A three-dimensional PIV-based non-contact dynamic testing method for propeller-fan aerodynamic performance

Abstract

The application provides a kind of three-dimensional PIV-based fan blade aerodynamic performance non-contact dynamic testing method, it is related to testing field, including the following steps: obtaining the real-time phase signal of rotating fan blade, and generating synchronous trigger instruction based on real-time phase signal;In the upstream of fan blade, the evenly distributed tracer particles are broadcast to the inflow area;In response to the synchronous trigger instruction, control the laser system to project the sheet light source to the fan blade measurement area, and control at least two cameras to shoot synchronously to obtain the illuminated flow field particle image;Combine real-time phase signal, three-dimensional space coordinate calibration is carried out to rotating fan blade measurement area, to correct the optical distortion caused by fan blade movement and window structure, to obtain calibration result;Based on flow field particle image and calibration result, the three-dimensional velocity vector distribution of fan blade flow field is calculated;According to three-dimensional velocity vector distribution, the aerodynamic performance parameters of fan blade are calculated.The high-precision real-time correction of dynamic optical distortion is realized.

Description

Technical Field

[0001] This application relates to the field of propeller testing technology, specifically to a non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV. Background Technology

[0002] In the research and performance evaluation of propellers (such as propellers and fans), accurately acquiring the velocity distribution of the surrounding three-dimensional flow field during their rotation is crucial. Particle image velocimetry (PIV), as a non-contact, full-field measurement method, has been attempted for application in such scenarios. However, when applied to the measurement of the external flow field of a high-speed rotating propeller, a fundamental technical challenge emerges: the dynamic and time-varying optical distortion introduced by the high-speed rotation of the propeller and its observation window (such as the optical window on the wind tunnel wall) severely disrupts the spatial coordinate reference of the three-dimensional PIV measurement, making it impossible to reconstruct a true and accurate three-dimensional velocity field from the acquired particle images.

[0003] Specifically, most existing technical solutions are based on static or pre-set calibration models, which cannot respond in real time to changes in the viewing angle, blade attitude, and phase variation caused by propeller rotation. This results in spatial distortion of the measurement results, especially in critical areas such as the blade tip, root, and wake, where errors are significant. The obtained velocity vector field has low reliability, making it difficult to support high-precision quantitative analysis of aerodynamic forces / torques and performance evaluation. This greatly limits the in-depth application of 3D PIV technology in advanced propeller design and fault diagnosis. Summary of the Invention

[0004] In view of the above-mentioned defects or deficiencies in the prior art, this application aims to provide a non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV, including the following steps: Acquire the real-time phase signal of the rotating propeller and generate a synchronization trigger command based on the real-time phase signal; Upstream of the propfan, tracer particles are uniformly distributed into the inflow region; In response to the synchronous trigger command, the laser system is controlled to project a sheet light source onto the paddle fan measurement area, and at least two cameras are controlled to capture images synchronously to obtain an image of the illuminated flow field particles. By combining the real-time phase signal, the three-dimensional spatial coordinates of the rotating propeller measurement area are calibrated to correct the optical distortion caused by the propeller motion and the window structure, and the calibration result is obtained. Based on the flow field particle image and the calibration results, the three-dimensional velocity vector distribution of the propeller flow field is calculated; Based on the three-dimensional velocity vector distribution, the aerodynamic performance parameters of the propeller are calculated.

[0005] According to the technical solution provided in this application, the step of calibrating the three-dimensional spatial coordinates of the rotating propeller measurement area by combining the real-time phase signal includes the following steps: Based on the three-dimensional geometric model of the propeller and the real-time phase signal, a virtual three-dimensional calibration field that completely corresponds to the current rotation phase and the spatial position of the blade is generated in real time. The laser system is controlled to project an auxiliary structured light pattern with specific coding features onto the paddle fan measurement area, and a calibration image containing the auxiliary structured light pattern is obtained by synchronously capturing the image with a camera. The auxiliary structured light feature points extracted from the calibration image are matched with the theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field; Based on the matching results, the comprehensive distortion field caused by propeller motion, window refraction and optical system is dynamically solved to establish an accurate mapping relationship between image coordinates and three-dimensional spatial coordinates under the current phase.

[0006] According to the technical solution provided in this application, matching the auxiliary structured light feature points extracted from the calibration image with the theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field includes the following steps: Based on the real-time phase signal, the auxiliary structured light feature points collected in each rotation cycle are divided into feature point subsets corresponding to each blade according to the blade passing frequency of the propeller. For each subset of the feature points, based on the pre-stored blade aeroelastic deformation model, the estimated elastic deformation of the corresponding blade under the current rotational speed and aerodynamic load is calculated. The theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field are superimposed with the estimated elastic deformation to generate the compensated theoretical projection points after blade deformation compensation. With the goal of minimizing the matching residual between the subset of feature points corresponding to each blade and the compensated theoretical projection points, the spatial pose and local deformation of each blade in the current phase are solved to achieve high-precision feature matching.

[0007] According to the technical solution provided in this application, the step of minimizing the matching residual between the subset of feature points corresponding to each blade and the compensated theoretical projection points, and solving for the spatial pose and local deformation of each blade in the current phase, includes the following steps: The swept area of ​​each blade is divided into several flow field grid cells along the radial and / or chordal direction, and a correlation model is established between the blade spatial pose parameters, local deformation parameters and the coordinate correction of each flow field grid cell. The objective function is to minimize the sum of squared weighted residuals between the measured coordinates of the subset of feature points and the compensated theoretical projected coordinates of all blades on all corresponding flow field grid cells. The weighting coefficients are dynamically adjusted according to the relative positions of the grid cells and the leading edge, trailing edge and tip region of the blade. The fixed support constraints at the blade root and the aerodynamic interference correlation between adjacent blades are used as physical constraints. These are substituted into the objective function, and a constrained nonlinear least squares algorithm is used for iterative solution. The precise spatial pose parameters, local deformation parameters, and target coordinate correction of each flow field grid cell are output simultaneously for the current phase of each blade.

[0008] According to the technical solution provided in this application, the step of seeding uniformly distributed tracer particles into the inflow region upstream of the paddle fan includes the following steps: The background flow field upstream of the propfan without particle injection is predicted to obtain the background flow field parameters of the inflow region, including the average velocity distribution and turbulence intensity distribution. Based on the average velocity distribution and turbulence intensity distribution, the particle seeding parameters are dynamically planned so that when the seeded particle cloud arrives at the propeller measurement plane, its spatial concentration distribution matches the background flow field parameters, thereby achieving uniform particle distribution within the propeller measurement area.

[0009] According to the technical solution provided in this application, controlling at least two cameras to simultaneously capture images to obtain illuminated flow field particle images includes the following steps: Keeping the camera parameters and the synchronization trigger command unchanged, only turning off the sheet light source of the laser system, controlling all cameras to synchronously capture the paddle fan scene in the same phase, and obtaining the background noise light intensity map corresponding to each camera; Control at least two cameras to capture images simultaneously, obtaining the initial particle image for each frame of the formal acquisition; For each frame of the initial particle image, based on the background noise intensity map of its corresponding camera and corresponding phase, the background noise pixels caused by the reflection of solid surface light from each propeller structure are identified and subtracted to obtain the flow field particle image.

