A hybrid method for predicting the shock detachment distance of a vee-lip straight-finned strut

Abstract

A hybrid method for predicting the shock detachment distance of V-shaped lip straight leading edge is disclosed, which relates to the field of aerodynamics. The method combines neural network model and theoretical model to solve the problem of predicting the shock detachment distance under different inflow conditions, especially when there is a sideslip angle. A data set is constructed through numerical simulation to train the neural network model to predict the detachment distance. A theoretical model is constructed to predict the fully developed detachment distance. By comparing the outputs of the two models, the boundary conditions for the fully developed detachment shock are determined. According to the input conditions, the appropriate model is selected for calculation to achieve efficient and accurate prediction. Based on the simplified continuity equation and the neural network model, the shock detachment distance of V-shaped lip straight leading edge under different inflow conditions is predicted, and the trend of the shock structure is analyzed when the inflow conditions change. The method has higher prediction efficiency and accuracy, and can obtain the main features in the flow field in a shorter time than computer simulation. It is suitable for various inflow conditions, including the case of sideslip angle.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of aerodynamics, in particular to a hybrid method for predicting the shock detachment distance of a straight front edge section of a V-shaped lip. BACKGROUND

[0002] In recent years, with the development of three-dimensional internal turning inlet, the complex three-dimensional shock wave interference phenomenon caused by the detached shock wave at the V-shaped lip has become a hot issue. Due to the mutual interference of the detached shock wave, three types of main shock wave interference structures will be generated on the symmetry plane of the V-shaped lip, which are regular reflection, Mach reflection and same-side regular reflection. Among them, the flow phenomenon generated by Mach reflection is the most complex, and the corresponding heat flow generation mechanism also presents diversity, so it becomes the focus of research. In order to optimize the design of the V-shaped lip, in recent years, experts and scholars have devoted themselves to modeling the flow field of the V-shaped lip, so as to provide scientific basis and reference for the related design, and the prediction of the detached shock wave detachment distance at the three-wave point of the V-shaped lip is an important part of the model.

[0003] In the aspect of V-shaped lip modeling, KANG D, YAN C, LI Z, et al. (2023) combined the three-wave theory, the inviscid shock theory and the shock polar method to model the wave structure of V-shaped lip under the condition of Mach reflection, and realized the estimation of the flow field parameters in each region of V-shaped lip. ZHANG Z, LI Z, YANG J. (2021) firstly calculated the standoff distance of detached shock based on the theoretical approximation method proposed by SINCLAIR J, CUI X. (2017). Then, the iterative method proposed by EMANUEL G. (1982) was used to solve the position of curved shock generated by the rounded region. Finally, the position of triple point under different conditions was determined by fitting the position of triple point in the numerical simulation results, so as to reproduce the wave structure on the symmetry plane of V-shaped lip. Although this method can accurately reproduce the flow field on the symmetry plane, it still depends on a series of numerical simulations of V-shaped lip due to the need to fit the position of triple point.To obtain the approximate position of the shock wave under various working conditions more quickly, ZHANG T, CHENG J, SHI C, et al. (ZHANG T, CHENG J, SHI C, et al. Mach reflection of three-dimensional curved shock waves on V-shaped blunt leading edges[J / OL]. Journal of Fluid Mechanics, 2023, 975: A45. DOI:10.1017 / jfm.2023.866) extended the method proposed by Moeckel for predicting two-dimensional detached distance based on the simplified continuity equation (MOECKEL W E. Approximate method for predicting form and location of detached shock waves ahead of plane or axially symmetric bodies[R / OL]. (1949)) to three-dimensional space, thereby approximately solving the position of the three-wave point by a theoretical method, so as to achieve the goal of reproducing the entire V-shaped lip port symmetry surface wave system structure by a pure theoretical method. However, when the compression effect of the airflow perpendicular to the wall surface is weak, the detached shock wave on the finite straight leading edge may not develop fully, which will increase the prediction error of the detached distance of the straight leading edge segment by the theoretical method, and then affect the overall prediction accuracy of the model.

