An open-rotor flow field measurement method and system

Abstract

The application relates to the field of aviation propulsion systems, and provides an open rotor flow field measurement method and system.The method comprises the following steps: predicting the gradient distribution of a flow field based on a fluid mechanics neural network, determining high and low gradient regions, arranging measurement points, and obtaining a measurement point array; capturing sound pressure signals of a blade passing frequency, calculating phase angle information of the blade, and establishing a phase synchronization system; under the triggering of the phase synchronization system, performing time-series data acquisition on the measurement point array, and obtaining sparse flow field data; reconstructing the sparse flow field data, and inversely obtaining a flow field database; decomposing the flow field database, extracting a periodic unsteady flow component and a time-averaged flow field component, calculating a turbulent pulsation component from the difference between instantaneous data and the average component, and obtaining a multi-scale flow field feature data set; and extracting and calculating rotor performance parameters, and obtaining an open rotor flow field performance evaluation result.The application improves the accuracy of smooth measurement of the open rotor.

Description

Technical Field

[0001] This invention relates to the field of aviation propulsion system technology, and in particular to an open rotor flow field measurement method and system. Background Technology

[0002] Open rotors, as an advanced aerospace propulsion system, possess a unique aerodynamic configuration with counter-rotating front and rear rows of blades and no outer bypass casing, significantly reducing fuel consumption compared to traditional turbofan engines. However, the mutual interference between the front and rear rows of blades in an open rotor generates a complex three-dimensional unsteady flow field, including multi-scale flow phenomena such as tip vortices, wake disturbances, and shock wave interference. These flow field characteristics directly affect the rotor's aerodynamic efficiency and noise level. Accurately measuring the spatiotemporal evolution characteristics of the open rotor flow field is a key technological foundation for optimizing blade design and reducing aerodynamic losses and noise; therefore, open rotor flow field measurement methods have become an important research direction in the field of aerospace propulsion.

[0003] Existing open rotor flow field measurement technologies mainly employ a dense sensor array global arrangement scheme, which involves uniformly distributing a large number of measurement points in all key areas around the rotor to capture complex flow field structures. While this method can obtain relatively complete flow field information, it has the following drawbacks: First, a dense sensor array requires hundreds or even thousands of measurement points, resulting in high cost of measurement equipment and a huge workload for data acquisition and processing. Second, since high-gradient regions such as tip vortices and shock waves in the open rotor flow field only occupy a local space, while the flow in most low-gradient regions is relatively gentle, a uniform arrangement will generate a large number of redundant measurement points in the low-gradient regions, resulting in a serious waste of measurement resources. Third, existing technologies typically install physical sensors on the blades to obtain phase reference signals. This contact method will disturb the flow field and increase the blade load, and it is difficult to apply to existing test scenarios.

[0004] Therefore, how to reduce the number of measurement points, lower measurement costs, and achieve non-contact phase synchronization while ensuring measurement accuracy has become a key technical problem that urgently needs to be solved in this field. Summary of the Invention

[0005] This invention provides an open rotor flow field measurement method and system to overcome the shortcomings of the prior art.

[0006] This invention provides a method for measuring the flow field of an open rotor, comprising: S1: Based on the fluid dynamics neural network, the gradient distribution of the flow field is predicted. Based on the prediction results, the high gradient region and the low gradient region are determined. Through an adaptive sampling strategy, measurement points are arranged on multiple axial sections of the open rotor to obtain a sparsely distributed measurement point array. S2: Use an acoustic probe array to capture the sound pressure signal at the frequency of the blade passage, and calculate the phase angle information of the blade through the cross-correlation function to establish a non-contact phase synchronization system. S3: Under the triggering of the non-contact phase synchronization system, time-series data acquisition is performed on the measurement points in the measurement array to obtain sparse flow field data carrying spatiotemporal coordinates and phase information; S4: Input the sparse flow field data into the compressed sensing algorithm for reconstruction processing, and obtain a complete flow field database by inversion from sparse sampling points; S5: The flow field database is decomposed according to the time scale. Periodic unsteady flow components are extracted by phase averaging algorithm, and time-averaged flow field components are extracted by time averaging algorithm. Turbulent fluctuation components are calculated from the difference between instantaneous data and average components to obtain a multi-scale flow field feature dataset. S6: Extract and calculate the rotor performance parameters based on the multi-scale flow field feature dataset to obtain the open rotor flow field performance evaluation results.

[0007] According to the open rotor flow field measurement method provided by the present invention, step S1 further includes: S11: Construct a pre-trained fluid dynamics neural network model, input the flow field measurement data of the previous moment into the neural network model, predict and calculate the flow field gradient distribution of the next moment, and obtain the flow field gradient prediction result. S12: The measurement area is classified according to the flow field gradient prediction results, and high gradient areas and low gradient areas are obtained by gradient threshold determination. S13: Multiple measurement sections are set in front of the rotor, on the rotor plane and behind the rotor. For high gradient regions, dense sampling mode is used to arrange measurement points, and for low gradient regions, sparse sampling mode is used to arrange measurement points to obtain a measurement point array.

[0008] According to the open rotor flow field measurement method provided by the present invention, the process of constructing the fluid dynamics neural network model in step S11 includes: S111: Collect historical flow field data of open rotors under various working conditions, and use the velocity, pressure and temperature components of the flow field as training samples. S112: Establish a physical information neural network and embed the Navier-Stokes equation as a physical constraint term in the network's loss function; S113: The physical information neural network is trained using the training samples, and the network parameters are continuously optimized through the backpropagation algorithm. The pre-trained fluid dynamics neural network model is obtained by judging the convergence of the validation set error.

[0009] According to the open rotor flow field measurement method provided by the present invention, step S2 further includes: S21: Arrange an array of acoustic probes around the rotor; S22: The acoustic pressure signal generated during rotor rotation is synchronously acquired through the acoustic probe array, and the characteristic peak value of the blade passing frequency is extracted by the spectrum analysis of the acoustic pressure signal; S23: Perform cross-correlation calculation on the sound pressure signals collected by multiple acoustic probes, calculate the time difference of the sound wave arriving at different probes, and back-calculate the spatial position of the blade based on the time difference and the probe spacing. S24: Combine the tachometer signal to perform time series calibration of the blade's spatial position, establish the correspondence between the time when the blade passes through the fixed reference point and the phase angle, and obtain a non-contact phase synchronization system.

