time data is input into the digital twin, the axis offset trend is dynamically calculated, and deviation prediction results are generated; and a dynamic adjustment strategy for the axis is generated based on the deviation prediction results. By constructing the digital twin, the present application realizes real-time mapping, deviation prediction and dynamic optimization of the axis state, and significantly improves the accuracy and efficiency of the adjustment of the axis of the generating set.

Description

Technical Field

[0001] This invention relates to the field of hydro-generator installation and operation and maintenance technology, and more specifically, to a method and system for adjusting the shaft of a hydro-generator based on digital twins. Background Technology

[0002] Hydropower turbine generator units are the core equipment of hydropower stations, and their axis condition directly affects the unit's operational stability, vibration level, and service life. Axis adjustment is a critical process in unit installation and operation maintenance. Traditional axis adjustment methods mainly rely on manual measurement and experience-based judgment. During unit installation, various monitoring data are involved, including foundation settlement, concrete temperature, and unit strain. However, these data are scattered across different systems, lacking unified data fusion and correlation analysis, making it difficult to comprehensively reflect the evolution of the axis condition. Traditional axis adjustment typically involves reactive correction after deviations occur, failing to predict axis offset trends and proactively compensate during construction. This results in some deviations being solidified in subsequent processes, increasing the difficulty of adjustment. Existing methods primarily focus on axis adjustment during the installation phase, lacking effective dynamic optimization methods for axis offsets caused by factors such as foundation settlement, temperature changes, and long-term operational wear during operation. Traditional axis calculations rely on ideal models from the design phase, failing to reflect real-time changes in the actual state during construction and operation, leading to discrepancies between simulation results and reality.

[0003] With the development of digital twin technology, how to construct a digital twin of the unit's axis and realize real-time mapping, deviation prediction and dynamic adjustment of the axis status has become a technical problem that urgently needs to be solved in this field. Summary of the Invention

[0004] The purpose of this invention is to provide a method and system for adjusting the shaft of a hydro-generator unit based on digital twins, so as to solve the problems of lag in shaft adjustment, data silos, and lack of dynamic optimization throughout the entire life cycle in the prior art.

[0005] In some embodiments of this application, a method for adjusting the shaft of a hydro-generator unit based on digital twins is provided, including: During the unit installation phase, sensors are deployed at key parts of the unit, and environmental monitoring data during the construction period is also connected to build a full-element sensing network covering the unit itself and the construction environment. Based on the unit design data and sensor deployment information, a geometric model of the unit is established; by integrating material properties and mechanical characteristics, a coupled dynamic model reflecting the foundation-frame-shaft system is constructed, forming a digital twin of the unit's axis; During the construction and operation phases of the unit, the real-time data collected by the full-element sensing network is input into the digital twin to dynamically calculate the axis offset trend and generate deviation prediction results for the axis status. Based on the deviation prediction results, a dynamic axis adjustment strategy is generated.

[0006] In some embodiments of this application, during the unit installation phase, sensors are deployed at key parts of the unit, and environmental monitoring data from the construction period is simultaneously integrated to construct a comprehensive sensing network covering both the unit itself and the construction environment, including: Fiber optic strain sensors, vibration sensors, and displacement sensors are embedded in the generator rotor, turbine runner, and upper and lower frames to collect strain, vibration, and displacement data in real time during the unit installation process. Simultaneously, it integrates foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data during the construction period, forming the aforementioned all-element sensing network covering the unit's physical status and construction environment parameters.

[0007] In some embodiments of this application, when fiber optic strain sensors, vibration sensors, and displacement sensors are embedded in the generator rotor, turbine runner, and upper and lower frames, the following are included: The fiber optic strain sensors are spirally arranged along the axis of the unit at key sections of the rotor, impeller and upper and lower frames to form a three-dimensional strain field monitoring array. The vibration sensor and displacement sensor are arranged in pairs at the upper and lower guide bearings of the generator rotor, the guide bearing of the turbine and the thrust bearing, forming a synchronous monitoring unit for radial and axial displacement. Low-power smart sensor nodes are deployed in a wireless network at the foundation settlement monitoring points and temperature measurement points inside the concrete pouring chamber. Each node synchronizes with the central control platform through a self-organizing network protocol.

[0008] In some embodiments of this application, the formation of the full-element sensing network covering the unit's physical state and construction environment parameters includes: Establish a unified timestamp benchmark and synchronize various sensors through a high-precision time synchronization module; The fiber optic strain sensor, vibration sensor, and displacement sensor collect high-frequency dynamic data in real time at a preset sampling frequency. The foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data are collected in the form of low-frequency steady-state data according to the construction process and are event-triggered. The central control platform performs spatiotemporal registration of high-frequency dynamic data and low-frequency steady-state data based on a unified timestamp, and constructs a multi-source heterogeneous data fusion matrix that includes time dimension, spatial dimension and physical quantity dimension.

[0009] In some embodiments of this application, the step of establishing a geometric model of the unit based on unit design data and sensor deployment information; and integrating material properties and mechanical characteristics to construct a coupled dynamic model reflecting the foundation-frame-shaft system, forming a digital twin of the unit's axis, includes: Based on the unit design drawings and the sensor placement locations, a high-precision three-dimensional geometric model including the rotor, impeller, frame, and foundation structure was established. By integrating material properties, stiffness matrix and boundary conditions, a finite element mechanism model containing the coupling characteristics of foundation-frame-shaft system is constructed, wherein the boundary conditions are dynamically updated based on real-time data collected by the full-element sensing network. The geometric model and the finite element mechanism model are fused through bidirectional data mapping to generate a digital twin that can reflect the changes in the unit's axis state in real time and has simulation and deduction capabilities.

