Similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data

Abstract

According to the similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data, train number clusters and curve clusters are creatively proposed and designed, and aerodynamic laws of aerodynamic data can be represented and highlighted from multiple dimensions; ordered feature vectors representing aerodynamic data are designed in a four-order Chebyshev coefficient mode, and the method solves the problem that aerodynamic data formats are diversified due to inconsistent wind tunnel test states of aircrafts; the problem of diversification of aerodynamic data along state variable definition caused by inconsistent wind tunnel coordinate system definition and aircraft attitude definition is solved, and the range of available historical aerodynamic data is widened; and realizing similarity measurement of aerodynamic rules through an Euclidean distance method, and finally realizing historical data matching recommendation with similar aerodynamic rules.

Description

Technical Field

[0001] The present invention relates to a method for matching and recommending analysis of similar aerodynamic laws based on historical wind tunnel data, belonging to the field of aerodynamics. Background Art

[0002] First, briefly introduce the basic knowledge and industry terms involved in this patent. Wind tunnel test types include force measurement tests, pressure measurement tests, inlet tests, capture trajectory tests, thermal ablation tests, free flight tests, etc. Among them, force measurement tests are a basic and large proportion of wind tunnel tests, resulting in the largest amount of historical data accumulated in force measurement tests, which is a treasure trove waiting to be developed, but no effective method has been formed to fully explore and utilize this treasure trove.

[0003] The state of wind tunnel force measurement tests generally refers to the angle of attack α, sideslip angle β, roll angle , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r , and flight Mach number Ma.

[0004] The aerodynamic data obtained from wind tunnel force measurement tests generally refers to the six-component force and moment coefficients obtained by converting the electrical signals measured by the balance through formulas under different wind tunnel force measurement test states, specifically the forces in the X, Y, and Z directions and the moments in the Mx, My, and Mz directions.

[0005] Secondly, as a scarce resource, wind tunnels can no longer meet the needs of aircraft development. It is urgent to improve the efficiency of wind tunnel tests, and the rapid analysis of wind tunnel test data is the key to improving the efficiency of wind tunnel tests. At present, the analysis of wind tunnel test data mainly relies on aerodynamic experts with many years of wind tunnel work experience. With the sharp increase in the amount of wind tunnel tests, the purely manual method is inefficient, and there are problems such as large workload, insufficient personnel, and analysis misjudgment caused by fatigue operation, which seriously restricts the production efficiency of wind tunnels. Therefore, there is an urgent need for a fast and effective analysis method for the aerodynamic data obtained from wind tunnel tests.

[0006] Traditional aerodynamic data analysis methods usually conduct wind tunnel tests on a new aircraft shape and obtain aerodynamic data. When analyzing the aerodynamic data, it is hoped to find similar shapes in historical tests. The aerodynamic data of these similar shapes can be used to evaluate the correctness and reliability of the laws of newly obtained aerodynamic data, and at the same time as an important resource for tracing complex aerodynamic laws. This analysis method greatly improves the efficiency of data analysis and supports the smooth progress of a large number of new shape tests.

[0007] However, since the mapping relationship between aircraft shape and aerodynamic data is a high-dimensional (aerodynamic shape belongs to high-dimensional geometric features) to low-dimensional (usually six aerodynamic forces and moment coefficients) mapping, even if there are thousands of aerodynamic data of aircraft shapes, it still belongs to the category of "small sample". This leads to the limitation of the application scenarios of the above analysis method, and in most cases similar shapes cannot be found.

[0008] In fact, there are many cases where aircraft with very different appearances have similar aerodynamic laws. That is, searching for historical aerodynamic data with similar aerodynamic laws through the aerodynamic laws presented by the aerodynamic data is an analysis method purely based on aerodynamic laws. This is more valuable to some extent, because the analysis and mining of common aerodynamic laws is not only beneficial for wind tunnel data analysts to conduct discriminant analysis on newly generated aerodynamic data, but also can guide the overall design department of the aircraft to optimize the design of the aircraft shape, and even explore potential physical laws to achieve more valuable knowledge discovery.

[0009] The key to the wind tunnel test data analysis method based on similar aerodynamic laws is how to characterize the aerodynamic laws and measure the similarity of the aerodynamic laws. The characterization method of the aerodynamic laws contained in the traditional aerodynamic data generally takes one wind tunnel test train as the minimum research unit. The horizontal axis is generally the angle of attack α of the aircraft, and the vertical axis is the force and moment coefficient of a certain component. The other test states of the aircraft are the sideslip angle β, the roll angle , pitch rudder deflection d p 、Yaw rudder deflection d y 、Roll rudder deflection d r , the influence of flight Mach number Ma on aerodynamic force cannot be reflected in one wind tunnel test train, that is, when studying aerodynamic data analysis based on data-driven and similar aerodynamic law matching and recommendation, the traditional aerodynamic law characterization method cannot fully characterize the aerodynamic law, and there is randomness in a single wind tunnel test train. Therefore, there is an urgent need for an effective aerodynamic law characterization and similarity measurement method, which is an aerodynamic data analysis method based on the mining of historical wind tunnel big data and the matching and recommendation of similar aerodynamic laws. Summary of the invention

[0010] The technical problem solved by the present invention is: in view of the problem that the aerodynamic data of wind tunnel tests cannot be analyzed quickly and effectively in the current prior art, a similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data is proposed.

[0011] The present invention solves the above technical problems by the following technical solutions:

[0012] A similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data, including:

[0013] Construct an aerodynamic law feature database that characterizes aerodynamic laws using train trip clusters, curve clusters, and feature vectors;

[0014] According to the types of aerodynamic law feature data, perform single-component matching recommendation analysis and multi-component matching recommendation analysis on the aerodynamic law feature data of the same type, and obtain the single-component matching recommendation analysis results and multi-component matching recommendation analysis results of the selected type of aerodynamic law feature data;

[0015] Perform matching promotion analysis on the next type of aerodynamic law feature data until all types of aerodynamic law feature data in the aerodynamic law feature database are traversed.

[0016] The determination methods for train trip clusters, curve clusters, and feature vectors are as follows:

[0017] Number the historical aerodynamic data from wind tunnel force measurement tests according to the aircraft shape;

[0018] For the historical aerodynamic data of the same aircraft shape, group them according to the single-factor variable method based on the wind tunnel test state. For the aerodynamic data within the same group, select 2 train trips in a combined manner to construct Class A train trip clusters, select 3 train trips in a combined manner to construct Class B train trip clusters, and sort the train trips within the train trip clusters in ascending order of the value of the variable factor θ;

[0019] For the train trip data within the same train trip cluster, while keeping the order of the train trips within the train trip cluster unchanged, extract the corresponding aerodynamic data according to the six-component force and moment units X, Y, Z, Mx, My, and Mz to construct a curve cluster;

[0020] For the aerodynamic data curves within the curve cluster, use the fourth-order Chebyshev method for fitting to obtain the Chebyshev coefficients of each curve And construct a feature vector;

[0021] According to the obtained feature vectors and the six aerodynamic unit components X, Y, Z, Mx, My, and Mz, construct a feature database that characterizes aerodynamic laws.

