Sub-orbital aircraft carrying performance optimization method considering flight state constraint
Through an approximate optimization strategy based on the initial sample expansion mechanism and the Kriging agent model, combined with the constraint differential evolution algorithm, the problems of high computing cost and low design efficiency in the design of suborbital aircraft are solved, efficient load optimization is achieved, and payload is improved Mass and meets flight status constraints.
Patent Information
- Application Number
- CN202410816801.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-24
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-06-24
AI Technical Summary
The existing technology is difficult to effectively optimize the design of suborbital aircraft, resulting in high computational cost, low design efficiency, and difficult to quickly converge to the global optimal solution. Especially in the multidisciplinary design optimization problem, traditional optimization methods are difficult to take into account both overload and dynamic pressure. Wait for flight status constraints.
Using an approximate optimization strategy based on the initial sample expansion mechanism, combined with the Kriging proxy model and the constraint differential evolution algorithm, instead of the high-time-consuming analysis model, the proxy model management and update are realized, and the suborbital is efficiently guided. The vehicle carrying performance optimization process converges to the global optimal solution.
It significantly improves the design efficiency of suborbital vehicles, shortens the design cycle, and improves the payload quality. The payload of the optimized solution can be increased by more than 171kg compared with the initial solution, and maximizes the load performance while meeting the flight status constraints.
Smart Images

Figure CN118917037A_ABST
Abstract
Description
Technical Field
[0001] The invention relates to a suborbital vehicle carrying performance optimization method taking flight state constraints into consideration, and belongs to the technical field of suborbital vehicle design optimization. Background Art
[0002] A suborbital vehicle is a new type of vehicle that can reach the boundary of space (about 100 km). Due to its reusability, easy maintenance and high cost-effectiveness, suborbital vehicles have attracted widespread attention from industries such as tourism and transportation (Chang, 2020). In recent years, suborbital vehicles such as Spaceship One, SpaceLiner and Lynx have been fully developed. However, suborbital vehicles often face multiple challenges such as strong overload, large heat flow and high dynamic pressure during their reentry into the Earth's atmosphere. In order to meet the above challenges, domestic researchers have done a lot of research on the design optimization of suborbital vehicle subsystems. Wang (2018) used the direct search region method to optimize the vehicle thermal protection system to minimize the stagnation heat flow during flight. Zhang (2014) designed an optimal angle of attack program for the reentry process of suborbital vehicles based on the particle swarm optimization algorithm. Luo (2023) established the RBCC engine performance model based on the quasi-one-dimensional thermodynamic method, and explored the Pareto frontier of specific impulse enhancement ratio and thrust-area ratio based on the NSGA-Ⅱ method. However, the overall performance of suborbital vehicles is affected by multiple subsystems such as geometric features and aerodynamics, so the design optimization of suborbital vehicles is a typical multidisciplinary design optimization (MDO) problem. In addition, suborbital vehicle design optimization often involves time-consuming analysis models such as aerodynamics and heat. If traditional evolutionary optimization algorithms are still used to solve suborbital vehicle design optimization problems, the computational cost will increase dramatically.
[0003] In order to improve the design efficiency of the suborbital vehicle in the overall design stage, it is very necessary to develop a suborbital vehicle carrying performance optimization method that takes into account flight state constraints, so as to improve the design optimization efficiency and shorten the design cycle, so that the suborbital vehicle design scheme can be quickly optimized and modified in the overall design stage, providing a scientific basis and reference for the suborbital vehicle system scheme demonstration and overall design.
[0004] In order to better illustrate the technical solution of the present invention, the mathematical basis related to the Kriging proxy model involved is introduced below.
[0005] The Kriging surrogate model is an unbiased optimal estimation interpolation model, which is composed of the global model and the local deviation term, as shown in formula (1):
[0006]
[0007] Where μ(x) is a polynomial global approximation model in the design space. When the numerical characteristics of the approximated object are unknown, it is generally taken as a constant μ. The local deviation term Z(x) has a mean of zero and a variance of σ. 2 , a random process with non-zero covariance, characterizing the local deviation based on the global approximate model. The correlation function and covariance matrix can be expressed by equation (2) and equation (3) respectively:
[0008]
[0009] Cov[Z(x i ),Z(x j )]=σ 2 R[R(x i ,x j )] (3)
[0010] Where R is the symmetric correlation matrix; R(·,·) is the correlation function; n v is the design variable dimension; x i are training sample points. μ and σ 2 The estimated value of can be obtained by the least squares method according to formula (4):
[0011]
[0012] Where n KRG Indicates the number of sample points for building the proxy model; I represents n KRG The correlation coefficient θ can be determined by solving the sub-optimization problem in equation (5) using the maximum likelihood method. k
[0013]
[0014] In addition, the Kriging surrogate model can estimate the variance s of the predicted value at any point x 2 (x), thereby evaluating the approximation error of the surrogate model. 2 (x) can be expressed as:
[0015]
[0016] In the formula, :r represents the correlation function vector, and its specific form is as follows
[0017] Summary of the invention
[0018] In order to solve the problem of improving the carrying performance of suborbital vehicles, the purpose of the present invention is to provide a suborbital vehicle carrying performance optimization method that takes into account flight state constraints. This method guides the suborbital vehicle design optimization process to quickly converge to the optimal solution while ensuring that the suborbital vehicle solution meets the constraints of overload, heat flow, dynamic pressure, etc., thereby improving the suborbital vehicle payload. The present invention can be applied to the overall design optimization of suborbital vehicles with different mission requirements, and improve the performance and design efficiency of suborbital vehicles.