[0010] According to the technical solution provided in this application, the process of identifying and deducting background noise pixels caused by solid surface reflections from each paddle fan component structure includes the following steps: For each background noise intensity map, an adaptive threshold map is generated based on the brightness statistical characteristics of the neighborhood of each pixel. The brightness value of each pixel in the initial particle image is compared with the set threshold in the adaptive threshold map at its corresponding position; If the brightness value is lower than or equal to the set threshold, it is determined to be a valid particle scattering signal and retained; if the brightness value is higher than the set threshold, it is determined to be a background noise pixel caused by solid surface reflection light, and the brightness of the pixel is set to zero or set to the background value.

[0011] According to the technical solution provided in this application, the step of calculating the aerodynamic performance parameters of the propeller based on the three-dimensional velocity vector distribution includes the following steps: Based on the three-dimensional velocity vector distribution, at least one key vortex structure generated by the rotation of the propeller is identified and located in three-dimensional space. The key vortex structure includes tip vortex, wake vortex or hub vortex. For each identified key vortex structure, its induced velocity field and corresponding vortex moment are calculated, and then the aerodynamic contribution of the key vortex structure to the overall propfan induced drag, additional thrust or torque is quantified. Based on the aerodynamic contribution component of the key vortex structure, the total thrust, total torque, and aerodynamic efficiency of the propfan are obtained.

[0012] The aerodynamic contribution component of the key vortex structure is vector-synthesized with the basic thrust and torque components calculated from the main field in the three-dimensional velocity vector distribution using the control volume momentum method to obtain the corrected propfan total thrust, total torque, and aerodynamic efficiency.

[0013] According to the technical solution provided in this application, the process of obtaining the total thrust, total torque, and aerodynamic efficiency of the propfan based on the aerodynamic contribution component of the key vortex structure includes the following steps: Based on the three-dimensional velocity vector distribution, a control body containing a propeller is selected, and the change in momentum flux flowing through the control body is calculated to obtain the basic thrust component and basic torque component generated by the main current field. The aerodynamic contribution component of the key vortex structure is vector-synthesized with the basic thrust component and the basic torque component in three-dimensional space to obtain the total thrust and total torque of the propfan. Based on the total thrust and torque of the propeller, and combined with the current rotational speed, the aerodynamic efficiency of the propeller is calculated.

[0014] According to the technical solution provided in this application, the step of calculating the three-dimensional velocity vector distribution of the propeller flow field based on the flow field particle image and the calibration result includes the following steps: Based on the calibration results, the flow field particle image is geometrically corrected to eliminate optical distortion caused by propeller motion and window structure; Based on the corrected flow field particle image, the initial vector distribution of the paddle fan flow field in the absolute coordinate system is calculated; Based on the real-time phase signal, the initial vector distribution is transformed into a rotating relative coordinate system fixed to the propeller blades to obtain the three-dimensional velocity vector distribution of the propeller flow field.

[0015] Compared with the prior art, the beneficial effects of this application are as follows: First, it achieves high-precision real-time correction of dynamic optical distortion, fundamentally ensuring the accuracy of the measurement benchmark. By combining real-time phase signals for online dynamic calibration, it can accurately characterize and compensate for the optical distortion field caused by the propeller motion and the fixed viewing window at every instant. This ensures that the entire calculation process from the flow field particle image to the three-dimensional velocity vector distribution is based on an accurate spatial coordinate mapping relationship, laying a reliable foundation for obtaining the real flow field structure.

[0016] Second, it significantly improves the accuracy and reliability of three-dimensional velocity field measurement results. By solving the dynamic distortion problem from the data source (image coordinate-spatial coordinate mapping relationship), the accuracy of the spatial position and velocity values ​​of the three-dimensional velocity vector distribution calculated based on the calibration results is significantly improved, especially near blades with large flow field gradients and complex spatial structures, and in the wake vortex region. This provides high-quality data support for subsequent in-depth flow mechanism analysis and performance quantification. Attached Figure Description

[0017] Figure 1 The flowchart illustrates the steps of the non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV provided in this application. Detailed Implementation

[0018] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and not intended to limit it. Furthermore, it should be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings.

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] Example 1 As mentioned in the background section, to address the problems in the existing technology, this application proposes a non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV, such as... Figure 1 As shown, it includes the following steps: S1. Obtain the real-time phase signal of the rotating propeller, and generate a synchronization trigger command based on the real-time phase signal; S2. Upstream of the propeller fan, uniformly distributed tracer particles are seeded into the inflow region; S3. In response to the synchronous triggering command, control the laser system to project a light source onto the paddle fan measurement area, and control at least two cameras to capture images synchronously to obtain an image of the illuminated flow field particles. S4. Combine the real-time phase signal to perform three-dimensional spatial coordinate calibration on the rotating propeller measurement area to correct the optical distortion caused by the propeller motion and the window structure, and obtain the calibration result. S5. Based on the flow field particle image and the calibration results, the three-dimensional velocity vector distribution of the propeller flow field is calculated; S6. Calculate the aerodynamic performance parameters of the propeller based on the three-dimensional velocity vector distribution.

[0021] Specifically, the real-time phase signal is an electrical signal characterizing the angular position of the propeller's rotating shaft at any instant. This signal is typically acquired by a high-precision photoelectric encoder or a magnetoelectric angle sensor, which is mounted on the propeller's drive shaft or a component rigidly connected to the shaft, ensuring that its output pulses are strictly synchronized with the rotational position of the propeller blades. The synchronization trigger command is an electronic command generated by the central timing controller based on the received real-time phase signal, after frequency multiplication and time window calculation, used to uniformly control the laser and camera actions. In this implementation, the timing controller locks the pulse frequency output by the sensor and calculates the corresponding trigger timing according to a preset acquisition frequency (e.g., acquiring 36 phase points per revolution). This command ensures that the laser pulse emission and camera exposure occur precisely at the instant the propeller rotates to a specific phase angle.

[0022] Tracer particles are tiny particles capable of following fluid movement and scattering laser light, such as silica or DEHS oil mist particles with a diameter of 1-5 micrometers. The seeding device is located in the upstream stable section of the paddle fan flow direction. It typically uses a Laskin nozzle-based atomizer to generate a high concentration of particles, which are then injected into the airflow through multiple injection holes surrounding the air intake duct. Uniform distribution means that the spatial concentration of particles remains statistically consistent as they flow through the laser-illuminated area, without obvious areas of aggregation or sparseness. During implementation, the air pressure and flow rate of the seeder need to be adjusted, and a diffusion net or mixing chamber should be used to ensure thorough mixing of the particles with the incoming airflow.

[0023] The laser system typically includes a dual-pulse Nd:YAG laser, whose beam, after passing through a cylindrical lens group, forms a sheet light source approximately 1 mm thick. The propeller measurement area is a pre-defined physical space where flow field information needs to be acquired, usually covering the swept surfaces of one or more blade channels and part of the wake region. In practice, the sheet light source is incident vertically from the side or above the propeller, illuminating the measurement area. At least two high-speed CMOS cameras are positioned at a specific stereo angle (typically 20-60 degrees) on the same side of the sheet light source, with the lens planes tilted according to the Scheimpflug condition to obtain a clear image of the entire field. When the timing controller issues a trigger command, the laser emits a pair of pulses with a known interval, and the two cameras simultaneously perform two exposures, recording the particle positions at two instantaneous moments, thus obtaining a set of two-frame particle images for subsequent cross-correlation analysis.