[0004] The above-mentioned theoretical modeling methods are all established under the condition of no side slip angle. However, in the actual flight process, the appearance of the side slip angle is inevitable. The appearance of the side slip angle will further weaken the compression effect of the V-shaped lip port leeward airflow perpendicular to the straight leading edge segment wall surface, and then cause the problem of airflow not developing fully on the finite straight leading edge segment to be more significant, which will significantly increase the error of the theoretical prediction of the detached distance of the detached shock wave of the straight leading edge segment. SUMMARY

[0005] The purpose of the present application is to provide a hybrid method for predicting the shock wave detached distance of the straight leading edge segment of the V-shaped lip port, and to realize the prediction of the shock wave detached distance of the straight leading edge segment of the V-shaped lip port under various incoming flow conditions. The present application introduces a hybrid method of neural network model and theoretical model for predicting the detached distance of the fully developed / non-developed detached shock wave at the three-wave point. Thus, a high-efficiency and widely applicable means is provided for the wave system structure analysis and prediction of complex aerodynamic components such as V-shaped lip ports and aerodynamic design.

[0006] The present application comprises the following steps:

[0007] 1) Constructing dataset: Numerical simulation is performed on the swept-back blunt body under different incoming flow conditions, and the shock point on the symmetry plane is extracted from the simulation results to construct the dataset; wherein the features of each sample include Mach number, sweepback angle and development distance, and the label is the detachment distance of the detached shock under the corresponding feature condition;

[0008] 2) Constructing and training neural network model (ResMLP), the trained neural network model takes Mach number, sweepback angle and development distance as input parameters, and outputs the predicted detachment distance of the detached shock under the corresponding input condition;

[0009] 3) Constructing a theoretical model for predicting the detachment distance of the detached shock of the straight leading edge section before full development; using the assumed shape of the detached shock and the wave parameter column, a simplified continuity equation is written, and the detachment distance is solved by Newton iteration method;

[0010] 4) Traverse different incoming flow conditions, respectively use the theoretical model and the neural network model to calculate the detachment distance under the corresponding working condition, compare the outputs of the two; when the absolute value of the relative error of the two is less than 5%, it is considered that the detached shock has fully developed; thus the development distance corresponding to the fully developed detached shock under each incoming Mach number and sweepback angle is determined, and the fully developed boundary about development distance and sweepback angle under each Mach number is obtained;

[0011] 5) Determine the model for calculating the detachment distance according to the input: judge whether the input condition exceeds the fully developed boundary obtained in step 4); when the input condition exceeds the fully developed boundary obtained in step 4), it means that the detached shock has fully developed, and the detachment distance of the detached shock of the straight leading edge section at this time is calculated by the theoretical method; on the contrary, if the input condition does not exceed the boundary, it means that the detached shock has not fully developed, and the detachment distance is predicted by the neural network hybrid model.

[0012] In step 2), the constructed neural network model is divided into three parts, which are encoder, processor and decoder;

[0013] The encoder sets Fourier encoding layer and MLP layer in the input channel of Mach number and sweepback angle, and only sets MLP layer in the input channel of development distance; after combining the three variables into a tensor, the standardization layer and MLP layer are input in turn;

[0014] The processor includes 4 ResMLP layers, which adopt residual connection;

[0015] The decoder includes a standardization layer and a linear layer, which realizes the goal of outputting one-dimensional data;

[0016] Such an arrangement can enable the model to better extract high-dimensional features of the data and improve the stability of training; in addition, for the characteristics of small sample label values, a relative error is used as a loss function to train the model to prevent problems such as gradient disappearance during model training.

[0017] In step 3), the specific steps of constructing the theoretical model for predicting the shedding distance of the shedding shock wave of the straight leading edge section just before full development can be:

[0018] 3.1) Assuming that the two-dimensional shape of the shedding shock wave is a branch of a parabola, the coordinates of each point in the flow field are represented by an equation containing only the shedding distance as an unknown number;

[0019] 3.2) Calculate the parameters behind the shedding shock wave according to the inviscid shock wave theory;

[0020] 3.3) Use the parameters obtained in steps 3.1) and 3.2) to write a simplified continuity equation, which is a one-variable equation only about the shedding distance, and the shedding distance can be solved by using the Newton iteration method.