[0019] The purpose of the present invention is achieved through the following technical solutions.
[0020] The suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in the present invention comprehensively considers the coupling relationship between key disciplines such as suborbital vehicle geometry and aerodynamics, and establishes a suborbital vehicle multidisciplinary analysis model and a multidisciplinary design optimization problem model. The present invention adopts an approximate optimization strategy based on the initial sample expansion mechanism, takes the maximum payload of the suborbital vehicle as the objective function, and considers flight state constraints such as overload and dynamic pressure to optimize the suborbital vehicle design variables. In the optimization process, the Kriging proxy model is used to replace the time-consuming analysis model of the suborbital vehicle, and the constrained differential evolution algorithm is used to solve the pseudo-optimal solution of the proxy model. Combined with the comprehensive improvement probability and the key design space method, the proxy model management and update are realized, and the suborbital vehicle carrying performance optimization process is efficiently guided to converge to the global optimal solution, that is, the suborbital vehicle carrying performance optimization considering flight state constraints is realized. This method overcomes the problems of low solution efficiency and poor global convergence of traditional optimization methods, can efficiently mine high-performance design solutions for suborbital vehicles, and shorten the design cycle.
[0021] The suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in the present invention comprises the following steps:
[0022] Step 1: Determine the suborbital vehicle launch performance optimization model and determine the initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism.
[0023] The specific implementation method of step one is as follows:
[0024] Suborbital vehicle payload mass m pl The maximum is the objective function, and the take-off mass m is considered takeoff , axial overload N a , Normal overload N n , reentry end velocity V end , reentry range R, lift-to-drag ratio at the end of the reentry phase C l / C d , flight pressure Q, reentry end velocity inclination Θ touch , stagnation point heat flux qstag and the heat flux density q at the leading edge of the wing wing Constraints are set to establish the suborbital vehicle carrying performance optimization model, as shown in formula (8):
[0025] find X=[c v,1 ,c v,2 ,c h,1 ,c h,2 ,L int ,L ext ,b root ,b tip ,x int ,x ext ,L h ,W r ,f u ,f a ,k α ]
[0026] min f(X)=-m pl
[0027]
[0028] Where X represents the design variables, including the coordinates of the control point 1 in the vertical direction of the aircraft head contour c v,1 , the coordinates of the control point 2 in the vertical direction of the aircraft head contour c v,2 , horizontal direction control point 1 coordinate c of the aircraft head contour h,1 , the coordinates of the control point 2 in the vertical direction of the aircraft head contour c h,2 、Inner wing section half span L int 、Outer wing section half span L ext , wing root chord length b root , wing tip chord length b tip , inner wing section sweep angle χ int , outer wing section sweep angle χ ext , head length L h , rear body width W r , Unexpected fuel mass ratio f u , additional fuel mass ratio f a and the angle of attack decrease rate k α As the design variable. LB With X UB represent the lower and upper bounds of the design variables respectively.
[0029] The initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism include the initial sampling number n doe , constraint threshold scaling factor η, initial sample expansion number n l , the number of new sample points n in the local search stage a And the maximum number of model calls
[0030] Step 2: Based on the suborbital vehicle design variables, a parametric geometric master model of the suborbital vehicle is established, and the geometric characteristics of the suborbital vehicle are obtained. The geometric characteristics of the suborbital vehicle include reference area, maximum curvature radius of the head stagnation point, minimum curvature radius of the head stagnation point, wing leading edge curvature radius, wing leading edge sweep angle, total volume, total surface area, wing area and tail area.
[0031] The specific implementation method of step 2 is as follows:
[0032] Based on the NURBS method shown in formula (9), the head, fuselage, wings, rear body and tail of the suborbital vehicle are parametrically modeled to obtain the parametric geometric main model:
[0033]
[0034] Where, d i is the curve control point, R i,p (u) is a rational basis function defined on u∈[0,1], ω i is the weight coefficient, N i,p (u) is the i-th p-order B-spline basis function, and n is the number of nodes. i,p (u) can be expressed by equation (10) and equation (11); when the node u and the order p are determined, the B-spline basis function can be uniquely determined
[0035]
[0036] Where N i,p (u) is the i-th p-order B-spline basis function, u i is the i-th node.