[0022] The aircraft shape numbering uses 4 digits, with a range of [0000, 9999]. The aerodynamic data of the same aircraft shape uses the same number, and the aerodynamic data of different aircraft shapes uses different numbers, without order requirements;

[0023] The wind tunnel test states include the sideslip angle β, roll angle , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r , and six variables of Mach number Ma. The angle of attack α of the aircraft does not participate in the single-factor variable grouping;

[0024] The variable factor θ of any constructed train trip cluster is the sideslip angle β, the roll angle , the pitch rudder deflection d o , the yaw rudder deflection d y , the roll rudder deflection d r , and the Mach number Ma, any one of the six variables; the train trip cluster includes two specifications, A and B. The A train trip cluster contains 2 wind tunnel test train trips, and the B train trip cluster contains 3 wind tunnel test train trips.

[0025] The order of the curves within the curve cluster is consistent with the order of the train trips within the train trip cluster, and they are all sorted in ascending order according to the value of the variable factor θ; the curve cluster includes two specifications, A and B, and is divided into six aerodynamic components: X, Y, Z, Mx, My, and Mz;

[0026] The Chebyshev coefficient The subscript i represents the order of the Chebyshev coefficient, is the constant term, is the coefficient of the first-order term, is the coefficient of the second-order term, is the coefficient of the third-order term, is the coefficient of the fourth-order term;

[0027] The Chebyshev coefficient The superscript j represents the train trip sequence number sorted in ascending order of the variable factor θ within the curve cluster;

[0028] The eigenvectors include two specifications, A and B, corresponding to the A and B specifications of the train trip cluster and the curve cluster, where:

[0029] Type A eigenvector:

[0030] Type B eigenvector:

[0031] The characteristic database includes a type A aerodynamic law characteristic database, corresponding to the type A eigenvector, and contains six databases: A-X, A-Y, A-Z, A-Mx, A-My, and A-Mz;

[0032] The type B aerodynamic law characteristic database, corresponding to the type B eigenvector, contains six databases: B-X, B-Y, B-Z, B-Mx, B-My, and B-Mz.

[0033] The method for performing single-component matching recommendation analysis is:

[0034] Taking the type A curve cluster corresponding to any aerodynamic component as the minimum research object, call the type A aerodynamic law characteristic data through the aerodynamic law characteristic database;

[0035] Calculate the Euclidean distances between the feature vector T of the curve cluster to be matched and the feature vectors of all other curve clusters, and take the top N with the Euclidean distances sorted from small to large, denoted as the preliminary result R1;

[0036] Re - sort the feature vector T of the curve cluster to be matched in descending order according to the variable factor θ. The adjusted feature vector is denoted as the reverse - ordered feature vector Tr, and determine the type - A reverse - ordered feature vector and type - B reverse - ordered feature vector;

[0037] Calculate the Euclidean distances between the reverse - ordered feature vector Tr of the curve cluster to be matched and the feature vectors of all other curve clusters, and take the top M with the Euclidean distances sorted from small to large, denoted as the preliminary result R2;

[0038] Merge R1 and R2, and take the top K with the Euclidean distances sorted from small to large, denoted as the matching recommendation result R of a certain single - component aerodynamic data.

[0039] The N takes an integer between [1, 10]; M takes an integer between [1, 10]; a small Euclidean distance represents a high degree of similarity of aerodynamic laws. The type - A reverse - ordered feature vector Type - B reverse - ordered feature vector:

[0040] The method for performing multi - component matching recommendation analysis is as follows:

[0041] Taking the train - trip cluster as the smallest research object, perform matching recommendations for the six groups of curve clusters X, Y, Z, Mx, My, and Mz included in the train - trip cluster according to the similarity of aerodynamic laws respectively, and combine them according to the principle that the six matching results come from the same train - trip cluster as the matching recommendation result of the train - trip cluster;

[0042] Calculate the corresponding feature vectors T x ,T y ,T z ,T mx ,T my ,T mz of the six groups of curve clusters X, Y, Z, Mx, My, and Mz included in the train - trip cluster to be matched respectively with the feature vectors of the six groups of curve clusters included in other train - trip clusters, and ensure that the six Euclidean distances come from the calculation results of the same train - trip cluster and the train - trip cluster to be matched;

[0043] Weight - sum the six Euclidean distances to obtain the metric result dis of the train - trip cluster to be matched and this train - trip cluster;

[0044] Take the top N with the metric result dis of the train - trip cluster sorted from small to large, denoted as the preliminary result RR1;

[0045] The six groups of curve clusters X, Y, Z, Mx, My, and Mz included in the train trip cluster to be matched correspond to the feature vector T x , T y , T z , T mx , T my , T mz Reorder according to the variable factor θ from large to small, and the adjusted feature vector is denoted as the reverse-order feature vector T x _r, T y _r, T z _r, T mx _r, T my_r , T mz _r;

[0046] Calculate the reverse-order feature vector T x _r, T y _r, T z _r, T mx _r, T my_r , T mz _r and the six Euclidean distances of the feature vectors of other train trip clusters, and weighted sum to obtain the measurement results with other train trip clusters. The weighting coefficients are the same as those set in the single-component matching recommendation analysis. Take the top M names sorted from small to large of the measurement results and denote them as the preliminary result RR2;

[0047] Merge RR1 and RR2, and take the top K names sorted from small to large of the measurement results of the train trip clusters and denote them as the multi-component aerodynamic data matching recommendation result RR.

[0048] The determination method of the measurement result dis between the train trip cluster to be matched and this train trip cluster is:

[0049] dis = α1 * dis_x + α2 * dis_y + α3 * dis_z + α4 * dis_mx + α5 * dis_m + α6 * dis_mz

[0050] Among them, the weighting coefficients α1, α2, α3, α4, α5, α6 take values from [8, 4, 2, 1, 0, 0.5, 0.25, 0.125], and the same values are allowed to be taken;

[0051] After the weighting coefficients α1, α2, α3, α4, α5, α6 are selected, the same set of weighting coefficients is used for the calculation of all train trip clusters; the weighting coefficient can be selected as 0, indicating that this aerodynamic component does not participate in the matching measurement.

