A digital twin-based dynamic performance digital design method of a blower

Abstract

This invention belongs to the technical field of digital design of Roots blower equipment, specifically involving a digital twin-based digital design method for the dynamic performance of blowers. This method includes constructing a parametric structural model of multiple components of the blower, establishing a multi-physics coupled model encompassing aerodynamics, structure, heat, electromagnetics, and acoustics, applying dynamic boundary conditions such as flow disturbances, inertial forces, and base disturbances, constructing a digital twin model synchronized with the physical blower state, performing a master-slave strongly coupled joint solution, extracting multi-field response indices and performing normalization analysis, and finally outputting optimized design parameters for the blower through a multi-objective optimization feedback mechanism. This invention enables multi-domain balanced design among the blower's structural performance, electromagnetic response, thermal stability, and acoustic characteristics, improving modeling accuracy and parameter optimization capabilities, and is suitable for the digital development of high-reliability Roots blower equipment under complex operating conditions.

Description

Technical Field

[0001] This invention belongs to the technical field of digital design of Roots blower equipment, specifically relating to a digital design method for the dynamic performance of blowers based on digital twins. Background Technology

[0002] As a critical gas conveying and pressure regulating device, blowers play a core role in many industries such as petrochemicals, power, metallurgy, building materials, and environmental protection. With the increasing demands of industrial systems for efficient, intelligent, and reliable operation, the performance design of blowers must not only meet static indicators, but also focus on response characteristics under different operating conditions, dynamic loads, and multi-field coupling effects.

[0003] Currently, blower performance design mostly employs static CFD simulation or structural mechanics analysis, supplemented by experimental data correction, to evaluate key parameters such as aerodynamic efficiency, strength, and vibration characteristics. Although some companies have begun to explore the use of multiphysics coupling tools to improve simulation accuracy, most remain at the stage of single design model-based approaches, lacking real-time modeling and feedback of dynamic changes during operation. Furthermore, there is a significant information gap between traditional simulation models and actual equipment, failing to synchronously reflect the health status and performance fluctuations of the equipment under different operating conditions. This leads to delayed optimization processes, increased maintenance costs, and design results that struggle to cover the reliability and stability requirements throughout the entire lifecycle. Therefore, existing design methods still have significant limitations in responding to dynamic operating conditions, reflecting the true state of the system, and supporting full lifecycle management.

[0004] To achieve intelligent and systematic development in blower design, it is urgent to build a digital twin model that integrates virtual and real mapping and real-time feedback, and promote the transformation from static design to a dynamic performance-driven digital design mode. Summary of the Invention

[0005] To address the above problems, the present invention aims to propose a digital design method for the dynamic performance of a blower based on digital twins, comprising the following steps:

[0006] S1. Construct a parametric structural model: Construct a parametric structural model of the wind turbine. The structural model includes components such as rotor, shell, wall panel, oil tank, hub, base and motor. By setting the geometric dimensions, material properties and boundary connection relationships of each component, assembly constraints and structural topology are established in a unified three-dimensional coordinate system.

[0007] S2. Establish a multiphysics simulation model: Establish a multiphysics simulation model, which includes aerodynamic field, structural field, thermal field, electromagnetic field and acoustic field. Define the governing equations, boundary conditions and initial values ​​respectively, and set up coupling interfaces in the boundary region to realize variable interaction.

[0008] S3. Input multi-source dynamic boundary loads: Construct multi-source dynamic boundary loads under wind turbine operating conditions. The multi-source dynamic boundary loads include flow disturbance, turbulent impact, inertial force, gravity change and base disturbance. The loads are generated and applied to the boundary of the simulation model through time domain function and spatial mapping function.

[0009] S4. Constructing a digital twin model and synchronizing it with the state: Establish a digital twin model of the wind turbine. The model synchronously inputs the structural response, temperature field, electromagnetic characteristics and sound pressure information of the physical wind turbine into the simulation system through a state mapping function, and dynamically updates the simulation state according to the collected data.

