A quantitative analysis method for complex flow dynamics in body-fitted grid CFD calculations

Abstract

The present invention discloses a method for quantitative analysis of complex flow driven mechanism for body-fitting grid CFD calculation, belonging to the technical field of fluid dynamics analysis. In the method of the present invention, a quantitative analysis expression of complex flow aerodynamic load based on weight function mapping of Navier-Stokes equation is first derived, and a vortex identification method based on threshold adaptation, a vortex segmentation method based on local grid spacing adaptation-variable density clustering algorithm and a vortex tracking method based on vortex structure position prediction matching are combined, and finally the aerodynamic load induced by each vortex structure is obtained, and a quantitative analysis of the driven mechanism of each local vortex structure for body-fitting grid CFD calculation results is performed. The method of the present invention effectively solves the defects of existing complex flow driven mechanism quantitative analysis methods such as vortex moment theory, which are difficult to quantitatively analyze the driven mechanism of local vortex structure and cannot consider the driven mechanism of other factors such as surface motion effect of object, viscosity effect and friction effect in pressure boundary.

Description

Technical Field

[0001] The present invention belongs to the technical field of fluid dynamics analysis, and relates to a complex flow mechanism analysis method in the field of fluid dynamics, and specifically to a complex flow mechanism quantitative analysis method for body-fitting grid CFD calculation. Technical Background

[0002] Complex flow problems are widely present in important engineering fields related to national security and national economy, such as aerospace, ships and ships, and wind power generation. Complex flow fields often have complex flow behaviors such as flow separation, boundary motion, and vortex evolution. A large number of studies have found that there is a significant correlation between flow field evolution, vortex development, and aerodynamic characteristics of objects in complex flows. In order to deeply understand the influence of vortex structure and its dynamic evolution on aerodynamic characteristics, it is crucial to establish a method for the mechanism of vortex structure.

[0003] The commonly used quantitative analysis method for the driving mechanism of vortex structure is the vortex moment theory, which reveals that the aerodynamic force or moment depends only on the rate of change of the first-order and second-order moments of the vortex field. It can be applied to the quantitative analysis of the driving mechanism of unsteady flow. Later, relevant scholars applied the derivative moment change method to develop the finite domain vortex moment theory, which solved the problem that the classical vortex moment theory can only be applied to infinite domains. Although the vortex moment theory has been continuously improved in the process of development, it still has the defects of being difficult to quantitatively analyze the driving mechanism of local vortex structure and being unable to consider the driving mechanism of other factors such as the surface motion effect of the object, the viscosity effect and friction effect in the pressure boundary.

[0004] If we want to quantitatively analyze the driving mechanism of local vortex structure, we first need to identify the local vortex structure. At present, the vorticity, Q criterion and λ 2 There are many vortex identification methods such as criterion, but these vortex identification methods need to set a threshold of relevant criteria. When this threshold is selected to be small, there are many entanglements between local vortex structures, which causes great trouble to the segmentation of local vortex structures; but when this threshold is selected to be large, the sum of the aerodynamic forces induced by the extracted local vortex structure will have a significant deviation from the aerodynamic resultant force generated by the vortex structure of the entire flow field. Therefore, the selection of the vortex identification threshold is crucial, but the selection of this threshold currently depends mainly on experience and lacks relevant threshold selection criteria.