[0052] The train trip cluster, curve cluster, and feature vector only include the aerodynamic law characteristic data of type A and type B. When data matching recommendation analysis is required, the required data is directly called from the aerodynamic law characteristic database.

[0053] The advantages of the present invention compared with the prior art are as follows:

[0054] (1) A similarity aerodynamic law matching recommendation analysis method based on wind tunnel historical data provided by the present invention proposes a design method for train trip clusters and curve clusters, and a new method for characterizing aerodynamic laws using train trip clusters and curve clusters. Since the train trip clusters and curve clusters are cyclically constructed in a single-factor variable manner according to the aircraft test state, it can be ensured that a train trip cluster can not only reflect the variation law of aerodynamic data along the angle of attack α, but also reflect the variation law of aerodynamic data along the variable factor θ. And the cyclically constructed train trip clusters and curve clusters will traverse all wind tunnel test states, so all train trip clusters and curve clusters will reflect the variation law of aerodynamic data along all wind tunnel test states;

[0055] (2) The train trip clusters and curve clusters cyclically constructed by the present invention in a single-factor variable manner according to the aircraft test state enrich the aerodynamic data samples, highlight the details of the aerodynamic laws that the aerodynamic data can present along each test variable, and enrich the aerodynamic laws;

[0056] (3) In the present invention, the train trips or curves contained in the train trip clusters and curve clusters are sorted according to the variable factor θ instead of being randomly combined. Only the ordered train trip clusters and curve clusters can present the variation trend and degree of the aerodynamic laws along the variable factor θ, and it is also convenient to perform operations and comparisons on the aerodynamic laws;

[0057] (4) The train trip clusters of the present invention have two specifications, A and B, which contain 2 and 3 wind tunnel test train trips respectively, and the constructed train trip clusters need to be sorted in ascending order according to the variable factor θ. The above design is based on the following considerations: 2 wind tunnel test train trips can determine the vector change direction of the aerodynamic data along the variable factor θ, and 3 wind tunnel test train trips are a further confirmation of this vector direction, and can measure the degree of influence of the variable factor θ on the aerodynamic data, that is, the step size of the change of the aerodynamic data along the variable factor θ direction; It is not necessary to include train trip clusters with 4 or more train trips because wind tunnel tests are high-cost tests. In order to reduce costs, after grouping according to the wind tunnel test state, the number of wind tunnel test train trips included in each group is very small. If the train trip clusters contain too many train trips, it will cause a sharp reduction in the matchable data, and instead miss many aerodynamic laws;

[0058] (5) The present invention selects the fitting coefficients of Chebyshev to construct the feature vector, solves the problem of unified measurement of aerodynamic data, and enables the value of a large amount of wind tunnel test data to be utilized; at the same time, the Chebyshev fitting method is selected, which can effectively reduce the fluctuation of high-order polynomial fitting and effectively avoid the Runge phenomenon compared with polynomial and other fitting methods, and increase the stability;

[0059] (6) The Chebyshev fitting method selected in the present invention can better control the distribution of errors in the entire angle of attack range compared to polynomial and other fitting methods, avoid excessive local errors, and achieve the effect of minimizing the maximum error;

[0060] (7) In the present invention, the constant term and the first-order slope trend term in the linear region can characterize the main aerodynamic laws. The second-order coefficient can describe the aerodynamic laws in a more refined manner. In the non-linear region, the second-order coefficient and the third-order coefficient can characterize the main aerodynamic laws, and the fourth-order coefficient can describe them in a more refined manner. The present invention selects the fourth-order Chebyshev fitting method based on the experience of aerodynamic experts and the fitting test of aerodynamic data curves; the similarity of aerodynamic laws can be measured by a simple Euclidean distance, and the smaller the distance, the higher the similarity of aerodynamic laws; and considering the sideslip angle β, roll angle , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r are defined differently, and there are also differences in different wind tunnel test coordinate systems, resulting in complex wind tunnel historical aerodynamic data and inconsistent definitions according to a unified standard. The present invention can adapt to all wind tunnel historical aerodynamic data through eigenvectors and their reverse operations, broadening the scope of application of historical aerodynamic data and enabling the discovery and utilization of more valuable aerodynamic laws;

[0061] (8) The setting of weighting coefficients in the present invention facilitates highlighting valuable aerodynamic components and weakening aerodynamic components with low value or even interference, because the influence degrees of different wind tunnel test state variables on aerodynamic components are different. When analyzing aerodynamic data, if focusing on the influence of pitch rudder deflection d p on aerodynamic data, it is necessary to focus on the Y and Mz components. If wanting to explore how yaw rudder deflection d y affects aerodynamic data, it is necessary to focus on the Z and My components, and roll rudder deflection d r has a greater influence on Mx, so the weighting coefficients can be formulated based on the aerodynamic components concerned by the new shape aircraft, and more suitable aerodynamic laws can be matched from the database. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 is the flowchart of the similar aerodynamic law matching and recommendation analysis method based on wind tunnel historical big data provided by the present invention;

[0063] Figure 2 is the simple topological structure diagram of the aerodynamic law feature database provided by the present invention;

[0064] Figure 3 is the flowchart of the matching and recommendation analysis of the Mz component corresponding to the A-specification train set curve cluster and eigenvector provided by the present invention;

[0065] Figure 4 This is the multi-component matching recommendation analysis flowchart corresponding to the train trip cluster curve cluster and feature vector provided by the present invention. Detailed implementation manners

[0066] A method for matching and recommending analysis of similar aerodynamic laws based on wind tunnel historical data creatively proposes and designs a train trip cluster and a curve cluster, which can characterize and highlight the aerodynamic laws of aerodynamic data from multiple dimensions; designs an ordered feature vector for characterizing aerodynamic data by using the fourth-order Chebyshev coefficient. This method not only solves the problem of diverse aerodynamic data formats caused by inconsistent wind tunnel test states of aircraft, but also solves the problem of diverse definitions of aerodynamic data along state variables caused by inconsistent definitions of wind tunnel coordinate systems and aircraft attitudes, broadening the scope of available historical aerodynamic data; realizes the similarity measurement of aerodynamic laws through the Euclidean distance method, and finally realizes the matching recommendation of historical data with similar aerodynamic laws.

[0067] The method for matching and recommending analysis of similar aerodynamic laws based on wind tunnel historical data includes:

[0068] Construct an aerodynamic law feature database for characterizing aerodynamic laws with a train trip cluster, a curve cluster, and a feature vector;

[0069] Conduct single-component matching recommendation analysis and multi-component matching recommendation analysis for aerodynamic law feature data of the same type respectively according to the type of aerodynamic law feature data, and obtain the single-component matching recommendation analysis result and multi-component matching recommendation analysis result of the selected type of aerodynamic law feature data;

[0070] Conduct the matching promotion analysis for the next type of aerodynamic law feature data until all types of aerodynamic law feature data in the aerodynamic law feature database are traversed.