[0037] The parameterized shapes of the head, fuselage, wing, rear body and tail are generated by equations (9)-(11); and the geometric characteristics (reference area, maximum curvature radius of the head stagnation point, minimum curvature radius of the head stagnation point, wing leading edge curvature radius, wing leading edge sweep angle, total volume, total surface area, wing area and tail area) are measured as inputs for obtaining the lift coefficient, drag coefficient, head stagnation point heat flux density, wing inner wing leading edge heat flux density, wing outer wing leading edge heat flux density, reentry mass and ballistic model.
[0038] Step 3: Based on the suborbital vehicle geometric main model and reference area in step 2, obtain the lift coefficient and drag coefficient of the suborbital vehicle.
[0039] The specific implementation method of step three is as follows:
[0040] ①. Based on the parametric main geometric model and computational domain in step 2, divide the unstructured grid; and add local refinement processing at the head stationary point and wing.
[0041] Add boundary layer grids at the head and wings, and the first layer grid height is shown in formula (12):
[0042]
[0043] Where, the dimensionless wall distance y is + ≈1, L r is the reference length of the aircraft, Re is the Reynolds number, which is expressed by formula (13):
[0044]
[0045] In the formula, ρ ∞ is the atmospheric density, v ∞ is the incoming flow velocity in flight conditions, μ is the atmospheric viscosity;
[0046] After local refinement and adding boundary layer mesh, mesh A is obtained.
[0047] ②. Based on the reference area of the suborbital vehicle measured in step 2 and the grid A in ①, the lift coefficient and drag coefficient of the suborbital vehicle are solved by CFD method. In the CFD solution, the density-based solver and energy equation are used, and the incoming flow is set as the pressure far-field boundary condition; the convective upwind splitting format is used for spatial discretization, and the Green-Gauss node method is used to calculate the gradient; in addition, a 3-level FMG initialization is used before starting the solution to improve the solution convergence speed.
[0048] Step 4: According to the maximum curvature radius of the head stagnation point of the suborbital vehicle, the minimum curvature radius of the head stagnation point, the curvature radius of the wing leading edge, the wing leading edge sweep angle, and the flight conditions in step 2, obtain the heat flux density of the head stagnation point of the suborbital vehicle, the heat flux density of the leading edge of the inner wing section, and the heat flux density of the leading edge of the outer wing section.
[0049] The specific implementation method of step 4 is as follows:
[0050] ①. Calculate the stagnation point gas sensible enthalpy and gas wall enthalpy based on equations (14) and (15)
[0051]
[0052] In the formula, c p is the specific heat ratio of air, T ∞ is the incoming flow temperature, R is the ideal gas constant, W is the atmospheric constant, T w is the wall temperature;
[0053] ② Based on formula (16), calculate the stagnation point heat flux value
[0054]
[0055] In the formula, r min is the minimum head stagnation point curvature radius, r max is the maximum head stagnation point curvature radius, ρ sea is the sea level atmospheric density, ρ ∞ is the incoming flow density;
[0056] ③. Based on equations (17) and (18), calculate the heat flux density values of the leading edges of the inner and outer wing sections:
[0057]
[0058] In the formula, q sp is the heat flow of a sphere of equal radius, χ int is the leading edge sweep angle of the inner wing section, χ ext is the sweep angle of the leading edge of the outer wing section, n a is the correction index, α is the angle of attack in the flight condition; q sp With n a The specific value of is given by formula (19) and formula (20)
[0059]
[0060] In the formula, r l is the radius of curvature of the wing leading edge, h 300K is the enthalpy of air at a temperature of 300K, and χ is the wing sweep angle.
[0061] Step 5: Obtain the reentry mass of the suborbital vehicle based on the total volume, total surface area, wing area, and tail area of the suborbital vehicle in step 2.
[0062] The specific implementation method of step five is as follows:
[0063] ①. Based on formula (21), calculate the dry weight of the suborbital vehicle
[0064] m dry =m ext +m p +m tps +m pl +m uc +m es +m of (twenty one)
[0065] In the formula, m ext is the external structure mass, m p To propel the system mass, m tps is the mass of the thermal protection system, muc is the landing gear mass, m es is the mass of the onboard electronic system, m of is the mass of other liquids. ext It is expressed by formula (22)
[0066]
[0067] In the formula, m fu is the fuselage mass, m wing is the wing mass, m vt is the tail mass, w fu is the fuselage unit mass factor, w wing is the wing unit mass factor, w vt is the tail unit mass factor, S fu is the fuselage surface area, S wing is the wing surface area, S vt is the tail surface area, g 0 is the gravitational acceleration at sea level;
[0068] Propulsion system mass m p , thermal protection system mass m tps 、Fuel mass m fuel and other liquid mass m of It is expressed by equation (23)-(26)
[0069] m p =n p × p ×S in (twenty three)
[0070] m tps =w tps ×S wing (twenty four)
[0071] m fuel =ρ fuel ×V tank ×k fuel (25)
[0072] m of =m fuel ×k of (26)
[0073] Where n p is the number of engines, w p is the engine unit mass factor, S in is the engine intake area, w tps is the unit quality factor of the thermal protection system, ρ fuel is the fuel density, V tank is the volume of the fuel tank, kfuel is the fuel loading factor, k of is the fuel-other liquid ratio coefficient;
[0074] ② Solve the linear equations in equation (27) to obtain the suborbital vehicle takeoff mass m takeoff and dry weight m dry
[0075]
[0076] ③. Calculate the reentry mass of the suborbital spacecraft based on formula (28)
[0077] m entry =m dry +(f u +f a )·m fuel (28)
[0078] Step 6: Establish a ballistic model of the suborbital vehicle based on the reference area of the suborbital vehicle in step 2, the lift coefficient and drag coefficient in step 3, the re-entry mass of the suborbital vehicle, and the flight conditions.