[0010] In some embodiments of this application, during the unit construction and operation phases, real-time data collected by the all-element sensing network is input into the digital twin to dynamically calculate the axis offset trend and generate deviation prediction results for the axis status, including: The strain, vibration, and displacement data, as well as the foundation settlement monitoring data and concrete temperature monitoring data, collected in real time by the full-element sensing network are input into the digital twin. Using the finite element mechanism model and the real-time data as boundary conditions, the offset, offset direction, and deviation distribution of the axis of the dynamic computer group are measured throughout the construction process. Based on the calculation results, deviation prediction results for the axis state are generated.

[0011] In some embodiments of this application, a dynamic axis adjustment strategy is generated based on the deviation prediction results, including: Before the unit is turned around, a virtual turning simulation is performed based on the digital twin to simulate the axle state under different operating conditions. Using a multi-objective optimization algorithm, with the optimization objectives of minimizing axis offset and minimizing shim adjustment amount, the adjustment amount of the shim adjustment scheme is determined; Based on the adjustment amount, an installation period adjustment instruction is generated to guide the on-site shim adjustment work.

[0012] Some embodiments of this application also include: During the concrete pouring process, real-time monitoring of foundation settlement data and concrete temperature data is conducted. When the detected axial offset caused by uneven foundation settlement or concrete temperature stress exceeds the preset threshold, the compensation adjustment mechanism is triggered. The compensation adjustment mechanism includes: Based on the current axis offset and offset direction, the compensation adjustment amount is calculated in reverse through the digital twin; Based on the compensation adjustment amount, a pouring compensation adjustment instruction is generated, which includes local grouting in the area to be poured or pre-tightening adjustment in the already poured area. The compensation adjustment is completed before the concrete continues to be poured to prevent the axial offset from being solidified by subsequent pouring.

[0013] Some embodiments of this application also include: After the unit is filled with water and put into trial operation, the measured vibration data and swing data of the unit are collected; The model of the digital twin is corrected based on the measured data, and the model correction includes: The measured vibration data and sway data are used as feedback signals and compared with the simulation output of the digital twin. The stiffness and damping parameters of the finite element mechanism model are inverted and corrected using the Kalman filter algorithm, so that the corrected digital twin matches the actual unit state. Based on the corrected digital twin, the force distribution of the thrust bearing or the clearance of the guide bearing is dynamically adjusted to generate operational optimization instructions, thereby achieving closed-loop dynamic optimization of the shaft state.

[0014] In some embodiments of this application, a digital twin-based hydro-generator shaft adjustment system includes: The all-element sensing module is used to deploy sensors at key parts of the unit during the unit installation phase, and at the same time access environmental monitoring data during the construction period to build an all-element sensing network covering the unit itself and the construction environment. The digital twin construction module is used to establish a geometric model of the unit based on the unit design data and sensor deployment information; it integrates material properties and mechanical characteristics to construct a coupled dynamic model that reflects the foundation-frame-shaft system, forming a digital twin of the unit's axis; The deviation prediction module is used to input the real-time data collected by the full-element sensing network into the digital twin during the construction and operation phases of the unit, dynamically calculate the axis offset trend, and generate deviation prediction results of the axis status. The dynamic adjustment strategy generation module is used to generate a dynamic adjustment strategy for the axis based on the deviation prediction results.

[0015] The method and system for adjusting the shaft of a hydro-generator unit based on digital twins, as described in this application, have the following advantages compared to the prior art: By deploying multiple sensors at key locations within the unit and integrating environmental monitoring data from the construction period, a comprehensive sensing network covering both the unit itself and the construction environment is constructed. Employing unified timestamps and multi-source heterogeneous data fusion technology, all-round real-time sensing of the unit's axis status is achieved. By integrating geometric models and finite element mechanism models, a digital twin reflecting the coupling characteristics of the foundation-frame-shaft system is constructed. Real-time data-driven dynamic mapping and simulation of the axis status are then implemented, providing precise decision-making basis for axis adjustments.

[0016] Covering the installation, pouring, and operation phases, a closed-loop optimization system for axis adjustment is formed throughout the entire lifecycle, significantly improving the long-term stability of the unit's axis. Through the simulation and extrapolation capabilities of digital twins, the offset trend can be predicted before axis deviation occurs, and proactive compensation adjustments can be made before key processes such as concrete pouring are completed, preventing deviations from being solidified by subsequent processes and greatly reducing the difficulty and cost of adjustment. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a method for adjusting the shaft of a hydro-generator unit based on digital twins, as described in an embodiment of this application. Detailed Implementation

[0018] The specific embodiments of this application will be described in further detail below with reference to the accompanying drawings and examples. The following examples are used to illustrate this application, but are not intended to limit the scope of this application.

[0019] like Figure 1 As shown in the embodiments of this application, some embodiments of this application provide a method for adjusting the shaft of a hydro-generator unit based on digital twins, including: During the unit installation phase, sensors are deployed at key parts of the unit, and environmental monitoring data during the construction period is also connected to build a full-element sensing network covering the unit itself and the construction environment. Based on the unit design data and sensor deployment information, a geometric model of the unit is established; by integrating material properties and mechanical characteristics, a coupled dynamic model reflecting the foundation-frame-shaft system is constructed, forming a digital twin of the unit's axis; During the construction and operation phases of the unit, the real-time data collected by the full-element sensing network is input into the digital twin to dynamically calculate the axis offset trend and generate deviation prediction results for the axis status. Based on the deviation prediction results, a dynamic axis adjustment strategy is generated.

[0020] In some embodiments of this application, during the unit installation phase, sensors are deployed at key parts of the unit, and environmental monitoring data from the construction period is simultaneously integrated to construct a comprehensive sensing network covering both the unit itself and the construction environment, including: Fiber optic strain sensors, vibration sensors, and displacement sensors are embedded in the generator rotor, turbine runner, and upper and lower frames to collect strain, vibration, and displacement data in real time during the unit installation process. Simultaneously, it integrates foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data during the construction period, forming the aforementioned all-element sensing network covering the unit's physical status and construction environment parameters.