[0071] The determination methods for the train trip cluster, the curve cluster, and the feature vector are as follows:

[0072] Number the historical aerodynamic data of the wind tunnel force measurement test according to the aircraft shape;

[0073] For the historical aerodynamic data of the same aircraft shape, group them according to the single-factor variable method based on the wind tunnel test state. For the aerodynamic data in the same group, select 2 trips to construct an A-type train trip cluster in a combined manner, select 3 trips to construct a B-type train trip cluster in a combined manner, and sort the trips in the train trip cluster in ascending order of the value of the variable factor θ;

[0074] For the trip data in the same train trip cluster, while keeping the order of the trips in the train trip cluster unchanged, extract the corresponding aerodynamic data according to the six-component force and moment units X, Y, Z, Mx, My, and Mz respectively to construct a curve cluster;

[0075] For the aerodynamic data curves within the curve cluster, the fourth-order Chebyshev method is used for fitting to obtain the Chebyshev coefficients of each curve respectively. And construct eigenvectors;

[0076] According to the obtained eigenvectors and the six aerodynamic unit components of X, Y, Z, Mx, My, and Mz, construct a characteristic database representing the aerodynamic law.

[0077] The aircraft shape number is a 4-digit number, ranging from [0000, 9999]. The aerodynamic data of the same aircraft shape uses the same number, and the aerodynamic data of different aircraft shapes uses different numbers, without order requirements;

[0078] The wind tunnel test state includes the sideslip angle β and roll angle of the aircraft , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r , and six variables of Mach number Ma. The angle of attack α of the aircraft is not involved in the single-factor variable grouping;

[0079] The variable factor θ of any constructed train trip cluster is the sideslip angle β, roll angle , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r , and any one of the six variables of Mach number Ma; the train trip cluster includes two specifications, A and B. The A train trip cluster contains 2 wind tunnel test trips, and the B train trip cluster contains 3 wind tunnel test trips.

[0080] The order of the curves within the curve cluster is the same as the order of the train trips within the train trip cluster, both sorted in ascending order according to the value of the variable factor θ; the curve cluster includes two specifications, A and B, and is divided into six types according to the aerodynamic components: X, Y, Z, Mx, My, and Mz;

[0081] The said Chebyshev coefficient The subscript i represents the order of the Chebyshev coefficient, is the constant term, is the first-order term coefficient, is the second-order term coefficient, is the third-order term coefficient, is the fourth-order term coefficient;

[0082] The Chebyshev coefficient The superscript j represents the train trip sequence number sorted in ascending order according to the variable factor θ within the curve cluster;

[0083] The eigenvectors include two specifications, A and B, corresponding to the two specifications, A and B, of the train trip cluster and the curve cluster, where:

[0084] Type A feature vector:

[0085] Type B feature vector:

[0086] The feature database includes a Type A aerodynamic law feature database, corresponding to the Type A feature vector, which contains six databases: A-X, A-Y, A-Z, A-Mx, A-My, and A-Mz;

[0087] A Type B aerodynamic law feature database, corresponding to the Type B feature vector, which contains six databases: B-X, B-Y, B-Z, B-Mx, B-My, and B-Mz.

[0088] The method for performing single-component matching recommendation analysis is as follows:

[0089] Taking the Type A curve cluster corresponding to any aerodynamic component as the minimum research object, call the Type A aerodynamic law feature data through the aerodynamic law feature database;

[0090] Calculate the Euclidean distance between the feature vector T of the curve cluster to be matched and the feature vectors of all other curve clusters, and take the top N in the ascending order of the Euclidean distance, denoted as the preliminary result R1;

[0091] Re-sort the feature vector T of the curve cluster to be matched in descending order of the variable factor θ, and the adjusted feature vector is denoted as the reverse-order feature vector T_r, and determine the Type A reverse-order feature vector and the Type B reverse-order feature vector;

[0092] Calculate the Euclidean distance between the reverse-order feature vector T_r of the curve cluster to be matched and the feature vectors of all other curve clusters, and take the top M in the ascending order of the Euclidean distance, denoted as the preliminary result R2;

[0093] Merge R1 and R2, and take the top K in the ascending order of the Euclidean distance, denoted as the matching recommendation result R of a certain single-component aerodynamic data.

[0094] N takes an integer between [1, 10]; M takes an integer between [1, 10]; a small Euclidean distance represents a high degree of aerodynamic law similarity. Type A reverse-order feature vector: Type B reverse-order feature vector:

[0095] The method for performing multi-component matching recommendation analysis is as follows:

[0096] Taking the train trip cluster as the smallest research object, six groups of curve clusters X, Y, Z, Mx, My, and Mz included in the train trip cluster are respectively matched and recommended according to the similarity of aerodynamic laws, and combined based on the principle that the six matching results come from the same train trip cluster as the matching recommendation result of the train trip cluster;

[0097] Calculate the feature vectors T corresponding to the six groups of curve clusters X, Y, Z, Mx, My, and Mz included in the train trip cluster to be matched x , T y , T z , T mx , T my , T mz and the six Euclidean distances dis_x, dis_y, dis_z, dis_mx, dis_my, and dis_mz of the feature vectors corresponding to the six groups of curve clusters included in other train trip clusters, ensuring that the six Euclidean distances come from the calculation results of the same train trip cluster and the train trip cluster to be matched;

[0098] Weighted sum the six Euclidean distances to obtain the metric result dis of the train trip cluster to be matched and this train trip cluster;

[0099] Take the top N of the metric results dis of the train trip cluster sorted from small to large, denoted as the preliminary result RR1;

[0100] The feature vectors T corresponding to the six groups of curve clusters X, Y, Z, Mx, My, and Mz included in the train trip cluster to be matched x , T y , T z , T mx , T my , T mz Re - sort in descending order according to the variable factor θ, and the adjusted feature vectors are denoted as the reverse - order feature vectors T x _r, T y _r, T z _r, T mx _r, T my_r , T mz _r;

[0101] Calculate the six Euclidean distances between the reverse - order feature vectors T x _r, T y _r, T z _r, T mx _r, T my_r , T mz _r of the train trip cluster to be matched and the feature vectors of other train trip clusters, and weighted sum to obtain the metric result with other train trip clusters. The weighting coefficients are the same as those set in the single - component matching recommendation analysis. Take the top M of the metric results sorted from small to large, denoted as the preliminary result RR2;

[0102] Merge RR1 and RR2, and select the top K results of the measured values of the train trip clusters sorted from small to large, denoted as the multi-component aerodynamic data matching recommendation result RR.