[0079] The specific implementation method of step six is as follows:
[0080] ①. Based on formula (29), determine the reentry angle of attack of the suborbital vehicle:
[0081]
[0082] Where v(t) is the flight speed; α max is the initial reentry angle of attack, α min is the final reentry angle, k α is the rate of decrease of angle of attack, v max is the maximum speed before pneumatic deceleration begins, t 1 The time required to reach the maximum speed, t 2 is the time it takes for the velocity change to be Δv; Δv is expressed by formula (30):
[0083]
[0084] ②. Solve the partial differential equations in equation (31) to obtain the velocity, velocity inclination, heading angle, latitude, longitude and distance from the center of the earth during the reentry of the suborbital spacecraft
[0085]
[0086] Where V is the flight speed, θ is the velocity inclination, σ is the yaw angle, r is the distance from the center of the earth, φ is the latitude, λ is the longitude, m is the reentry mass of the spacecraft, g ris the component of gravity acceleration in the direction of the center of the earth, ω e is the angular velocity of the Earth's rotation, g ω is the component of gravity acceleration in the direction of the earth's rotation, ν is the inclination angle of the suborbital vehicle, L(v,α) is the lift, and D(v,α) is the drag.
[0087] Step 7: Based on the suborbital vehicle performance optimization model in step 1, with the maximum payload mass of the suborbital vehicle as the optimization goal, the initial sample expansion mechanism is used to expand the suborbital vehicle sample, and all sample information in the sample database is used to construct the Kriging proxy model of the objective function and the constraint function. The Kriging proxy model is used to replace the time-consuming analysis model of the suborbital vehicle, and the constrained differential evolution algorithm is used to solve the pseudo-optimal solution of the proxy model; combined with the comprehensive improvement probability and key design space method, the proxy model management and update are realized while meeting the design requirements of various disciplines of the suborbital vehicle, and the suborbital vehicle performance optimization process is efficiently guided to converge to the global optimal solution.
[0088] The specific implementation method of step seven is as follows:
[0089] ①. Determine the initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism, including the initial sampling number n doe , constraint threshold scaling factor η, initial sample expansion number n l , the number of new sample points n in the local search stage a And the maximum number of model calls
[0090] ②Use Latin hypersquare experimental design method to construct n in the design space doe Initial sample points are created, and the real analysis model is called to calculate the model response value at the initial sample point; all sample points and response values are added to the sample database; the number of optimization iterations iter is set to 1;
[0091] ③. Based on the initial sample expansion mechanism, obtain n l Sample points;
[0092] (1) Based on the sample database in step ②, establish a random forest classifier as shown in formula (32)
[0093]
[0094] Where n RF The number of sample points for constructing the random forest classifier, is the classification value of the i-th sample point, c i is the constraint violation degree of the i-th sample point, is the normalized value of the jth constraint function, is the constraint threshold of the jth constraint function, P thresh is the constraint violation threshold; P thresh Expressed as
[0095] P thresh =c i,min +η·(c i,max -c i,min ) (33)
[0096] In the formula, c i,min is the minimum constraint violation, c i,max is the maximum constraint violation degree, η is the constraint threshold scaling factor;
[0097] Generate a large number of simple sample points in the design space, and classify the simple sample points by the constructed random forest classifier; if the classification value is 1, the simple sample point is recorded as a high-quality sample point;
[0098] (2) Count the number of high-quality sample points; if there are no high-quality sample points, the constraint threshold is increased by 10%; otherwise, the K-means method is used to obtain the cluster center of high-quality sample points;
[0099] (3) Randomly generate a sample point between the cluster center of high-quality sample points and the sample point with the minimum objective function response value in the sample database of step ②; call the true analysis model to calculate the true model response value of the sample point, and add the sample information to the sample database of step ②; check whether the number of newly added sample points reaches n l , if yes, the expansion mechanism terminates; otherwise, it returns to step (1) and the expansion process continues;
[0100] ④. Use all sample information in the sample database to construct the Kriging proxy model of the objective function and constraint function, and optimize the constructed Kriging proxy model by the constrained differential evolution method to obtain the pseudo optimal solution Call the real analysis model to calculate the real model response value of the pseudo optimal solution, and add the pseudo optimal solution sample information to the sample database;
[0101] ⑤. Solve the sub-optimization problem shown in equation (34) to obtain the sample point x with the maximum probability of comprehensive improvement in the design space PI
[0102]