[0021] In some embodiments of this application, when fiber optic strain sensors, vibration sensors, and displacement sensors are embedded in the generator rotor, turbine runner, and upper and lower frames, the following are included: The fiber optic strain sensors are spirally arranged along the axis of the unit at key sections of the rotor, impeller and upper and lower frames to form a three-dimensional strain field monitoring array. The vibration sensor and displacement sensor are arranged in pairs at the upper and lower guide bearings of the generator rotor, the guide bearing of the turbine and the thrust bearing, forming a synchronous monitoring unit for radial and axial displacement. Low-power smart sensor nodes are deployed in a wireless network at the foundation settlement monitoring points and temperature measurement points inside the concrete pouring chamber. Each node synchronizes with the central control platform through a self-organizing network protocol.

[0022] In this embodiment, multiple strain measurement points are uniformly arranged circumferentially on the upper, middle, and lower end faces of the generator rotor. These measurement points are connected in series via optical fibers, forming a continuously distributed three-dimensional strain monitoring network along the axial direction. The helical arrangement allows for the simultaneous capture of axial strain, circumferential strain, and shear strain components. By monitoring the strain evolution of the rotor, impeller, and frame during installation in real time, the deformation trends of each component caused by its own weight, assembly stress, and temperature changes can be accurately identified, providing fundamental data support for axial offset calculations.

[0023] In this embodiment, a pair of eddy current displacement sensors are installed at the upper guide bearing of the generator along the orthogonal X and Y directions to monitor the radial displacement of the rotor relative to the bearing center in real time. The same arrangement is used at the lower guide bearing and the turbine guide bearing to form a multi-section radial displacement synchronous monitoring network. At the thrust bearing, multiple axial displacement sensors are evenly arranged along the circumference to monitor the force distribution and axial displacement changes of the thrust bearing in real time.

[0024] In this embodiment, differential pressure settlement sensors are embedded at key settlement monitoring points on the unit foundation, and digital temperature sensors are embedded at different elevations within the concrete pouring chamber. Each sensor is equipped with a low-power wireless communication module, establishing a communication link with the central control platform via Zigbee or LoRa self-organizing network protocols. The central control platform has a built-in high-precision time synchronization module, which performs periodic time synchronization calibration on each sensor node through wireless synchronization signals, ensuring that the data collected by all sensors have a unified time reference and achieving precise alignment between construction environment data and unit body data in the time dimension.

[0025] In some embodiments of this application, the formation of the full-element sensing network covering the unit's physical state and construction environment parameters includes: Establish a unified timestamp benchmark and synchronize various sensors through a high-precision time synchronization module; The fiber optic strain sensor, vibration sensor, and displacement sensor collect high-frequency dynamic data in real time at a preset sampling frequency. The foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data are collected in the form of low-frequency steady-state data according to the construction process and are event-triggered. The central control platform performs spatiotemporal registration of high-frequency dynamic data and low-frequency steady-state data based on a unified timestamp, and constructs a multi-source heterogeneous data fusion matrix that includes time dimension, spatial dimension and physical quantity dimension.

[0026] In this embodiment, the all-element sensing network includes various types of sensors, such as fiber optic strain sensors, vibration sensors, displacement sensors, foundation settlement sensors, and concrete temperature sensors. These sensors belong to different data acquisition systems, and their internal clock sources differ. Without a unified time reference, time misalignment will occur during subsequent data fusion, thus affecting the accurate mapping of the axis state by the digital twin. To solve the above technical problem, this application sets up a high-precision time synchronization module in the central control platform. This module uses both the Global Positioning System (GPS) and the BeiDou Navigation Satellite System (BDS) for time synchronization to achieve precise time synchronization. The central control platform periodically sends a unified timestamp reference signal to each sensor node via wired or wireless communication links. After receiving the time synchronization signal, each sensor node synchronously calibrates its internal clock. For sensor devices that do not support external time synchronization, a timestamp marking unit is added to the front end of their data acquisition channel to add a unified timestamp when the data enters the central control platform.

[0027] In this embodiment, the central control platform continuously acquires data from the fiber optic strain sensor, vibration sensor, and displacement sensor in real time according to a preset sampling frequency, forming a high-frequency dynamic data stream. This high-frequency dynamic data can accurately reflect the transient response characteristics of the unit during installation and operation caused by factors such as assembly stress, temperature changes, and vibration excitation.

[0028] For foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data, the central control platform adopts a low-frequency steady-state data acquisition strategy. Specifically, foundation settlement monitoring data is collected periodically at preset time intervals; concrete temperature monitoring data is collected more frequently in the early stages of pouring, following the development pattern of hydration heat after concrete pouring, and the collection frequency is gradually reduced after the peak of hydration heat; concrete pouring progress data is collected in an event-triggered manner according to construction procedures. That is, whenever a pouring layer or key process is completed, on-site personnel enter the pouring progress information via mobile terminals or the system interface, forming an event record synchronized with the construction progress.

[0029] In this embodiment, a spatiotemporal registration module is set up in the central control platform to perform unified processing of multi-source heterogeneous data. Specifically, this includes: For high-frequency dynamic data, the central control platform establishes a continuous time series based on the data stream with the highest sampling frequency. For low-frequency steady-state data, interpolation algorithms are used to complete the time dimension of data between two adjacent acquisition times, forming a continuous data stream aligned with the high-frequency data time series. For example, for foundation settlement data, linear interpolation is used to estimate the settlement at the intermediate time between two actual acquisition times, ensuring that the time density of the settlement data is consistent with that of the vibration data. Through time registration, it is ensured that all data have complete data values ​​at every calculation time, providing a continuous time series input for the digital twin.