[0010] According to the open rotor flow field measurement method provided by the present invention, step S22 further includes: S221: The acoustic pressure signal generated during rotor rotation is synchronously acquired through the acoustic probe array; S222: Perform a fast Fourier transform on the sound pressure signal to obtain the frequency domain spectrum; S223: Determine the fundamental frequency from the frequency corresponding to the spectral peak with the largest amplitude in the frequency domain spectrum, and mark the spectral peaks corresponding to the fundamental frequency and integer multiples of the fundamental frequency as the characteristic peaks of the blade passage frequency.

[0011] According to the open rotor flow field measurement method provided by the present invention, the calculation method of cross-correlation operation in step S23 is as follows: S231: Select any two acoustic probes in the acoustic probe array as a reference probe pair, and extract the sound pressure signal sequence of the two probes within the same time window; S232: Calculate the cross-correlation function of the two sound pressure signal sequences, and determine the time delay by the peak position of the cross-correlation function; S233: The sound wave propagation path difference is calculated based on the time delay and sound speed. The instantaneous position coordinates of the blade are then inferred by the triangulation algorithm in combination with the spatial spacing of the reference probe pair to obtain the spatial position of the blade.

[0012] According to the open rotor flow field measurement method provided by the present invention, step S4 further includes: S41: Establish the observation matrix of the sparse flow field data, wherein the number of rows of the observation matrix corresponds to the number of sparse sampling points, and the number of columns of the observation matrix corresponds to the number of spatial grids of the complete flow field; S42: Choose wavelet transform or Fourier transform as the sparse basis; S43: Construct an L1 norm minimization optimization problem, using the observation matrix, the sparse basis, and the sparse flow field data as inputs, and solve the optimization problem using the orthogonal matching pursuit algorithm or the basis pursuit algorithm. The complete flow field database is obtained by inverse transformation of the optimization solution.

[0013] According to the open rotor flow field measurement method provided by the present invention, in step S43, the solution process of the orthogonal matching pursuit algorithm is as follows: S431: Initialize the residual vector to equal the sparse flow field data vector. In each iteration, select the column vector with the largest absolute value of the inner product with the residual vector from the column vectors of the sparse basis and add it to the support set. S432: Update the coefficients corresponding to the support set using the least squares method, calculate the updated residual vector, and terminate the iteration when the norm of the updated residual vector is less than the preset threshold or the number of iterations reaches the upper limit. The complete flow field data is obtained by reconstructing the final support set and coefficients.

[0014] According to the open rotor flow field measurement method provided by the present invention, S6 specifically includes: Based on the multi-scale flow field feature dataset, the overall aerodynamic performance index is obtained through integral calculation. Based on the multi-scale flow field feature dataset, the tip vortex characteristic parameters are obtained through gradient calculation. Based on the multi-scale flow field characteristic dataset, the aerodynamic interference intensity coefficient is calculated by comparing and analyzing the front and rear cross-sectional data.

[0015] According to the open rotor flow field measurement method provided by the present invention, the calculation process of the aerodynamic interference intensity coefficient includes: S631: Extract velocity deficit data of the front rotor wake at multiple axial sections from a multi-scale flow field feature dataset; S632: Based on velocity loss data, an exponential function fitting is used to establish the relationship curve between wake attenuation rate and axial distance; S633: Extract the phase-averaged flow field component and time-averaged flow field component of the rear rotor inlet section, calculate the root mean square deviation, and define the obtained root mean square deviation as the unsteady disturbance intensity coefficient. S634: Perform statistical analysis on the circumferential velocity distribution at the inlet section of the rear rotor, and calculate the flow field non-uniformity by the ratio of the standard deviation to the average value of the velocity distribution. S635: The aerodynamic disturbance intensity coefficient is obtained by weighted summation of the unsteady disturbance intensity coefficient and the flow field nonuniformity.

[0016] The present invention also provides an open rotor flow field measurement system for performing an open rotor flow field measurement method as described in any of the preceding claims, comprising: Prediction module: used to predict the gradient distribution of the flow field based on a fluid dynamics neural network and obtain the prediction results; Point placement module: Used to determine high gradient regions and low gradient regions based on prediction results, and to arrange measurement points on multiple axial sections of the open rotor through an adaptive sampling strategy to obtain a sparsely distributed measurement point array. Establishment module: Used to receive the sound pressure signal of the blade passage frequency captured by the acoustic probe array, and calculate the phase angle information of the blade through the cross-correlation function to establish a non-contact phase synchronization system; Acquisition module: used to acquire time-series data from the measurement points in the measurement array under the trigger of the non-contact phase synchronization system, and obtain sparse flow field data carrying spatiotemporal coordinates and phase information; Reconstruction module: used to input the sparse flow field data into the compressed sensing algorithm for reconstruction processing, and obtain a complete flow field database by inversion from sparse sampling points; Decomposition module: used to decompose the flow field database according to the time scale, extract periodic unsteady flow components through phase averaging algorithm, extract time-averaged flow field components through time averaging algorithm, and calculate turbulent fluctuation components from the difference between instantaneous data and average components to obtain multi-scale flow field feature dataset; Analysis module: used to extract and calculate rotor performance parameters based on the multi-scale flow field feature dataset to obtain the open rotor flow field performance evaluation results.

[0017] This invention introduces a fluid dynamics neural network to predict the flow field gradient distribution, enabling the identification of high-gradient regions such as tip vortices and shock waves, as well as low-gradient regions with smooth flow, before measurement. This allows for targeted adjustment of the measurement point density, significantly reducing the number of measurement points compared to traditional uniform arrangement schemes. This drastically lowers sensor procurement costs and the scanning time of the 3D moving platform, while also reducing data storage and post-processing computation, thus significantly improving measurement efficiency. Secondly, this invention uses an acoustic probe array to capture the sound pressure signal at the blade's passing frequency and achieves non-contact phase locking through cross-correlation function calculations. This avoids the flow field disturbance problems caused by installing physical sensors on high-speed rotating blades, allowing the measurement results to more accurately reflect the rotor's original aerodynamic characteristics. Furthermore, this non-contact solution can be directly applied to existing systems. The open rotor in practical testing requires no modification to the blades, greatly expanding the applicability of the technology. Secondly, this invention reconstructs the complete flow field from sparse sampling points using a compressed sensing algorithm. This not only compensates for the information loss that may result from fewer measurement points but also effectively suppresses the impact of measurement noise. The reconstructed flow field data achieves accuracy comparable to or even better than dense measurements in key parameters such as tip vortex intensity and wake attenuation rate. Furthermore, the prediction model based on a physical information neural network continuously optimizes itself with the accumulation of measurement data, making the measurement point placement strategy increasingly precise. This forms a positive cycle of measurement-learning-optimization, resulting in sustained improvement in both measurement efficiency and data quality over long-term use. This provides an efficient, accurate, and economical flow field diagnostic method for the aerodynamic optimization design of open rotors. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This invention provides a schematic flowchart of an open rotor flow field measurement method. Figure 2 This is a schematic diagram of an open rotor flow field measurement system provided by the present invention.