[0103] Where PI(·) is the probability of objective function improvement, P(gi(·)≤0) is the probability of satisfying the i-th constraint, and d m (X,X P-min ) is the Manhattan distance between the sample point and the optimal comprehensive improvement probability sample point in the current sample database; PI(·), P(gi(·)≤0) and dm (X,X P-min ) are respectively expressed by equation (35), equation (36) and equation (37)
[0104]
[0105] Where Φ(·) is the distribution function of the standard normal distribution, f min is the minimum value of the objective function in the sample database, is the Kriging surrogate model prediction value of the objective function, is the predicted value of the Kriging surrogate model of the i-th constraint function, s(·) is the predicted variance of the Kriging surrogate model of the objective function, s g,i (·) is the predicted variance of the Kriging proxy model of the i-th constraint function; the real analysis model is called to calculate the sample point x with the maximum probability of comprehensive improvement. PI The true model response value, and x PI The sample information is added to the sample database;
[0106] ⑥. Based on the current sample database, construct a key sampling space; the radius of the key sampling space is shown in formula (38)
[0107]
[0108] In the formula, x opt,k is the optimal feasible solution for the kth iteration, x opt,k-1 is the optimal feasible solution for the k-1th iteration, is the sample point with the largest error obtained by the one-by-one verification method when the number of iterations is 1; if x opt,k The optimality of x opt,k-1 If there is improvement, then x opt,k As the focus of the design space center, otherwise x opt,k-1 As a key design space center;
[0109] Randomly generate a sample point x in the key sampling space SSS , call the real analysis model to calculate x SSS The true model response value, and x SSS The sample information is added to the sample database;
[0110] ⑦. Determine whether the number of sample points in the current sample database has reached If it is reached, the optimization is stopped; otherwise, the number of optimization iterations iter is set to iter+1, and go to step 7④ to update the Kriging proxy model of the objective function and the constraint function, and the optimization process continues.
[0111] Step 8: Output the feasible optimal solution in the current sample database as the suborbital vehicle optimization result, that is, to achieve the optimization of the suborbital vehicle carrying performance considering the flight state constraints.
[0112] It also includes step nine: applying the suborbital vehicle carrying performance optimization method considering flight state constraints described in steps one to eight to the suborbital vehicle carrying performance optimization to improve the performance and design efficiency of the suborbital vehicle; according to the suborbital vehicle optimization result obtained in step eight, the suborbital vehicle is made to maximize its carrying performance under the constraints of take-off mass, maximum axial overload, maximum normal overload, reentry stage terminal speed, reentry stage range, reentry stage terminal lift-to-drag ratio, maximum flight operating pressure, reentry stage terminal speed inclination, maximum stagnation point heat flux density, and maximum wing leading edge heat flux density. The suborbital vehicle carrying performance optimization includes the fields of single-stage horizontal take-off and landing suborbital vehicle carrying performance optimization, two-stage horizontal take-off and landing suborbital vehicle carrying performance optimization, and vertical take-off and landing suborbital vehicle carrying performance optimization.
[0113] Beneficial effects:
[0114] 1. Aiming at the problem that the calculation of multidisciplinary analysis models is time-consuming and data mining is insufficient under limited computing costs, the suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in the present invention comprehensively considers the parameter transfer relationship between the suborbital vehicle geometric main model, aerodynamic thermal model, mass estimation model and ballistic model, and establishes a suborbital vehicle multidisciplinary analysis model considering flight state constraints such as overload, dynamic pressure, and heat flow. An approximate optimization strategy based on the initial sample expansion mechanism is adopted to achieve efficient mining of high-quality design solutions for suborbital vehicles. The suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in the present invention combines an approximate optimization strategy based on the initial sample expansion mechanism to expand the suborbital vehicle design variables, and uses all sample information in the sample database to construct a Kriging proxy model of the objective function and the constraint function. The Kriging proxy model is used to replace the high-time analysis model of the suborbital vehicle, and the constrained differential evolution algorithm is used to solve the pseudo-optimal solution of the proxy model; combined with the comprehensive improvement probability and the key design space method, the proxy model management and update are realized while meeting the design requirements of various disciplines of the suborbital vehicle, and the suborbital vehicle carrying performance optimization process is efficiently guided to converge to the global optimal solution. The payload of the optimization scheme obtained by this method can be increased by more than 171kg compared with the initial scheme.