[0021] In step 4), when the full development boundary is determined, the development distance corresponding to the fully developed shedding shock wave is obtained under the given Mach number and sweep angle by an interpolation method.

[0022] In step 5), the shedding distance of the shedding shock wave of the straight leading edge section at this time is calculated by a theoretical method, and the shedding distance of the fully developed shedding shock wave is calculated according to the input Mach number and sweep angle by using the theoretical model constructed in step 3).

[0023] Compared with the prior art, the present application has the following advantages:

[0024] 1) The present application proposes a hybrid prediction method for predicting the shedding distance of the shedding shock wave of the straight leading edge section of the V-shaped lip, especially under complex incoming flow conditions such as side slip angle. This method effectively deals with the problem that the airflow compression effect is weakened due to the side slip angle, and the shedding shock wave may not be fully developed on the limited length of the straight leading edge, thereby improving the accuracy and applicability of the prediction.

[0025] 2) The present application combines a neural network model and a theoretical model to achieve efficient prediction of the shedding distance of the shedding shock wave. The neural network model can learn and extract high-dimensional features in the data, and the theoretical model is based on a simplified continuity equation. The combination of the two makes the prediction process fast and accurate.

[0026] 3、Through the method of the application, the detached shock wave structure can be efficiently predicted, so as to analyze the change trend of the detached shock wave structure when the inflow condition changes. And the detached distance of the triple point can be accurately predicted, which is a key parameter in the modeling of the V-shaped lip flow field. It is of great significance for understanding the interaction between the shock wave and the flow field, and optimizing the V-shaped lip design. Accurate prediction of the detached distance helps to more accurately reproduce the wave system structure at the symmetry plane, and provides a scientific basis for aerodynamic design.

[0027] 4、Compared with the traditional theoretical method, the application has obvious advantages in accuracy. By comparing the computer numerical simulation results, the traditional theoretical results and the results calculated by the application, it is found that the method of the application is more accurate in predicting the detached distance. Compared with computer simulation, the method of the application can obtain the main features in the flow field in a shorter time. This is of great significance for rapid design and optimization of aerodynamic components, and helps to improve design efficiency and reduce cost.

[0028] 5、The application is based on a simplified continuity equation and a neural network model to predict the detached distance of the V-shaped lip straight front edge segment shock wave under different inflow conditions. It has the characteristics of rapidity, efficiency and accuracy, and provides an effective way for wave system structure analysis, prediction and aerodynamic design of V-shaped lip and other complex aerodynamic components, has wide application prospect, and can be applied to the fields of aerospace and high-speed aircraft design. BRIEF DESCRIPTION OF DRAWINGS

[0029] Figure 1 It is a schematic view of the detached shock wave and the wall surface of the V-shaped lip straight front edge segment in supersonic flow.

[0030] Figure 2 It is a schematic view of the ResMLP neural network model in the detached shock wave prediction model.

[0031] Figure 3 It is Figure 1 It is a schematic view of the V-shaped blunt front edge wall surface and the detached shock wave at the A-A cross section.

[0032] Figure 4 It is a flowchart of the theoretical and neural network hybrid model.

[0033] Figure 5 It is a comparison chart of the calculation results of the application, the traditional theoretical results and the computer simulation results. DETAILED DESCRIPTION

[0034] In order to make the technical problems, technical solutions and beneficial effects of the application more clear, specific and understandable, the application will be further described in detail in combination with the drawings.

[0035] First, combined with Figure 1The basic configuration, parameters and shock wave structure of the straight leading edge section are introduced. Figure 1 In this case, the straight leading edge section 1 of the V-shaped lip is essentially a swept-back blunt body, and the sweep angle is denoted by υ. Under the action of the incoming flow with Mach number Ma0, a detached shock wave 2 is generated. The detachment distance of the detached shock wave gradually increases along the flow direction from the starting position of the blunt body until the distance L d of the swept-back blunt body is equal to L fd , the detachment distance of the detached shock wave reaches the maximum value l fd and no longer changes along the flow direction. At this time, the detached shock wave is said to be fully developed.