[0020] Attached reference numerals: 100, Prediction module; 200, Point placement module; 300, Establishment module; 400, Data acquisition module; 500, Reconstruction module; 600, Decomposition module; 700, Analysis module. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions 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, embodiments of this invention, and should not be construed as limiting the 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. In the description of this invention, it should be understood that the terminology used is for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0022] The embodiments of the present invention are described below with reference to the figures.

[0023] like Figure 1 As shown, the present invention provides an open rotor flow field measurement method, comprising: S1: Based on the fluid dynamics neural network, the gradient distribution of the flow field is predicted. Based on the prediction results, the high gradient region and the low gradient region are determined. Through an adaptive sampling strategy, measurement points are arranged on multiple axial sections of the open rotor to obtain a sparsely distributed measurement point array.

[0024] Step S1 further includes: S11: Construct a pre-trained fluid dynamics neural network model, input the flow field measurement data of the previous moment into the neural network model, predict and calculate the flow field gradient distribution of the next moment, and obtain the flow field gradient prediction result.

[0025] The process of constructing the fluid dynamics neural network model in step S11 includes: S111: Collect historical flow field data of open rotor under various working conditions, and use the velocity, pressure and temperature components of the flow field as training samples.

[0026] This invention first collects historical flow field data of an open rotor under various operating conditions, including different rotational speeds, blade angles of attack, and incoming Mach numbers. For each operating condition, this invention extracts the three-dimensional velocity components (axial velocity u, radial velocity v, circumferential velocity w), static pressure p, and static temperature T at each spatial grid point in the flow field, forming a five-dimensional feature vector. After collection, this invention organizes the data for all operating conditions according to a time series, with the flow field state at each time step constituting a training sample, typically numbering in the thousands to tens of thousands. Subsequently, this invention divides these samples into a training set, a validation set, and a test set, with a ratio of 7:2:1. The training set is used for network parameter learning, the validation set is used to monitor the training process to prevent overfitting, and the test set is used for final performance evaluation.

[0027] S112: Establish a physical information neural network and embed the Navier-Stokes equation as a physical constraint term in the network's loss function.

[0028] The physical information neural network established in this invention includes an input layer, multiple hidden layers, and an output layer, wherein the input layer receives the flow field state at the current time t. and spatial coordinates The hidden layers employ a fully connected structure, with each layer containing 128-256 neurons. The tanh activation function is used, and the output layer predicts the next time step. The flow field gradient components include velocity gradient components and pressure gradient components. It should be noted that the loss function in the physical information network designed in this invention decomposes into two parts: a data fitting term and a physical constraint term. The data fitting term calculates the mean square error between the network's predicted values ​​and the true values ​​of the training samples. The physical constraint term embeds the continuity and momentum equations of the Navier-Stokes equations, calculates the sum of squares of the equation residuals, and the total loss function is a weighted sum of the two terms. The weight coefficients are determined through cross-validation.

[0029] S113: The physical information neural network is trained using the training samples, and the network parameters are continuously optimized through the backpropagation algorithm. The pre-trained fluid dynamics neural network model is obtained by judging the convergence of the validation set error.

[0030] In step S113, this invention trains the physical information neural network using training set data. The training process employs mini-batch gradient descent, with each batch containing 64-128 samples. In each iteration, this invention forward-propagates the input data of one batch through each layer of the network, calculates the predicted gradient value of the output layer, and then calculates the total value of the loss function, including the mean squared error between the predicted gradient and the true gradient, as well as the residual of the Navier-Stokes equation. Subsequently, the gradient of the loss function with respect to each parameter (weight and bias) in the network is calculated using the backpropagation algorithm, and the parameters are updated using the Adam optimizer. The learning rate is initially set to 0.001 and decays to 0.5 times its original value every 50 training epochs. During training, this invention evaluates the network performance on the validation set every 10 epochs, records the loss function value of the validation set, and determines convergence when the validation set loss no longer decreases for 20 consecutive epochs. Training is then stopped, and the network parameters at this point are saved, resulting in a pre-trained fluid dynamics neural network model.

[0031] S12: The measurement area is classified according to the flow field gradient prediction results, and high gradient areas and low gradient areas are obtained by gradient threshold determination. Furthermore, after obtaining the flow field gradient prediction results, this invention calculates the gradient magnitude of each spatial grid point. The gradient magnitude is defined as the square root of the sum of the squares of all gradient components. Specifically, this invention statistically analyzes the gradient magnitudes of all grid points, calculates their maximum and minimum values, and normalizes the gradient magnitudes to map them to the 0-1 interval. Subsequently, a gradient threshold is set to 0.7 times the normalized value. All grid points are traversed, and grid points with normalized gradient magnitudes greater than 0.7 are marked as high-gradient regions, while those less than 0.7 are marked as low-gradient regions. The high-gradient regions described in this invention are concentrated in locations such as the tip vortex core, the boundary between the front rotor wake and the rear rotor leading edge, and near the shock wave. The low-gradient regions are mainly distributed in the free flow region far from the rotor and in the boundary layer in the middle of the blade.

[0032] S13: Multiple measurement sections are set in front of the rotor, on the rotor plane and behind the rotor. For high gradient regions, dense sampling mode is used to arrange measurement points, and for low gradient regions, sparse sampling mode is used to arrange measurement points to obtain a measurement point array.

[0033] In a specific embodiment, the present invention sets a first measurement section at 0.5 times the rotor diameter in front of the open rotor, a second measurement section on the rotor plane, and third and fourth measurement sections at 1.0 and 2.0 times the rotor diameter behind the rotor, respectively. For each measurement section, the present invention arranges measurement points according to the regional classification results obtained in step S12. In the high gradient region, the radial distance extends outward from the blade tip to 1.5 times the blade tip radius, the measurement point spacing is set to 1 / 18 of the blade height, and the circumferential measurement point spacing is set to 2.5 degrees, forming a dense grid. In the low gradient region, the radial measurement point spacing is widened to 1 / 7 of the blade height, and the circumferential measurement point spacing is expanded to 9 degrees, forming a sparse grid. Finally, the present invention stores the measurement point coordinates of each section as a measurement point matrix data table, which contains the three-dimensional spatial coordinates of each measurement point, the region type (high gradient or low gradient), and the corresponding sampling density parameters.