[0024] Because the propeller rotates at high speed, its blade position and attitude are constantly changing. Simultaneously, light refracts when passing through a fixed circular or planar observation window. These two factors combined result in dynamically changing geometric distortions in the camera-captured images, which vary with phase. Traditional static calibration board methods are completely ineffective in this regard. The implementation method in this step is as follows: using the acquired real-time phase signal as an index, a calibration calculation is dynamically performed for each specific rotational phase angle, tailored to that transient geometric configuration. The calibration result is not a fixed set of parameters, but rather a series of phase-varying mapping functions or parameter matrices. It establishes a precise correspondence between the pixel coordinates of each camera image at each phase and the real 3D world coordinates at that phase.

[0025] This implementation solves the calibration problem of dynamic optical distortion in rotating machinery by introducing real-time phase signals and integrating them throughout the triggering, calibration, and data analysis processes. Its technical principle lies in parameterizing rotational motion as a phase variable, and using this variable as a link to mathematically connect the dynamic physical world with the static measurement system. The achieved technical effect is the ability to reconstruct a realistic, spatially accurate three-dimensional velocity field from severely distorted original images with high precision. This makes it possible to quantitatively test the performance of high-speed rotating propellers using three-dimensional PIV (Portable Image Verification), providing unprecedented detailed data support for propeller design optimization and fault diagnosis.

[0026] In a preferred embodiment, the step of calibrating the three-dimensional spatial coordinates of the rotating propeller measurement area in conjunction with the real-time phase signal includes the following steps: Based on the three-dimensional geometric model of the propeller and the real-time phase signal, a virtual three-dimensional calibration field that completely corresponds to the current rotation phase and the spatial position of the blade is generated in real time. The laser system is controlled to project an auxiliary structured light pattern with specific coding features onto the paddle fan measurement area, and a calibration image containing the auxiliary structured light pattern is obtained by synchronously capturing the image with a camera. The auxiliary structured light feature points extracted from the calibration image are matched with the theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field; Based on the matching results, the comprehensive distortion field caused by propeller motion, window refraction and optical system is dynamically solved to establish an accurate mapping relationship between image coordinates and three-dimensional spatial coordinates under the current phase.

[0027] Specifically, the three-dimensional geometric model refers to the computer-aided design model of the propeller, which is a digital file containing precise three-dimensional surface definitions of all components such as blades and hub. The virtual three-dimensional calibration field is a digital scene rendered in real-time in computer memory, conforming to the current physical state. Its implementation involves loading the propeller's CAD model into the computer and, based on real-time input phase signals (e.g., angle values ​​from 0 to 360 degrees), driving the digital propeller model to synchronously rotate to a spatial position and attitude completely consistent with the actual physical propeller through three-dimensional graphical transformations (rotation and translation). At this point, the propeller model surface in the computer at that phase constitutes a virtual calibration reference volume with known precise three-dimensional coordinates of every point on its surface.

[0028] Then, the auxiliary structured light pattern is an optical pattern projected to create easily identifiable feature points in a real physical scene, such as a regular dot matrix, intersecting grids, or pseudo-random speckle. This pattern is projected by a separate projector or generated by guiding a portion of the laser beam through a grating via a beam splitter. The implementation is as follows: during the intervals between acquiring flow field particle images, a timing controller issues a special trigger signal, at which point the laser switches to structured light generation mode (or activates an independent projector), projecting the coded pattern onto the rotating propeller surface and its surrounding space. Two cameras simultaneously capture this scene, resulting in a calibration image that includes not only the propeller but also the area covered by the projected pattern. Because the propeller is rotating, this step must be performed using the same phase-triggered logic as the flow field measurement to ensure the pattern is projected at the same spatial location.

[0029] Next, feature points refer to the centers or vertices of the auxiliary structured light patterns automatically identified from the calibration images using image processing algorithms (such as center detection and corner detection). Theoretical projection points refer to the expected two-dimensional coordinate positions of known three-dimensional coordinate points on the virtual three-dimensional calibration field (i.e., the surface of the paddle fan digital model at this phase) projected onto the image planes of the two cameras, based on the camera's pinhole imaging model. In practice, the two-dimensional image coordinates of the feature points are first extracted from the two calibration images. Simultaneously, the computer calculates the theoretical projection coordinates of a series of sampling points (or pattern points) on the model surface onto the two image planes based on the virtual model at the current phase and the camera's intrinsic and extrinsic parameters (obtained through initial calibration). Then, a stereo matching algorithm is used to establish the correspondence between the feature points extracted from the images and the set of theoretical projection points.

[0030] Finally, the overall distortion field refers to the sum of all geometric errors that cause the actual image point position to deviate from the theoretical projection point position. The matching result is a set of corresponding point pairs: the theoretical 3D point coordinates (X, Y, Z) and the actual captured 2D image point coordinates (u, v). Since the theoretical projection does not consider deviations caused by window refraction and dynamic motion details, there is a difference between these two. In practice, a parameterized distortion model (e.g., a model including lens distortion coefficients and window refraction plane parameters) is used to describe this difference. By matching a large number of point pairs, optimization algorithms such as the least squares method are used to solve for the distortion model parameters that minimize the difference between the theoretical projection point and the actual image point. Once the parameters are determined, for that specific phase, an accurate mapping function from any 3D spatial coordinates to the image coordinates of the two cameras is established, thus completing the dynamic calibration of that phase.

[0031] In a preferred embodiment, matching the auxiliary structured light feature points extracted from the calibration image with the theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field includes the following steps: Based on the real-time phase signal, the auxiliary structured light feature points collected in each rotation cycle are divided into feature point subsets corresponding to each blade according to the blade passing frequency of the propeller. For each subset of the feature points, based on the pre-stored blade aeroelastic deformation model, the estimated elastic deformation of the corresponding blade under the current rotational speed and aerodynamic load is calculated. The theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field are superimposed with the estimated elastic deformation to generate the compensated theoretical projection points after blade deformation compensation. With the goal of minimizing the matching residual between the subset of feature points corresponding to each blade and the compensated theoretical projection points, the spatial pose and local deformation of each blade in the current phase are solved to achieve high-precision feature matching.

[0032] Specifically, based on the real-time phase signal, the auxiliary structured light feature points acquired in each rotation cycle are divided into subsets corresponding to each blade, according to the blade passage frequency. The blade passage frequency is the blade rotation frequency multiplied by the number of blades, representing how many blades pass a fixed point per unit time. Since multiple blades may have slight manufacturing tolerances geometrically, and their deformation under aerodynamic loads may differ, mixing their data for overall matching would introduce averaging errors. This step requires knowledge of the number of blades, N. For a large number of feature points acquired at a certain phase, based on the circumferential distribution of their image coordinates and their relationship to the real-time phase signal, combined with the blade rotation direction, these points are divided into N subsets using a clustering algorithm (such as angle-based K-means clustering). Each subset corresponds to a specific blade (e.g., blade 1, blade 2, ..., blade N). In this way, subsequent matching and deformation calculations can be performed independently for each blade.