[0005] After the vortex identification threshold is determined, the flow field area greater than the threshold is identified. In order to analyze the quantitative impact of local vortex structure on aerodynamics, the identified flow field area needs to be segmented according to different vortex structures (such as leading edge vortex, trailing edge vortex, secondary vortex, wingtip vortex, etc.). Traditional density clustering algorithms, such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise), are widely used in vortex segmentation in fluid mechanics. This is a density-based clustering method that mainly identifies clustering points of vortex structure by setting global parameters of minimum number of neighbors (ns) and distance threshold (eps). Specifically, the parameter eps is the maximum distance between two sample points that are considered to be the same cluster, and ns is the minimum number of samples required to form a cluster. Its advantage is that it can effectively handle noise and identify clusters of arbitrary shapes. However, its clustering effect depends on the global parameter setting, and its clustering effect will be significantly reduced when processing data sets with non-uniform density. However, in body-fitting CFD calculations, in order to balance calculation accuracy and efficiency, the distribution of the calculation grid is usually uneven. The grid density in the boundary layer area close to the object wall is usually much higher than that in the far field area. This grid inhomogeneity poses a huge challenge to the segmentation of the vortex structure. Figure 1 An example of the vortex segmentation effect of DBSCAN is given. If the parameter eps value is set too small, a single vortex structure far away from the object boundary may be segmented into multiple clusters or even marked as a noise point; if the eps value is set too large, multiple vortex structures close to the object boundary may be merged into one cluster. Therefore, how to accurately segment vortex structures under non-uniform body-fitting grids is still a difficult problem that needs to be solved urgently in the existing technology. Summary of the invention

[0006] The problems existing in the current quantitative analysis methods of complex flow mechanisms for body-fitted grid CFD calculations include:

[0007] a. It is difficult to quantitatively analyze the driving mechanism of the local vortex structure and cannot consider other factors such as the surface motion effect, the viscosity effect and friction effect in the pressure boundary;

[0008] b. The selection of vortex identification threshold mainly depends on experience and lacks relevant threshold selection criteria; and the selection of vortex identification threshold is crucial to the quantitative analysis of the local vortex structure mechanism.

[0009] c. The grid density distribution of the body-fitting CFD grid is uneven, and the density clustering algorithm in the current method is difficult to accurately segment each local vortex structure.

[0010] The present invention proposes a quantitative analysis method for complex flow force mechanism for body-fitted grid CFD calculation, and the specific steps are as follows:

[0011] Step 1: Acquisition of body-fitted grid CFD calculation data.

[0012] First, the aerodynamic load data, body-fitting grid files, and flow field information at different times are obtained through body-fitting grid CFD calculations. Then, the body-fitting grid files at each time are read, and according to the load type being analyzed, the weight function values ​​of the center points of different grid units at each time are obtained by solving the weight function equation corresponding to the aerodynamic load;

[0013] Step 2: Observe the evolution of the vortex structure in the flow field.

[0014] The flow field calculation results at different times are visualized, and the changes in aerodynamic loads are qualitatively analyzed. At the same time, the Q criterion is used as the vortex criterion to identify the vortex structure in the flow field and obtain the maximum Q value of the entire flow field at different times. max By calculating the cumulative distribution curve of the vortex-induced aerodynamic load at different times, the Q criterion threshold Q at different times can be determined. th , ensuring that the aerodynamic load deviation of the extracted vortex structure does not exceed 5%.

[0015] Step 3: Vortex structure identification.

[0016] Based on the Q criterion threshold Q at different times determined in step 2 th , the vortex structure is identified at each moment, and Q>Q th The grid points are marked as the vortex structure area Label_Q=1, and the remaining grid points are marked as other areas.

[0017] The vortex identification method based on threshold adaptation is realized through steps 2 and 3.

[0018] Step 4: Vortex structure segmentation.

[0019] The vortex structure is segmented for the grid points marked as vortex structure areas using a local grid spacing adaptive-variable density clustering algorithm. The method is as follows:

[0020] 1) Initialization: Set the minimum number of neighbors ns and the distance threshold eps parameter and the minimum distance d between local cells of the body-fitting CFD grid min The proportionality coefficient A eps .

[0021] 2) Traverse each unmarked unit center point p in the extracted vortex structure area.

[0022] Read the body-fitting CFD grid file, determine the other grid cells adjacent to the grid cell, calculate the distance between the center of the grid cell and the centers of these adjacent cells, and take the minimum value, which is d min, the distance threshold eps of this grid cell is equal to A eps ×d min .

[0023] Calculate the number of grid cell center points N in the vortex structure area within the neighborhood of the grid cell center point eps eps (p), if N eps (p)≥ns, the center of the unit is marked as the core point.

