Method for predicting rotor aerodynamic force based on DMD and prediction system

Abstract

The invention belongs to the field of aerodynamic analysis of unmanned aerial vehicles, and particularly relates to a method and system for predicting rotor aerodynamic force based on DMD, and the method comprises the following steps: obtaining key parameter variables of a rotor structure and blade aerodynamic performance, and setting a key parameter variation range including rotation; adopting a Latin hypercube sampling method to construct a uniformly distributed multi-dimensional sampling space in a parameter range; transient flow field data of each sample point in the sampling space is obtained through CFD simulation; based on the flow field time sequence data, utilizing a dynamic mode decomposition (DMD) algorithm to extract dominant mode features, and constructing a flow field order reduction model; transient flow fields under different parameter combinations are rapidly reconstructed through the model, and then rotor aerodynamic force, fuselage downward load and aerodynamic power are calculated. According to the method, on the basis that CFD precision is kept, rapid prediction and dynamic response analysis of rotor aerodynamic force are achieved, and the iteration efficiency and optimization capacity of overall parameter design of the tilt rotorcraft are remarkably improved.

Description

Technical Field

[0001] This invention belongs to the field of UAV aerodynamic analysis, specifically a method and prediction system for predicting rotor aerodynamic forces based on DMD. Background Technology

[0002] Tiltrotor aircraft, as a novel type of aircraft combining the vertical takeoff and landing and hovering capabilities of helicopters with the high-speed forward flight performance of fixed-wing aircraft, have shown broad application potential in fields such as military reconnaissance, material transport, and emergency rescue. However, during flight, complex aerodynamic interference effects exist between the rotor and fuselage of tiltrotor aircraft. Especially during hovering and low-speed forward flight, the rotor downwash strongly impacts the fuselage surface, causing significant downward loads on the fuselage, directly affecting flight quality, structural strength, and handling stability. Therefore, in the overall design phase of tiltrotor aircraft, accurately and efficiently predicting the impact of different rotor parameters (such as rotor radius, collective pitch, spacing, and center height) on rotor thrust, torque, and downward fuselage loads is of great significance for optimizing its aerodynamic layout and flight performance.

[0003] Currently, computational fluid dynamics (CFD) is the primary technique for analyzing rotor flow fields and aerodynamic characteristics. While high-precision CFD simulations can accurately capture flow field details, their high computational resource consumption and long processing time limit their application in rapid design and optimization under multiple parameters and operating conditions. Especially in the stages of overall parameter iteration and scheme comparison, traditional CFD methods struggle to meet the engineering requirements for rapid evaluation and real-time response.

[0004] In recent years, reduced-order model (ROM) techniques have provided a new approach for the rapid analysis of complex flow field problems. Among them, the Dynamic Mode Decomposition (DMD) method has been widely used in fluid dynamics modeling and prediction because it can extract dominant dynamic modes from unsteady flow field data and construct linear low-order models that can describe the evolution of the flow field. However, there are no reports in existing research on applying the DMD reduced-order modeling system to rotor / fuselage aerodynamic disturbance problems, especially in the rapid prediction of overall parameters of tiltrotor aircraft, where a complete, efficient, and accurate aerodynamic prediction method is still lacking.

[0005] Therefore, developing a rapid prediction method for rotor aerodynamics based on the DMD reduced-order model can significantly improve the efficiency of aerodynamic performance evaluation of tiltrotor aircraft under various design parameters while ensuring computational accuracy. This has important engineering application value for promoting the overall design and optimization process of tiltrotor aircraft. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for predicting rotor aerodynamic forces based on DMD, in order to solve the technical problem that traditional CFD methods cannot meet the requirements for rapid iterative design of overall parameters of tiltrotor aircraft. This method achieves rapid and accurate prediction of rotor aerodynamic forces and fuselage downward loads by integrating Latin hypercube sampling, CFD simulation, and DMD order reduction modeling techniques.