[0103] The method for determining the train trip cluster to be matched and the measurement result dis of this train trip cluster is as follows:

[0104] dis = α1 * dis_x + α2 * dis_y + α3 * dis_z + α4 * dis_mx + α5 * dis_m + α6 * dis_mz

[0105] Among them, the weighting coefficients α1, α2, α3, α4, α5, α6 are selected from [8, 4, 2, 1, 0, 0.5, 0.25, 0.125], and the same value is allowed to be taken;

[0106] After the weighting coefficients α1, α2, α3, α4, α5, α6 are selected, the same set of weighting coefficients is used for the calculation of all train trip clusters; the weighting coefficient can be selected as 0, indicating that this aerodynamic component does not participate in the matching measurement.

[0107] The train trip clusters, curve clusters, and eigenvectors only include type A and type B aerodynamic law characteristic data. When data matching recommendation analysis is required, the required data is directly called from the aerodynamic law characteristic database.

[0108] The following is a further description in conjunction with the accompanying drawings of the specification and the preferred embodiments:

[0109] In the current embodiment, the similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data is as Figure 1 shown:

[0110] The first step: Design a method to characterize the aerodynamic law with train trip clusters, curve clusters, and eigenvectors, and construct an aerodynamic law characteristic database;

[0111] S1. Number the historical aerodynamic data accumulated from wind tunnel force measurement tests according to the aircraft shape;

[0112] The numbering uses 4-digit numbers, and the range is [0000, 9999]. As long as the numbering ensures that the aerodynamic data of the same aircraft shape uses the same number, and the aerodynamic data of different aircraft shapes uses different numbers, there is no order requirement;

[0113] Selecting 4 significant figures can encode 10,000 aircraft shapes, which is sufficient to cover the aircraft wind tunnel test shapes in recent decades;

[0114] S2. Group the historical aerodynamic data of the same aircraft configuration according to the single-factor variable method based on the wind tunnel test conditions. For the aerodynamic data within the same group, select 2 test runs in a combined manner to construct Class A test run clusters, select 3 test runs in a combined manner to construct Class B test run clusters, and sort the test runs within the test run clusters in ascending order of the value of the variable factor θ.

[0115] The wind tunnel test conditions include the sideslip angle β, roll angle , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r , and Mach number Ma, a total of six variables. The angle of attack α of the aircraft is not involved in the single-factor variable grouping.

[0116] For a constructed test run cluster, the variable factor θ of the test run cluster is one of the six variables: sideslip angle β, roll angle , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r , and Mach number Ma.

[0117] The test run clusters have two specifications, A and B. Among them, Class A test run clusters contain 2 wind tunnel test runs, and Class B test run clusters contain 3 wind tunnel test runs.

[0118] Table 1 shows a schematic diagram of the variable representation of the wind tunnel test conditions for an aircraft configuration. The variables for each state have been arranged in ascending order. Taking the Mach number state as an example, there are a total of four Mach numbers, and the values are ma1 < ma2 < ma3 < ma4.

[0119] Table 1 Variable Table of Wind Tunnel Test Conditions for a Certain Aircraft Configuration

[0120]

[0121] The first example of constructing test run clusters by the single-factor variable method is shown in Table 2: The variable factor is the yaw rudder deflection. There are 6 Class A test run clusters: The variable factor values are [dy1, dy2], [dy1, dy3], [dy1, dy4], [dy2, dy3], [dy2, dy4], [dy3, dy4], and the variable values of other wind tunnel test conditions are the bold italic large-sized values in Table 2.

[0122] Table 2 The First Example of Constructing Test Run Clusters by the Single-Factor Variable Method

[0123]

[0124] The second example of constructing train trip clusters by the single-factor variable method is shown in Table 3: The variable factor is Mach number, and there are 4 train trip clusters in Cluster B: The variable factor values are [ma1, ma2, ma3], [ma1, ma2, ma4], [ma1, ma3, ma4], [ma2, ma3, ma4], and the variable values of other wind tunnel test states are the bold, italic, and large-sized values in Table 3;

[0125] Table 3 The first example of constructing train trip clusters by the single-factor variable method

[0126]

[0127]

[0128] The present invention creatively designs the concept of train trip clusters. Since the train trip clusters are circularly constructed in the way of single-factor variables according to the aircraft test states, it can ensure that a train trip cluster can not only reflect the variation law of aerodynamic data along the angle of attack α, but also reflect the variation law of aerodynamic data along the variable factor θ. And the circularly constructed train trip clusters will traverse all wind tunnel test states, so all the train trip clusters will reflect the variation law of aerodynamic data along all wind tunnel test states;

[0129] There are two specifications for train trip clusters, namely A and B, which contain 2 and 3 wind tunnel test train trips respectively. And the constructed train trip clusters need to be sorted in ascending order according to the variable factor θ. The above design is based on the following considerations: 2 wind tunnel test train trips can determine the vector change direction of aerodynamic data along the variable factor θ, and 3 wind tunnel test train trips are a further confirmation of this vector direction and can measure the degree of influence of the variable factor θ on aerodynamic data, that is, the step size of the change of aerodynamic data along the variable factor θ direction can be determined; It is not necessary to have train trip clusters containing 4 or more train trips because wind tunnel test is a high-cost test. In order to reduce costs, after grouping according to wind tunnel test states, the number of wind tunnel test train trips included in each group is very small. If the train trip clusters contain too many train trips, it will lead to a sharp reduction in matchable data and instead miss many aerodynamic laws.

[0130] S3. For the train trip data within the same train trip cluster, while ensuring the order of the train trips within the train trip cluster remains unchanged, extract the corresponding aerodynamic data according to the six-component forces and moments units X, Y, Z, Mx, My, and Mz respectively to construct a curve cluster. The order of the curves within the curve cluster is the same as the order of the train trips within the train trip cluster, and they are all sorted in ascending order according to the value of the variable factor θ;

[0131] The curve cluster also has two specifications, namely A and B, and is divided into six types according to aerodynamic components: X, Y, Z, Mx, My, and Mz;

[0132] By introducing the concept of a curve cluster, since aerodynamic data analysis is generally carried out for aerodynamic components, it is meaningful to analyze the same aerodynamic components, and it is almost meaningless to conduct cross-analysis on different aerodynamic components;

[0133] S4. For the aerodynamic data curves within the curve cluster, use the fourth-order Chebyshev method for fitting to obtain the Chebyshev coefficients of each curve respectively and construct the eigenvector;

[0134] Chebyshev coefficient The subscript i represents the order of the Chebyshev coefficient, is the constant term, is the coefficient of the first-order term, is the coefficient of the second-order term, is the coefficient of the third-order term, is the coefficient of the fourth-order term;