[0024] 3) If the unit center point p is a core point, create a cluster and add the unit center point p to this cluster. For each vortex structure area unit center point q in the eps-neighborhood of p:

[0025] If q is a boundary point, mark the point and add it to the current cluster;

[0026] If q is a core point, q and all points in its eps-neighborhood are marked and added to the current cluster, and then the process is repeated for each point in the eps-neighborhood of q. After the process is repeated, return to step 2) until the traversal is completed.

[0027] 4) After the traversal process is completed, each cluster obtained corresponds to a local vortex structure area.

[0028] The result after segmentation is: for the grid with Label_Q=1, the noise point is marked as Label_S=-1, and each vortex structure cluster in the clustering result is marked as Label_S=1,...,N in turn. s .

[0029] Step 5: Vortex structure tracking and matching.

[0030] For the labels Label_S=1,...,N s The vortex structure cluster is tracked in time series by using the method based on the prediction and matching of the center of mass position; the center of mass position X and velocity V of each vortex structure are calculated by the volume weighted method, and the vortex clustering result (X old ,V old ), predict the vortex structure position X at the current time step pre =X old +V old ×dt, and the actual position X new Make a comparison.

[0031] After tracking, label it as Label_T=1,...,N t The vortex structure can achieve consistency in the time domain.

[0032] Step 6: Calculation of vortex-induced aerodynamic loads.

[0033] Finally, the regional positions of each vortex structure at different times were obtained. The aerodynamic loads induced by each vortex structure at different times were calculated using the quantitative analysis expression of the aerodynamic loads. The relationship between the vortex dynamic characteristics such as the vortex regional position, size and strength and the aerodynamic characteristics was quantitatively established.

[0034] The advantages of the present invention are:

[0035] 1. The present invention derives a quantitative analysis expression of aerodynamic loads of complex flows based on the weight function mapping of the Navier-Stokes equation, which solves the defects of existing quantitative analysis methods of complex flow driving mechanisms such as vortex moment theory, which are difficult to quantitatively analyze the driving mechanism of local vortex structures and cannot consider the driving mechanism of other factors such as the surface motion effect of objects, the viscosity effect and friction effect in the pressure boundary;

[0036] 2. The present invention proposes a vortex identification method based on threshold adaptation, which determines the Q criterion threshold by calculating the cumulative distribution curve of the vortex-induced aerodynamic load. On the premise of ensuring the aerodynamic accuracy of the extracted vortex structure, it reduces the entanglement between different vortex structures and improves the accuracy of subsequent vortex segmentation;

[0037] 3. The present invention proposes a vortex segmentation method based on local grid spacing adaptive-variable density clustering algorithm, by establishing a distance threshold eps parameter and the minimum distance d between local units of the body-fitting CFD grid min The relationship between the local unit density information of the body-fitting CFD grid is introduced to realize the spatial adaptation of the distance threshold EPS parameter in the density clustering algorithm. It solves the defects of the traditional density clustering algorithm such as the incorrect merging of vortex structures in the near-object surface area when the EPS parameter is large and the incorrect segmentation of vortex structures in the far-object surface area when the EPS parameter is small, and significantly improves the accuracy of vortex structure segmentation. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 The limitation of DBSCAN clustering algorithm in eddy segmentation problem;

[0039] Figure 2 M is the threshold-adaptive eddy identification method. Q Cumulative proportion distribution curve;

[0040] Figure 3 The vortex segmentation effect of AVDSC clustering algorithm on non-uniform CFD body-fitting grid;

[0041] Figure 4 The flow chart and effect diagram of the aerodynamic load decomposition method of the present invention are as follows DETAILED DESCRIPTION

[0042] The present invention provides a quantitative analysis method for complex flow force mechanisms for body-fitting grid CFD calculations. First, a quantitative analysis expression for complex flow aerodynamic loads based on weight function mapping of the Navier-Stokes equation is derived, and then a vortex identification method based on threshold adaptation, a vortex segmentation method based on local grid spacing adaptation-variable density clustering algorithm, and a vortex tracking method based on vortex structure position prediction and matching are combined to finally obtain the aerodynamic loads induced by each vortex structure, and perform a quantitative analysis of the force mechanisms of each local vortex structure for body-fitting grid CFD calculation results.