[0007] The technical solution adopted by the present invention to achieve the above objectives is: a method for predicting rotor aerodynamic forces based on DMD, comprising the following steps:

[0008] Step S1: Obtain the key parameter variables of the rotor structure and blade aerodynamic performance, and establish constraints based on the rotor configuration characteristics, including setting the variation range of key rotor parameters;

[0009] Step S2: Based on the set parameter range, a multidimensional sampling space is constructed using the Latin hypercube sampling method, and a set of uniformly distributed sample points are generated in the sampling space to cover the diversity of rotor parameters;

[0010] Step S3: Using fluid dynamics simulation software, solve the steady-state and transient flow fields at the sample points in the sampling space to obtain the velocity and pressure field data around the rotor;

[0011] Step S4: Based on the flow field time series data obtained from CFD simulation, extract the main modal features using the dynamic mode decomposition algorithm and establish a flow field order reduction model;

[0012] Step S5: Reconstruct the flow field distribution under different parameters using a flow field reduction model, and calculate the rotor aerodynamic force, fuselage downward load and aerodynamic power to achieve rapid prediction and dynamic response analysis of rotor aerodynamic force.

[0013] In step S1, the key parameter variables are determined based on the configuration constraints and operating characteristics of the tiltrotor aircraft, including: rotor radius, rotor center height, rotor pitch, blade twist angle, rotor collective pitch, and inflow velocity.

[0014] Among them, the rotor radius and rotor center height determine the flow field scale; the rotor spacing, blade twist angle and rotor collective pitch determine the disk load distribution; and the inflow velocity determines the aerodynamic response time.

[0015] The setting of the range of variation of the key parameters is specifically as follows: based on the rotorcraft design envelope and flight state envelope, the upper and lower limits of the parameters are determined through statistical analysis, and a multi-dimensional parameter space is generated based on mechanical structure constraints.

[0016] Step S4 includes the following steps:

[0017] Step S4-1: Define the input-output matrix, that is, define the parameter variables and key flow field information, extract the key flow field information, and form matrix X;

[0018] Step S4-2: Perform singular value decomposition on the matrix to extract the main features of the data;

[0019]

[0020] Where U is a left singular matrix, It is a singular value diagonal matrix. The conjugate transpose of a right singular matrix;

[0021] Step S4-3: Construct the linear mapping matrix and calculate the low-dimensional approximate linear mapping matrix;

[0022] Step S4-4: Perform eigenvalue decomposition on the low-dimensional approximation linear mapping matrix, i.e.:

[0023]

[0024] Where W is the eigenvector matrix, It is a diagonal matrix, and the elements on its diagonal are eigenvalues;

[0025] Step S4-5: Reconstruct the flow field and construct a reduced-order flow field model in the sampling space based on the basic eigenvectors.

[0026] Step S4-1 specifically includes:

[0027] Step S4-1-1: Define the mapping relationship between parameter variables and key flow field information, and obtain key flow field information based on transient flow field time series data obtained from CFD simulation;

[0028] Step S4-1-2: Spatial discretization and vectorization of the physical field data of the entire flow field or region of interest at each simulation time step, that is, flattening the flow field data of each time step into a column vector;

[0029] Step S4-1-3: Combine all these column vectors arranged in chronological order into a high-dimensional data matrix X, that is:

[0030] ;

[0031] in This represents the flow field snapshot at the k-th time step, where m is the total number of time steps.

[0032] The high-dimensional data matrix X is used as input for dynamic mode decomposition to capture the dynamic characteristics of the flow field as it evolves over time.

[0033] In step S4-3, the construction of the linear mapping matrix specifically involves:

[0034] Step S4-3-1: Define the data matrix:

[0035] Divide the original data matrix Remove the last data snapshot, and Remove the first data snapshot to establish relationships between data from adjacent time points;

[0036] Step S4-3-2: Construct the linear mapping matrix:

[0037] Based on the assumption of local linear approximation, there exists a linear mapping matrix A such that... ;

[0038] By utilizing the relationship between the results of singular value decomposition and the data matrix, we can obtain... ,in, This indicates a pseudo-inverse operation;

[0039] In step S4-3, the calculation of the low-dimensional approximate linear mapping matrix is ​​as follows:

[0040] Calculate the truncated low-dimensional approximate linear mapping matrix ,Right now:

[0041]

[0042] in, Let U be the first r orders of the matrix, i.e., the truncated left singular vector matrix. It means The conjugate transpose of .