[0010] S5. Multiphysics joint solution: The multiphysics simulation model is jointly solved, with the structural field as the main control domain and the others as slave control domains. An iterative coupling solution strategy is adopted to control coupling error and ensure solution convergence.

[0011] S6. Dynamic performance index extraction and normalization: Process the simulation output data to extract and normalize performance indices including stress distribution, modal frequency shift, heat accumulation degree, electromagnetic interference intensity and sound pressure level.

[0012] S7. Optimization Feedback Generation and Design Parameter Output: Based on performance indicators, a multi-objective optimization function is constructed. Combining structural safety, thermal stability, electromagnetic compatibility, and acoustic environment requirements, a set of feedback control parameters is generated. The control structural parameters and boundary conditions are iteratively reconstructed, and finally, the optimized design parameters of the wind turbine are given.

[0013] As a preferred technical solution, in the parameterized structural model:

[0014] The impeller's three-dimensional shape is defined by setting the profile curve, segment thickness, and local twist angle;

[0015] The base adopts a segmented taper and wall thickness function to express a variable diameter structure;

[0016] The housing defines the arrangement and connection sequence of each sub-component, while the motor defines the magnetic pole distribution and axial dimensions.

[0017] As a preferred technical solution, in the multiphysics simulation model, the aerodynamic field uses unsteady-state control equations to solve for pressure and velocity distribution, the structural field is modeled based on geometrically nonlinear shell elements, the thermal field uses a set of steady-state heat conduction equations, the electromagnetic field uses vector potential control equations, and the acoustic field is simulated for sound wave propagation using the finite difference method.

[0018] As a preferred technical solution, in the boundary load construction, the flow disturbance generates a disturbance sequence based on the measured flow rate and turbulence intensity function and is applied to the impeller node; the inertial force is calculated by the change in rotational speed to determine the centrifugal load caused by the impeller rotation; and the base disturbance is constructed from ground acceleration sensor data to form a dynamic input signal.

[0019] As a preferred technical solution, the digital twin model obtains the structural response through strain gauges, obtains the temperature distribution through thermocouples, measures the change in magnetic flux density through magnetic sensors, and obtains the operating sound field characteristics through an acoustic sensor array. All types of data are processed and synchronously transmitted to the corresponding modules of the simulation model.

[0020] As a preferred technical solution, the joint solution adopts a strongly coupled strategy. After each round of solution, the structural field is passed to other fields to update the boundary conditions. The convergence is jointly controlled by residual constraints and relaxation factors. The maximum number of iterations does not exceed a set upper limit, and the error threshold is lower than a set boundary.

[0021] As a preferred technical solution, the joint solution adopts a strongly coupled strategy, in which the structural field is set as the main control domain, and the other aerodynamic field, thermal field, electromagnetic field and acoustic field are respectively used as subordinate control domains to participate in the iterative calculation.

[0022] In each solution iteration, the structural field is first solved using the finite element method to obtain the nodal displacements and stress responses under the current conditions. Then, the structural response data is transferred to each subordinate physical field, and the coupling states are synchronized by updating their boundary conditions or load input states. To control the iterative stability during data transfer between different physical domains, a residual constraint mechanism is introduced to monitor the changes of each physical quantity in consecutive iterations, and the update magnitude of the solution variables in each subordinate field is adjusted in conjunction with a relaxation factor.

[0023] The iterative process sets a maximum number of iterations, ranging from 20 to 50 rounds, and sets a convergence error threshold of no more than 10‰. When the residuals of all coupled variables are lower than this threshold and the rate of change of three consecutive iterations is lower than the preset threshold, the multiphysics coupling solution is determined to have converged.

[0024] As a preferred technical solution, among the performance indicators, the structural stress index is obtained by extracting the maximum value node from the equivalent stress cloud map, the modal frequency offset is normalized by the frequency difference under loaded and unloaded conditions, the heat accumulation index is calculated by the temperature rise integral and specific heat product, the electromagnetic interference index is obtained by combining the magnetic flux change rate with the current density change, and the sound pressure level is calculated by the root mean square value of the sound pressure at the simulated receiving point.