[0043] The derivation of the quantitative analysis expression of the complex flow aerodynamic load mapped by the weight function of the Navier-Stokes equation in the present invention is divided into the derivation of the quantitative analysis expression of the aerodynamic force and the derivation of the quantitative analysis expression of the aerodynamic moment based on the type of aerodynamic load, as follows:

[0044] A. Derivation of the expression for quantitative analysis of aerodynamic forces

[0045] Aerodynamic force includes pressure term (pressure force) and viscosity term (friction force):

[0046]

[0047] Among them, B represents the surface boundary of the object, p is the pressure, ρ is the fluid density, and ν is the kinematic viscosity coefficient. ω is the vorticity vector, and n is the unit vector. Therefore, the magnitude of the aerodynamic force F in the i direction is:

[0048]

[0049] Where S is the area of ​​the object; e i is the Cartesian basis vector in the i direction, n i is the component of the unit vector n in the i direction. If i = 1, that is, the direction of i is the x flow direction, then F i represents the resistance; if i = 2, that is, the direction of i is y perpendicular to the flow direction, then F i Represents lift.

[0050] When the aerodynamic load type is aerodynamic force, its weight function φ i Defined as:

[0051]

[0052] Where, Σ is the outer boundary of the fluid domain;

[0053] It can be seen that the weight function is related to the instantaneous position and geometric shape of the object and the boundary geometry information of the external flow domain.

[0054] In order to further decompose the pressure term, the weight function is mapped, the Navier-Stokes equation (NS equation) is mapped to the gradient field of the weight function, and the entire calculation domain V f Perform volume integration:

[0055]

[0056] In the formula, V is the volume of the object; t is the time; u is the speed of the object.

[0057] By performing a series of transformation operations on each term in equation (4) using Gaussian divergence theorem, continuity equation and other identities, the quantitative analysis expression of aerodynamic force is obtained:

[0058]

[0059] Where Q is the second Galilean invariant of the velocity gradient tensor.

[0060] Since the outer boundary Σ of the fluid domain is far away from the surface of the object, the Σ related term is almost zero and can be ignored. The simplified quantitative analysis expression of the aerodynamic force is:

[0061]

[0062] In the formula, F k 、F vp and F v They represent the aerodynamic force generated by the acceleration effect of the object surface, the viscosity effect in the pressure boundary, and the friction effect; F Q Represents the aerodynamic force generated by the vortex structure. By integrating the above formula over the spatial region where the local vortex structure is located, the aerodynamic force generated by the local vortex structure can be obtained.

[0063] B. Derivation of the expression for quantitative analysis of dynamic torque:

[0064] Aerodynamic moment M ea It is composed of pressure term and friction term. In the derivation process, a force arm is added compared with the aforementioned aerodynamic force, and the aerodynamic torque is expressed as:

[0065]

[0066] Among them, e k is the Cartesian basis vector in the k direction; X is the position vector of the node on the surface of the object, X ea is the position vector of the rotation center. M is the aerodynamic torque; F is the aerodynamic force.

[0067] When the aerodynamic load is aerodynamic moment, its weight function ψ k Defined as:

[0068]

[0069] Map the Navier-Stokes equations (NS equations) to the gradient field of the weight function and calculate the entire flow domain V f Perform volume integration:

[0070]

[0071] Then the quantitative analysis expression of aerodynamic torque is obtained:

[0072]

[0073] Since the outer boundary Σ of the fluid domain is far away from the surface of the object, the Σ related term is almost zero and can be ignored. The simplified quantitative analysis expression of the aerodynamic torque is:

[0074]

[0075] Where M k 、M vp and M v They represent the aerodynamic torque generated by the acceleration effect of the object surface, the viscosity effect in the pressure boundary, and the friction effect; M Q represents the aerodynamic torque generated by the vortex structure. Therefore, by integrating the above formula over the spatial region where the local vortex structure is located, the aerodynamic torque generated by the local vortex structure can be obtained.