[0043] Steps S4-5 are specifically as follows:

[0044] Step S4-5-1: Based on the obtained eigenvector matrix W and diagonal eigenvalue matrix Λ, reconstruct the dynamic modes of the flow field; the reconstructed eigenvectors are obtained by calculating the DMD mode matrix. To achieve this, its expression is:

[0045] ;

[0046] in, The original data matrix after removing the first time-step snapshot is the basis eigenvector of the reduced-order model. Let V be the truncation matrix formed by the first r columns of the right singular matrix V in the singular value decomposition. W is a truncated diagonal matrix consisting of the first r rows and r columns of the singular value matrix Σ; W is a low-dimensional approximate linear mapping matrix. The feature vector matrix; each DMD mode This corresponds to a specific dynamic frequency and growth rate;

[0047] Step S4-5-2: Utilize the calculated DMD modes Using the eigenvalue Λ, a reduced-order model of the flow field evolution over time is constructed; the flow field snapshot at any time k is approximately reconstructed by a linear combination of DMD modes, i.e.:

[0048] ;

[0049] in, It is the initial amplitude vector, obtained by projecting an initial flow field snapshot onto the DMD mode;

[0050] Step S4-5-3: Verify the accuracy of the model by comparing the error between the flow field reconstructed by the reduced-order model and the original CFD simulation flow field; and determine the optimal truncation order r based on the energy contribution rate of the singular value spectrum or the preset error tolerance, thus forming the flow field reduced-order model.

[0051] Step S5 includes the following steps:

[0052] Step S5-1: Substitute the given new key parameter combination into the flow field reduction model constructed based on DMD modes to obtain the transient velocity field and pressure field sequence under these parameters within one or more rotor rotation cycles;

[0053] Step S5-2: Calculation of rotor aerodynamic forces and torques: Based on the reconstructed transient pressure field, at each time step, pressure integration is performed on the surface of all rotor blades to solve for the aerodynamic loads acting on each blade, and vector synthesis is used to obtain the instantaneous thrust, drag and torque of the entire rotor.

[0054] Calculation of fuselage downward load: Also based on the reconstructed pressure field, at each time step, the pressure is integrated on the upper surface of the fuselage, and the vertical components of the integrated normal force are superimposed to obtain the instantaneous downward load acting on the fuselage.

[0055] Step S5-3: Aerodynamic Power Calculation: Using the rotor torque calculated in Step S5-2, combined with the rotor speed, the aerodynamic power is calculated using the formula... Calculate the instantaneous aerodynamic power consumed by the rotor, where M is the torque and Ω is the rotational angular velocity;

[0056] Step S5-4: Time-average the transient thrust, downward load, and power data obtained in steps S5-2 and S5-3 for a complete cycle to obtain their steady-state values; at the same time, analyze their pulsation range and frequency to evaluate the dynamic response characteristics of the rotor aerodynamics.

[0057] A DMD-based rotor aerodynamic prediction system is provided to implement the DMD-based rotor aerodynamic prediction method described in claim 1. The system comprises:

[0058] The parameter definition and sampling module is used to obtain key parameter variables of rotor structure and blade aerodynamic performance, establish constraints based on rotor configuration characteristics and set parameter variation range; the module has a built-in Latin hypercube sampling algorithm to construct a multi-dimensional sampling space based on the set parameter range and generate a set of uniformly distributed sample points.

[0059] The computational fluid dynamics simulation module, connected to the parameter definition and sampling module, is used to receive sample point data within the sampling space. This module integrates or calls fluid dynamics simulation software to perform steady-state and transient flow field solutions for each sample point and outputs velocity field, pressure field, and time series data around the rotor.

[0060] The dynamic mode decomposition and order reduction model construction module, connected to the CFD simulation module, is used to receive flow field time series data and perform the following operations:

[0061] Snapshot extraction and matrixing of flow field time series data are performed to form a high-dimensional data matrix;

[0062] Singular value decomposition is performed on the data matrix to extract key features;

[0063] Construct a linear mapping matrix and compute its low-dimensional approximation;

[0064] Perform eigenvalue decomposition on the low-dimensional linear mapping matrix to calculate the DMD modes;

[0065] A reduced-order model of flow field evolution over time is constructed based on DMD modes and eigenvalues, and its core expression is:

[0066] A rapid aerodynamic prediction and analysis module is connected to the reduced-order model construction module; this module includes:

[0067] The flow field reconstruction unit is used to input new combinations of key parameters into the flow field reduction model to quickly generate transient velocity and pressure field sequences.

[0068] The load calculation unit is used to perform integral calculations on the reconstructed pressure field to solve for the instantaneous thrust, drag, torque of the rotor, and downward load on the fuselage.