[0025] As a preferred technical solution, in step S7, the evaluation function is constructed by combining various performance indicators in a linear weighted manner. Each indicator is given a normalized interval and a constraint range. The structural strength shall not exceed 70% of the material yield strength, the modal frequency shall be at least 10% away from the excitation frequency, the maximum temperature rise shall be less than 85% of the material's allowable limit, the magnetic flux change intensity shall be consistent with the environmental standard, and the sound pressure level shall be less than the limit at 30 meters.

[0026] As a preferred technical solution, in step S7, the feedback control parameter set includes impeller length, base wall thickness, shell layout parameters, motor pole pitch, boundary load amplitude and disturbance spectrum. The optimization algorithm is based on a genetic evolution mechanism, and generates a new parameter combination in each round to update the structural model and simulation model.

[0027] As a preferred technical solution, in step S7, the optimization process iteration termination conditions include: the objective function value has not improved within three consecutive rounds, or all performance indicators simultaneously meet the preset constraint range, and the finally obtained optimized design parameters are output to the design platform for subsequent wind turbine prototype modeling and verification analysis.

[0028] Beneficial effects

[0029] 1. This invention significantly enhances the adjustable modeling capability of wind turbine structures by constructing a parametric model system covering the entire structural level, including blade rotor, shell, wall panel, oil tank, hub, and base, and realizing refined assembly constraints between components under a unified three-dimensional coordinate system.

[0030] Unlike traditional isolated modeling methods oriented towards component levels, this structural model has the ability to define segmented local attributes and associate global assembly topology, enabling accurate structural response capture while maintaining geometric consistency during subsequent multiphysics coupling simulations, thus improving the adaptability and decoupling of the model during reconstruction.

[0031] 2. This invention introduces a dynamic synchronization mechanism based on a state mapping function, which maps the structural response, thermal distribution, electromagnetic parameters and acoustic field data collected during the actual operation of the wind turbine to the node level of the simulation model, thus constructing a three-dimensional coupled digital twin system of structure-field-time.

[0032] Unlike existing virtual-real synchronization methods that rely on fixed-point monitoring value calibration or static field reference, this invention supports dynamic fusion of multi-domain data at the variable level, which significantly improves the fitting accuracy of the simulation state to the physical operating state and provides a guarantee for the simulation accuracy of high-frequency disturbances, load fluctuations and thermal coupling effects.

[0033] 3. The multi-objective optimization feedback mechanism proposed in this invention constructs a multi-objective constraint evaluation function based on five types of normalized performance indicators. Combined with the output results of strongly coupled iterative solution, it drives the automatic update of structural parameters and boundary variables, and outputs parameter combinations that can directly guide the optimization design of wind turbines.

[0034] Compared to existing optimization paths driven by a single indicator (such as stress or efficiency), this mechanism can achieve a multi-dimensional balance between structural strength, electromagnetic compatibility, thermal stability and acoustic performance, thereby improving overall operational reliability and adaptability to operating conditions while ensuring system safety margin. Attached Figure Description

[0035] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0036] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0037] Example 1

[0038] This embodiment focuses on a Roots blower motor unit, applying the method proposed in this invention to digitally model, couple, and optimize its structural, aerodynamic, electromagnetic, thermal, and acoustic properties. Specific method steps are as follows: Figure 1 As shown, it includes:

[0039] S1. Construct a parameterized structural model:

[0040] In this embodiment, a parametric structural model is first constructed on a CAD platform based on the wind turbine design manual and existing prototype data.

[0041] The model covers the seven major subsystems of the wind turbine, including: rotor, casing, wall panels, oil tank, hub, base structure and motor system.

[0042] The rotor system is divided into multiple blade units along the axial direction. Each segment is defined with airfoil parameters, section thickness, relative chord length and installation angle. High-strength metal alloys are selected as materials, and mechanical performance parameters are entered into the model library as adjustable variables.

[0043] The shell system adopts an integral shell structure modeling approach. The outer shell geometry is controlled by parametric surfaces, and key internal components such as cooling channels and support rib structures are represented by Boolean blocks, which have the ability to map thermal and mechanical boundaries.