[0115] 2. The suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in the present invention adopts an approximate optimization strategy based on the initial sample expansion mechanism to expand the suborbital vehicle design variables, uses all sample information in the sample database to construct the Kriging proxy model of the objective function and the constraint function, and uses the Kriging proxy model to replace the high-time-consuming analysis model of the suborbital vehicle, and solves the pseudo-optimal solution of the proxy model by the constrained differential evolution algorithm; combined with the comprehensive improvement probability and the key design space method, the proxy model management and update are realized under the condition of meeting the design requirements of various disciplines of the suborbital vehicle, and the suborbital vehicle carrying performance optimization process is efficiently guided to converge to the global optimal solution, further improving the efficiency of suborbital vehicle carrying performance optimization. The global optimal solution corresponds to the suborbital vehicle optimization result, and according to the suborbital vehicle optimization result, the suborbital vehicle realizes the maximization of carrying performance under the flight state constraint. BRIEF DESCRIPTION OF THE DRAWINGS
[0116] Figure 1 This is a schematic diagram of the suborbital vehicle carrying performance optimization method considering flight state constraints of the present invention;
[0117] Figure 2 It is the geometric configuration of the suborbital vehicle;
[0118] Figure 3 This is a flow chart of the approximate optimization strategy based on the initial sample expansion mechanism;
[0119] Figure 4 Optimizing convergence graphs for suborbital vehicles;
[0120] Figure 5 Comparison of the configuration before and after optimization for suborbital vehicles;
[0121] Figure 6 This is a comparison chart of the flight status of suborbital spacecraft, Figure 6 (a) is the suborbital spacecraft altitude-time curve. Figure 6 (b) is the Mach number curve of the suborbital vehicle. Figure 6 (c) is the axial overload curve of the suborbital spacecraft. Figure 6 (d) is the normal overload curve of the suborbital vehicle. Figure 6 (e) is the flight pressure curve of the suborbital spacecraft. Figure 6 (f) is the heat flux density curve of the suborbital spacecraft stationary point. Figure 6 (g) is the heat flux density curve of the leading edge of the suborbital vehicle wing. DETAILED DESCRIPTION
[0122] In order to further illustrate the purpose and advantages of the present invention, the present invention is further described below in conjunction with the accompanying drawings and embodiments.
[0123] Example 1: Multidisciplinary Design Optimization Example of Commercial Suborbital Vehicle
[0124] The suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in this embodiment is applicable to the multidisciplinary design optimization problem of suborbital vehicles for commercial transportation. It can ensure that a high-performance design solution for the suborbital vehicle is obtained in the overall design stage and shorten the design cycle of the suborbital vehicle.
[0125] The suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in this embodiment has the following specific implementation steps:
[0126] Step 1: Determine the suborbital vehicle launch performance optimization model and the initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism.
[0127] The specific implementation method of step one is as follows:
[0128] Suborbital vehicle payload mass m pl The maximum is the objective function, and the take-off mass m is considered takeoff , axial overload N a , Normal overload N n , reentry end velocity V end , reentry range R, lift-to-drag ratio at the end of the reentry phase C l / C d , flight pressure Q, reentry end velocity inclination Θ touch , stagnation point heat flux q stag and the heat flux density q at the leading edge of the wing wing Constraints are set to establish the suborbital vehicle carrying performance optimization model, as shown in formula (39):
[0129] find X=[c v,1 ,c v,2 ,c h,1 ,c h,2 ,L int ,L ext ,b root ,b tip ,x int ,x ext ,L h ,W r ,f u ,f a ,k α ]
[0130] min f(X)=-m pl
[0131]
[0132] Where X represents the design variables, including the coordinates of the control point 1 in the vertical direction of the aircraft head contour c v,1 , the coordinates of the control point 2 in the vertical direction of the aircraft head contour c v,2 , horizontal direction control point 1 coordinate c of the aircraft head contour h,1 , the coordinates of the control point 2 in the vertical direction of the aircraft head contour c h,2 、Inner wing section half span L int 、Outer wing section half span L ext , wing root chord length b root , wing tip chord length b tip , inner wing section sweep angle χ int , outer wing section sweep angle χ ext , head length L h , rear body width W r , Unexpected fuel mass ratio f u , additional fuel mass ratio f a and the angle of attack decrease rate k α As the design variable. LB With X UB represent the lower and upper bounds of the design variables respectively.
[0133] The initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism include the initial sampling number n doe , constraint threshold scaling factor η, initial sample expansion number n l , the number of new sample points n in the local search stage a And the maximum number of model calls
[0134] Step 2: Based on the suborbital vehicle design variables, a parametric geometric master model of the suborbital vehicle is established, and the geometric characteristics of the suborbital vehicle (reference area, maximum curvature radius of the head stagnation point, minimum curvature radius of the head stagnation point, wing leading edge curvature radius, wing leading edge sweep angle, total volume, total surface area, wing area and tail area) are obtained.