[0069] The power calculation unit is used to calculate instantaneous aerodynamic power based on rotor torque and rotational speed;

[0070] The data analysis unit is used to perform time averaging on transient calculation results to obtain steady-state values ​​and to analyze the dynamic response characteristics of aerodynamic forces.

[0071] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a method for predicting rotor aerodynamics based on DMD as described above.

[0072] The present invention has the following beneficial effects and advantages:

[0073] 1. This invention combines high-fidelity CFD simulation with DMD reduced-order modeling, which not only retains the accuracy advantages of traditional CFD methods, but also improves the flow field prediction speed by several orders of magnitude by using the reduced-order model, realizing the "second-level" rapid prediction of rotor aerodynamics, effectively supporting parameter iteration and scheme selection in the early stage of design.

[0074] 2. The Latin hypercube sampling method used in this invention can maximize the coverage of the diversity of parameter combinations in a multidimensional parameter space with the fewest number of sample points, which greatly reduces the number of basic CFD simulations required to build the reduced-order model, thereby significantly saving computing resources and time costs.

[0075] 3. The present invention employs a dynamic mode decomposition algorithm to extract the dominant dynamic modes from unsteady flow field data. Therefore, the constructed reduced-order model can not only predict time-averaged aerodynamic forces, but also accurately reconstruct the transient evolution process of the flow field and analyze the pulsation characteristics and frequency components of aerodynamic loads, providing key data for evaluating the aerodynamic stability and dynamic response of the rotor.

[0076] 4. The parameterized reduced-order model constructed in this invention can quickly respond to continuous changes in key design parameters such as rotor radius, collective pitch, and spacing, providing an efficient and universal analysis tool for aerodynamic performance evaluation and multi-objective optimization of tiltrotor aircraft in multiple flight states (such as hovering, transition, and forward flight).

[0077] 5. This invention can be integrated into existing aircraft design processes or automated optimization platforms to form a complete closed loop from parameter sampling and CFD calculation to model building and rapid prediction, which greatly improves the automation and intelligence level of the overall aerodynamic design of tiltrotor aircraft. Attached Figure Description

[0078] Figure 1 The flowchart of the method for predicting rotor aerodynamics based on DMD of the present invention is shown in the figure. Detailed Implementation

[0079] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of the embodiments of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0080] This invention provides a method for predicting rotor aerodynamics based on DMD, the overall process of which can be found in [reference needed]. Figure 1 As shown, the main steps include:

[0081] Step S1: Determine the key parameter variables and constraints, and set the parameter variation range.

[0082] First, based on the specific configuration and operating conditions of the tiltrotor aircraft, the key design parameters affecting rotor performance and fuselage aerodynamic loads are determined. These parameters mainly include: rotor radius (R), rotor center height (H), rotor pitch (D), and blade twist angle (R / H). ), rotor collective pitch ( ) and the inflow velocity ( ).

[0083] Among them, the rotor radius and rotor center height together determine the scale and spatial range of the flow field calculation; the rotor pitch, blade twist angle and rotor collective pitch directly affect the load distribution characteristics on the rotor disk; and the inflow velocity determines the aerodynamic response time scale of the system.

[0084] Subsequently, based on the rotorcraft's design envelope and expected flight state envelope, the upper and lower limits of the aforementioned key parameters are determined through statistical analysis or engineering experience. Simultaneously, physical constraints such as the feasibility of the mechanical structure must be comprehensively considered, ultimately generating a multi-dimensional parameter design space.

[0085] Step S2: Construct a multidimensional sampling space using the Latin hypercube sampling method.

[0086] To efficiently generate representative sample points within the multidimensional parameter space determined in step S1, this invention employs the Latin Hypercube Sampling (LHS) method. LHS is a stratified random sampling technique whose core idea is to achieve uniform coverage of the parameter space even with a small sample size. The specific steps are as follows:

[0087] S2-1: Divide the range of variation of each key parameter variable (i.e. each dimension) independently and equally into m non-overlapping intervals.

[0088] S2-2: Randomly select a sample point within each interval of each dimension parameter.

[0089] S2-3: Map all randomly selected sample points to conform to a normal distribution by applying the inverse function of the standard normal distribution.

[0090] S2-4: Randomly select a sample point from each dimension parameter after the mapping in step S2-3, and combine these sample points from different dimensions to form a sample vector.

[0091] S2-5: Sort the generated sample vectors according to their numerical values ​​to form an X-dimensional sample space containing n key parameter variables, which is the required sampling space.