[0044] The wall panel system is defined as a multi-layer composite panel structure. The panel thickness distribution, fixed boundary shape and flange arrangement matrix are set according to the installation location, and it has rigidity and thermal stability control parameters.

[0045] The modeling of the oil tank system takes into account the shell shape, volume parameters and oil inlet and outlet layout, while defining the internal oil baffle structure, sensor mounting port and heat sink arrangement area, and adapting to the dynamic boundary input of liquid level.

[0046] The hub system is modeled using a spherical shell-connecting disc assembly, with parameters including spherical shell thickness, shaft hole size, and multi-hole flange connection array, to meet the requirements of multi-condition linkage rigidity simulation.

[0047] The base structure is divided into three conical cylinders: upper, middle and lower. The conicity, wall thickness function, distribution of connecting bolts and foot constraint method are defined respectively to support the load transfer path of the overall wind turbine structure.

[0048] The electric motor system takes the stator-rotor coupled structure as the core modeling object, and the parameters cover the core size, electromagnetic clearance, winding structure and cooling plate arrangement, and support synchronous mapping with the structure and electromagnetic field coupled input and output.

[0049] After all components are modeled, they are uniformly registered in 3D space using the coordinate origin, and boundary relationships are defined, including rigid connections, elastic supports, and contact surface definitions. Variable-driven reconstruction of the entire structural model is achieved through parameter binding.

[0050] S2. Establish a multiphysics simulation model:

[0051] Based on the established structural model, a multiphysics simulation system for the wind turbine was further constructed. Integrated collaborative modeling using finite element analysis software and a CFD platform was employed to construct five types of physical fields:

[0052] The aerodynamic field model uses the incompressible Reynolds-averaged Navier-Stokes equations, and the turbulence model is k-ω SST. The pressure and lift distribution on the impeller surface are calculated. In the boundary settings, the inflow uses a flow distribution profile function (highly correlated), the outflow is set as a static pressure boundary, and the impeller surface is a moving wall.

[0053] The structural field model uses shell elements to model the impeller and base, while the hub and main shaft are modeled as solid elements. A nonlinear solver is used, considering material plasticity and connection stiffness. Constraints include a fixed base, hub bearing support, and impeller root flange linkage.

[0054] The thermal field model defines the heat source points inside the motor and establishes a transient heat conduction network based on thermal conductivity, specific heat capacity, and contact thermal resistance. A convective heat transfer coefficient is applied to the cooling air duct to simulate the heat generation and dissipation process under variable load conditions.

[0055] The electromagnetic field model is based on the electromagnetic vector potential control equation, defining the magnetic pole arrangement, core permeability and air gap inductance, and analyzing the magnetic flux density, rotor induced current density and iron loss heat source distribution.

[0056] The sound field model uses the FDTD method, with sound sources placed at the blade tip, gear meshing, and cooling duct outlet to analyze the sound pressure level distribution in the main frequency band (20–1000 Hz).

[0057] The physical domains are coupled through shared grid nodes and boundary variable interfaces, and a unified time step (0.01s) and simulation duration (600s) are defined.

[0058] S3. Input multi-source dynamic boundary loads:

[0059] During wind turbine operation, it is subject to the combined effects of various environmental disturbances and structural responses. This embodiment introduces multi-source dynamic boundary loads:

[0060] The flow disturbance was established using a three-dimensional stochastic wind field generation program. The input was based on the Kaimal spectrum and the Coherence model, generating a spatially correlated flow time series. A three-component flow vector was applied at the inflow boundary.

[0061] Turbulent disturbances are generated by superimposing cosine waves with multiple frequency components, reflecting flow rate and random fluctuations.

[0062] The centrifugal inertial load is calculated using the fan operating curve, taking into account the combined effects of speed and pitch angle changes, and acts on the concentrated mass node at the impeller root.

[0063] Ground disturbances were constructed using accelerometer measured data to create a six-degree-of-freedom acceleration sequence for the base, and forced displacement was applied through the rigid bottom surface nodes.