[0033] Then, the aeroelastic deformation model of the blade is a mathematical model or data lookup table describing the structural deformation (including bending and torsion) of the blade under the coupled action of centrifugal force, aerodynamic force, and inertial force. This model can be pre-calculated through finite element analysis and stored in the computer. During implementation, the computer reads the currently measured rotational speed signal and incoming flow conditions (such as dynamic pressure) as input parameters to query or call the deformation model. The model outputs the deformation displacement vector (Δx, Δy, Δz) of each calculation node (or the surface position to which the feature point may be attached) on the blade relative to its rigid design position under the current operating conditions. This deformation is theoretically predicted.

[0034] Next, the theoretical projection points in the virtual 3D calibration field are based on a rigid, undeformed CAD model. To more accurately match physical reality, these theoretical 3D points need to be corrected. The implementation method is as follows: for each 3D point belonging to a certain blade in the virtual model, its corresponding estimated deformation amount is obtained from the aeroelastic deformation model. Then, the original coordinates of the 3D point are added to the deformation displacement vector to obtain a new 3D coordinate that is closer to the actual deformed state of the blade, i.e., the compensated theoretical 3D point. These compensated 3D points are then projected onto the image plane according to the camera model to obtain the compensated theoretical projection points.

[0035] Finally, the objective function is to minimize the total distance (residual) between the coordinates of the feature points extracted from the image and the coordinates of the compensated theoretical projection points on all blades. The optimization variables include not only the overall rigid body pose of each blade (three translations and three rotations to correct for installation tolerances and overall sway), but also local fine-tuning parameters for the estimated elastic deformation (to correct for model errors and instantaneous dynamic deformation). In implementation, a nonlinear least squares optimization algorithm (such as the Levenberg-Marquardt algorithm) is used to iteratively solve for these pose and deformation parameters of each blade, independently or jointly, until the optimal parameter combination that minimizes the residuals is found. At this point, the algorithm not only completes feature point matching but also accurately calculates the actual spatial state of each blade.

[0036] In a preferred embodiment, the step of minimizing the matching residual between the subset of feature points corresponding to each blade and the compensated theoretical projection points, and solving for the spatial pose and local deformation of each blade in the current phase, includes the following steps: The swept area of ​​each blade is divided into several flow field grid cells along the radial and / or chordal direction, and a correlation model is established between the blade spatial pose parameters, local deformation parameters and the coordinate correction of each flow field grid cell. The objective function is to minimize the sum of squared weighted residuals between the measured coordinates of the subset of feature points and the compensated theoretical projected coordinates of all blades on all corresponding flow field grid cells. The weighting coefficients are dynamically adjusted according to the relative positions of the grid cells and the leading edge, trailing edge and tip region of the blade. The fixed support constraints at the blade root and the aerodynamic interference correlation between adjacent blades are used as physical constraints. These are substituted into the objective function, and a constrained nonlinear least squares algorithm is used for iterative solution. The precise spatial pose parameters, local deformation parameters, and target coordinate correction of each flow field grid cell are output simultaneously for the current phase of each blade.

[0037] Specifically, the flow field mesh element is a small volume element or surface element that covers the flow field around the blade, divided for subsequent PIV cross-correlation calculations and flow field data display. The correlation model is a mathematical function that describes how to derive the correction amount required for the coordinates of any mesh node in the flow field from known macroscopic blade pose parameters (such as center of gravity position and attitude angle) and local deformation parameters describing blade bending and torsion (such as modal coordinates). In practice, a structured or unstructured computational mesh is first defined in the flow field space around the blade. Then, based on the structural dynamic characteristics of the blade (such as mode shapes), a transfer function or interpolation function is constructed. For example, the displacement field of the blade can be represented as a linear superposition of several mode shapes, and its coefficients are the local deformation parameters. Then, the displacement (i.e., the coordinate correction amount) of a certain mesh node in the flow field can be obtained by interpolation of the same mode shape through the spatial relationship between that node and the nodes on the blade surface. In this way, the pose deformation parameters of the blade are correlated with the geometric deformation of the entire mesh.

[0038] Then, the objective function is to minimize the sum of squared weighted residuals between the measured coordinates and the compensated theoretical projected coordinates of the subset of feature points on all blades across all corresponding flow field grid cells. The weight coefficients are dynamically adjusted based on the relative positions of the grid cells and the leading, trailing, and tip regions of the blade. The objective function is the core of the optimization algorithm; it quantifies the difference between the current solution and the ideal state. This step introduces grid cell weights and regional adjustments. The implementation is as follows: when constructing the objective function, not all matching points are treated equally; instead, a weight coefficient is assigned to the flow field grid cell associated with each matching point. Grid cells located near the leading edge (critical flow separation region), trailing edge (vortex shedding region), and tip (tip vortex initiation region) are assigned higher weights because ensuring the geometric accuracy of these critical flow regions is crucial for the final flow field analysis. Regions with relatively gentle flow, such as the pressure surface in the blade, can be assigned lower weights. The weights can be dynamically calculated based on the distance between the center coordinates of the grid cell and the blade's geometric feature lines. In this way, the optimization algorithm prioritizes ensuring the matching accuracy of critical flow regions during the solution process.

[0039] Next, the fixed support constraints at the blade root and the aerodynamic interference correlation between adjacent blades are used as physical constraints. These are substituted into the objective function, and a constrained nonlinear least squares algorithm is used for iterative solution. The algorithm simultaneously outputs the precise spatial pose parameters, local deformation parameters, and target coordinate corrections for each flow field grid cell in the current phase of each blade. The physical constraints introduce known physical laws as limitations into the optimization problem, making the solution more realistic. The fixed support constraint refers to the requirement that the displacement and rotation angle at the blade root connection to the hub should be zero or conform to the connection stiffness model; this is an equality or inequality constraint. The aerodynamic interference correlation refers to the mutual influence of deformation between adjacent blades through aerodynamic forces due to their presence. This can be simplified into a constraint equation representing the correlation of deformation parameters between adjacent blades. In practice, these constraints, along with the aforementioned weighted objective function, are constructed into a constrained nonlinear optimization problem. A specialized optimization algorithm (such as sequential quadratic programming) is used to solve this problem. After the algorithm converges iteratively, it not only outputs the optimal pose and deformation parameters for each blade, but more importantly, through the established correlation model, it can immediately calculate the three-dimensional coordinate correction amount required for each flow field grid node relative to its standard design position, i.e., the target coordinate correction amount.

[0040] In a preferred embodiment, the step of seeding uniformly distributed tracer particles upstream of the paddle fan into the incoming flow region includes the following steps: The background flow field upstream of the propfan without particle injection is predicted to obtain the background flow field parameters of the inflow region, including the average velocity distribution and turbulence intensity distribution. Based on the average velocity distribution and turbulence intensity distribution, the particle seeding parameters are dynamically planned so that when the seeded particle cloud arrives at the propeller measurement plane, its spatial concentration distribution matches the background flow field parameters, thereby achieving uniform particle distribution within the propeller measurement area.