[0092] Step S3: Perform CFD numerical simulation to obtain flow field time series data.

[0093] Using computational fluid dynamics (CFD) software (such as Fluent, Star-CCM+, OpenFOAM, etc.), numerical simulations are performed on the rotor configuration represented by each sample point in the sampling space of step S2. The simulations must include both steady-state and transient calculations to obtain detailed evolution data of the flow field around the rotor over one or more rotation cycles. Finally, the velocity and pressure field data corresponding to each sample point, arranged in time series, are output as inputs for subsequent dynamic mode decomposition.

[0094] Step S4: Construct a flow field order reduction model based on the DMD algorithm.

[0095] Step S4 is the core of this invention, which uses Dynamic Mode Decomposition (DMD) to reduce the order of the transient flow field data obtained from CFD simulation. The specific process is as follows:

[0096] Step S4-1: Define the input-output matrix, that is, define the parameter variables and key flow field information, extract the key flow field information, and form matrix X;

[0097] Step S4-1 is as follows:

[0098] Extract flow field snapshot data from all time steps obtained from CFD simulation. For each time step, spatially discretize the physical quantities (such as velocity or pressure) in the entire flow field or region of interest, and flatten them into a column vector. Arrange the column vectors of all m time steps in chronological order to form a high-dimensional data matrix. .

[0099] Step S4-2: Perform singular value decomposition on the matrix to extract the main features of the data;

[0100]

[0101] Where U is a left singular matrix, It is a singular value diagonal matrix. The conjugate transpose of a right singular matrix;

[0102] Step S4-3: Construct the linear mapping matrix and calculate the low-dimensional approximate linear mapping matrix;

[0103] (1) Construct the linear mapping matrix, specifically:

[0104] Step S4-3-1: Define the data matrix:

[0105] Divide the original data matrix Remove the last data snapshot, and Remove the first data snapshot to establish relationships between data from adjacent time points;

[0106] Step S4-3-2: Construct the linear mapping matrix:

[0107] Based on the assumption of local linear approximation, there exists a linear mapping matrix A such that... ;

[0108] By utilizing the relationship between the results of singular value decomposition and the data matrix, we can obtain... ,in, This indicates a pseudo-inverse operation;

[0109] (2) Calculate the low-dimensional approximate linear mapping matrix:

[0110] Calculate the truncated low-dimensional approximate linear mapping matrix ,Right now:

[0111]

[0112] in, Let U be the first r orders of the matrix, i.e., the truncated left singular vector matrix. It means The conjugate transpose of .

[0113] Step S4-4: Perform eigenvalue decomposition on the low-dimensional approximation linear mapping matrix, i.e.:

[0114]

[0115] Where W is the eigenvector matrix, It is a diagonal matrix, and the elements on its diagonal are eigenvalues;

[0116] Step S4-5: Reconstruct the flow field and construct a reduced-order flow field model in the sampling space based on the basic eigenvectors.

[0117] Step S4-5-1: Calculate the DMD mode matrix :

[0118] ;

[0119] in, and These are the first r columns / first r order truncation matrices of V and Σ, respectively.

[0120] Step S4-5-2: Construct a reduced-order flow field model. A snapshot of the flow field at any time k. It can be reconstructed by linear approximation of DMD modes:

[0121] ;

[0122] in, It is the initial amplitude vector, obtained by taking the initial snapshot. Projecting onto the DMD mode yields ( ).

[0123] Step S4-5-3: Verify the model accuracy by comparing the error between the reconstructed flow field and the original CFD flow field. Based on the singular value energy contribution rate or a preset error tolerance, determine the optimal truncation order r to form the final flow field order reduction model.

[0124] Step S5: Rapidly predict aerodynamic performance based on the reduced-order model.

[0125] Step S5-1: Input a new set of key parameters into the constructed flow field reduction model to quickly reconstruct the transient velocity field and pressure field sequence within at least one complete rotor rotation cycle under these parameters.

[0126] Step S5-2: Load Calculation

[0127] Rotor aerodynamic forces and torques: At each time step, the reconstructed transient pressure field is integrated on the surface of all rotor blades to solve for the aerodynamic load of each blade, and then vectorized to form the instantaneous thrust, drag and torque of the entire rotor.

[0128] Downward load on fuselage: At each time step, the pressure field of the fuselage (especially the upper surface) is integrated, and the components of the normal force in the vertical direction are superimposed to obtain the instantaneous downward load acting on the fuselage.