[0135] The specific implementation method of step 2 is as follows:
[0136] Based on the NURBS method shown in formula (40), the suborbital vehicle head, fuselage, wings, rear body and tail are parametrically modeled to obtain the parametric geometric main model:
[0137]
[0138] Where, d i is the curve control point, R i,p (u) is a rational basis function defined on u∈[0,1], ω i is the weight coefficient, N i,p(u) is the i-th p-order B-spline basis function, and n is the number of nodes. i,p (u) can be expressed by equation (41) and equation (42); when the node u and the order p are determined, the B-spline basis function can be uniquely determined
[0139]
[0140] Where N i,p (u) is the i-th p-order B-spline basis function, u i is the i-th node.
[0141] The parameterized shapes of the head, fuselage, wing, rear body and tail are generated by equations (40)-(42); and the geometric characteristics (reference area, maximum curvature radius of the head stagnation point, minimum curvature radius of the head stagnation point, wing leading edge curvature radius, wing leading edge sweep angle, total volume, total surface area, wing area and tail area) are measured as inputs for obtaining the lift coefficient, drag coefficient, head stagnation point heat flux density, wing inner wing section leading edge heat flux density, wing outer wing section leading edge heat flux density, reentry mass and ballistic model.
[0142] Step 3: Based on the suborbital vehicle geometric main model and reference area in step 2, obtain the lift coefficient and drag coefficient of the suborbital vehicle.
[0143] The specific implementation method of step three is as follows:
[0144] ①. Based on the parametric main geometric model and computational domain in step 2, divide the unstructured grid with a grid size of 0.5m; and add local refinement processing at the head stationary point and wing, with a grid size of 0.15m.
[0145] Add boundary layer grids at the head and wings, and the first layer grid height is shown in formula (43):
[0146]
[0147] Where, the dimensionless wall distance y is + ≈1, L r is the reference length of the aircraft, Re is the Reynolds number, which is expressed by formula (44):
[0148]
[0149] In the formula, ρ ∞ is the atmospheric density, v ∞ is the incoming flow velocity in flight conditions, μ is the atmospheric viscosity;
[0150] After local refinement and adding boundary layer mesh, mesh A is obtained.
[0151] ②. Based on the reference area of the suborbital vehicle measured in step 2 and the grid A in ①, the lift coefficient and drag coefficient of the suborbital vehicle are solved by CFD method. In the CFD solution, the density-based solver and energy equation are used, and the incoming flow is set as the pressure far-field boundary condition; the convective upwind splitting format is used for spatial discretization, and the Green-Gauss node method is used to calculate the gradient; in addition, a 3-level FMG initialization is used before starting the solution to improve the solution convergence speed.
[0152] Step 4: According to the maximum curvature radius of the head stagnation point of the suborbital vehicle, the minimum curvature radius of the head stagnation point, the curvature radius of the wing leading edge, the wing leading edge sweep angle, and the flight conditions in step 2, obtain the heat flux density of the head stagnation point of the suborbital vehicle, the heat flux density of the leading edge of the inner wing section, and the heat flux density of the leading edge of the outer wing section.
[0153] The specific implementation method of step 4 is as follows:
[0154] ①. Calculate the stagnation point gas sensible enthalpy and gas wall enthalpy based on equations (45) and (46)
[0155]
[0156] In the formula, c p is the specific heat ratio of air, T ∞ is the incoming flow temperature, R is the ideal gas constant, W is the atmospheric constant, T w is the wall temperature;
[0157] ② Based on formula (47), calculate the stagnation point heat flux value
[0158]
[0159] In the formula, r min is the minimum head stagnation point curvature radius, r max is the maximum head stagnation point curvature radius, ρ sea is the sea level atmospheric density, ρ ∞ is the incoming flow density;
[0160] ③. Based on equations (48) and (49), calculate the heat flux density values of the leading edges of the inner and outer wing sections:
[0161]
[0162]
[0163] In the formula, q sp is the heat flow of a sphere of equal radius, χ int is the leading edge sweep angle of the inner wing section, χ ext is the sweep angle of the leading edge of the outer wing section, n ais the correction index, α is the angle of attack in the flight condition; q sp With n a The specific value of is given by equation (50) and equation (51)
[0164]
[0165] In the formula, r l is the radius of curvature of the wing leading edge, h 300K is the enthalpy of air at a temperature of 300K, and χ is the wing sweep angle.
[0166] Step 5: Obtain the reentry mass of the suborbital vehicle based on the total volume, total surface area, wing area, and tail area of the suborbital vehicle in step 2.