[0030] This application uses the three-dimensional spatial coordinate system defined in the unit design drawings as a unified spatial reference. For fiber optic strain sensors, vibration sensors, and displacement sensors, a mapping relationship between the sensor spatial coordinates and the unit's geometric model is established based on their coordinate values ​​in the design drawings. For foundation settlement monitoring points and concrete temperature measuring points, their actual spatial coordinates are determined using a total station or a 3D laser scanner and then transformed to the unified spatial coordinate system. Based on this, for data requirements at any spatial location, the data can be calculated from the measured data of adjacent sensor measuring points using spatial interpolation methods, forming a continuous spatial data field covering the entire unit and construction area.

[0031] After completing time and spatial registration, the central control platform organizes the processed data into a multi-source heterogeneous data fusion matrix. This fusion matrix includes: a time dimension representing the time series of data acquisition, indexed by a unified timestamp; a spatial dimension representing the spatial location of the data, indexed by coordinate points in a unified spatial coordinate system; and a physical quantity dimension representing the physical type of the data, including different physical parameters such as strain, vibration, displacement, settlement, temperature, and progress. During each iteration of calculation, the digital twin quickly extracts the corresponding data values ​​from the fusion matrix based on the current timestamp and the required spatial location, using these values ​​as boundary conditions or load inputs to drive the finite element mechanism model, thereby achieving dynamic calculation and simulation of the unit's axis state.

[0032] In some embodiments of this application, the step of establishing a geometric model of the unit based on unit design data and sensor deployment information; and integrating material properties and mechanical characteristics to construct a coupled dynamic model reflecting the foundation-frame-shaft system, forming a digital twin of the unit's axis, includes: Based on the unit design drawings and the sensor placement locations, a high-precision three-dimensional geometric model including the rotor, impeller, frame, and foundation structure was established. By integrating material properties, stiffness matrix and boundary conditions, a finite element mechanism model containing the coupling characteristics of foundation-frame-shaft system is constructed, wherein the boundary conditions are dynamically updated based on real-time data collected by the full-element sensing network. The geometric model and the finite element mechanism model are fused through bidirectional data mapping to generate a digital twin that can reflect the changes in the unit's axis state in real time and has simulation and deduction capabilities.

[0033] In this embodiment, constructing a digital twin requires establishing a high-precision three-dimensional geometric model of the unit. Based on the unit's design drawings, this application extracts the geometric dimensions, assembly relationships, and spatial location information of key components such as the rotor, impeller, upper and lower frames, thrust bearing, guide bearing, and foundation structure. In 3D modeling software, high-precision three-dimensional solid models of each component are established at a 1:1 scale, and the entire unit is assembled according to the assembly constraints in the design drawings. During the geometric model construction process, the aforementioned sensor deployment location information needs to be embedded into the model. Specifically, based on the spatial coordinates of each sensor during actual deployment, sensor marker points are added at corresponding positions in the geometric model, and attribute information such as sensor type, measurement direction, and measurement range is recorded. The established high-precision three-dimensional geometric model includes at least the following structural units: rotor unit: including generator rotor body, magnetic poles, rotor support, etc.; runner unit: including turbine runner, blades, runner body, etc.; frame unit: including upper frame, lower frame, foundation frame, etc.; bearing unit: including upper guide bearing, lower guide bearing, water guide bearing, thrust bearing, etc.; foundation unit: including concrete foundation, foundation plate, anchor bolts, etc.

[0034] In this embodiment, based on the geometric model, this application further constructs a finite element mechanism model reflecting the physical and mechanical behavior of the unit. The specific construction process is as follows: According to the actual material composition of each component of the unit, corresponding material property parameters are assigned to different parts in the geometric model. The material property parameters include, but are not limited to: elastic modulus: reflecting the material's ability to resist elastic deformation; Poisson's ratio: reflecting the ratio of lateral deformation to axial deformation of the material under uniaxial tension or compression; density: reflecting the mass distribution of the material, used for gravity load calculation; coefficient of thermal expansion: reflecting the thermal deformation characteristics of the material under temperature changes; damping coefficient: reflecting the energy dissipation characteristics of the material and structure. For steel components such as rotors and impellers, the material properties are determined according to the material grade marked in the design drawings; for concrete foundations, the material properties are determined according to the concrete grade and mix proportion; for special components such as thrust bearing oil film, equivalent stiffness and equivalent damping parameters are used for characterization.

[0035] In this embodiment, based on material properties and geometry, foundation stiffness matrices, frame stiffness matrices, rotor-wheel shaft system stiffness matrices, and connection stiffness matrices between components are constructed. Specifically: the foundation stiffness matrix reflects the constraint effect of the concrete foundation on the superstructure; the frame stiffness matrix reflects the supporting effect of the upper and lower frames on the bearings; the shaft system stiffness matrix reflects the overall bending and torsional stiffness of the rotor, wheel, and connecting shafts; and the connection stiffness matrix reflects nonlinear connection characteristics such as bearing oil film stiffness, bolt connection stiffness, and contact surface stiffness. By assembling these stiffness matrices, an overall system stiffness matrix is ​​formed, achieving coupled dynamic modeling of the foundation-frame-wheel system.

[0036] In this embodiment, the boundary conditions are dynamically updated, including: displacement boundary conditions: based on foundation settlement monitoring data, the displacement constraints of the foundation structure are updated in real time to reflect the actual impact of foundation settlement on the unit axis; temperature boundary conditions: based on concrete temperature monitoring data, the temperature field distribution of the foundation structure is updated in real time to reflect the heat of hydration and thermal deformation caused by changes in ambient temperature; load boundary conditions: based on concrete pouring progress data, the construction loads applied to the foundation structure during pouring are updated in real time; constraint boundary conditions: based on bolt preload monitoring data during unit installation, the constraint status of the connection surfaces is updated in real time. The central control platform processes the real-time data collected by the full-element sensing network through a data fusion matrix and transmits it to the finite element solver. The solver reassembles the boundary conditions within each calculation step, driving the finite element mechanism model to perform calculations. In this embodiment, the geometric information in the geometric model is mapped to the finite element mechanism model, providing a computational grid and spatial positioning for finite element analysis. Specifically, this includes: using the mesh generation result of the geometric model as the discrete unit for finite element calculation; mapping the spatial coordinates of sensor marker points in the geometric model to the corresponding nodes of the finite element mesh, realizing the spatial association between measured data and computational nodes; and mapping the assembly relationships of various components in the geometric model to contact pairs and constraint relationships in the finite element mechanism model.