[0129] Step S5-3: Aerodynamic Power Calculation: Using the rotor torque (M) calculated in step S5-2 and the known rotor speed (Ω), the aerodynamic power is calculated using the formula... Calculate the instantaneous aerodynamic power consumed by the rotor.

[0130] Step S5-4: Data Analysis: Time-average the transient thrust, downward load, and power data over a complete cycle to obtain their steady-state values. Simultaneously, analyze the pulsation range and dominant frequency to evaluate the dynamic response characteristics of the rotor aerodynamics.

[0131] This embodiment also provides a system for implementing the above method, the system comprising:

[0132] The parameter definition and sampling module is used to obtain key parameter variables of rotor structure and blade aerodynamic performance, establish constraints based on rotor configuration characteristics and set parameter variation range; the module has a built-in Latin hypercube sampling algorithm to construct a multi-dimensional sampling space based on the set parameter range and generate a set of uniformly distributed sample points.

[0133] The computational fluid dynamics simulation module, connected to the parameter definition and sampling module, is used to receive sample point data within the sampling space. This module integrates or calls fluid dynamics simulation software to perform steady-state and transient flow field solutions for each sample point and outputs velocity field, pressure field, and time series data around the rotor.

[0134] The dynamic mode decomposition and order reduction model construction module, connected to the CFD simulation module, is used to receive flow field time series data and perform the following operations:

[0135] Snapshot extraction and matrixing of flow field time series data are performed to form a high-dimensional data matrix;

[0136] Singular value decomposition is performed on the data matrix to extract key features;

[0137] Construct a linear mapping matrix and compute its low-dimensional approximation;

[0138] Perform eigenvalue decomposition on the low-dimensional linear mapping matrix to calculate the DMD modes;

[0139] A reduced-order model of flow field evolution over time is constructed based on DMD modes and eigenvalues, and its core expression is:

[0140] A rapid aerodynamic prediction and analysis module is connected to the reduced-order model construction module; this module includes:

[0141] The flow field reconstruction unit is used to input new combinations of key parameters into the flow field reduction model to quickly generate transient velocity and pressure field sequences.

[0142] The load calculation unit is used to perform integral calculations on the reconstructed pressure field to solve for the instantaneous thrust, drag, torque of the rotor, and downward load on the fuselage.

[0143] The power calculation unit is used to calculate instantaneous aerodynamic power based on rotor torque and rotational speed;

[0144] The data analysis unit is used to perform time averaging on transient calculation results to obtain steady-state values ​​and to analyze the dynamic response characteristics of aerodynamic forces.

[0145] Furthermore, the present invention also relates to a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, enables the implementation of the above-described method for predicting rotor aerodynamics based on DMD.

[0146] In summary, this invention constructs a complete rapid prediction system for rotor aerodynamics by integrating advanced technologies such as Latin hypercube sampling, CFD simulation, and dynamic mode decomposition. This method, while maintaining the accuracy of traditional CFD calculations, achieves rapid prediction and dynamic characteristic analysis of rotor thrust, torque, and fuselage downward loads, effectively overcoming the bottlenecks of low computational efficiency and difficulty in supporting rapid design iterations inherent in traditional methods. The implementation of this invention provides an efficient and reliable technical means for optimizing the overall parameters of tiltrotor aircraft, significantly improving design efficiency and quality, and possessing significant engineering application value and promising prospects in the field of aircraft aerodynamic design.

[0147] Those skilled in the art will understand that the above description is merely a preferred embodiment of the present invention, and the features described in the various embodiments and / or claims of this disclosure can be combined or combined in various ways, even if such combinations or combinations are not explicitly described in this disclosure. This is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0148] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention. Clearly, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if these modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include these modifications and modifications.

Claims

1. A method for predicting rotor aerodynamics based on DMD, characterized in that, Includes the following steps: Step S1: Obtain the key parameter variables of the rotor structure and blade aerodynamic performance, and establish constraints based on the rotor configuration characteristics, including setting the variation range of key rotor parameters; Step S2: Based on the set parameter range, a multidimensional sampling space is constructed using the Latin hypercube sampling method, and a set of uniformly distributed sample points are generated in the sampling space to cover the diversity of rotor parameters; Step S3: Using fluid dynamics simulation software, solve the steady-state and transient flow fields at the sample points in the sampling space to obtain the velocity and pressure field data around the rotor; Step S4: Based on the flow field time series data obtained from CFD simulation, extract the main modal features using the dynamic mode decomposition algorithm and establish a flow field order reduction model; Step S5: Reconstruct the flow field distribution under different parameters using a flow field reduction model, and calculate the rotor aerodynamic force, fuselage downward load and aerodynamic power to achieve rapid prediction and dynamic response analysis of rotor aerodynamic force.