[0135] Chebyshev coefficient The superscript j represents the train number sequence number sorted from small to large by the variable factor θ within the curve cluster;

[0136] The finally constructed eigenvector corresponds to the two specifications A and B of the train cluster and the curve cluster, and there are also two specifications:

[0137] Type A eigenvector:

[0138] Type B eigenvector:

[0139] Due to the wind tunnel tests of different aircraft configurations, the length of the angle of attack sequence and the minimum and maximum angles of attack may all be different, resulting in the inability to perform mathematical operation-based comparisons on the aerodynamic data of different aircraft configurations. The present invention selects the Chebyshev fitting coefficients to construct the eigenvector, solving the problem of unified measurement of aerodynamic data and enabling the utilization of the value of a large amount of wind tunnel test data;

[0140] One of the purposes of wind tunnel tests is to obtain the safe flight envelope of the aircraft, and it rarely explores the extreme boundaries. The situation where the aerodynamic data generated by wind tunnel tests fluctuates violently can be almost ignored. The present invention selects the Chebyshev fitting method, which can effectively reduce the fluctuations of high-order polynomial fitting and effectively avoid the Runge phenomenon and increase stability compared with polynomial and other fitting methods;

[0141] Since the values of aerodynamic data at each angle of attack are very important, the Chebyshev fitting method selected by the present invention can better control the distribution of errors in the entire angle of attack interval, avoid locally excessive errors, and obtain the effect of minimizing the maximum error compared with polynomial and other fitting methods;

[0142] Since the aerodynamic data can be roughly divided into a small angle of attack linear region and a large angle of attack non-linear region according to the curve shape, for the linear region, the constant term and the first-order slope trend term can characterize the main aerodynamic laws, and the second-order coefficient can describe the aerodynamic laws more precisely. For the non-linear region, the second-order coefficient and the third-order coefficient can characterize the main aerodynamic laws, and the fourth-order coefficient can describe them more precisely. The present invention selects the fourth-order Chebyshev fitting method based on the experience of aerodynamic experts and the fitting test of the aerodynamic data curve;

[0143] S5. Based on the above A and B eigenvectors and the six aerodynamic unit components of X, Y, Z, Mx, My, and Mz, construct a feature database representing the aerodynamic laws;

[0144] The A-class aerodynamic law feature database, corresponding to the A-class eigenvector, includes six databases: A-X, A-Y, A-Z, A-Mx, A-My, and A-Mz;

[0145] The B-class aerodynamic law feature database, corresponding to the B-class eigenvector, includes six databases: B-X, B-Y, B-Z, B-Mx, B-My, and B-Mz:

[0146] For the comparative analysis of the aerodynamic data of different aircraft shapes, it is necessary to analyze them one by one according to the aerodynamic unit components, and cross-comparative analysis between the aerodynamic unit components is not allowed;

[0147] Such as Figure 2 shown, is the simple topological structure diagram of the aerodynamic law feature database of the embodiment of the present invention. In the figure, taking the aircraft shape numbered 0000 as an example, it shows all the aerodynamic data of the aircraft shape, the train trip clusters, the curve clusters corresponding to the six aerodynamic components, and the topological structure relationship between the eigenvectors corresponding to the six aerodynamic components;

[0148] The second step: For the aerodynamic laws characterized by the A-class train trip clusters, A-class curve clusters, and the corresponding A-class eigenvectors constructed by 2 train trips, conduct single-component matching recommendation analysis according to the similarity degree of the aerodynamic laws;

[0149] S6. Taking the A-class curve cluster of a certain aerodynamic component as the smallest research object, conduct matching recommendation in the A-class aerodynamic law feature database to which the aerodynamic component belongs;

[0150] S7. Calculate the Euclidean distance between the eigenvector T of the curve cluster to be matched and the eigenvectors of all other curve clusters, and take the top N in the ascending order of the Euclidean distance, denoted as the preliminary result R1;

[0151] Among them, N can take an integer between [1, 10];

[0152] A small Euclidean distance represents a high degree of similarity of aerodynamic laws;

[0153] S8. Reorder the eigenvectors T of the curve cluster to be matched in descending order according to the variable factor θ. The adjusted eigenvector is denoted as the reverse-order eigenvector Tr. The reverse-order eigenvectors corresponding to specifications A and B are as follows:

[0154] Reverse-order eigenvector of type A:

[0155] Reverse-order eigenvector of type B:

[0156] S9. Calculate the Euclidean distances between the reverse-order eigenvector Tr of the curve cluster to be matched and the eigenvectors of all other curve clusters, and select the top M with the Euclidean distances sorted from small to large, denoted as the preliminary result R2;

[0157] Among them, M can be an integer between [1, 10];

[0158] Considering that the sideslip angle β, roll angle , pitch rudder deflection d p , yaw rudder deflection d y , roll rudder deflection d r in the wind tunnel test state of different aircraft shapes are defined differently, and there are also differences in different wind tunnel test coordinate systems. Matching and measuring the eigenvectors of the curve cluster only by sorting them in ascending order along the variable factor θ will lose important aerodynamic laws. It is necessary to sort the eigenvectors of the curve cluster in descending order along the variable factor θ and measure them again to adapt to the differences brought by the definition of the wind tunnel coordinate system and the definition of the test state direction;

[0159] S10. Merge R1 and R2, and select the top K with the Euclidean distances sorted from small to large, denoted as the matching recommendation result R of a single-component aerodynamic data;

[0160] Among them, K can be an integer between [1, 10];

[0161] As Figure 3 shown, the flowchart of the matching recommendation analysis of the Mz component corresponding to the curve cluster and eigenvector of the vehicle type cluster of specification A in the embodiment of the present invention;

[0162] The third step: Carry out multi-component matching recommendation analysis on the aerodynamic laws characterized by the type A vehicle type cluster, type A curve cluster, and corresponding type A eigenvectors constructed by 2 vehicle types according to the similarity of aerodynamic laws;

[0163] S11. Take the vehicle type cluster as the smallest research object, perform matching recommendations on the six groups of curve clusters of X, Y, Z, Mx, My, and Mz included in the vehicle type cluster according to the similarity of aerodynamic laws, and then combine them together according to the principle that the six matching results come from the same vehicle type cluster as the matching recommendation result of the vehicle type cluster;

[0164] S12. Calculate the feature vectors T corresponding to the six groups of curve clusters X, Y, Z, Mx, My, and Mz included in the train trip cluster to be matched respectively. x , T y , T z , T mx , T my , T mz Calculate the six Euclidean distances dis_x, dis_y, dis_z, dis_mx, dis_my, and dis_mz between the feature vectors corresponding to the six groups of curve clusters included in this train trip cluster and those of other train trip clusters, ensuring that the six Euclidean distances are the calculation results from the same train trip cluster and the train trip cluster to be matched.