[0036] The method for predicting the detachment distance of the detached shock wave is introduced as follows. The first step is to construct a ResMLP neural network model as shown in Figure 2 , which mainly includes the following 5 steps:

[0037] 1) Constructing a data set: Numerical simulation is performed on swept-back blunt bodies under different incoming flow conditions, and the shock points are extracted as samples to construct a data set. The features of each sample include Mach number Ma0, sweep angle υ, development distance L d , and the label is the detachment distance of the detached shock wave under the corresponding feature condition.

[0038] 2) Constructing a model encoder (Encoder): Fourier encoding layers and MLP layers are set in the input channels of Mach number Ma and sweep angle υ. For the development distance L d , since the numerical difference between samples is small, if Fourier encoding is performed, the difference will be further reduced, which may lead to problems such as gradient disappearance and unstable training. Therefore, only MLP layers are set in the input channel of L d . Next, the three variables are combined into a tensor, which is input into a standardization layer and an MLP layer in turn.

[0039] 3) Constructing a model processor (Processor): The processor includes 4 ResMLP layers. Since residual connection is used, ResMLP has better training ability and stability than traditional MLP.

[0040] 4) Constructing a model decoder (Decoder): The decoder includes a standardization layer and a linear layer to achieve the goal of outputting one-dimensional data.

[0041] 5) Using the data set constructed in step 1), the relative error function as shown in equation (1) is used as the loss function to train the above model, and finally the construction of the entire neural network model is completed.

[0042] (1)

[0043] A method for predicting detachment distance based on a simplified continuity equation is introduced.

[0044] Figure 3 yes Figure 1 A cross-sectional view of the flow field structure and wall surface at section AA. In the figure, the blunt-nosed body 1 with a V-shaped lip and straight leading edge generates a detached shock wave 2 under the influence of the incoming flow. Points S and B correspond to the sound velocity points on the shock wave and wall surface, respectively. Line segment 3 connecting S and B is an imaginary sound velocity line. The angles of the airflow at points S and B relative to the horizontal direction are δ... s and δ d Since the detached shock wave is perpendicular to the incoming flow at point D, and the detached shock wave approaches the asymptote 4 as the x value approaches infinity, the detached shock wave 2 can be represented by one branch of the hyperbola shown in equation (2):

[0045] (2)

[0046] Where k = tanα, α represents the Mach angle; x and y represent the coordinates of the shock point on the detached shock wave; the shock angle θ at each point can be expressed as:

[0047] (3)

[0048] To formulate the simplified continuity equation for the subsonic region after the detached shock wave, it is first necessary to determine the coordinates of each point. Based on the incoming flow conditions, the shock angle at which the Mach number behind the wave is the speed of sound can be determined (i.e., the shock angle at point S). Therefore, the coordinates of point S can be deduced by combining equation (3):

[0049] (4)

[0050] Where, θ S This represents the shock angle of the detached shock wave at point S. The wall angle at point B is approximately assumed to be equal to the maximum angle that could deflect the airflow when the shock wave is still intact under the same incoming flow conditions. Specifically, when the blunt body is an arc:

[0051] (5)

[0052] Assuming the sound velocity line is perpendicular to the flow direction, then points S and B have the following geometric relationship:

[0053] (6)

[0054] in, Substituting equation (4) into equation (6), we get:

[0055] (7)

[0056] At this time, the coordinates of the sonic region SBOD are determined except for point D. The simplified continuity equation can be used to obtain:

[0057] (8)

[0058] where ; ; represents the area ratio required for the isentropic expansion of the incoming flow to the sonic speed; P0 represents the total pressure of the incoming flow; P C represents the total pressure after the detached shock wave at x . The above parameters can be expressed by parameters containing x D , or obtained by the inviscid shock wave theory method. Therefore, equation (8) is an equation only about the unknown x D , which can be solved by using the Newton iteration method, so as to obtain the theoretical value of the detached distance.