[0034] S2: Using an acoustic probe array to capture the sound pressure signal at the frequency of the blade passage, the phase angle information of the blade is obtained by calculating the cross-correlation function in order to establish a non-contact phase synchronization system.

[0035] Step S2 further includes: S21: Arrange an array of acoustic probes around the rotor.

[0036] In step S21, the present invention arranges an acoustic probe array around the rotor. The arranged acoustic probe array includes at least three acoustic probes, and the spatial positions of each acoustic probe are non-collinearly distributed. In specific implementation, the present invention arranges one acoustic probe at each of the three circumferential positions of 120 degrees, 240 degrees and 360 degrees around the rotor. The radial distance of the three probes from the rotor center is 1.2 times the rotor radius, and the axial position is flush with the rotor plane, forming a non-collinear triangular array.

[0037] S22: The acoustic pressure signal generated during rotor rotation is synchronously acquired through the acoustic probe array, and the characteristic peak value of the blade passing frequency is extracted by the spectrum analysis of the acoustic pressure signal.

[0038] Step S22 further includes: S221: The acoustic pressure signal generated during rotor rotation is synchronously acquired through the acoustic probe array.

[0039] Furthermore, as the rotor rotates, the blades cut the airflow, generating periodic pressure pulsations, which propagate outward in the form of sound waves. In step S221, the present invention synchronously triggers three acoustic probes via a data acquisition card to acquire sound pressure signals within the same time window. The length of the time window is set to one rotor rotation cycle, and each probe obtains approximately 50,000 sampling points within this time window.

[0040] S222: Perform a fast Fourier transform on the sound pressure signal to obtain the frequency domain spectrum.

[0041] In step S222, the present invention performs Fast Fourier Transform (FFT) processing on the sound pressure signals collected by the three acoustic probes. The FFT converts the time-domain signal into a frequency-domain spectrum, revealing the frequency components and their amplitudes contained in the signal. Specifically, before the transformation, the present invention applies a Hanning window function to the sound pressure signal to reduce spectral leakage. After the transformation, a frequency-amplitude spectrum is obtained, with the horizontal axis representing frequency and the vertical axis representing the amplitude of the corresponding frequency. Since the frequency of blade passage is equal to the rotational speed multiplied by the number of blades, the frequency and its harmonics appear as significant peaks in the spectrum.

[0042] S223: Determine the fundamental frequency from the frequency corresponding to the spectral peak with the largest amplitude in the frequency domain spectrum, and mark the spectral peaks corresponding to the fundamental frequency and integer multiples of the fundamental frequency as the characteristic peaks of the blade passage frequency.

[0043] This invention searches for the spectral peak with the largest amplitude in the frequency domain spectrum. The frequency corresponding to this peak is the fundamental frequency, denoted as . In a physical sense, the fundamental frequency is equal to the passing frequency of the blade, reflecting the repetition frequency of a single blade passing a fixed measuring point. Subsequently, this invention marks the frequency in the spectrum as... , , The positions of the fundamental frequency that are integer multiples of the fundamental frequency are identified, and the corresponding spectral peaks are the harmonic components of the blade's passing frequency. This invention records the fundamental frequency and the frequencies and amplitudes of its first five harmonics as characteristic peak data of the blade's passing frequency. This data includes frequency values, amplitudes, and phase information.

[0044] S23: Perform cross-correlation calculation on the sound pressure signals collected by multiple acoustic probes, calculate the time difference of the sound wave arriving at different probes, and infer the spatial position of the blade based on the time difference and the probe spacing.

[0045] The calculation method for the cross-correlation operation in step S23 is as follows: S231: Select any two acoustic probes in the acoustic probe array as a reference probe pair, and extract the sound pressure signal sequence of the two probes within the same time window.

[0046] In step S231, the present invention selects probe 1 and probe 2 from three acoustic probes as the first reference probe pair, and extracts the sound pressure signal sequences acquired by both probes within the same time window. The signal sequence of probe 1 is denoted as... The signal sequence of probe 2 is denoted as Both sequences have a length of 50,000 sampling points and a time resolution of 0.02 milliseconds. They record the sound pressure changes generated by the same sound source (blade pressure pulsation) at different spatial locations. Due to different propagation distances, Compared to There is a time delay.

[0047] S232: Calculate the cross-correlation function of the two sound pressure signal sequences, and determine the time delay by the peak position of the cross-correlation function.

[0048] In step S232, the present invention calculates and cross-correlation function The cross-correlation function is defined as the cross-correlation function of two signals at different time offsets. The similarity metric is as follows. During the calculation process, this invention will... Translate along the time axis Calculate the translation and The pointwise product of the samples is summed over all sampling points and normalized to obtain the product. Corresponding cross-correlation function value This invention traverses... Generate a cross-correlation function curve for all possible values ​​from -10 milliseconds to +10 milliseconds, with a step size of 0.02 milliseconds. This curve then... The value reaches its peak at a certain point, and the peak position corresponds to... That is, the time delay. This indicates the time it takes for a sound wave to travel from the sound source to probe 2 compared to the time it takes to travel to probe 1.

[0049] S233: The sound wave propagation path difference is calculated based on the time delay and sound speed. The instantaneous position coordinates of the blade are then inferred by the triangulation algorithm in combination with the spatial spacing of the reference probe pair to obtain the spatial position of the blade.

[0050] This invention is based on the amount of time delay and speed of sound Calculate the sound wave propagation path difference (taking 340 m / s under standard atmospheric conditions). Since the spatial coordinates of probe 1 and probe 2 are known, the distance between the two probes can be determined. The Euclidean distance is calculated as the coordinate difference. The distances from the sound source (blade position) to the two probes are denoted as follows: and ,satisfy This invention establishes a system of triangulation equations, using the sound source coordinates... The unknowns are the distance equations for the two probes and the path difference equation. Since three probes are used, this invention repeats the above calculations for probe 1 and probe 3, and probe 2 and probe 3 respectively, obtaining three sets of equations. By solving the overdetermined system of equations (three equations with three unknowns), the instantaneous position coordinates of the blade are obtained using the least squares method. .

[0051] S24: Combine the tachometer signal to perform time series calibration of the blade's spatial position, establish the correspondence between the time when the blade passes through the fixed reference point and the phase angle, and obtain a non-contact phase synchronization system.