2. The method for predicting rotor aerodynamics based on DMD according to claim 1, characterized in that, In step S1, the key parameter variables are determined based on the configuration constraints and operating characteristics of the tiltrotor aircraft, including: rotor radius, rotor center height, rotor pitch, blade twist angle, rotor collective pitch, and inflow velocity. Among them, the rotor radius and rotor center height determine the flow field scale; the rotor spacing, blade twist angle and rotor collective pitch determine the disk load distribution; and the inflow velocity determines the aerodynamic response time. The setting of the range of variation of the key parameters is specifically as follows: based on the rotorcraft design envelope and flight state envelope, the upper and lower limits of the parameters are determined through statistical analysis, and a multi-dimensional parameter space is generated based on mechanical structure constraints.

3. The method for predicting rotor aerodynamics based on DMD according to claim 1, characterized in that, Step S4 includes the following steps: Step S4-1: Define the input-output matrix, that is, define the parameter variables and key flow field information, extract the key flow field information, and form matrix X; Step S4-2: Perform singular value decomposition on the matrix to extract the main features of the data; ； Where U is a left singular matrix, It is a singular value diagonal matrix. The conjugate transpose of a right singular matrix; Step S4-3: Construct the linear mapping matrix and calculate the low-dimensional approximate linear mapping matrix; Step S4-4: Perform eigenvalue decomposition on the low-dimensional approximation linear mapping matrix, i.e.: ； Where W is the eigenvector matrix, It is a diagonal matrix, and the elements on its diagonal are eigenvalues; Step S4-5: Reconstruct the flow field and construct a reduced-order flow field model in the sampling space based on the basic eigenvectors.

4. The method for predicting rotor aerodynamics based on DMD according to claim 3, characterized in that, Step S4-1 specifically includes: Step S4-1-1: Define the mapping relationship between parameter variables and key flow field information, and obtain key flow field information based on transient flow field time series data obtained from CFD simulation; Step S4-1-2: Spatial discretization and vectorization of the physical field data of the entire flow field or region of interest at each simulation time step, that is, flattening the flow field data of each time step into a column vector; Step S4-1-3: Combine all these column vectors arranged in chronological order into a high-dimensional data matrix X, that is: ； in This represents the flow field snapshot at the k-th time step, where m is the total number of time steps. The high-dimensional data matrix X is used as input for dynamic mode decomposition to capture the dynamic characteristics of the flow field as it evolves over time.

5. The method for predicting rotor aerodynamics based on DMD according to claim 3, characterized in that, In step S4-3, the construction of the linear mapping matrix specifically involves: Step S4-3-1: Define the data matrix: Divide the original data matrix Remove the last data snapshot, and Remove the first data snapshot to establish relationships between data from adjacent time points; Step S4-3-2: Construct the linear mapping matrix: Based on the assumption of local linear approximation, there exists a linear mapping matrix A such that... ; By utilizing the relationship between the results of singular value decomposition and the data matrix, we can obtain... ,in, This indicates a pseudo-inverse operation.

6. The method for predicting rotor aerodynamics based on DMD according to claim 3, characterized in that, In step S4-3, the calculation of the low-dimensional approximate linear mapping matrix is ​​as follows: Calculate the truncated low-dimensional approximate linear mapping matrix ,Right now: ； in, Let U be the first r orders of the matrix, i.e., the truncated left singular vector matrix. It means The conjugate transpose of .