[0059] Next, a hybrid prediction method of the theoretical model and the neural network model of the detached distance is introduced, and a flow chart thereof is shown in Figure 4 , which mainly includes the following steps:

[0060] (1) Determining the fully developed boundary: first, traverse each working condition, and for a given working condition range (including different Mach numbers Ma and sweepback angles υ), use the theoretical model and the neural network model to calculate the detached distance under the corresponding working condition. When the relative error of the detached distance calculated by the theory and the detached distance predicted by the neural network model is less than 5%, it is considered that the detached shock wave is fully developed. Thus, the corresponding sweepback angle and development distance of the fully developed detached shock wave under each Mach number can be obtained. Further, the fully developed boundary about the development distance and the sweepback angle under each Mach number can be obtained.

[0061] (2) Predicting the development distance by using the interpolation method: when the incoming flow parameters are input, Ma0 and υ among them are input to the fully developed boundary determined in step (1), and the development distance L fd required for the detached shock wave to reach the fully developed state under the condition is given by the computer by using the interpolation method.

[0062] (3) Judging whether the detached shock wave is fully developed and calculating the detached distance: according to the predicted development distance L fd in step (2) and the given incoming flow parameters, it is judged whether the detached shock wave is fully developed under the parameters. If it is fully developed, the detached distance l fd of the fully developed detached shock wave is calculated according to Ma0 and υ by using the theoretical method; otherwise, if the detached shock wave is not fully developed, the detached distance l is calculated according to the Mach number Ma0, the sweepback angle υ, the development distance L d by using the neural network model.

[0063] Specifically, when using theoretical iteration to solve for the positions of the three wave points of the V-shaped lip, the development distance of the detached shock wave at the three wave points changes continuously with the iteration process. The hybrid prediction method described above can incorporate this change in development distance, updating the detachment distance of the detached shock wave at the three wave points in each iteration based on the current Mach number, sweep angle, and development distance.

[0064] To verify the effectiveness and accuracy of the proposed hybrid model in predicting the wave system structure of the V-shaped lip rounded region, the results of the proposed hybrid model and the pure theoretical model were used as inputs to the prediction model, resulting in two sets of prediction results. These results were then compared with simulation results. A simulation was conducted using a V-shaped lip with a rounded radius to straight leading edge passivation radius ratio of 4 and a straight leading edge sweep angle υ = 24°, under incoming flow conditions of Mach number 6 and a sideslip angle of 5°. The results are as follows: Figure 5 The results are shown.

[0065] in, Figure 5 Figure (a) shows a comparison between the calculation results and simulation results of this invention. It can be observed that the hybrid model proposed in this invention has a good prediction effect on the detachment distance on both the windward and leeward sides. This indicates that the hybrid model can accurately capture the main features of the wave system structure in the V-shaped lip rounded region, including the position and shape of the detached shock wave. Figure 5 Figure (b) shows a comparison between the calculation results of the traditional theoretical model and the simulation results. Since the compression effect perpendicular to the wall is stronger on the windward side than on the leeward side, the detached shock wave on the windward side has reached a fully developed state. Therefore, the theory has a better prediction effect on the detached shock wave on the windward side. However, since the detached shock wave on the leeward side has not reached a fully developed state, the theoretically predicted detachment distance is greater than the actual detachment distance of the detached shock wave. This error not only affects the prediction accuracy of the detachment distance, but also causes a large error in the prediction of the Mach rod position.

[0066] In summary, the hybrid model proposed in this invention demonstrates higher accuracy in predicting the wave system structure of the V-lip rounded region compared to traditional theoretical methods. The hybrid model can accurately predict the detachment distance on both the windward and leeward sides, providing valuable guidance for the design and optimization of the V-lip. Traditional theoretical models perform reasonably well in predicting the detachment shock wave on the windward side, but exhibit significant errors in predicting the detachment shock wave on the leeward side. This is mainly because the detachment shock wave on the leeward side is not fully developed, leading to a discrepancy between theoretical and actual results.

[0067] Compared with the traditional theoretical method, the method has higher accuracy, and can greatly improve the precision of the V-shaped lip inverted round area wave system structure prediction model. Therefore, the method has good guiding significance in predicting and analyzing the shock wave structure generated by the V-shaped lip and other complex aerodynamic shapes, and provides a more accurate and reliable prediction tool for aerodynamic design and optimization.