[0052] In step S24, the present invention installs a tachometer on the rotor shaft. The tachometer outputs a pulse signal every revolution, recording the time when the rotor completes one revolution. Subsequently, the present invention aligns the tachometer signal with the blade position sequence obtained by acoustic positioning to establish a time-position-phase mapping relationship. In specific operation, the present invention uses one rotation cycle... It is divided into 360 phase angles, each phase angle corresponding to 1 degree. At time... The tachometer outputs a pulse, with the phase angle defined as 0 degrees at this moment. The pulse is output again, and the phase angle returns to 0 for any given time. Its phase angle Calculated as The present invention uses each moment in the instantaneous position coordinate sequence of the blade obtained in step S233. Substitute into the above formula to calculate the corresponding phase angle. Since the open rotor contains two rows of blades rotating in opposite directions, this invention installs tachometers on the front and rear rotors respectively, establishing two independent time-phase angle correspondences. By statistically analyzing data from multiple rotation cycles, this invention ultimately averages the phase angle calculation results to eliminate random errors, resulting in a stable non-contact phase synchronization system. The output data includes the front row phase angle at each moment. and rear phase angle .

[0053] S3: Under the triggering of the non-contact phase synchronization system, time-series data acquisition is performed on the measurement points in the measurement array to obtain sparse flow field data carrying spatiotemporal coordinates and phase information.

[0054] In step S3, the present invention uses the phase angle signal output by the non-contact phase synchronization system as a trigger source to start flow field data acquisition. At each measuring point in the measurement array, the present invention uses a three-dimensional moving platform to precisely position the sensor (five-hole probe or hot-wire anemometer). Current rotor phase angle When a preset trigger value is reached (e.g., triggered every 5 degrees), the data acquisition system records the coordinates of the measuring point at that moment. Flow field parameters (velocity) , , and pressure Front row phase angle and rear phase angle This invention continuously collects data covering 200 rotation cycles of the front rotor at each measuring point, with 72 triggers per cycle (360 degrees / 5 degrees). A total of 14,400 data samples are collected at a single measuring point. All data are stored in a five-tuple format of "measuring point number-time stamp-front phase angle-rear phase angle-flow field parameter" to form a sparse flow field data table.

[0055] S4: Input the sparse flow field data into the compressed sensing algorithm for reconstruction processing, and obtain a complete flow field database by inversion from sparse sampling points.

[0056] Step S4 further includes: S41: Establish the observation matrix of the sparse flow field data, wherein the number of rows in the observation matrix corresponds to the number of sparse sampling points, and the number of columns in the observation matrix corresponds to the number of spatial grids of the complete flow field.

[0057] Furthermore, this invention establishes an observation matrix based on the sparse flow field data obtained in step S3. The sparse flow field data includes flow field parameters from N sparse sampling points. This invention arranges the velocity and pressure data from these N sampling points according to their measurement point numbers, forming an observation vector y of length N. Subsequently, this invention defines a spatial grid for the complete flow field, covering the entire measurement area around the open rotor. The grid is divided into grid cells in the radial, circumferential, and axial directions, with a total of M grid cells (M is much larger than N). Observation Matrix The dimension is N×M, and its physical meaning describes the mapping relationship between sparse sampling points and complete grid points. Specifically, the observation matrix is ​​constructed as follows: the i-th row corresponds to the i-th sparse sampling point; if this sampling point happens to be located at the j-th grid point of the complete grid, then... The remaining positions are 0. If the sampling point is located between multiple grid points, this invention uses bilinear interpolation to calculate the weighting coefficient based on the distance from the sampling point to the adjacent grid points, so that... The final observation matrix is ​​equal to this weight value. It is a sparse matrix, where sparse means that each row has only a few non-zero elements.

[0058] S42: Choose wavelet transform or Fourier transform as the sparse basis.

[0059] In step S42, the present invention aims to select wavelet transform as a sparse basis. Wavelet transform can decompose signals simultaneously in the time and frequency domains, making it particularly suitable for characterizing multi-scale vortex structures in flow fields. Therefore, this invention employs the Daubechies wavelet db4 to construct a two-dimensional wavelet transform matrix. The column vectors of this matrix consist of wavelet basis functions at different scales and positions, constructing a sparse basis... The dimension is M×M, representing the transformation relationship of the complete flow field data in the wavelet domain. Furthermore, the sparsity of the flow field data is reflected in the fact that although the flow field data in the spatial domain has M components, after transformation in the wavelet domain... After the transformation, most coefficients are close to zero, and only a few K significant coefficients (K is much smaller than M) can characterize the main features of the flow field. Ultimately, this invention achieves this through... transpose matrix Achieve the inverse transformation from wavelet coefficients to spatial flow field.

[0060] S43: Construct an L1 norm minimization optimization problem, using the observation matrix, the sparse basis, and the sparse flow field data as inputs, and solve the optimization problem using the orthogonal matching pursuit algorithm or the basis pursuit algorithm. The complete flow field database is obtained by inverse transformation of the optimization solution.

[0061] In step S43, the solution process of the orthogonal matching pursuit algorithm is as follows: S431: Initialize the residual vector to equal the sparse flow field data vector. In each iteration, select the column vector with the largest absolute value of the inner product with the residual vector from the column vectors of the sparse basis and add it to the support set. S432: Update the coefficients corresponding to the support set by the least squares method, calculate the updated residual vector, and terminate the iteration when the norm of the updated residual vector is less than the preset threshold or the number of iterations reaches the upper limit. The complete flow field data is obtained by reconstructing the final support set and coefficients.

[0062] In step S43, the present invention constructs an L1 norm minimization optimization problem, the objective of which is to solve for the sparse coefficient vector in the wavelet domain. This makes the observation equation To be satisfied, at the same time The optimization problem is to minimize the L1 norm (the sum of the absolute values ​​of all components) under the constraints. Below, request smallest .

[0063] After conceiving the problem, this invention employs the orthogonal matching pursuit algorithm to solve it. At the start of the algorithm, the residual vector is first initialized. equal to the observation vector Support set For an empty set, iteration counting In the first In the next iteration, this invention calculates the residual vector. With the perception matrix The inner product of each column, the perception matrix The dimension is N×M, and its th List as .

[0064] Subsequently, this invention traverses Calculate the inner product value from 1 to M. Select the column index with the largest absolute value of the inner product. , this column vector Add support set Subsequently, the present invention utilizes a perception matrix. Extract all column vectors corresponding to the support set and form a submatrix. Solve the equation using the least squares method This yields the sparse vectors corresponding to the support set. .

[0065] After obtaining the sparse vector, this invention calculates a new residual vector. Determine the L2 norm of the residual vector Is it less than the preset threshold? (Set as 1% of the norm of the observed vector), or the number of iterations Whether the upper limit K (set to twice the sparsity) has been reached.