7. The method for predicting rotor aerodynamics based on DMD according to claim 3, characterized in that, Steps S4-5 are specifically as follows: Step S4-5-1: Based on the obtained eigenvector matrix W and diagonal eigenvalue matrix Λ, reconstruct the dynamic modes of the flow field; the reconstructed eigenvectors are obtained by calculating the DMD mode matrix. To achieve this, its expression is: ； in, The original data matrix after removing the first time-step snapshot is the basis eigenvector of the reduced-order model. Let V be the truncation matrix formed by the first r columns of the right singular matrix V in the singular value decomposition. W is a truncated diagonal matrix consisting of the first r rows and r columns of the singular value matrix Σ; W is a low-dimensional approximate linear mapping matrix. The feature vector matrix; each DMD mode This corresponds to a specific dynamic frequency and growth rate; Step S4-5-2: Utilize the calculated DMD modes Using the eigenvalue Λ, a reduced-order model of the flow field evolution over time is constructed; the flow field snapshot at any time k is approximately reconstructed by a linear combination of DMD modes, i.e.: ； in, It is the initial amplitude vector, obtained by projecting an initial flow field snapshot onto the DMD mode; Step S4-5-3: Verify the accuracy of the model by comparing the error between the flow field reconstructed by the reduced-order model and the original CFD simulation flow field; and determine the optimal truncation order r based on the energy contribution rate of the singular value spectrum or the preset error tolerance, thus forming the flow field reduced-order model.

8. The method for predicting rotor aerodynamics based on DMD according to claim 3, characterized in that, Step S5 includes the following steps: Step S5-1: Substitute the given new key parameter combination into the flow field reduction model constructed based on DMD modes to obtain the transient velocity field and pressure field sequence under these parameters within one or more rotor rotation cycles; Step S5-2: Calculation of rotor aerodynamic forces and torques: Based on the reconstructed transient pressure field, at each time step, pressure integration is performed on the surface of all rotor blades to solve for the aerodynamic loads acting on each blade, and vector synthesis is used to obtain the instantaneous thrust, drag and torque of the entire rotor. Calculation of fuselage downward load: Also based on the reconstructed pressure field, at each time step, the pressure is integrated on the upper surface of the fuselage, and the vertical components of the integrated normal force are superimposed to obtain the instantaneous downward load acting on the fuselage. Step S5-3: Aerodynamic Power Calculation: Using the rotor torque calculated in Step S5-2, combined with the rotor speed, the aerodynamic power is calculated using the formula... Calculate the instantaneous aerodynamic power consumed by the rotor, where M is the torque and Ω is the rotational angular velocity; Step S5-4: Time-average the transient thrust, downward load, and power data obtained in steps S5-2 and S5-3 for a complete cycle to obtain their steady-state values; at the same time, analyze their pulsation range and frequency to evaluate the dynamic response characteristics of the rotor aerodynamics.

9. A rotor aerodynamics prediction system based on DMD, used to implement the method for predicting rotor aerodynamics based on DMD as described in claim 1, characterized in that, The system includes: The parameter definition and sampling module is used to obtain key parameter variables of rotor structure and blade aerodynamic performance, establish constraints based on rotor configuration characteristics and set parameter variation range; the module has a built-in Latin hypercube sampling algorithm to construct a multi-dimensional sampling space based on the set parameter range and generate a set of uniformly distributed sample points. The computational fluid dynamics simulation module, connected to the parameter definition and sampling module, is used to receive sample point data within the sampling space. This module integrates or calls fluid dynamics simulation software to perform steady-state and transient flow field solutions for each sample point and outputs velocity field, pressure field, and time series data around the rotor. The dynamic mode decomposition and order reduction model construction module, connected to the CFD simulation module, is used to receive flow field time series data and perform the following operations: Snapshot extraction and matrixing of flow field time series data are performed to form a high-dimensional data matrix; Singular value decomposition is performed on the data matrix to extract key features; Construct a linear mapping matrix and compute its low-dimensional approximation; Perform eigenvalue decomposition on the low-dimensional linear mapping matrix to calculate the DMD modes; A reduced-order model of flow field evolution over time is constructed based on DMD modes and eigenvalues, and its core expression is: A rapid aerodynamic prediction and analysis module is connected to the reduced-order model construction module; this module includes: The flow field reconstruction unit is used to input new combinations of key parameters into the flow field reduction model to quickly generate transient velocity and pressure field sequences. The load calculation unit is used to perform integral calculations on the reconstructed pressure field to solve for the instantaneous thrust, drag, torque of the rotor, and downward load on the fuselage. The power calculation unit is used to calculate instantaneous aerodynamic power based on rotor torque and rotational speed; The data analysis unit is used to perform time averaging on transient calculation results to obtain steady-state values ​​and to analyze the dynamic response characteristics of aerodynamic forces.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements a method for predicting rotor aerodynamics based on DMD as described in any one of claims 1 to 8.

Images