[0076] The threshold adaptive vortex identification method in the present invention adopts the Q criterion as the vortex criterion to identify the vortex structure in the flow field. Under the premise of ensuring that the aerodynamic load deviation of the extracted vortex structure does not exceed 5%, in order to minimize the entanglement between different vortex structures and facilitate the subsequent vortex segmentation, the present invention adopts the following Q criterion threshold Q th The adaptive method avoids the empiricism of threshold selection. The specific method is as follows:

[0077] (1) Determine the maximum value Q of the entire flow field max , the interval [0,Q max ] is divided into N small interval;

[0078] (2) For each Q value interval, perform volume integration on the area where the Q value is within the interval to obtain the M corresponding to the area. Q value;

[0079] (3) Draw the Q 累计 The value is the horizontal axis, M Q The cumulative proportion is M on the vertical axis Q Cumulative proportion distribution curve, such as Figure 2 As shown, the independent variable is Q 累计, M Q The cumulative percentage calculation formula is:

[0080]

[0081] Get command M Q Cumulative share = 5% of Q 累计 value, as the threshold Q th =Q 累计 ; Finally, Q>Q th The area is taken as the extracted vortex structure area.

[0082] In order to solve the problem of inaccurate segmentation of traditional density clustering algorithms under non-uniform body-fitting grids, this paper proposes an eddy segmentation method based on adaptive variable density clustering algorithm (AVDSC). The distance threshold eps parameter in the traditional density clustering algorithm is a constant value, while the AVDSC algorithm establishes a distance threshold eps parameter and the minimum distance d between the local units of the body-fitting CFD grid. min The relationship between the local cell density information of the body-fitting CFD grid is introduced to achieve spatial adaptation of the distance threshold eps parameter in the density clustering algorithm. The specific process of this method is as follows:

[0083] 1) Initialization: Set the minimum number of neighbors ns and the distance threshold eps parameter and the minimum distance d between local cells of the body-fitting CFD grid min The proportionality coefficient A eps .

[0084] 2) Traverse each unmarked unit center point p in the extracted vortex structure area.

[0085] Read the body-fitting CFD grid file, determine the other grid cells adjacent to the grid cell, calculate the distance between the center of the grid cell and the centers of these adjacent cells, and take the minimum value, which is d min , the distance threshold eps of this grid cell is equal to A eps ×d min ;

[0086] Calculate the number of grid cell center points N in the vortex structure area within the neighborhood of the grid cell center point eps eps (p), if N eps (p)≥ns, the center of the unit is marked as the core point.

[0087] 3) If the unit center point p is a core point, create a cluster and add the unit center point p to this cluster. For each vortex structure area unit center point q in the eps-neighborhood of p:

[0088] If q is a boundary point (i.e., not a non-core point, but within the eps-neighborhood of a core point), the point is marked and added to the current cluster;

[0089] If q is a core point, q and all points in its eps-neighborhood are marked and added to the current cluster, and then the process is repeated for each point in the eps-neighborhood of q. After the process is repeated, return to step 2) until the traversal is completed.

[0090] 4) After the traversal process is completed, each cluster obtained corresponds to a local vortex structure area.

[0091] The specific effects are as follows: Figure 3 As shown, compared Figure 1 The AVDSC algorithm can effectively solve the defects of the traditional density clustering algorithm, such as the incorrect merging of vortex structures in the near-object surface area when the EPS parameter is large and the incorrect segmentation of vortex structures in the far-object surface area when the EPS parameter is small, and significantly improves the accuracy of vortex structure segmentation.

[0092] Using the aforementioned vortex segmentation method based on local grid spacing adaptive-variable density clustering algorithm, the vortex structure area extracted at each moment can be segmented into local vortex structure areas, but the correspondence between local vortex structures at different moments is lacking. Therefore, the present invention proposes a vortex tracking method based on vortex structure position prediction and matching, which uses the cluster (vortex structure) centroid obtained by clustering to predict the position of the cluster centroid at the next moment:

[0093] X pre =X old +V old ×dt

[0094] Among them, X pre is the position of the cluster centroid at the next moment; X old is the position of the centroid of the cluster at the current time t; V old is the velocity of the cluster center at the current moment t; dt is the time interval, that is, the next moment is t+dt.