[0066] If the termination condition is met, the invention stops iterating and the support set is... Corresponding coefficients Fill in the complete coefficient vector Set the corresponding position and the rest to 0 to obtain the sparse coefficient vector. Finally, this invention utilizes inverse transformation. wavelet coefficients Transform back into the spatial domain to obtain the complete flow field data vector. This vector contains flow field parameters for M grid points, forming a complete flow field database.

[0067] S5: The flow field database is decomposed according to the time scale. Periodic unsteady flow components are extracted by phase averaging algorithm, and time-averaged flow field components are extracted by time averaging algorithm. Turbulent fluctuation components are calculated from the difference between instantaneous data and average components to obtain a multi-scale flow field feature dataset.

[0068] In step S5, the present invention performs multi-scale decomposition on the complete flow field database. The flow field database contains a time series, with each time step corresponding to specific front and rear phase angles. During the decomposition, the present invention first performs time averaging for each spatial grid point. The time-averaged flow field component is obtained by taking the arithmetic mean of all time-step data, denoted as . , , , This component reflects the steady-state characteristics of the flow field.

[0069] Subsequently, the present invention performs phase averaging, averaging the flow field data according to the front row phase angle. The data is grouped, with a phase window width of 5 degrees, and each phase window contains data from multiple time steps. Subsequently, this invention performs a ensemble averaging on all data within each phase window to obtain the phase-averaged flow field component, denoted as... , , , This component varies with the phase angle The changes reflect the characteristics of periodic unsteady flow.

[0070] Next, this invention calculates the difference between the instantaneous flow field data and the time-averaged flow field component and the phase-averaged flow field component to obtain the turbulent fluctuation component. For time... ,Location instantaneous velocity turbulent pulsation component Calculated as After simplification, it becomes This invention performs the above calculations at all time steps and all grid points to obtain statistics on turbulent fluctuation components, including the root mean square value of the fluctuation velocity and the Reynolds stress component. Finally, this invention integrates the time-averaged flow field component, the phase-averaged flow field component, and the turbulent fluctuation statistics into a multi-scale flow field feature dataset.

[0071] S6: Extract and calculate the rotor performance parameters based on the multi-scale flow field feature dataset to obtain the open rotor flow field performance evaluation results.

[0072] Specifically, S6 includes: Based on the multi-scale flow field feature dataset, the overall aerodynamic performance index is obtained through integral calculation.

[0073] Furthermore, this invention extracts the time-averaged flow field components from a multi-scale flow field feature dataset and reads the time-averaged static pressure of all grid points on a measurement cross-section located 2.0 times the diameter behind the rotor axis. and time-averaged velocity components , , This invention calculates the total pressure at each grid point, defined as the sum of static pressure and dynamic pressure, where dynamic pressure equals... ,in This refers to air density.

[0074] Subsequently, this invention performs area integration on the total pressure across the cross-section to calculate the average total pressure of the cross-section. The invention reads the average total pressure at the inlet cross-section located 0.5 times the diameter in front of the rotor's axial direction. The difference between the average total pressures of the two cross-sections is divided by the inlet dynamic pressure to obtain the total pressure loss coefficient. This invention also calculates the mass flow rate of the cross-section, which is equal to the integral of the product of density, axial velocity, and area over the cross-section. Kinetic efficiency is defined as the ratio of the kinetic energy at the outlet cross-section to the kinetic energy at the inlet cross-section, where kinetic energy is equal to 0.5 × mass flow rate × velocity squared. Finally, this invention outputs the total pressure loss coefficient and kinetic efficiency as overall aerodynamic performance indicators.

[0075] Based on the multi-scale flow field feature dataset, the tip vortex characteristic parameters are obtained through gradient calculation.

[0076] Furthermore, this invention extracts the time-averaged velocity field data of the blade tip region (i.e., radial positions greater than 0.95 times the blade tip radius) from a multi-scale flow field feature dataset. After acquiring the data, the spatial gradient of the velocity field is first calculated based on the obtained data, and numerical differentiation is performed using a central difference scheme. Subsequently, this invention calculates the vorticity vector, where the vorticity modulus is the square root of the sum of the squares of its three components. Following this, this invention searches for local maxima of the vorticity modulus in the blade tip region, marking points where the vorticity modulus exceeds five times the overall field average as blade tip vortex cores.

[0077] This invention extracts the peak vorticity of the vortex core as the tip vortex intensity. After extraction, this invention measures the radial velocity component in the tip clearance region, i.e., the annular space between the tip and the hypothetical bypass. Area integration is performed, and the integration result is the tip leakage flow rate. Finally, this invention combines the tip vortex intensity and tip leakage flow rate into a tip vortex characteristic parameter output.

[0078] Based on the multi-scale flow field characteristic dataset, the aerodynamic interference intensity coefficient is calculated by comparing and analyzing the front and rear cross-sectional data.

[0079] The calculation process for the aerodynamic interference intensity coefficient includes: S631: Extract velocity deficit data of the front rotor wake at multiple axial sections from a multi-scale flow field feature dataset.

[0080] In step S631, the present invention extracts each axial section behind the front rotor from the multi-scale flow field feature dataset, including the rotor plane, and the time-averaged flow field components at 1.0 times and 2.0 times the diameter axially behind. On each section, the present invention identifies the wake region of the front rotor blades, the wake feature being the axial velocity. The velocity is lower than the free-flow velocity. Subsequently, this invention takes a wake profile at each blade pitch in the circumferential direction, extracts the axial velocity distribution along the radial direction on this profile, and defines the velocity deficit as the difference between the free-flow velocity and the local axial velocity. Based on the calculation method, this invention calculates the velocity deficit value at each radial position. This invention also calculates the wake width, defined as the radial range where the velocity deficit is greater than half of the maximum value. Finally, this invention correlates the peak velocity deficit and wake width data of different axial sections with the corresponding axial distances to form a velocity deficit data table.

[0081] S632: Based on velocity loss data, an exponential function fitting is used to establish the relationship curve between wake attenuation rate and axial distance.

[0082] Furthermore, the present invention performs fitting processing on the velocity loss data obtained in step S631. The fitting function used in the present invention is an exponential decay model, which represents the velocity loss. With axial distance The relationship of change is expressed as ,in Due to initial speed loss, This refers to the attenuation rate coefficient. In this invention, the axial distance from the speed loss data table is used as the independent variable. Peak speed loss as the dependent variable The parameters are solved using the nonlinear least squares method. and .