[0041] Specifically, the background flow field refers to the original incoming flow field during propeller operation, undisturbed by tracer particle seeding. This is typically implemented before the formal PIV test begins. Predictions can be made using other flow diagnostic tools, such as a one-dimensional or multi-point hot-wire anemometer array with the particle seeding device off, or by directly using a two-dimensional PIV system to scan and measure a specified cross-section upstream of the propeller. The average velocity distribution describes the spatial variation of the time-averaged velocity at each point on this cross-section, usually represented as a velocity field contour map or data matrix. The turbulence intensity distribution describes the spatial variation of the velocity fluctuation intensity at each point, calculated by dividing the standard deviation of the velocity by the average velocity. These distribution data collectively characterize the flow field environment features where particles will be injected and transported to the measurement area, serving as the fundamental input for subsequent intelligent seeding planning.

[0042] Then, particle seeding parameters typically include the particle output rate of different nozzles or spouts on the seeding device, the pressure or flow rate of the carrier gas (high-pressure gas used to transport particles), and the spatial pointing angle of the nozzle. Dynamic programming refers to automatically calculating a set of optimal seeding parameters based on the specific flow field characteristics measured upstream. Its implementation principle is to achieve spatially uniform particle concentration upon arrival at the downstream laser measurement plane. Due to the uneven velocity of the upstream flow field (e.g., fast at the center, slow at the sides), if particles are injected uniformly at a constant rate, particles in the fast flow areas will be quickly carried away, leading to sparse particles in the downstream region, while particles may accumulate in the slow flow areas. Therefore, the planning algorithm needs to work in reverse. Based on the measured average velocity distribution, the time for particles to be transported from the seeding surface to the measurement surface can be estimated. To achieve uniform downstream concentration, higher density particles need to be injected into the flow path with shorter transport time (high-speed region), and lower density particles need to be injected into the flow path with longer transport time (low-speed region). Turbulence distribution is used for correction, as particle diffusion is more intense in highly turbulent regions, potentially requiring fine-tuning of the injection strategy. During implementation, the control computer runs a planning algorithm to calculate the required particle output velocity and carrier gas pressure for each nozzle, and automatically adjusts the corresponding valves on the seeding device. Simultaneously, it can control the angle adjustment of nozzles with directional functions, allowing the particle jet to better integrate into the background streamlines. Through this pre-planned, flow field feedback-based approach, the particle cloud is actively shaped during transport and mixing in the background flow field, ultimately achieving an ideal, uniform distribution on the measurement plane.

[0043] In a preferred embodiment, controlling at least two cameras to simultaneously capture images to obtain illuminated flow field particle images includes the following steps: Keeping the camera parameters and the synchronization trigger command unchanged, only turning off the sheet light source of the laser system, controlling all cameras to synchronously capture the paddle fan scene in the same phase, and obtaining the background noise light intensity map corresponding to each camera; Control at least two cameras to capture images simultaneously, obtaining the initial particle image for each frame of the formal acquisition; For each frame of the initial particle image, based on the background noise intensity map of its corresponding camera and corresponding phase, the background noise pixels caused by the reflection of solid surface light from each propeller structure are identified and subtracted to obtain the flow field particle image.

[0044] Specifically, the background noise intensity map is an image that does not contain tracer particle scattering signals, but only records noise light such as ambient light, laboratory stray light, and light reflected from the propeller surface itself, fluorescence, etc., which may be excited by the laser sheet light. The key to its implementation lies in phase synchronization. The specific operation is as follows: before the formal acquisition of flow field data, or during a calibration cycle alternating with the formal acquisition, the timing control system generates a synchronization trigger command sequence that is exactly the same as the formal trigger. The only variable is to control the laser to not output a sheet light source illuminating the particles at this moment (or to reduce its power to an extremely low level). Two or more cameras are synchronously exposed at the instant the propeller rotates to a specific phase angle (e.g., the blades are in a vertical position) with identical exposure time and gain settings. Due to the high-speed rotation of the propeller, the reflective pattern on its surface changes drastically with the phase. Only by shooting at exactly the same phase can the obtained background noise image strictly correspond in spatial intensity distribution to the noise background in subsequent images with particles. Each camera sees a different reflective pattern due to its different viewing angle; therefore, it is necessary to acquire a unique background noise intensity map for each camera.

[0045] Then, at least two cameras are controlled to capture images simultaneously, obtaining the initial particle image for each frame of the formal acquisition. This step is the standard image acquisition process for 3D PIV. When the flow field needs to be measured, the timing controller issues a trigger command, and the laser emits a high-power sheet light source pulse to illuminate the tracer particles in the measurement area. All cameras are exposed in strict synchronization, capturing the image of the illuminated particles, i.e., the initial particle image. At this time, the image simultaneously contains the desired particle scattering signal and unwanted background noise signals such as propeller surface reflection.

[0046] Finally, for each initial particle image captured, the system retrieves a background noise intensity map taken by the same camera at the same phase, based on the real-time phase signal corresponding to its acquisition. Since the two images are identical in camera parameters and shooting phase, the positions and grayscale intensities of the background noise structures (such as bright spots on blades and reflections from wheel hubs) contained in them should theoretically be exactly the same. Subtraction is typically performed using direct image subtraction. The grayscale value of each pixel in the initial particle image is subtracted from the grayscale value of the corresponding pixel in the background noise intensity map. Ideally, after subtraction, the background noise is eliminated, leaving only the grayscale values ​​of the particle scattering signal, thus obtaining the purified flow field particle image. In practice, small-scale image registration fine-tuning may be necessary to compensate for any minute mechanical drift that may exist.

[0047] In a preferred embodiment, identifying and deducting background noise pixels caused by solid surface reflections from each propeller component structure includes the following steps: For each background noise intensity map, an adaptive threshold map is generated based on the brightness statistical characteristics of the neighborhood of each pixel. The brightness value of each pixel in the initial particle image is compared with the set threshold in the adaptive threshold map at its corresponding position; If the brightness value is lower than or equal to the set threshold, it is determined to be a valid particle scattering signal and retained; if the brightness value is higher than the set threshold, it is determined to be a background noise pixel caused by solid surface reflection light, and the brightness of the pixel is set to zero or set to the background value.

[0048] Specifically, the adaptive threshold map is an image of the same size as the background noise intensity map, but the value of each pixel is not a fixed gray level, but a threshold dynamically calculated based on the brightness characteristics of the local area surrounding that pixel. The direct subtraction method assumes that the background noise is strictly subtractable. However, in reality, due to the nonlinearity of the camera sensor, uneven coating on the blade surface, or slight changes in the illumination angle, the intensity of the background noise may spatially vary within the same image, and its superposition with the particle signal may not be a simple linear relationship. To implement this step, the system performs local statistical analysis on each background noise intensity map. For example, for each pixel in the image, it examines the gray values ​​of all pixels within a rectangular window (e.g., 15×15 pixels) centered on that pixel, calculating the local average gray value and local standard deviation of that window. The adaptive threshold can then be calculated using a formula, such as: Threshold = Local Average + k × Local Standard Deviation, where k is an adjustable constant coefficient. Thus, in areas with high and varied background brightness (e.g., the center of a reflective bright spot), the threshold automatically increases; in areas with dark and uniform background, the threshold decreases. The resulting threshold map provides a spatially variable and intelligent criterion for subsequently determining the signal attributes of each pixel. When processing the initial particle image, for any pixel (i, j) in the image, the threshold T for that position (i, j) is read from the adaptive threshold map. For pixel (i, j), its grayscale value I(i, j) in the initial particle image is compared with the threshold T(i, j). If I(i, j) ≤ T(i, j), the pixel's brightness is considered to primarily originate from particle scattering, or although it contains background, its intensity is within a reasonable range; therefore, its grayscale value is retained (or retained after background subtraction). If I(i, j) > T(i, j), the pixel's brightness is considered abnormally high, most likely caused by specular reflection or strong diffuse reflection from the paddle fan's solid surface, and is not a valid flow tracer signal. For such pixels, their grayscale value is directly set to 0 (black) or to a preset background grayscale value (such as the average background value of the entire image), thus completely removing them from the valid data.