[0095] The cluster centroid positions obtained by clustering at the next moment are further matched one by one (the distance between the centroid position of each cluster and the predicted position is calculated, and the cluster with the smallest distance to the predicted position at the next moment is matched with the cluster) to achieve real-time tracking of the position of each vortex structure area.

[0096] After the above operation, the regional position of each vortex structure at different times can be obtained. According to the calculated load type, the expression given by formula (6) or formula (11) can be used to calculate the aerodynamic load induced by each vortex structure at different times, and the relationship between the vortex dynamic characteristics such as the vortex regional position, size and strength and the aerodynamic characteristics is quantitatively established, which facilitates a deep understanding of the influence of the vortex structure and its dynamic evolution on the aerodynamic characteristics.

[0097] Based on the above methods proposed in the present invention, the steps of the quantitative analysis method for complex flow mechanism for body-fitted grid CFD calculation are as follows:

[0098] Step 1: Acquisition of body-fitted grid CFD calculation data.

[0099] First, through body-fitting grid CFD calculation, the aerodynamic load data at different times, body-fitting grid files (including grid information such as unit center point position, grid unit surface normal vector, unit volume, etc.), and flow field information (including physical quantity data such as velocity, Q value, vorticity, etc.) are obtained;

[0100] Then, the body-fitting grid file at each moment is read, and according to the load type being analyzed, the weight function value of the center point of different grid units at each moment is obtained by solving the weight function equation shown in formula (3) or formula (8).

[0101] Step 2: Observe the evolution of the vortex structure in the flow field

[0102] Use post-processing software to visualize the flow field data at different times obtained in step 1 and conduct qualitative analysis of the changes in aerodynamic loads. At the same time, obtain the maximum Q value of the entire flow field at different times. max The cumulative distribution curve of the vortex-induced aerodynamic load at different times is calculated by the aforementioned threshold-adaptive vortex identification method to determine the Q criterion threshold Q at different times. th .

[0103] Step 3: Vortex structure identification

[0104] Based on the Q criterion threshold Q at different times determined in step 2 th , the vortex structure is identified at each moment, and Q>Q th The grid points are marked as vortex structure areas (Label_Q=1), and the remaining grid points are marked as other areas (Label_Q=0).

[0105] Step 4: Vortex structure segmentation

[0106] The above-mentioned local grid spacing adaptive-variable density clustering algorithm is used to segment the grid points marked with Label_Q=1. The present invention proposes a local grid spacing adaptive-variable density clustering algorithm (AVDSC), and the distance threshold eps parameter is characterized as the minimum distance d between local units of the body-fitting CFD grid. min The function realizes the spatial adaptation of the distance threshold eps parameter in the density clustering algorithm. Specifically, the pseudo code of the AVDSC algorithm is as follows:

[0107]

[0108]

[0109] The result after segmentation is: for the grid with Label_Q=1, the noise point is marked as Label_S=-1, and each vortex structure cluster in the clustering result is marked as Label_S=1,...,N in turn. s .

[0110] Step 5: Vortex structure tracking and matching

[0111] For the labels Label_S=1,...,N s The vortex structure cluster is tracked in time series using the method based on the prediction and matching of the centroid position. Specifically, the centroid position X and velocity V of each vortex structure are calculated by the volume weighted method, and the vortex clustering result (X old ,V old ), predict the vortex structure position X at the current time step pre =X old +V old ×dt, and the actual position X new During tracking, the vortex structure clusters with smaller deviations from the predicted position are matched as the same vortex structure to ensure the continuity of the vortex structure in the time series. If a vortex structure cannot be matched to the corresponding position, the system will assign a new label to it to distinguish the newly born vortex structure.

[0112] After tracking, label it as Label_T=1,...,N t The vortex structure can achieve consistency in the time domain.