[0083] During the fitting process, this invention initializes the parameter values ​​and iteratively adjusts the parameters to minimize the sum of squared residuals between the fitted curve and the measured data points. After the fitting is completed, this invention obtains the attenuation rate coefficient. The reciprocal of the coefficient Characterizing the wake recovery characteristic length, the smaller the value, the faster the wake decays. Ultimately, this invention will... The relationship between the axial distance and the wake attenuation rate is plotted as a wake attenuation rate curve.

[0084] S633: Extract the phase-averaged flow field component and time-averaged flow field component of the rear rotor inlet section, calculate the root mean square deviation, and define the obtained root mean square deviation as the unsteady disturbance intensity coefficient.

[0085] This invention extracts the phase-averaged and time-averaged flow field components of the rear rotor inlet section (located between the front and rear rotors, with an axial distance of 0.7 times the rotor diameter) from a multi-scale flow field feature dataset. The phase-averaged flow field component is a function of the phase angle, denoted as... The time-averaged flow field components are scalar fields, denoted as . .

[0086] The invention relates to each grid point of this cross section The average phase velocity values ​​corresponding to all phase angles are extracted to form a vector of length 72 (corresponding to 72 phase windows). Subsequently, this invention calculates the square of the difference between this vector and the average velocity value, sums the results over all phase angles, divides by the number of phase windows, and then takes the square root to obtain the root mean square deviation (RMS) of the grid point. After performing the same calculation on all grid points on the cross section, the area-weighted average of the RMS values ​​of all grid points is taken and defined as the unsteady disturbance intensity coefficient. .

[0087] S634: Perform statistical analysis on the circumferential velocity distribution at the inlet section of the rear rotor, and calculate the flow field nonuniformity by the ratio of the standard deviation to the average value of the velocity distribution.

[0088] In step S634, the present invention extracts the time-averaged circumferential velocity distribution of the rear rotor inlet section. After extraction, the present invention fixes the radial position. At the mid-height of the blade, extract all circumferential angles at that radial position. The circumferential velocity values ​​are used to form a circumferential velocity sequence.

[0089] This invention calculates the arithmetic mean of the sequence. and standard deviation The standard deviation is calculated as the square root of the sum of the squares of the differences between all velocity values ​​and the mean, divided by the number of samples. Finally, this invention calculates the ratio of the standard deviation to the mean. This ratio reflects the degree of dispersion of the circumferential velocity distribution and is defined as the flow field nonuniformity η. After repeating the above calculation at multiple radial positions in this invention, the radial average value is taken as the flow field nonuniformity of the cross section.

[0090] S635: The aerodynamic disturbance intensity coefficient is obtained by weighted summation of the unsteady disturbance intensity coefficient and the flow field nonuniformity.

[0091] Furthermore, in step S635, the present invention uses the unsteady disturbance intensity coefficient obtained in step S633. and the flow field nonuniformity obtained in step S634 Weighted summation was performed, with weighting coefficients set to 0.6 and 0.4 respectively, to reflect the more significant impact of unsteady disturbances on the performance of the rear rotor. The obtained aerodynamic disturbance intensity coefficient I was calculated as follows: Finally, this invention will use the calculated aerodynamic interference intensity coefficient As a comprehensive index output that quantifies the degree of mutual interference between the front and rear rotors, the larger the value of this index, the stronger the aerodynamic interference.

[0092] like Figure 2 As shown, the present invention also provides an open rotor flow field measurement system, comprising: Prediction module 100: used to predict the gradient distribution of the flow field based on a fluid dynamics neural network and obtain the prediction results; Point placement module 200: Used to determine high gradient regions and low gradient regions based on prediction results, and to arrange measurement points on multiple axial sections of the open rotor through an adaptive sampling strategy to obtain a sparsely distributed measurement point array. Module 300: Used to receive the sound pressure signal of the blade passage frequency captured by the acoustic probe array, and calculate the phase angle information of the blade through the cross-correlation function to establish a non-contact phase synchronization system; Acquisition module 400: used to acquire time-series data from the measurement points in the measurement array under the trigger of the non-contact phase synchronization system, and obtain sparse flow field data carrying spatiotemporal coordinates and phase information; Reconstruction module 500: used to input the sparse flow field data into the compressed sensing algorithm for reconstruction processing, and obtain a complete flow field database by inversion from sparse sampling points; Decomposition module 600: is used to decompose the flow field database according to the time scale, extract the periodic unsteady flow component by the phase averaging algorithm, extract the time-averaged flow field component by the time averaging algorithm, and obtain the turbulent fluctuation component by the difference between the instantaneous data and the average component, so as to obtain a multi-scale flow field feature dataset. Analysis module 700: used to extract and calculate rotor performance parameters based on the multi-scale flow field feature dataset to obtain open rotor flow field performance evaluation results.

[0093] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0094] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0095] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for measuring the flow field of an open rotor, characterized in that, include: S1: Based on the fluid dynamics neural network, the gradient distribution of the flow field is predicted. Based on the prediction results, the high gradient region and the low gradient region are determined. Through an adaptive sampling strategy, measurement points are arranged on multiple axial sections of the open rotor to obtain a sparsely distributed measurement point array. S2: Use an acoustic probe array to capture the sound pressure signal at the frequency of the blade passage, and calculate the phase angle information of the blade through the cross-correlation function to establish a non-contact phase synchronization system. S3: Under the triggering of the non-contact phase synchronization system, time-series data acquisition is performed on the measurement points in the measurement array to obtain sparse flow field data carrying spatiotemporal coordinates and phase information; S4: Input the sparse flow field data into the compressed sensing algorithm for reconstruction processing, and obtain a complete flow field database by inversion from sparse sampling points; S5: The flow field database is decomposed according to the time scale. Periodic unsteady flow components are extracted by phase averaging algorithm, and time-averaged flow field components are extracted by time averaging algorithm. Turbulent fluctuation components are calculated from the difference between instantaneous data and average components to obtain a multi-scale flow field feature dataset. S6: Extract and calculate the rotor performance parameters based on the multi-scale flow field feature dataset to obtain the open rotor flow field performance evaluation results.

2. The method for measuring the flow field of an open rotor according to claim 1, characterized in that, Step S1 further includes: S11: Construct a pre-trained fluid dynamics neural network model, input the flow field measurement data of the previous moment into the neural network model, predict and calculate the flow field gradient distribution of the next moment, and obtain the flow field gradient prediction result. S12: The measurement area is classified according to the flow field gradient prediction results, and high gradient areas and low gradient areas are obtained by gradient threshold determination. S13: Multiple measurement sections are set in front of the rotor, on the rotor plane and behind the rotor. For high gradient regions, dense sampling mode is used to arrange measurement points, and for low gradient regions, sparse sampling mode is used to arrange measurement points to obtain a measurement point array.