[0049] In a preferred embodiment, calculating the aerodynamic performance parameters of the propeller based on the three-dimensional velocity vector distribution includes the following steps: Based on the three-dimensional velocity vector distribution, at least one key vortex structure generated by the rotation of the propeller is identified and located in three-dimensional space. The key vortex structure includes tip vortex, wake vortex or hub vortex. For each identified key vortex structure, its induced velocity field and corresponding vortex moment are calculated, and then the aerodynamic contribution of the key vortex structure to the overall propfan induced drag, additional thrust or torque is quantified. Based on the aerodynamic contribution component of the key vortex structure, the total thrust, total torque, and aerodynamic efficiency of the propfan are obtained.

[0050] The aerodynamic contribution component of the key vortex structure is vector-synthesized with the basic thrust and torque components calculated from the main field in the three-dimensional velocity vector distribution using the control volume momentum method to obtain the corrected propfan total thrust, total torque, and aerodynamic efficiency.

[0051] Specifically, key vortex structures refer to ordered vortices with concentrated vorticity that significantly influence the aerodynamic performance of propellers. The three-dimensional velocity vector distribution provides the velocity components (u, v, w) at each point in space. Based on this, the vorticity field can be calculated; vorticity is a measure of fluid rotation intensity. In implementation, vortex identification criteria are used to automatically extract discrete vortex structures from the continuous vorticity field. Commonly used criteria include the Q criterion and the λ² criterion, which define the vortex core region through the invariant of the velocity gradient tensor. For example, the Q criterion states that the Q value is greater than 0 in the rotationally dominant region. The system scans the three-dimensional data field, clustering all connected voxels that satisfy Q>0 and are clustered, marking them as independent vortex structures. By analyzing the position, intensity, and spatial orientation of the vortex core, tip vortices detached from the blade tip, wake vortices detached from the blade trailing edge (which may roll up to form vortex streets), and hub vortices generated near the hub can be distinguished. The algorithm can output characteristic parameters such as the spatial envelope (three-dimensional coordinate range), vortex core line, and circulation intensity for each identified vortex structure.

[0052] Then, according to vortex dynamics theory, a vortex structure induces a velocity field in the space around it. For an identified vortex structure, the additional velocity field induced by the vortex in the entire flow field, especially at key locations such as the blade surface and far-field control surfaces, can be calculated using the numerical integral form of the Biot-Savart law, based on its spatial vortex distribution. The vortex moment is a physical quantity directly related to the force generated by the vortex structure. More specifically, parameters such as the first vortex moment of each vortex structure can be calculated. By analyzing the direction and intensity of the induced velocity field and combining it with the vortex moment, theoretical models (such as models based on vortex force theory) can be applied to quantitatively estimate the induced effect of the vortex structure on the propeller: for example, tip vortices typically lead to induced drag; certain specific wake vortices may generate slight additional thrust or torque. The calculation results are independent contribution components for each vortex structure, a vector that may contain force and moment components.

[0053] Finally, based on the aerodynamic contribution components of key vortex structures, the total thrust, total torque, and aerodynamic efficiency of the propeller are obtained. This step aims to integrate the contributions of all vortex structures and typically needs to be combined with the mainstream field contribution. One implementation method is to first use the traditional control volume momentum method to calculate the initial estimates of the total thrust and total torque of the propeller based on the three-dimensional velocity field (which can be regarded as a mixture of the mainstream field contribution and all vortex contributions). Then, the induced drag, additional thrust / torque, and other contribution components of all key vortex structures obtained previously are vectorized. According to the principle of action and reaction, the force exerted by the vortex structure on the fluid has a reaction force that acts on the propeller. Therefore, the contribution components of these vortex structures can be subtracted or added to the total estimate for physical correction. For example, subtracting the induced drag component generated by tip vortices from the initial estimate of the total thrust yields a corrected value that is closer to the net thrust of the propeller. Another implementation method is to directly estimate the total force based on vortex dynamics theory by calculating the sum of the vortex moments of all vortex structures. Finally, using the corrected total thrust and total torque, combined with the measured rotational speed, the aerodynamic efficiency is calculated according to the formula: efficiency = (thrust × forward speed) / (torque × angular velocity).

[0054] In a preferred embodiment, obtaining the total thrust, total torque, and aerodynamic efficiency of the propfan based on the aerodynamic contribution component of the key vortex structure includes the following steps: Based on the three-dimensional velocity vector distribution, a control body containing a propeller is selected, and the change in momentum flux flowing through the control body is calculated to obtain the basic thrust component and basic torque component generated by the main current field. The aerodynamic contribution component of the key vortex structure is vector-synthesized with the basic thrust component and the basic torque component in three-dimensional space to obtain the total thrust and total torque of the propfan. Based on the total thrust and torque of the propeller, and combined with the current rotational speed, the aerodynamic efficiency of the propeller is calculated.

[0055] Specifically, the control volume is a static, imaginary spatial volume artificially selected for the application of fluid dynamics conservation laws, typically a cylindrical or cuboid region enclosing the entire propeller. The basic thrust and torque components refer to the force and torque values ​​directly calculated using classical macroscopic conservation laws without considering the vortex structure-induced effects separately. Their physical meaning represents the effect of the overall flow field (including the mainstream and all vortex structures mixed together) on the propeller. This step involves first defining the boundary of the control volume in the three-dimensional velocity field data, for example, a cylindrical surface with an inlet face sufficiently far upstream of the propeller, an outlet face sufficiently far downstream, and sides coaxial with the propeller's rotation axis. According to the momentum equation in the form of Reynolds' transport theorem, the thrust on the propeller is equal to the axial component of the fluid momentum flowing out of the control volume outlet face per unit time, minus the axial component of the momentum flowing in from the inlet face, plus the axial integral of the pressure acting on the control volume's sides (if the pressure field is measured), or simplified through assumptions. Torque is calculated based on the angular momentum equation, which calculates the rate of change of angular momentum of the fluid flowing through the control volume. Since the three-dimensional PIV provides detailed velocity distributions on each boundary surface of the control volume, these momentum and angular momentum fluxes can be calculated with high precision by performing surface integration on the velocities at the boundary surfaces. This calculation yields a macroscopic, overall measurement of the forces acting on the propeller, denoted as F_macro and T_macro.