[0037] The calculation results of the finite element mechanism model are back-mapped to the geometric model to achieve a visual representation of the calculation data. Specifically, this includes: mapping the physical quantities such as displacement, stress, and strain of each node obtained from the finite element calculation to the corresponding surface of the geometric model through an interpolation algorithm, generating a color cloud map to intuitively display the spatial distribution of the axis offset; overlaying the calculated axis offset amount, offset direction, and other core indicators onto the geometric model to achieve a visual representation of the axis state; and presenting the deviation prediction results as a dynamic evolution curve on the time axis panel of the geometric model to show the trend of axis offset over time.

[0038] To achieve bidirectional data synchronization between the geometric model and the mechanistic model, this application sets up a data mapping engine in the central control platform. After the finite element mechanistic model completes a calculation, the data mapping engine synchronizes the calculation results to the geometric model; when the sensor placement or assembly relationship in the geometric model changes, the data mapping engine automatically updates the corresponding parameters in the finite element mechanistic model.

[0039] In some embodiments of this application, during the unit construction and operation phases, real-time data collected by the all-element sensing network is input into the digital twin to dynamically calculate the axis offset trend and generate deviation prediction results for the axis status, including: The strain, vibration, and displacement data, as well as the foundation settlement monitoring data and concrete temperature monitoring data, collected in real time by the full-element sensing network are input into the digital twin. Using the finite element mechanism model and the real-time data as boundary conditions, the offset, offset direction, and deviation distribution of the axis of the dynamic computer group are measured throughout the construction process. Based on the calculation results, deviation prediction results for the axis state are generated.

[0040] In this embodiment, the central control platform collects real-time dynamic data of the unit body from fiber optic strain sensors, vibration sensors, and displacement sensors, as well as construction environment parameters such as foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data. After spatiotemporal registration and data fusion processing, the data is pushed to the finite element solver of the digital twin through a data-driven engine. Among them, the dynamic data of the unit body reflects the strain distribution, vibration characteristics, and displacement state of the unit in the current state; the construction environment data reflects the changing patterns of external factors such as foundation settlement, temperature field changes, and the application of construction loads.

[0041] In this embodiment, after receiving real-time data, the finite element solver dynamically loads it into the corresponding positions of the model: foundation settlement monitoring data is used as displacement boundary conditions to simulate the constraint effect of uneven ground settlement on the unit foundation; concrete temperature monitoring data is used as temperature boundary conditions to calculate the temperature field distribution inside the foundation through the heat conduction equation, and thermal strain is calculated based on the material's thermal expansion coefficient; concrete pouring progress data is used as load boundary conditions to determine the application range and magnitude of the construction load at the current moment. Based on the overall system stiffness matrix and damping matrix, the solver performs stepwise integration of the dynamic equations within each calculation time step to calculate the displacement response of each node. Radial displacements of each node on the rotor, impeller, and shaft centerline are extracted along the unit's axial direction, and the offset, direction, and axial deviation distribution of the axis relative to the ideal axis are calculated, forming a dynamic sequence of the axis state evolution over time.

[0042] In this embodiment, the deviation prediction results include multiple aspects: generating a current axis status assessment, including the current maximum offset and its location, direction, and spatial distribution, and comparing it with the design allowable deviation to calculate the axis health score. Based on the historical dynamic sequence of axis offset, a time series prediction method is used to predict the offset trend in the short term and calculate the expected time to reach the warning threshold. Correlation analysis is performed using construction environment data to identify the main causes of the deviation. For example, if foundation settlement is highly correlated with the offset, it is determined to be a settlement-dominated deviation; if temperature data is highly correlated with the offset, it is determined to be a temperature stress-dominated deviation. A three-dimensional visualization interface is used to display the spatial distribution of the axis offset in a color temperature map format, and a dynamic curve graph is used to display the trend of the offset over time and the prediction results. When the prediction exceeds the limit, an early warning message is automatically generated and pushed to on-site management and technical personnel.

[0043] In some embodiments of this application, a dynamic axis adjustment strategy is generated based on the deviation prediction results, including: Before the unit is turned around, a virtual turning simulation is performed based on the digital twin to simulate the axle state under different operating conditions. Using a multi-objective optimization algorithm, with the optimization objectives of minimizing axis offset and minimizing shim adjustment amount, the adjustment amount of the shim adjustment scheme is determined; Based on the adjustment amount, an installation period adjustment instruction is generated to guide the on-site shim adjustment work.

[0044] In this embodiment, real-time data collected by the current all-element sensing network is input into the digital twin to drive the finite element mechanism model. Virtual turning gear simulation can simulate the shaft state of the unit under various operating conditions: in static conditions, it simulates the shaft attitude when the unit is stationary, the uniformity of the air gap between the rotor and stator, and the radial clearance distribution at each bearing; in dynamic conditions, it simulates the operating state of the unit under different speeds and loads, calculating the dynamic sway caused by centrifugal force, unbalanced force, and other factors during rotor rotation. Through virtual turning gear simulation, a comprehensive assessment of the shaft state is obtained.