[0068] The above embodiments are only preferred embodiments of the present application and should not be considered as limiting the scope of the present application. Any equivalent changes and improvements made in accordance with the scope of the present application should still belong to the patent scope of the present application.

Claims

1. A hybrid method for predicting V-shaped lip straight front-fillet shock detachment distance, characterized by The method comprises the following steps: 1) constructing a data set: performing numerical simulation on a swept-back blunt body under different incoming flow conditions, and extracting shock points at the symmetry plane from the simulation results to construct a data set; wherein the characteristics of each sample include Mach number, sweepback angle and development distance, and the label is the shedding distance of the detached shock under the corresponding characteristic condition; 2) constructing and training a neural network model, the trained neural network model takes Mach number, sweepback angle and development distance as input parameters, and outputs the predicted shedding distance of the detached shock under the corresponding input condition; The constructed neural network model is divided into three parts, which are encoder, processor and decoder; the encoder sets Fourier encoding layer and MLP layer in the input channel of Mach number and sweepback angle, and only sets MLP layer in the input channel of development distance, after merging the three variables into a tensor, inputting into the standardization layer and MLP layer in turn; the processor includes 4 ResMLP layers, which adopt residual connection; the decoder includes a standardization layer and a linear layer, which realize the goal of outputting one-dimensional data; this setting can better extract the high-dimensional features of the data and improve the stability of the training; in view of the small sample label value, the relative error is used as the loss function to train the model to prevent the gradient from disappearing during model training; 3) constructing a theoretical model for predicting the shedding distance of the detached shock of the straight leading edge section after full development; writing a simplified continuity equation by using the assumed shape of the detached shock and the post-shock parameter column, and solving the shedding distance by Newton iteration method; 4) traversing different incoming flow conditions, respectively using the theoretical model and the neural network model to calculate the shedding distance under the corresponding working condition, and comparing the outputs of the two; when the absolute value of the relative error of the two is less than 5%, it is considered that the detached shock has fully developed; thus the development distance corresponding to the full development of the detached shock under each incoming Mach number and sweepback angle is determined, and the full development boundary about development distance and sweepback angle under each Mach number is obtained; 5) determining the model for calculating the shedding distance according to the input: judging whether the input condition exceeds the full development boundary obtained in step 4); when the input condition exceeds the full development boundary, it indicates that the detached shock has fully developed, and the shedding distance of the detached shock of the straight leading edge section at this time is calculated by the theoretical method; otherwise, if the input condition does not exceed the boundary, it indicates that the detached shock has not fully developed, and the shedding distance is predicted by the neural network hybrid model.

2. The hybrid method of predicting V-shaped lip straight front-fillet segment shock detachment distance as claimed in claim 1, wherein In step 3), the specific steps of constructing a theoretical model for predicting the shedding distance of the detached shock of the straight leading edge section after full development are as follows: 3.1) assuming that the two-dimensional shape of the detached shock is a branch of parabola, and expressing the coordinates of each point in the flow field with an equation containing only one unknown number of shedding distance; 3.2) calculating the post-shock parameters of the detached shock according to the inviscid shock wave theory; 3.3) writing a simplified continuity equation by using the parameters obtained in steps 3.1) and 3.2), which is a one-variable equation only about shedding distance, and solving the shedding distance by Newton iteration method.

3. The hybrid method of predicting V-shaped lip straight front-fillet segment shock detachment distance as claimed in claim 1, wherein In step 4), when the fully developed boundary is determined, the development distance corresponding to the fully developed detached shock wave at the given Mach number and the given sweep angle is obtained by an interpolation method.

4. The hybrid method of predicting V-shaped lip straight front-fillet segment shock detachment distance as claimed in claim 1, wherein In step 5), the detachment distance of the detached shock wave of the straight leading edge section at this time is calculated by a theoretical method, and the detachment distance of the detached shock wave when the detached shock wave is fully developed is calculated according to the input Mach number and the sweep angle by using the theoretical model constructed in step 3).

Images