[0064] The motor load disturbance is set as a torque function that varies with time to simulate grid-connected load fluctuations.

[0065] S4. Construct a digital twin model and synchronize it with the state:

[0066] Deploy an integrated sensing system on the actual wind turbine platform:

[0067] Strain gauges were placed on wall panels, oil tanks, shells, etc., to collect structural stress and displacement responses;

[0068] Thermocouples are placed at the stator and rotor of the motor to capture the temperature rise of key components;

[0069] Hall sensors are deployed on the air gap and the magnetic yoke to measure changes in magnetic flux density;

[0070] An 8-channel MEMS microphone array was installed inside and outside the housing for sound pressure spectrum analysis.

[0071] All signals are amplified, filtered, and sampled by the on-site signal conditioning device, and preprocessed by the edge computing module;

[0072] The state mapping function is used to interpolate and map sensor data to the nodes and boundary regions of the simulation model for twin synchronization.

[0073] The synchronization cycle is set to 1 second. Residual evaluation is performed before each round of synchronization update. If the virtual and real state deviation exceeds the set threshold, model fine-tuning and heat source and load updates are triggered.

[0074] A digital twin is a high-fidelity mapping model of a physical object in virtual space, evolving synchronously with the actual object through real-time data. In this embodiment, the digital twin model is not merely a simple simulation copy, but a dynamic mapping entity with state awareness, real-time updates, and behavior prediction capabilities.

[0075] To achieve efficient coupling between the real wind turbine and the simulation model, the system was designed with the following key components:

[0076] Physical-virtual mapping mechanism: A correspondence table is established between each simulation model node and the sensor acquisition point. For example, the output of the wall panel strain gauge is mapped to the corresponding unit node on the structural model; the air gap magnetic induction intensity data is mapped to the corresponding magnetic flux channel in the electromagnetic field simulation mesh.

[0077] Edge computing support: Given the large number of sensors and high update frequency, the raw signals are first preprocessed by edge computing modules (such as NVIDIA Jetson or Intel Movidius), including filtering, anomaly removal and compression, to ensure the quality and timeliness of the data uploaded to the cloud simulation engine.

[0078] State residual-driven adaptive adjustment: During the synchronization cycle of once per second, the system first calculates the "virtual-real residual", which is the deviation between sensor data and simulation output (such as predicted temperature rise vs. measured temperature rise). If it exceeds a threshold, such as stress deviation >8%, the local fine-tuning mechanism of the simulation model is triggered, such as local mesh refinement and heat source power correction.

[0079] Objective: Dynamic closed-loop control and prediction: Through this mechanism, digital twins can not only reflect the "current state" but also be used to predict "future risks" in advance, such as identifying potential fatigue damage areas or overheating trends.

[0080] S5. Multiphysics joint solution:

[0081] In the co-simulation process, the structural field serves as the main control domain, and all other physical domains update their boundary conditions based on the current structural state after each round of structural solution.

[0082] The aerodynamic field reconstructs the velocity boundary based on the impeller surface after structural deformation;

[0083] The thermal field updates the heat flux distribution area based on structural displacement;

[0084] The electromagnetic field adjusts the air gap by deforming the magnetic slit;

[0085] The sound field updates the propagation path and reflecting surface based on deformation.

[0086] The solution employs a two-layer nested iterative structure: the inner layer solves the physical domain, and the outer layer calculates the residuals of the coupled variables. A residual threshold of 0.1% is set, and the relaxation factor is dynamically adjusted from 0.5 to 0.8, allowing a maximum of 30 rounds of coupled iterations.

[0087] Multiphysics digital twins are not merely a stack of multiple simulation modules, but a strongly coupled, feedback-based collaborative solution system. This step utilizes the digital twin model to capture the interconnected feedback of structural behavior in aerodynamics, thermal, electromagnetic, and acoustic domains, and dynamically adjusts the boundary conditions and initial states for solving each physical domain.

[0088] The reason for using structure as the primary control domain is that during the operation of the fan, structural deformation has a direct impact on airflow path, electromagnetic gap, cooling efficiency, and noise propagation path. Therefore, the structural field is the primary control domain, and the input of other physical domains is dynamically adjusted through structural response.