[0113] Step 6: Calculation of vortex-induced aerodynamic loads

[0114] Finally, the regional positions of each vortex structure at different times are obtained (the vortex structure region corresponds to the grid point marked as Label_T=i). Based on the analyzed type, the aerodynamic load induced by each vortex structure at different times can be calculated using the expression given in Equation (6) or Equation (11), and the relationship between the vortex dynamic characteristics such as the vortex region position, size and strength and the aerodynamic characteristics is quantitatively established, which facilitates a deep understanding of the influence of the vortex structure and its dynamic evolution on the aerodynamic characteristics.

Claims

1. A quantitative analysis method for complex flow dynamics in body-fitted grid CFD calculations, characterized by: The specific steps are as follows: Step 1: Acquisition of body-fitted grid CFD calculation data; First, the aerodynamic load data, body-fitting grid files, and flow field information at different times are obtained through body-fitting grid CFD calculations. Then, the body-fitting grid files at each time are read, and according to the load type being analyzed, the weight function values ​​of the center points of different grid units at each time are obtained by solving the weight function equation corresponding to the aerodynamic load. Step 2: Observe the evolution of the vortex structure in the flow field; The flow field calculation results at different times are visualized, and the changes in aerodynamic loads are qualitatively analyzed. At the same time, the Q criterion is used as the vortex criterion to identify the vortex structure in the flow field and obtain the maximum Q value of the entire flow field at different times. max By calculating the cumulative distribution curve of the vortex-induced aerodynamic load at different times, the Q criterion threshold Q at different times can be determined. th , ensuring that the aerodynamic load deviation of the extracted vortex structure does not exceed 5%; Step 3: vortex structure identification; Based on the Q criterion threshold Q at different times determined in step 2 th , the vortex structure is identified at each moment, and Q>Q th The grid points are marked as vortex structure areas Label_Q = 1, and the remaining grid points are marked as other areas; The vortex identification method based on threshold self-adaptation is realized through steps 2 and 3; Step 4: vortex structure segmentation; The vortex structure is segmented for the grid points marked as vortex structure areas using a local grid spacing adaptive-variable density clustering algorithm. The method is as follows: 1) Initialization: Set the minimum number of neighbors ns and the distance threshold eps parameter and the minimum distance d between local cells of the body-fitting CFD grid min The proportionality coefficient A eps ; 2) Traverse each unmarked unit center point p in the extracted vortex structure area: Read the body-fitting CFD grid file, determine the other grid cells adjacent to the grid cell, calculate the distance between the center of the grid cell and the centers of these adjacent cells, and take the minimum value, which is d min , the distance threshold eps of this grid cell is equal to A eps ×d min ; Calculate the number of grid cell center points N in the vortex structure area within the neighborhood of the grid cell center point eps eps (p), if N eps (p) ≥ ns, then the center of the unit is marked as the core point; 3) If the unit center point p is a core point, create a cluster and add the unit center point p to this cluster. For each vortex structure area unit center point q in the eps-neighborhood of p: If q is a boundary point, mark the point and add it to the current cluster; If q is a core point, mark q and all points in its eps-neighborhood and add them to the current cluster. Then repeat this process for each point in q's eps-neighborhood. After the process is repeated, return to step 2) until the traversal is completed. 4) After the traversal process is completed, each cluster obtained corresponds to a local vortex structure area; The result after segmentation is: for the grid with Label_Q=1, the noise point is marked as Label_S=-1, and each vortex structure cluster in the clustering result is marked as Label_S=1,...,N in turn. s ; Step 5: vortex structure tracking and matching; For the labels Label_S=1,...,N s The vortex structure cluster is tracked in time series by using the method based on the prediction and matching of the center of mass position; the center of mass position X and velocity V of each vortex structure are calculated by the volume weighted method, and the vortex clustering result (X old ,V old ), predict the vortex structure position X at the current time step pre =X old +V old ×dt, and the actual position X new Make a comparison; After tracking, label it as Label_T=1,...,N t The vortex structure can achieve consistency in the time domain; Step 6: Calculation of vortex-induced aerodynamic loads; Finally, the regional positions of each vortex structure at different times were obtained, and the aerodynamic loads induced by each vortex structure at different times were calculated using the quantitative analysis expression of the aerodynamic loads. The relationship between the vortex regional position, size, and strong and weak vortex dynamic characteristics and aerodynamic characteristics was quantitatively established. The aerodynamic load quantitative analysis expression includes: The quantitative analysis expression of aerodynamic force is: Among them, F k 、F vp and F v They represent the aerodynamic force generated by the acceleration effect of the object surface, the viscosity effect in the pressure boundary, and the friction effect; F Q represents the aerodynamic force generated by the vortex structure; ρ is the fluid density, ω is the vorticity vector, n is the unit vector; φ i is the weight function corresponding to the aerodynamic force; S is the area of ​​the object, u is the velocity of the object, and Q is the second Galilean invariant of the velocity gradient tensor; The quantitative analysis expression of dynamic torque is: Where M k 、M vp and M v They represent the aerodynamic torque generated by the acceleration effect of the object surface, the viscosity effect in the pressure boundary, and the friction effect; M Q represents the aerodynamic torque generated by the vortex structure; e k is the Cartesian basis vector in the k direction, X is the position vector of the node on the surface of the object, and X ea is the rotation center position vector, ψ k is the weight function corresponding to the aerodynamic moment.