[0056] Then, the aerodynamic contribution components of the vortex structures, as previously calculated, are the induced forces or moments independently generated by each identified vortex structure (such as tip vortices and wake vortices), denoted as F_vortex,i and T_vortex,i. It is important to understand that the force F_macro calculated by the macroscopic control volume implicitly includes all the effects of these vortex structures. However, due to the inherent uncertainty of PIV measurements in high-gradient regions such as the vortex core, the directly integrated F_macro may contain errors. The synthesis in this step is a physical decomposition and reconstruction. A preferred implementation is the vector subtraction correction method. According to vortex dynamics theory, the force contribution of a vortex structure to the far field can be calculated independently through its own dynamic characteristics (such as vortex moment), and this contribution is part of F_macro. Therefore, a more physically accurate propeller net aerodynamic force (e.g., the effective thrust primarily used to balance flight drag) should be: F_net = F_macro - ΣF_vortex, induced, where ΣF_vortex, induced is the vector sum of the equal components of induced drag generated by all vortex structures. A similar correction is used for torque. The synthesis must be performed in three-dimensional space because the induced forces generated by different vortex structures may have different directions (e.g., some vortices may generate lateral forces). During implementation, the computer performs vector addition and subtraction operations on the contribution components (vectorized data) of all vortex structures and the macroscopic fundamental components, and finally outputs the corrected total thrust vector F_total and total torque vector T_total.

[0057] Finally, force and torque are converted into key indicators for measuring energy conversion efficiency. The current rotational speed is directly measured by a phase signal source (such as an encoder) and denoted as n (revolutions per second) or ω (radians per second). The aerodynamic efficiency (η) of the propfan is defined as the ratio of the useful power output by the propfan to the input power. In typical forward flight conditions, the useful power is the product of thrust F_total (taking the axial component) and flight velocity V (the incoming flow velocity in wind tunnel experiments). The input power is the product of total torque T_total (taking the axial component) and angular velocity ω. Therefore, the efficiency calculation formula is: η = (F_total·V) / (T_total·ω). In practice, the known incoming flow velocity V is read from the data system, and the calculated F_total and T_total are substituted into the formula to obtain the aerodynamic efficiency value under that operating condition. This efficiency value, because it is based on force and torque corrected for vortex contribution, more accurately reflects the propfan's ability to convert rotational mechanical energy into propulsive work or lift.

[0058] In a preferred embodiment, calculating the three-dimensional velocity vector distribution of the propeller flow field based on the flow field particle image and the calibration result includes the following steps: Based on the calibration results, the flow field particle image is geometrically corrected to eliminate optical distortion caused by propeller motion and window structure; Based on the corrected flow field particle image, the initial vector distribution of the paddle fan flow field in the absolute coordinate system is calculated; Based on the real-time phase signal, the initial vector distribution is transformed into a rotating relative coordinate system fixed to the propeller blades to obtain the three-dimensional velocity vector distribution of the propeller flow field.

[0059] Specifically, the calibration result is the precise mapping relationship between the image coordinates and the true 3D spatial coordinates at the current phase, usually represented as a set of camera parameters (such as a projection matrix) and / or a distortion correction lookup table. The flow field particle image is the raw grayscale image captured by the camera, in which the particle image points are shifted due to dynamic distortion. The purpose of geometric correction is to map the particle signal recorded by each pixel in the image back to the spatial position where it should be traversed by the true 3D light rays. When performing this step, for each pair (or group) of synchronously acquired raw particle images, the system calls the parameters obtained from calibration at that phase, based on the phase corresponding to the acquisition time. Then, a reverse coordinate transformation is applied to the entire image. For example, using the calibrated mapping function, for each pixel coordinate (u, v) in the original image, its new coordinates (u', v') in the distortion-corrected image are calculated, and a new image is regenerated through grayscale interpolation (such as bilinear interpolation). After correction, the geometric shape of the particle swarm in the image is consistent with the spatial geometry of the particle distribution in the real flow field, laying a correct spatial reference for subsequent accurate velocity calculations.

[0060] Then, the absolute coordinate system is a stationary reference frame fixed in the laboratory or wind tunnel. The initial vector distribution refers to the first version of the three-dimensional velocity field calculated using the standard three-dimensional PIV algorithm. This step is implemented using the core of a mature three-dimensional particle image velocimetry algorithm. For a pair of synchronized images from at least two cameras that have undergone geometric correction, the tracer particles need to be identified in each image first. Then, a stereo matching algorithm is used to determine the image points of the same physical particle in space in different camera images. Next, using the calibrated camera parameters (these parameters are implicit in the calibration results or can be directly derived from them), the position coordinates of the particle in the real three-dimensional space corresponding to the first step of correction are reconstructed using the forward intersection principle. Finally, for two consecutively captured images, the particle swarms at adjacent time points are matched using three-dimensional particle tracking velocimetry or the more commonly used three-dimensional cross-correlation algorithm (performed within a predefined three-dimensional volume query window), and the displacement vector of the center point of each three-dimensional query window is calculated. Given the laser pulse time interval Δt, the three-dimensional position is subtracted by Δt to obtain the three-dimensional velocity vector (u, v, w) of that point in the absolute coordinate system. By performing such calculations on the entire field, we obtain the three-dimensional velocity vector distribution of the propeller flow field in the absolute coordinate system, which is the data basis for all subsequent analyses.

[0061] Then, the rotating relative coordinate system is a moving reference system with its origin at the propeller's rotation center and its coordinate axes rotating with the propeller blades. For the aerodynamic analysis of rotating machinery, the velocity field in the relative coordinate system has more direct physical meaning than that in the absolute coordinate system. This step requires a coordinate system transformation. First, the precise rotation axis direction of the propeller and its position in the absolute coordinate system, as well as the rotation angle θ at the current moment (defined by the real-time phase signal), need to be known. For each velocity vector point V_abs(x, y, z) = (u, v, w) in the absolute velocity field, its spatial location also needs to be transformed from absolute coordinates (x, y, z) to relative coordinates (x_r, y_r, z_r). Then, the velocity composition theorem from classical mechanics is applied: the relative velocity V_rel equals the absolute velocity V_abs minus the entrainment velocity at that point due to the coordinate system rotation. The entrainment velocity is given by the cross product of the rotational angular velocity vector ω (directed along the axis of rotation, its magnitude determined by the rotational speed) and the position vector r (the vector pointing from the axis of rotation to that point), i.e., V_ent = ω × r. Therefore, V_rel = V_abs - ω × r. The system traverses all points in the absolute velocity field, performs the above vector calculations, and finally generates a new three-dimensional velocity vector distribution map in a rotating relative coordinate system. In this coordinate system, the airflow angle, angle of attack, and relative velocity magnitude within the blade passage can be visually observed. These are the most direct input parameters for blade load analysis, separation judgment, and performance evaluation.

[0062] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this invention, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. A non-contact dynamic testing method for the aerodynamic performance of a propeller fan based on three-dimensional PIV, characterized in that, Includes the following steps: Acquire the real-time phase signal of the rotating propeller and generate a synchronization trigger command based on the real-time phase signal; Upstream of the propfan, tracer particles are uniformly distributed into the inflow region; In response to the synchronous trigger command, the laser system is controlled to project a sheet light source onto the paddle fan measurement area, and at least two cameras are controlled to capture images synchronously to obtain an image of the illuminated flow field particles. By combining the real-time phase signal, the three-dimensional spatial coordinates of the rotating propeller measurement area are calibrated to correct the optical distortion caused by the propeller motion and the window structure, and the calibration result is obtained. Based on the flow field particle image and the calibration results, the three-dimensional velocity vector distribution of the propeller flow field is calculated; Based on the three-dimensional velocity vector distribution, the aerodynamic performance parameters of the propeller are calculated.

2. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 1, characterized in that: The step of calibrating the three-dimensional spatial coordinates of the rotating propeller measurement area by combining the real-time phase signal includes the following steps: Based on the three-dimensional geometric model of the propeller and the real-time phase signal, a virtual three-dimensional calibration field that completely corresponds to the current rotation phase and the spatial position of the blade is generated in real time. The laser system is controlled to project an auxiliary structured light pattern with specific coding features onto the paddle fan measurement area, and a calibration image containing the auxiliary structured light pattern is obtained by synchronously capturing the image with a camera. The auxiliary structured light feature points extracted from the calibration image are matched with the theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field; Based on the matching results, the comprehensive distortion field caused by propeller motion, window refraction and optical system is dynamically solved to establish an accurate mapping relationship between image coordinates and three-dimensional spatial coordinates under the current phase.

3. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 2, characterized in that: The step of matching the auxiliary structured light feature points extracted from the calibration image with the theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field includes the following steps: Based on the real-time phase signal, the auxiliary structured light feature points collected in each rotation cycle are divided into feature point subsets corresponding to each blade according to the blade passing frequency of the propeller. For each subset of the feature points, based on the pre-stored blade aeroelastic deformation model, the estimated elastic deformation of the corresponding blade under the current rotational speed and aerodynamic load is calculated. The theoretical projection points of the corresponding phase in the virtual three-dimensional calibration field are superimposed with the estimated elastic deformation to generate the compensated theoretical projection points after blade deformation compensation. With the goal of minimizing the matching residual between the subset of feature points corresponding to each blade and the compensated theoretical projection points, the spatial pose and local deformation of each blade in the current phase are solved to achieve high-precision feature matching.

4. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 3, characterized in that: The step of minimizing the matching residual between the subset of feature points corresponding to each blade and the compensated theoretical projection points, and solving for the spatial pose and local deformation of each blade in the current phase, includes the following steps: The swept area of ​​each blade is divided into several flow field grid cells along the radial and / or chordal direction, and a correlation model is established between the blade spatial pose parameters, local deformation parameters and the coordinate correction of each flow field grid cell. The objective function is to minimize the sum of squared weighted residuals between the measured coordinates of the subset of feature points and the compensated theoretical projected coordinates of all blades on all corresponding flow field grid cells. The weighting coefficients are dynamically adjusted according to the relative positions of the grid cells and the leading edge, trailing edge and tip region of the blade. The fixed support constraints at the blade root and the aerodynamic interference correlation between adjacent blades are used as physical constraints. These are substituted into the objective function, and a constrained nonlinear least squares algorithm is used for iterative solution. The precise spatial pose parameters, local deformation parameters, and target coordinate correction of each flow field grid cell are output simultaneously for the current phase of each blade.

5. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 1, characterized in that: The process of seeding uniformly distributed tracer particles upstream of the propeller fan into the incoming flow region includes the following steps: The background flow field upstream of the propfan without particle injection is predicted to obtain the background flow field parameters of the inflow region, including the average velocity distribution and turbulence intensity distribution. Based on the average velocity distribution and turbulence intensity distribution, the particle seeding parameters are dynamically planned so that when the seeded particle cloud arrives at the propeller measurement plane, its spatial concentration distribution matches the background flow field parameters, thereby achieving uniform particle distribution within the propeller measurement area.

6. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 1, characterized in that: The control of at least two cameras to simultaneously capture images of the illuminated flow field particles includes the following steps: Keeping the camera parameters and the synchronization trigger command unchanged, only turning off the sheet light source of the laser system, controlling all cameras to synchronously capture the paddle fan scene in the same phase, and obtaining the background noise light intensity map corresponding to each camera; Control at least two cameras to capture images simultaneously, obtaining the initial particle image for each frame of the formal acquisition; For each frame of the initial particle image, based on the background noise intensity map of its corresponding camera and corresponding phase, the background noise pixels caused by the reflection of solid surface light from each propeller structure are identified and subtracted to obtain the flow field particle image.

7. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 6, characterized in that: The process of identifying and deducting background noise pixels caused by reflections from the solid surfaces of each propeller component structure includes the following steps: For each background noise intensity map, an adaptive threshold map is generated based on the brightness statistical characteristics of the neighborhood of each pixel. The brightness value of each pixel in the initial particle image is compared with the set threshold in the adaptive threshold map at its corresponding position; If the brightness value is lower than or equal to the set threshold, it is determined to be a valid particle scattering signal and retained; if the brightness value is higher than the set threshold, it is determined to be a background noise pixel caused by solid surface reflection light, and the brightness of the pixel is set to zero or set to the background value.

8. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 1, characterized in that: The calculation of the aerodynamic performance parameters of the propeller based on the three-dimensional velocity vector distribution includes the following steps: Based on the three-dimensional velocity vector distribution, at least one key vortex structure generated by the rotation of the propeller is identified and located in three-dimensional space. The key vortex structure includes tip vortex, wake vortex or hub vortex. For each identified key vortex structure, its induced velocity field and corresponding vortex moment are calculated, and then the aerodynamic contribution of the key vortex structure to the overall propfan induced drag, additional thrust or torque is quantified. Based on the aerodynamic contribution component of the key vortex structure, the total thrust, total torque, and aerodynamic efficiency of the propfan are obtained. The aerodynamic contribution component of the key vortex structure is vector-synthesized with the basic thrust and torque components calculated from the main field in the three-dimensional velocity vector distribution using the control volume momentum method to obtain the corrected propfan total thrust, total torque, and aerodynamic efficiency.

9. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 8, characterized in that: The process of obtaining the total thrust, total torque, and aerodynamic efficiency of the propfan based on the aerodynamic contribution component of the key vortex structure includes the following steps: Based on the three-dimensional velocity vector distribution, a control body containing a propeller is selected, and the change in momentum flux flowing through the control body is calculated to obtain the basic thrust component and basic torque component generated by the main current field. The aerodynamic contribution component of the key vortex structure is vector-synthesized with the basic thrust component and the basic torque component in three-dimensional space to obtain the total thrust and total torque of the propfan. Based on the total thrust and torque of the propeller, and combined with the current rotational speed, the aerodynamic efficiency of the propeller is calculated.

10. The non-contact dynamic testing method for propeller aerodynamic performance based on three-dimensional PIV according to claim 1, characterized in that: The calculation of the three-dimensional velocity vector distribution of the propeller flow field based on the flow field particle image and the calibration results includes the following steps: Based on the calibration results, the flow field particle image is geometrically corrected to eliminate optical distortion caused by propeller motion and window structure; Based on the corrected flow field particle image, the initial vector distribution of the paddle fan flow field in the absolute coordinate system is calculated; Based on the real-time phase signal, the initial vector distribution is transformed into a rotating relative coordinate system fixed to the propeller blades to obtain the three-dimensional velocity vector distribution of the propeller flow field.

Images