[0089] Explanation of nested solution logic:

[0090] Inner iteration: Each physics field is solved individually until its internal variables converge;

[0091] Outer layer iteration: After updating the boundary through interactive updates of the physical field results, check the "coupling residuals", such as the difference between the air gap used in the magnetic field calculation and the air gap in the structural calculation results;

[0092] Iterative control: If the residual is less than 0.1%, the coupling is considered complete; otherwise, the next nested loop is executed. The relaxation factor ensures stable convergence and prevents oscillations.

[0093] The advantages of digital twins are evident: all solution processes are based on the latest sensor data to dynamically adjust boundary conditions, rather than static initial values. Therefore, they have the ability to respond to real-time changes, dynamically predict results, and provide feedback corrections, which greatly improves the realism and engineering value of the model.

[0094] S6. Dynamic performance index extraction and normalization:

[0095] After the simulation, the following indicators were extracted and normalized:

[0096] Structural stress: Extracting the Von-Mises equivalent stress at the connection between the blade root and the base;

[0097] Modal frequencies: Record the frequency differences of the first three modes under loaded and unloaded conditions respectively;

[0098] Heat accumulation: Extracting the total temperature rise inside the rotor and the change in heat dissipation rate;

[0099] Electromagnetic disturbance: Calculate the maximum rate of change of magnetic flux density;

[0100] Sound pressure level: Normalized after integrating the sound pressure spectrum at 6 points within a 30m range.

[0101] The index normalization interval is [0,1], which is used to unify the input optimization module.

[0102] This stage is the core bridge for digital twins to achieve "performance mapping" and "design feedback," extracting multi-dimensional indicators that can be used for subsequent optimization through the relationship between structure, state, and performance.

[0103] Indicator Selection and Physical Meaning: Each indicator corresponds to a key engineering performance characteristic.

[0104] Von-Mises stress corresponds to structural safety; modal frequency variation corresponds to dynamic stability; temperature rise and heat dissipation rate correspond to thermal stability; magnetic flux density change rate corresponds to electromagnetic interference and heat generation; sound pressure level corresponds to noise pollution and comfort.

[0105] The significance and method of normalization: Since the original indicators have different dimensions (such as MPa, Hz, K, dB) and cannot be directly compared, the min-max normalization method is used to uniformly map them to the [0,1] interval, facilitating the construction of the objective function. The normalized data can also serve as a standard vector input for digital twin evaluation feedback, enabling rapid comparison with design targets, historical data, or warning thresholds.

[0106] Integration with digital twins: Digital twin systems can continuously update these metrics and compare them with historical simulation data to identify performance degradation trends or abnormal deviations, provide early warnings or trigger model reconstruction, and enhance the system's intelligent perception and autonomous judgment capabilities.

[0107] S7. Optimize feedback generation and design parameter output:

[0108] A five-objective linear weighted optimization function is established, where the weights for structural stress, modal frequency, thermal accumulation, electromagnetic disturbance, and sound pressure level are 30%, 20%, 20%, 15%, and 15%, respectively. Design variables include:

[0109] Impeller length, wall thickness, root connection radius; base segment taper ratio; number of motor pole pairs and air gap length; cooling channel layout angle.

[0110] A genetic algorithm was used with an initial population of 100 groups and a maximum of 50 iterations. Structural and manufacturing constraints were introduced (e.g., the maximum impeller length should not exceed 80m and the air gap should not be less than 3mm). Ten optimal solutions were selected from each generation for updating, and finally, a set of optimized wind turbine design parameters was output and its performance improvement effect was verified.

[0111] In this embodiment, a genetic algorithm is used for multi-objective evolutionary optimization during the solution of wind turbine optimization parameters. The initial population size is set to 100 groups, each containing 12 design variables, covering impeller length, profile thickness distribution function coefficient, number of base segments and taper ratio, number of motor pole pairs, air gap length and cooling channel arrangement angle, etc. The variable initialization range is set according to the wind turbine design specifications.