2. The method for quantitative analysis of complex flow dynamics for body-fitted grid CFD calculations according to claim 1, characterized in that: In step 1, the weight function corresponding to the aerodynamic load is: The aerodynamic load type is aerodynamic force, and the weight function is: Where B represents the surface boundary of the object; n is a unit vector; Σ is the outer boundary of the fluid domain; n i is the component of the unit vector n in the i direction; The aerodynamic load type is aerodynamic moment, and the weight function is: Among them, e k is the Cartesian basis vector in the k direction; X is the position vector of the node on the surface of the object, X ea is the rotation center position vector.

3. The method for quantitative analysis of complex flow dynamics for body-fitted grid CFD calculations according to claim 1, characterized in that: In step 2 and step 3, the vortex identification method based on threshold self-adaptation is specifically as follows: 1) Determine the maximum value Q of the entire flow field max , the interval [0,Q max ] is evenly divided into N small intervals; 2) For each Q value interval, perform volume integration on the area where the Q value is within the interval to obtain the M corresponding to the area Q value; 3) Plot the Q cumulative value as the horizontal axis, M Q The cumulative proportion is M on the vertical axis Q Cumulative proportion distribution curve; take M Q Cumulative share = 5% of Q 累计 value, as the threshold Q th =Q 累计 ; 4) Finally, Q>Q th The area is taken as the extracted vortex structure area and marked as the vortex structure area Label_Q=1.

4. The method for quantitative analysis of complex flow dynamics for body-fitted grid CFD calculations according to claim 1, characterized in that: In step 4, when performing vortex segmentation, a vortex segmentation method based on local grid spacing adaptive-variable density clustering algorithm is proposed; by establishing the distance threshold eps parameter and the minimum distance d between the local units of the body-fitting CFD grid min The relationship between the local unit density information of the body-fitting CFD grid is introduced to realize the spatial adaptation of the distance threshold eps parameter in the density clustering algorithm; the defects of the traditional density clustering algorithm such as the incorrect merging of vortex structures in the near-object surface area when the eps parameter is large and the incorrect segmentation of vortex structures in the far-object surface area when the eps parameter is small are well solved, and the accuracy of vortex structure segmentation is significantly improved.

5. The method for quantitative analysis of complex flow dynamics for body-fitted grid CFD calculations according to claim 1, characterized in that: In step 5, when tracking, the vortex structure clusters with smaller deviations from the predicted position are matched as the same vortex structure to ensure the continuity of the vortex structure in the time series; If a vortex structure cannot be matched to a corresponding position, a new label is assigned to it to distinguish the newly formed vortex structure.

Images