[0045] In this embodiment, after obtaining the virtual turning gear simulation results, the digital twin automatically calculates the optimal shim adjustment scheme using a multi-objective optimization algorithm. This application uses minimizing axis offset and minimizing shim adjustment amount as two core optimization objectives: the former aims to bring the unit axis as close to the ideal state as possible through shim adjustment, while the latter aims to reduce the range and number of shim adjustments to improve construction efficiency. The optimization algorithm iteratively solves within the feasible region of the shim adjustment amount, generating a Pareto optimal solution set. This solution set contains multiple non-dominated solutions, each corresponding to a set of shim adjustment schemes, each with its own emphasis among different optimization objectives. The digital twin presents the Pareto optimal solution set in a visual manner, for example, displaying the trade-off between axis offset and shim adjustment amount for each scheme in the form of a scatter plot, allowing technicians to select based on actual site conditions. If axis accuracy is prioritized, the scheme with the smallest axis offset can be selected; if construction efficiency is prioritized, the scheme with the smallest shim adjustment amount can be selected.

[0046] In this embodiment, after determining the shim adjustment plan, the digital twin automatically generates installation-period adjustment instructions to guide on-site personnel in performing shim adjustments. The adjustment instructions include at least the following: adjustment location identification, clearly identifying the location of the shims to be adjusted, and highlighting it in the 3D geometric model to ensure accurate location recognition; adjustment amount details, listing the shim thickness change at each adjustment location; suggested adjustment sequence, recommending the optimal operation sequence based on the interrelationships of the shim adjustments; and expected results after adjustment, providing the expected offset and health score of the axis after adjustment based on simulation calculations from the digital twin, facilitating personnel confirmation of the adjustment target.

[0047] The central control platform pushes adjustment instructions to the mobile terminals of on-site personnel, who can view the adjustment plan at any time and perform shim adjustment work according to the instructions. After the adjustment is completed, the full-element sensing network collects real-time data again, and the digital twin performs verification calculations based on the new data to confirm whether the axis status has met the expected target. If there are still deviations, iterative optimization is carried out until the axis status meets the design requirements.

[0048] Some embodiments of this application also include: During the concrete pouring process, real-time monitoring of foundation settlement data and concrete temperature data is conducted. When the detected axial offset caused by uneven foundation settlement or concrete temperature stress exceeds the preset threshold, the compensation adjustment mechanism is triggered. The compensation adjustment mechanism includes: Based on the current axis offset and offset direction, the compensation adjustment amount is calculated in reverse through the digital twin; Based on the compensation adjustment amount, a pouring compensation adjustment instruction is generated, which includes local grouting in the area to be poured or pre-tightening adjustment in the already poured area. The compensation adjustment is completed before the concrete continues to be poured to prevent the axial offset from being solidified by subsequent pouring.

[0049] In this embodiment, during the concrete pouring process, foundation settlement monitoring data and concrete temperature monitoring data are continuously collected in real time through a full-element sensing network. Foundation settlement monitoring data is collected in real time by differential pressure settlement sensors embedded in key locations of the foundation, reflecting the amount of settlement and the degree of uneven settlement at different pouring stages. Concrete temperature monitoring data is collected in real time by digital temperature sensors embedded at different elevations within the pouring chamber, reflecting the dynamic changes in the internal temperature field of the concrete. The central control platform synchronizes the above real-time data with the digital twin.

[0050] In this embodiment, the digital twin dynamically calculates the current axis offset based on real-time input monitoring data. The system automatically triggers a compensation adjustment mechanism when the calculation result meets any of the following conditions: Condition 1: The axis offset caused by uneven foundation settlement exceeds a preset settlement threshold. For example, if the settlement difference between the two sides of the foundation exceeds the design allowable value, and the resulting axis offset exceeds the set threshold, compensation adjustment is triggered. Condition 2: The axis offset caused by concrete temperature stress exceeds a preset temperature threshold. For example, if thermal deformation caused by temperature differences between the inside and outside of the concrete or between different areas results in an axis offset exceeding the set threshold, compensation adjustment is triggered. The preset threshold can be set according to the unit design requirements and construction specifications, and supports dynamic adjustment based on actual construction conditions.

[0051] In this embodiment, the core of the compensation adjustment mechanism lies in reverse deduction, that is, using the simulation capabilities of a digital twin to deduce the required compensation measures from the current axis offset state. Specifically, based on the currently measured axis offset, the digital twin, combined with the finite element mechanism model, uses an inverse problem-solving method to reverse deduce the root cause of the offset and the required compensation amount. The reverse deduction process includes: identifying the main factors causing the current offset; calculating the equivalent adjustment amount required to eliminate the current offset; and deducing specific construction compensation measures based on the adjustment amount. For example, if the offset is mainly caused by uneven foundation settlement, the digital twin calculates how much grout needs to be added in that area to restore the axis to the ideal state; if the offset is mainly caused by temperature stress, it calculates how much preload needs to be adjusted or what temperature control measures need to be taken to counteract the effects of thermal deformation.

[0052] In this embodiment, the digital twin generates specific pouring compensation and adjustment instructions based on the back-engineering results. These instructions include at least two types: Local grouting instructions: When the axial offset is mainly caused by uneven foundation settlement, the system generates a local grouting instruction. This instruction specifies the location of the area requiring grouting, the volume of grout, the required grout material, and a suggested grouting sequence. Workers follow the instructions to perform local grouting in the area to be poured, increasing the volume of concrete in that area to raise the settled side and thus correct the axial offset. Preload adjustment instructions: When the axial offset is mainly caused by temperature stress or structural deformation, the system generates a preload adjustment instruction. This instruction specifies the bolt location, adjustment direction, and adjustment amount for adjusting the preload. Workers follow the instructions to adjust the preload at the connecting bolts in the already poured area, changing the constraint state of the connection surface to counteract the deformation caused by temperature stress and restore the axial direction to its ideal state.

[0053] In this embodiment, the application requires all compensation and adjustment operations to be completed before the concrete pouring continues. The technical significance of this is that when the concrete has not yet hardened, the foundation structure and unit axis remain adjustable. At this time, local grouting or pre-tensioning adjustments can effectively correct any existing axis misalignment and restore the axis to the design allowable range. If pouring continues before the compensation and adjustment are completed, the subsequently poured concrete will solidify the current state, permanently locking the axis misalignment within the structure.