3. The method for measuring the flow field of an open rotor according to claim 2, characterized in that, The process of constructing the fluid dynamics neural network model in step S11 includes: S111: Collect historical flow field data of open rotors under various working conditions, and use the velocity, pressure and temperature components of the flow field as training samples. S112: Establish a physical information neural network and embed the Navier-Stokes equation as a physical constraint term in the network's loss function; S113: The physical information neural network is trained using the training samples, and the network parameters are continuously optimized through the backpropagation algorithm. The pre-trained fluid dynamics neural network model is obtained by judging the convergence of the validation set error.

4. The method for measuring the flow field of an open rotor according to claim 1, characterized in that, Step S2 further includes: S21: Arrange an array of acoustic probes around the rotor; S22: The acoustic pressure signal generated during rotor rotation is synchronously acquired through the acoustic probe array, and the characteristic peak value of the blade passing frequency is extracted by the spectrum analysis of the acoustic pressure signal; S23: Perform cross-correlation calculation on the sound pressure signals collected by multiple acoustic probes, calculate the time difference of the sound wave arriving at different probes, and back-calculate the spatial position of the blade based on the time difference and the probe spacing. S24: Combine the tachometer signal to perform time series calibration of the blade's spatial position, establish the correspondence between the time when the blade passes through the fixed reference point and the phase angle, and obtain a non-contact phase synchronization system.

5. The method for measuring the flow field of an open rotor according to claim 4, characterized in that, Step S22 further includes: S221: The acoustic pressure signal generated during rotor rotation is synchronously acquired through the acoustic probe array; S222: Perform a fast Fourier transform on the sound pressure signal to obtain the frequency domain spectrum; S223: Determine the fundamental frequency from the frequency corresponding to the spectral peak with the largest amplitude in the frequency domain spectrum, and mark the spectral peaks corresponding to the fundamental frequency and integer multiples of the fundamental frequency as the characteristic peaks of the blade passage frequency.

6. The method for measuring the flow field of an open rotor according to claim 4, characterized in that, The calculation method for the cross-correlation operation in step S23 is as follows: S231: Select any two acoustic probes in the acoustic probe array as a reference probe pair, and extract the sound pressure signal sequence of the two probes within the same time window; S232: Calculate the cross-correlation function of the two sound pressure signal sequences, and determine the time delay by the peak position of the cross-correlation function; S233: The sound wave propagation path difference is calculated based on the time delay and sound speed. The instantaneous position coordinates of the blade are then inferred by the triangulation algorithm in combination with the spatial spacing of the reference probe pair to obtain the spatial position of the blade.

7. The method for measuring the flow field of an open rotor according to claim 1, characterized in that, Step S4 further includes: S41: Establish the observation matrix of the sparse flow field data, wherein the number of rows of the observation matrix corresponds to the number of sparse sampling points, and the number of columns of the observation matrix corresponds to the number of spatial grids of the complete flow field; S42: Choose wavelet transform or Fourier transform as the sparse basis; S43: Construct an L1 norm minimization optimization problem, using the observation matrix, the sparse basis, and the sparse flow field data as inputs, and solve the optimization problem using the orthogonal matching pursuit algorithm or the basis pursuit algorithm. The complete flow field database is obtained by inverse transformation of the optimization solution.

8. The method for measuring the flow field of an open rotor according to claim 7, characterized in that, In step S43, the solution process of the orthogonal matching pursuit algorithm is as follows: S431: Initialize the residual vector to equal the sparse flow field data vector. In each iteration, select the column vector with the largest absolute value of the inner product with the residual vector from the column vectors of the sparse basis and add it to the support set. S432: Update the coefficients corresponding to the support set using the least squares method, calculate the updated residual vector, and terminate the iteration when the norm of the updated residual vector is less than the preset threshold or the number of iterations reaches the upper limit. The complete flow field data is obtained by reconstructing the final support set and coefficients.

9. The method for measuring the flow field of an open rotor according to claim 1, characterized in that, S6 specifically includes: Based on the multi-scale flow field feature dataset, the overall aerodynamic performance index is obtained through integral calculation. Based on the multi-scale flow field feature dataset, the tip vortex characteristic parameters are obtained through gradient calculation. Based on the multi-scale flow field feature dataset, the aerodynamic interference intensity coefficient is calculated by comparing and analyzing the front and rear cross-sectional data. The calculation process for the aerodynamic interference intensity coefficient includes: S631: Extract velocity deficit data of the front rotor wake at multiple axial sections from a multi-scale flow field feature dataset; S632: Based on velocity loss data, an exponential function fitting is used to establish the relationship curve between wake attenuation rate and axial distance; S633: Extract the phase-averaged flow field component and time-averaged flow field component of the rear rotor inlet section, calculate the root mean square deviation, and define the obtained root mean square deviation as the unsteady disturbance intensity coefficient. S634: Perform statistical analysis on the circumferential velocity distribution at the inlet section of the rear rotor, and calculate the flow field non-uniformity by the ratio of the standard deviation to the average value of the velocity distribution. S635: The aerodynamic disturbance intensity coefficient is obtained by weighted summation of the unsteady disturbance intensity coefficient and the flow field nonuniformity.

10. An open rotor flow field measurement system for performing an open rotor flow field measurement method as described in any one of claims 1 to 9, characterized in that, include: Prediction module: used to predict the gradient distribution of the flow field based on a fluid dynamics neural network and obtain the prediction results; Point placement module: Used to determine high gradient regions and low gradient regions based on prediction results, and to arrange measurement points on multiple axial sections of the open rotor through an adaptive sampling strategy to obtain a sparsely distributed measurement point array. Establishment module: Used to receive the sound pressure signal of the blade passage frequency captured by the acoustic probe array, and calculate the phase angle information of the blade through the cross-correlation function to establish a non-contact phase synchronization system; Acquisition module: used to acquire time-series data from the measurement points in the measurement array under the trigger of the non-contact phase synchronization system, and obtain sparse flow field data carrying spatiotemporal coordinates and phase information; Reconstruction module: used to input the sparse flow field data into the compressed sensing algorithm for reconstruction processing, and obtain a complete flow field database by inversion from sparse sampling points; Decomposition module: used to decompose the flow field database according to the time scale, extract periodic unsteady flow components through phase averaging algorithm, extract time-averaged flow field components through time averaging algorithm, and calculate turbulent fluctuation components from the difference between instantaneous data and average components to obtain multi-scale flow field feature dataset; Analysis module: used to extract and calculate rotor performance parameters based on the multi-scale flow field feature dataset to obtain the open rotor flow field performance evaluation results.

Images