[0165] S13. Perform a weighted sum of the six Euclidean distances to obtain the metric result dis between the train trip cluster to be matched and this train trip cluster.

[0166] dis = α1 * dis_x + α2 * dis_y + α3 * dis_z + α4 * dis_mx + α5 * dis_m + α6 * dis_mz

[0167] Among them, the weighting coefficients α1, α2, α3, α4, α5, and α6 are taken from [8, 4, 2, 1, 0, 0.5, 0.25, 0.125], and the same value is allowed to be taken.

[0168] After the weighting coefficients α1, α2, α3, α4, α5, and α6 are selected, the calculation of all train trip clusters must use the same set of weighting coefficients.

[0169] The setting of the weighting coefficients is convenient for highlighting valuable aerodynamic components and weakening aerodynamic components with low value or even interference, because different wind tunnel test state variables have different degrees of influence on aerodynamic components. When analyzing aerodynamic data, if focusing on the influence of the pitch rudder deflection d p on aerodynamic data, it is necessary to focus on the Y and Mz components. If you want to explore how the yaw rudder deflection d y affects aerodynamic data, it is necessary to focus on the Z and My components, while the roll rudder deflection d r has a relatively large influence on Mx. Therefore, the weighting coefficients can be formulated based on the aerodynamic components concerned by the new-shaped aircraft to match more appropriate aerodynamic laws from the database.

[0170] It should be particularly noted that the weighting coefficient can be 0. When the coefficient of this aerodynamic component is 0, it means that this aerodynamic component does not participate in the matching. Because the aerodynamic data of the Mx unit of many aircraft configurations are extremely small values, the curve shapes are messy and the repeatability of the results of multiple wind tunnel tests is poor. It can be considered that the Mx aerodynamic data of such aircraft are random interference quantities, and no available aerodynamic laws can be mined, and thus no matching metric can be carried out. Therefore, the design that the weighting coefficient can be selected as 0 is very important.

[0171] S14. Take the top N results of the metric results dis of the train number clusters sorted from small to large, and record them as the preliminary result RR1;

[0172] Among them, N can take an integer between [1, 10];

[0173] S15. Reorder the six groups of curve cluster corresponding eigenvectors T x , T y , T z , T mx , T my , T mz included in the train number cluster to be matched in descending order according to the variable factor θ. The adjusted eigenvector is recorded as the reverse-order eigenvector T x _r, T y _r, T z _r, T mx _r, T my_r , T mz _r;

[0174] S16. Calculate the six Euclidean distances between the reverse-order eigenvectors T x _r, T y _r, T z _r, T mx _r, T my_r , T mz _r of the train number cluster to be matched and the eigenvectors of other train number clusters, and sum them with weights to obtain the metric results with other train number clusters. The weighting coefficient is the same as that in step S13. Take the top M results of the metric results sorted from small to large, and record them as the preliminary result RR2;

[0175] Among them, M can take an integer between [1, 10];

[0176] S17. Combine RR1 and RR2, and take the top K results of the metric results of the train number clusters sorted from small to large, and record them as the multi-component aerodynamic data matching recommendation result RR;

[0177] Among them, K can take an integer between [1, 10];

[0178] Such as Figure 4As shown in the figure, it is the multi-component matching recommendation analysis flowchart corresponding to the B-specification train cluster curve cluster and feature vector in the embodiment of the present invention;

[0179] Fourthly, for the aerodynamic laws characterized by the B-class train cluster, B-class curve cluster, and corresponding B-class feature vectors constructed by 3 trains, conduct single-component matching recommendation analysis according to the similarity of aerodynamic laws, and the operation steps are the same as those of S6 to S10 above;

[0180] Fifthly, for the aerodynamic laws characterized by the B-class train cluster, B-class curve cluster, and corresponding B-class feature vectors constructed by 3 trains, conduct multi-component matching recommendation analysis according to the similarity of aerodynamic laws, and the operation steps are the same as those of S11 to S17 above;

[0181] In actual wind tunnel tests, for the new shape aircraft tests being carried out in the wind tunnel, after obtaining the aerodynamic data of a small number of trains, the train cluster, curve cluster, and feature vector to be matched for the new shape aircraft can be constructed according to the train cluster, curve cluster, and feature vector construction methods proposed by the present invention. According to the method of the present invention, the train cluster and curve cluster with similar aerodynamic laws in the wind tunnel historical database can be obtained, so that other important information of the historical aircraft with similar aerodynamic laws can be traced based on the matching recommendation results.

[0182] Because the historical aircraft has completed all state wind tunnel tests and has more aerodynamic data, and all these data are correct and reliable, while only a small amount of aerodynamic data has been obtained in the ongoing wind tunnel test of the new shape aircraft. Aerodynamic experts can assist in analyzing the aerodynamic data and aerodynamic laws of the new shape aircraft and the change trend of the laws along the test state variables based on the aerodynamic data of the historical aircraft and the aerodynamic laws and the change trend of the laws presented along the test state variables, improve the analysis efficiency of aerodynamic data, and thus improve the wind tunnel test efficiency.

[0183] Although the present invention has been disclosed above with preferred embodiments, it is not used to limit the present invention. Any person skilled in the art can make possible changes and modifications to the technical solution of the present invention without departing from the spirit and scope of the present invention. Therefore, any simple modification, equivalent change, and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention all belong to the protection scope of the technical solution of the present invention.

[0184] The content not detailedly described in the specification of the present invention belongs to the well-known technology of those skilled in the art.

Claims

1. A similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data, characterized in that include: The aerodynamic law feature database representing aerodynamic laws is constructed by train number clusters, curve clusters and feature vectors; According to the type of aerodynamic law characteristic data, single component matching recommendation analysis and multi-component matching recommendation analysis are respectively performed on the same type of aerodynamic law characteristic data to obtain single component matching recommendation analysis results and multi-component matching recommendation analysis results of the selected type of aerodynamic law characteristic data; The matching advancement analysis of the next type of aerodynamic law characteristic data is performed until all types of aerodynamic law characteristic data in the aerodynamic law characteristic database are traversed.

2. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 1 is characterized by: The method for determining train number clusters, curve clusters, and feature vectors is as follows: The historical aerodynamic data of wind tunnel force test are numbered according to the aircraft shape; For the historical aerodynamic data of the same aircraft shape, the data were grouped according to the single factor variable method based on the wind tunnel test status. For the aerodynamic data in the same group, two train trips were selected in combination to construct a Class A train cluster, and three train trips were selected in combination to construct a Class B train cluster. The train trips in the train cluster were sorted from small to large according to the value of the variable factor θ. For the train data in the same train cluster, while ensuring that the train order in the train cluster remains unchanged, the corresponding aerodynamic data are extracted according to the six-component force and moment units X, Y, Z, Mx, My, and Mz to construct a curve cluster; The aerodynamic data curves in the curve cluster are fitted using the fourth-order Chebyshev method to obtain the Chebyshev coefficients of each curve. And construct the feature vector; Based on the obtained characteristic vector and the six aerodynamic unit components X, Y, Z, Mx, My, and Mz, a characteristic database characterizing the aerodynamic laws is constructed.

3. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 2 is characterized by: The aircraft shape number is a 4-digit number in the range of [0000, 9999]. The aerodynamic data of the same aircraft shape uses the same number, and the aerodynamic data of different aircraft shapes use different numbers. There is no order requirement. The wind tunnel test conditions include the aircraft's sideslip angle β, roll angle Pitch rudder deflection d p 、Yaw rudder deflection d y 、Roll rudder deflection d r , six variables of Mach number Ma, the angle of attack α of the aircraft does not participate in the single factor variable grouping; The variable factors θ of any constructed train cluster are the sideslip angle β, the roll angle Pitch rudder deflection d p 、Yaw rudder deflection d y 、Roll rudder deflection d r , any one of the six variables of Mach number Ma; the train cluster includes two specifications, A and B. The train cluster A contains 2 wind tunnel test trains, and the train cluster B contains 3 wind tunnel test trains.

4. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 3 is characterized by: The order of curves in the curve cluster is consistent with the order of train numbers in the train number cluster, and both are sorted from small to large according to the value of the variable factor θ; the curve cluster includes two specifications, A and B, and is divided into six types according to the aerodynamic component, X, Y, Z, Mx, My, and Mz; The Chebyshev coefficient The subscript i represents the order of the Chebyshev coefficients, is a constant term, is the coefficient of the first-order term, is the coefficient of the quadratic term, is the cubic coefficient, is the coefficient of the fourth-order term; Chebyshev coefficient The superscript j represents the train sequence number sorted from small to large according to the variable factor θ within the curve cluster; The feature vector includes two specifications, A and B, corresponding to the two specifications A and B of the train number cluster and curve cluster, where: A class feature vector: Class B feature vector:

5. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 4 is characterized by: The feature database includes a Class A aerodynamic law feature database, corresponding to Class A feature vectors, including six databases: AX, AY, AZ, A-Mx, A-My, and A-Mz; The Class B aerodynamic law feature database corresponds to the Class B feature vector and includes six databases: BX, BY, BZ, B-Mx, B-My, and B-Mz.

6. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 5 is characterized by: The method for performing single component matching recommendation analysis is: According to the A-type curve cluster corresponding to any aerodynamic component as the minimum research object, the A-type aerodynamic law characteristic data is called through the aerodynamic law characteristic database; Calculate the Euclidean distance between the feature vector T of the curve cluster to be matched and the feature vectors of all other curve clusters, and take the top N Euclidean distances sorted from small to large, and record them as the preliminary result R1; The feature vectors T of the curve cluster to be matched are reordered from large to small according to the variable factor θ, and the adjusted feature vectors are recorded as reverse feature vectors T_r, and the reverse feature vectors of type A and type B are determined; Calculate the Euclidean distance between the reversed feature vector T_r of the curve cluster to be matched and the feature vectors of all other curve clusters, and select the top M in ascending order of Euclidean distance, which are recorded as the preliminary result R2; Merge R1 and R2, and take the top K names sorted from small to large in Euclidean distance, which are recorded as the recommended result R for a single-component aerodynamic data matching.

7. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 6 is characterized by: N is an integer between [1, 10]; M is an integer between [1, 10]; a small Euclidean distance represents a high degree of similarity in aerodynamic laws, and a reversed feature vector of type A B type reverse eigenvector:

8. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 6 is characterized by: The method for performing multi-component matching recommendation analysis is: Taking the train number cluster as the smallest research object, the six curve clusters X, Y, Z, Mx, My, and Mz contained in the train number cluster are matched and recommended according to the similarity of aerodynamic laws. Based on the principle that the six matching results come from the same train number cluster, they are combined as the matching recommendation results of the train number cluster. Calculate the feature vectors T corresponding to the six curve clusters X, Y, Z, Mx, My, and Mz contained in the train number cluster to be matched x , T y , T z , T mx , T my , T mz The six Euclidean distances dis_x, dis_y, dis_z, dis_mx, dis_my, and dis_mz of the feature vectors corresponding to the six curve clusters contained in other train clusters are ensured to be calculated from the same train cluster and the train cluster to be matched; The six Euclidean distances are weighted and summed to obtain the measurement result dis between the train number cluster to be matched and the train number cluster; Take the top N train number clusters’ measurement results dis, sorted from small to large, and record them as preliminary result RR1; The feature vectors T corresponding to the six curve clusters X, Y, Z, Mx, My, and Mz contained in the train number cluster to be matched x , T y , T z , T mx , T my , T mz Rearrange the variable factor θ from large to small, and the adjusted eigenvector is recorded as the reverse eigenvector T x _r, T y _r, T z _r, T mx _r, T my_r , T mz _r; Calculate the reverse feature vector T of the train number cluster to be matched x _r, T y _r, T z _r, T mx _r, T my_r , T mz The six Euclidean distances between _r and the feature vectors of other train number clusters are weighted and summed to obtain the measurement results with other train number clusters. The weight coefficient is the same as that of the single component matching recommendation analysis setting. The top M measurement results are sorted from small to large and recorded as the preliminary result RR2; Merge RR1 and RR2, and take the top K train number clusters sorted from small to large in terms of measurement results, which are recorded as the multi-component aerodynamic data matching recommendation result RR.

9. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 8 is characterized by: The method for determining the measurement result dis of the train number cluster to be matched and the train number cluster is as follows: dis=α1*dis_x+α2*dis_y+α3*dis_z+α4*dis_mx+α5*dis_m+α6*dis_mz Among them, the weighting coefficients α1, α2, α3, α4, α5, α6 are taken from [8, 4, 2, 1, 0, 0.5, 0.25, 0.125], and the same value is allowed; After the weighting coefficients α1, α2, α3, α4, α5, and α6 are selected, the calculation of all train clusters uses the same set of weighting coefficients; the weighting coefficient can be selected as 0, indicating that the aerodynamic component does not participate in the matching measurement.

10. The similar aerodynamic law matching recommendation analysis method based on wind tunnel historical data according to claim 9, characterized in that: Train clusters, curve clusters, and feature vectors only include Class A and Class B aerodynamic law feature data. When data matching recommendation analysis is required, the required data is directly called from the aerodynamic law feature database.

Images