[0112] In each iteration, the overall fitness of all individuals is calculated. The fitness function consists of five normalized performance indices: structural stress margin, modal frequency safety distance, thermal accumulation coefficient, electromagnetic disturbance intensity, and dominant frequency sound pressure level. These indices are combined linearly with weighted averages to form the overall objective function.

[0113] Multiple constraints were introduced during the optimization process to ensure the engineering feasibility of the solution, including structural dimension constraints (impeller length not exceeding 80 meters, maximum base cross-section not exceeding 6 meters), electromagnetic constraints (air gap length not less than 3 millimeters), and manufacturing process constraints (continuity of structural thickness variation not less than second-order polynomial smoothness). For individuals that did not meet the constraints, a penalty function strategy was used to weaken their fitness.

[0114] In each generation of optimization, an elite retention strategy is used to select the top 10 optimal solutions as the core population for the next generation, and the remaining individuals are generated through crossover and mutation operators. The iteration limit is set to 50 generations or the process automatically stops when the optimal fitness changes by less than 0.01% for 10 consecutive generations.

[0115] The final optimal solution is re-input into the simulation model for verification, and its performance improvement and reliability improvement under multiphysics are evaluated.

[0116] This embodiment employs a genetic algorithm for multi-objective evolutionary optimization during the wind turbine optimization parameter solution process. The initial population size is set to 100 groups, each containing 12 design variables, covering impeller length, profile thickness distribution function coefficient, number of base segments and taper ratio, number of motor pole pairs, air gap length, and cooling channel arrangement angle, etc. The variable initialization range is set according to the wind turbine design specifications. In each iteration, the overall fitness of all individuals is calculated. The fitness function consists of five normalized performance indices: structural stress margin, modal frequency safety distance, thermal accumulation coefficient, electromagnetic disturbance intensity, and dominant frequency sound pressure level. These are linearly weighted and combined to form the overall objective function.

[0117] Multiple constraints were introduced during the optimization process to ensure the engineering feasibility of the solution, including structural dimension constraints (impeller length not exceeding 80 meters, maximum base cross-section not exceeding 6 meters), electromagnetic constraints (air gap length not less than 3 millimeters), and manufacturing process constraints (continuity of structural thickness variation not less than second-order polynomial smoothness). For individuals that did not meet the constraints, a penalty function strategy was used to weaken their fitness.

[0118] In each generation of optimization, an elite retention strategy is used to select the top 10 optimal solutions as the core population for the next generation, and the remaining individuals are generated through crossover and mutation operators. The upper limit of iteration is set to 50 generations or the change in optimal fitness is less than 1×10 for 10 consecutive generations. -4 The simulation will automatically stop. The final optimal solution is then re-input into the simulation model for verification, evaluating its performance and reliability improvements under multiphysics conditions.

[0119] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A digital design method for the dynamic performance of a blower based on digital twins, characterized in that, Includes the following steps: S1. Construct a parametric structural model: Construct a parametric structural model of the wind turbine. The structural model includes components such as rotor, shell, wall panel, oil tank, hub, base and motor. By setting the geometric dimensions, material properties and boundary connection relationships of each component, assembly constraints and structural topology are established in a unified three-dimensional coordinate system. S2. Establish a multiphysics simulation model: Establish a multiphysics simulation model, which includes aerodynamic field, structural field, thermal field, electromagnetic field and acoustic field. Define the governing equations, boundary conditions and initial values ​​respectively, and set up coupling interfaces in the boundary region to realize variable interaction. S3. Input multi-source dynamic boundary loads: Construct multi-source dynamic boundary loads under wind turbine operating conditions. The multi-source dynamic boundary loads include flow disturbance, turbulent impact, inertial force, gravity change and base disturbance. The loads are generated and applied to the boundary of the simulation model through time domain function and spatial mapping function. S4. Constructing a digital twin model and synchronizing it with the state: Establish a digital twin model of the wind turbine. The model synchronously inputs the structural response, temperature field, electromagnetic characteristics and sound pressure information of the physical wind turbine into the simulation system through a state mapping function, and dynamically updates the simulation state according to the collected data. S5. Multiphysics joint solution: The multiphysics simulation model is jointly solved, with the structural field as the main control domain and the others as slave control domains. An iterative coupling solution strategy is adopted to control coupling error and ensure solution convergence. S6. Dynamic performance index extraction and normalization: Process the simulation output data to extract and normalize performance indices including stress distribution, modal frequency shift, heat accumulation degree, electromagnetic interference intensity and sound pressure level. S7. Optimization Feedback Generation and Design Parameter Output: Based on performance indicators, a multi-objective optimization function is constructed. Combining structural safety, thermal stability, electromagnetic compatibility, and acoustic environment requirements, a set of feedback control parameters is generated. The control structural parameters and boundary conditions are iteratively reconstructed, and finally, the optimized design parameters of the wind turbine are given.

2. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, In the parameterized structural model: The impeller's three-dimensional shape is defined by setting the profile curve, segment thickness, and local twist angle; The base adopts a segmented taper and wall thickness function to express a variable diameter structure; The housing defines the arrangement and connection sequence of each sub-component, while the motor defines the magnetic pole distribution and axial dimensions.

3. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, In the multiphysics simulation model, the aerodynamic field uses unsteady-state control equations to solve for pressure and velocity distribution, the structural field is modeled based on geometrically nonlinear shell elements, the thermal field uses steady-state heat conduction equations, the electromagnetic field uses vector potential control equations, and the acoustic field is simulated using the finite difference method for sound wave propagation.

4. The method for digital design of dynamic performance of a blower based on digital twin as described in claim 1, characterized in that, In the boundary load construction, the flow disturbance is generated into a disturbance sequence based on the measured flow rate and turbulence intensity function and applied to the impeller node; the inertial force is calculated by the change in rotational speed to determine the centrifugal load caused by the impeller rotation. The base disturbance is constructed from ground acceleration sensor data to form a dynamic input signal.

5. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, The digital twin model obtains the structural response through strain gauges, the temperature distribution through thermocouples, the magnetic flux density change through magnetic sensors, and the operating sound field characteristics through an acoustic sensor array. All types of data are processed and synchronously transmitted to the corresponding modules of the simulation model.

6. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, The joint solution adopts a strongly coupled strategy. After each round of solution, the structural field is passed to other fields to update the boundary conditions. The convergence is jointly controlled by residual constraints and relaxation factors. The maximum number of iterations does not exceed the set upper limit, and the error threshold is lower than the set boundary.

7. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, Among the performance indicators, the structural stress index is obtained by extracting the maximum value node from the equivalent stress cloud map, the modal frequency offset is normalized by the frequency difference under loaded and unloaded conditions, the heat accumulation index is calculated by the temperature rise integral and specific heat product, the electromagnetic interference index is obtained by combining the magnetic flux change rate with the current density change, and the sound pressure level is calculated by the root mean square value of the sound pressure at the simulated receiving point.

8. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, In step S7, the evaluation function is constructed by combining various performance indicators in a linear weighted manner. Each indicator is given a normalized interval and a constraint range. The structural strength shall not exceed 70% of the material yield strength, the modal frequency shall be at least 10% away from the excitation frequency, the maximum temperature rise shall be less than 85% of the material's allowable limit, the magnetic flux change intensity shall be consistent with the environmental standard, and the sound pressure level shall be less than the limit at 30 meters.

9. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, In step S7, the feedback control parameter set includes impeller length, base wall thickness, shell layout parameters, motor pole pitch, boundary load amplitude and disturbance spectrum. The optimization algorithm is based on a genetic evolution mechanism, generating a new parameter combination in each round to update the structural model and simulation model.

10. The method for digital design of dynamic performance of a blower based on digital twins according to claim 1, characterized in that, In step S7, the optimization process iteration termination conditions include: the objective function value has not improved within three consecutive rounds, or all performance indicators simultaneously meet the preset constraint range. The final optimized design parameters are output to the design platform for subsequent wind turbine prototype modeling and verification analysis.

Images