[0054] Therefore, after the digital twin generates compensation and adjustment instructions, the central control platform monitors the completion of the adjustment work in real time. Only after the operators confirm that the compensation and adjustment have been completed and the digital twin verifies that the axis status has returned to the allowable range will the system allow concrete pouring to continue. If the axis status still does not meet the standards after adjustment, the system will prompt for continued adjustment or recalculation of the compensation plan until the requirements are met.

[0055] Some embodiments of this application also include: After the unit is filled with water and put into trial operation, the measured vibration data and swing data of the unit are collected; The model of the digital twin is corrected based on the measured data, and the model correction includes: The measured vibration data and sway data are used as feedback signals and compared with the simulation output of the digital twin. The stiffness and damping parameters of the finite element mechanism model are inverted and corrected using the Kalman filter algorithm, so that the corrected digital twin matches the actual unit state. Based on the corrected digital twin, the force distribution of the thrust bearing or the clearance of the guide bearing is dynamically adjusted to generate operational optimization instructions, thereby achieving closed-loop dynamic optimization of the shaft state.

[0056] In this embodiment, after the unit is filled with water and undergoes trial operation, measured vibration and sway data of the unit under actual operating conditions are collected through a full-element sensing network. The vibration data is collected by vibration sensors installed at each bearing, reflecting the vibration amplitude and frequency characteristics of each part of the unit; the sway data is collected by displacement sensors installed at key parts of the rotor, runner, and shaft system, reflecting the dynamic sway trajectory of the axis of the unit during rotation.

[0057] In this embodiment, the digital twin model is corrected based on measured data. Specifically, the measured vibration and sway data are used as feedback signals and compared with the simulation output of the digital twin under the same operating conditions to calculate the error index at each measuring point. The stiffness and damping parameters of the finite element mechanism model are corrected by inverting the Kalman filter algorithm. This algorithm estimates the optimal parameter values ​​from the noisy measured data through recursive calculations in two steps: prediction and update. This ensures that the corrected digital twin highly matches the dynamic response characteristics of the actual unit.

[0058] In this embodiment, based on the corrected digital twin, the system can accurately analyze the current force distribution of the thrust bearing and the clearance state of the guide bearing pads. For the thrust bearing, when uneven force distribution is detected among the pads, the system generates a force adjustment command to guide maintenance personnel to adjust the support height of the thrust pads or replace the shims to ensure uniform force distribution among the pads, thereby optimizing the axial state of the shaft. For the guide bearing pad clearance, the system accurately calculates the optimal clearance value at each guide bearing based on measured vibration data and model simulation results. When there is a deviation between the actual clearance and the optimal clearance, a clearance adjustment command is generated, specifying the bearing position, adjustment direction, and adjustment amount that need to be adjusted, guiding maintenance personnel to optimize the clearance.

[0059] Through the aforementioned optimizations and adjustments, this application achieves closed-loop dynamic optimization of the unit's axis status. After maintenance personnel execute operations according to instructions, the full-element sensing network collects optimized measured data again. The digital twin performs verification calculations based on the new data to confirm whether the axis status has reached the optimization target. If it does not meet expectations, the system will perform model correction and optimization calculations again based on the new measured data, forming a complete closed loop of perception-modeling-prediction-adjustment-verification. As the unit operates for a period of time, the digital twin can continuously perform model corrections and parameter updates, always maintaining a high degree of matching with the actual unit status, ensuring that the axis is always in the optimal state and effectively improving the unit's operational stability.

[0060] In some embodiments of this application, a digital twin-based hydro-generator shaft adjustment system includes: The all-element sensing module is used to deploy sensors at key parts of the unit during the unit installation phase, and at the same time access environmental monitoring data during the construction period to build an all-element sensing network covering the unit itself and the construction environment. The digital twin construction module is used to establish a geometric model of the unit based on the unit design data and sensor deployment information; it integrates material properties and mechanical characteristics to construct a coupled dynamic model that reflects the foundation-frame-shaft system, forming a digital twin of the unit's axis; The deviation prediction module is used to input the real-time data collected by the full-element sensing network into the digital twin during the construction and operation phases of the unit, dynamically calculate the axis offset trend, and generate deviation prediction results of the axis status. The dynamic adjustment strategy generation module is used to generate a dynamic adjustment strategy for the axis based on the deviation prediction results.

[0061] The above description is only a preferred embodiment of this application. It should be noted that for those skilled in the art, several improvements and substitutions can be made without departing from the technical principles of this application, and these improvements and substitutions should also be considered within the scope of protection of this application.

Claims

1. A method for adjusting the shaft of a hydro-generator unit based on digital twins, characterized in that, include: During the unit installation phase, sensors are deployed at key parts of the unit, and environmental monitoring data during the construction period is also connected to build a full-element sensing network covering the unit itself and the construction environment. Based on the unit design data and sensor deployment information, a geometric model of the unit is established; by integrating material properties and mechanical characteristics, a coupled dynamic model reflecting the foundation-frame-shaft system is constructed, forming a digital twin of the unit's axis; During the construction and operation phases of the unit, the real-time data collected by the full-element sensing network is input into the digital twin to dynamically calculate the axis offset trend and generate deviation prediction results for the axis status. Based on the deviation prediction results, a dynamic axis adjustment strategy is generated.

2. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 1, characterized in that, During the unit installation phase, sensors are deployed at key locations on the unit, and environmental monitoring data from the construction period is simultaneously integrated to construct a comprehensive sensing network covering both the unit itself and the construction environment, including: Fiber optic strain sensors, vibration sensors, and displacement sensors are embedded in the generator rotor, turbine runner, and upper and lower frames to collect strain, vibration, and displacement data in real time during the unit installation process. Simultaneously, it integrates foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data during the construction period, forming the aforementioned all-element sensing network covering the unit's physical status and construction environment parameters.

3. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 2, characterized in that, When embedding fiber optic strain sensors, vibration sensors, and displacement sensors in the generator rotor, turbine runner, and upper and lower frames, the following applies: The fiber optic strain sensors are spirally arranged along the axis of the unit at key sections of the rotor, impeller and upper and lower frames to form a three-dimensional strain field monitoring array. The vibration sensor and displacement sensor are arranged in pairs at the upper and lower guide bearings of the generator rotor, the guide bearing of the turbine and the thrust bearing, forming a synchronous monitoring unit for radial and axial displacement. Low-power smart sensor nodes are deployed in a wireless network at the foundation settlement monitoring points and temperature measurement points inside the concrete pouring chamber. Each node synchronizes with the central control platform through a self-organizing network protocol.

4. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 2, characterized in that, The comprehensive sensing network that forms a coverage of the unit's physical status and construction environment parameters includes: Establish a unified timestamp benchmark and synchronize various sensors through a high-precision time synchronization module; The fiber optic strain sensor, vibration sensor, and displacement sensor collect high-frequency dynamic data in real time at a preset sampling frequency. The foundation settlement monitoring data, concrete temperature monitoring data, and concrete pouring progress data are collected in the form of low-frequency steady-state data according to the construction process and are event-triggered. The central control platform performs spatiotemporal registration of high-frequency dynamic data and low-frequency steady-state data based on a unified timestamp, and constructs a multi-source heterogeneous data fusion matrix that includes time dimension, spatial dimension and physical quantity dimension.

5. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 2, characterized in that, Based on unit design data and sensor deployment information, a geometric model of the unit is established; by integrating material properties and mechanical characteristics, a coupled dynamic model reflecting the foundation-frame-shaft system is constructed, forming a digital twin of the unit's axis, including: Based on the unit design drawings and the sensor placement locations, a high-precision three-dimensional geometric model including the rotor, impeller, frame, and foundation structure was established. By integrating material properties, stiffness matrix and boundary conditions, a finite element mechanism model containing the coupling characteristics of foundation-frame-shaft system is constructed, wherein the boundary conditions are dynamically updated based on real-time data collected by the full-element sensing network. The geometric model and the finite element mechanism model are fused through bidirectional data mapping to generate a digital twin that can reflect the changes in the unit's axis state in real time and has simulation and deduction capabilities.

6. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 5, characterized in that, During the unit's construction and operation phases, real-time data collected by the all-element sensing network is input into the digital twin to dynamically calculate the axis offset trend and generate deviation prediction results for the axis status, including: The strain, vibration, and displacement data, as well as the foundation settlement monitoring data and concrete temperature monitoring data, collected in real time by the full-element sensing network are input into the digital twin. Using the finite element mechanism model and the real-time data as boundary conditions, the offset, offset direction, and deviation distribution of the axis of the dynamic computer group are measured throughout the construction process. Based on the calculation results, deviation prediction results for the axis state are generated.

7. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 6, characterized in that, Based on the deviation prediction results, a dynamic axis adjustment strategy is generated, including: Before the unit is turned around, a virtual turning simulation is performed based on the digital twin to simulate the axle state under different operating conditions. Using a multi-objective optimization algorithm, with the optimization objectives of minimizing axis offset and minimizing shim adjustment amount, the adjustment amount of the shim adjustment scheme is determined; Based on the adjustment amount, an installation period adjustment instruction is generated to guide the on-site shim adjustment work.

8. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 7, characterized in that, Also includes: During the concrete pouring process, real-time monitoring of foundation settlement data and concrete temperature data is conducted. When the detected axial offset caused by uneven foundation settlement or concrete temperature stress exceeds the preset threshold, the compensation adjustment mechanism is triggered. The compensation adjustment mechanism includes: Based on the current axis offset and offset direction, the compensation adjustment amount is calculated in reverse through the digital twin; Based on the compensation adjustment amount, a pouring compensation adjustment instruction is generated, which includes local grouting in the area to be poured or pre-tightening adjustment in the already poured area. The compensation adjustment is completed before the concrete continues to be poured to prevent the axial offset from being solidified by subsequent pouring.

9. The method for adjusting the shaft of a hydro-generator unit based on digital twin as described in claim 8, characterized in that, Also includes: After the unit is filled with water and put into trial operation, the measured vibration data and swing data of the unit are collected; The model of the digital twin is corrected based on the measured data, and the model correction includes: The measured vibration data and sway data are used as feedback signals and compared with the simulation output of the digital twin. The stiffness and damping parameters of the finite element mechanism model are inverted and corrected using the Kalman filter algorithm, so that the corrected digital twin matches the actual unit state. Based on the corrected digital twin, the force distribution of the thrust bearing or the clearance of the guide bearing is dynamically adjusted to generate operational optimization commands, thereby achieving closed-loop dynamic optimization of the shaft state.

10. A shaft adjustment system for a hydro-generator unit based on digital twins, characterized in that, include: The all-element sensing module is used to deploy sensors at key parts of the unit during the unit installation phase, and at the same time access environmental monitoring data during the construction period to build an all-element sensing network covering the unit itself and the construction environment. The digital twin construction module is used to establish a geometric model of the unit based on the unit design data and sensor deployment information; it integrates material properties and mechanical characteristics to construct a coupled dynamic model that reflects the foundation-frame-shaft system, forming a digital twin of the unit's axis; The deviation prediction module is used to input the real-time data collected by the full-element sensing network into the digital twin during the construction and operation phases of the unit, dynamically calculate the axis offset trend, and generate deviation prediction results of the axis status. The dynamic adjustment strategy generation module is used to generate a dynamic adjustment strategy for the axis based on the deviation prediction results.

Images