[0167] The specific implementation method of step five is as follows:
[0168] ①. Based on formula (52), calculate the dry weight of the suborbital vehicle:
[0169] m dry =m ext +m p +m tps +m pl +m uc +m es +m of (52)
[0170] In the formula, m ext is the external structure mass, m p To propel the system mass, m tps is the mass of the thermal protection system, m uc is the landing gear mass, m es is the mass of the onboard electronic system, m of is the mass of other liquids. ext It is expressed by formula (53)
[0171]
[0172] In the formula, m fu is the fuselage mass, m wing is the wing mass, m vt is the tail mass, w fu is the fuselage unit mass factor, w wing is the wing unit mass factor, w vt is the tail unit mass factor, S fu is the fuselage surface area, S wing is the wing surface area, S vt is the tail surface area, g 0 is the gravitational acceleration at sea level;
[0173] Propulsion system mass m p , thermal protection system mass m tps 、Fuel mass m fuel and other liquid mass m of It is expressed by equation (54)-equation (57)
[0174] m p =n p × p ×S in (54)
[0175] m tps =w tps ×S wing (55)
[0176] m fuel =ρ fuel ×V tank ×k fuel (56)
[0177] m of =m fuel ×k of (57)
[0178] Where n p is the number of engines, w p is the engine unit mass factor, S in is the engine intake area, w tps is the unit quality factor of the thermal protection system, ρ fuel is the fuel density, V tank is the volume of the fuel tank, k fuel is the fuel loading factor, k of is the fuel-other liquid ratio coefficient;
[0179] ② Solve the linear equations in equation (58) to obtain the suborbital vehicle takeoff mass m takeoff and dry weight m dry
[0180]
[0181] ③. Calculate the reentry mass of the suborbital spacecraft based on formula (59)
[0182] m entry =m dry +(f u +f a )·m fuel (59)
[0183] Step 6: Establish a ballistic model of the suborbital vehicle based on the reference area of the suborbital vehicle in step 2, the lift coefficient and drag coefficient in step 3, the re-entry mass of the suborbital vehicle, and the flight conditions.
[0184] The specific implementation method of step six is as follows:
[0185] ①. Based on formula (60), determine the reentry angle of attack of the suborbital vehicle:
[0186]
[0187] Where v(t) is the flight speed; α max is the initial reentry angle of attack, α min is the final reentry angle, k α is the rate of decrease of angle of attack, v max is the maximum speed before pneumatic deceleration begins, t 1 The time required to reach the maximum speed, t 2 is the time it takes for the velocity change to be Δv; Δv is expressed by equation (61):
[0188]
[0189] ②. Solve the partial differential equations in equation (62) to obtain the velocity, velocity inclination, heading angle, latitude, longitude and distance from the center of the earth during the reentry of the suborbital spacecraft:
[0190]
[0191] Where V is the flight speed, θ is the velocity inclination, σ is the yaw angle, r is the distance from the center of the earth, φ is the latitude, λ is the longitude, m is the reentry mass of the spacecraft, g r is the component of gravity acceleration in the direction of the center of the earth, ω e is the angular velocity of the Earth's rotation, g ω is the component of gravity acceleration in the direction of the earth's rotation, ν is the inclination angle of the suborbital vehicle, L(v,α) is the lift, and D(v,α) is the drag.
[0192] Step 7: Based on the suborbital vehicle carrying performance optimization model in step 1, with the maximum payload mass of the suborbital vehicle as the optimization goal, the suborbital vehicle design variables are optimized using an approximate optimization strategy based on the initial sample expansion mechanism. This allows for efficient exploration of the maximum payload solution while meeting the design requirements of various disciplines of the suborbital vehicle.
[0193] The specific implementation method of step seven is as follows:
[0194] ①. Determine the initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism, including the initial sampling number n doe, constraint threshold scaling factor η, initial sample expansion number n l , the number of new sample points n in the local search stage a And the maximum number of model calls
[0195] ②Use Latin hypersquare experimental design method to construct n in the design space doe Initial sample points are created, and the real analysis model is called to calculate the model response value at the initial sample point; all sample points and response values are added to the sample database; the number of optimization iterations iter is set to 1;
[0196] ③. Based on the initial sample expansion mechanism, obtain n l Sample points;
[0197] (1) Based on the sample database in step ②, establish a random forest classifier as shown in formula (63):
[0198]
[0199] Where n RF The number of sample points for constructing the random forest classifier, is the classification value of the i-th sample point, c i is the constraint violation degree of the i-th sample point, is the normalized value of the jth constraint function, is the constraint threshold of the jth constraint function, P thresh is the constraint violation threshold; P thresh Expressed as
[0200] P thresh =c i,min +η·(c i,max -c i,min ) (64)
[0201] In the formula, c i,min is the minimum constraint violation, c i,max is the maximum constraint violation degree, η is the constraint threshold scaling factor;
[0202] Generate 10,000 simple sample points in the design space, and classify the simple sample points by the constructed random forest classifier; if the classification value is 1, the simple sample point is recorded as a high-quality sample point;
[0203] (2) Count the number of high-quality sample points; if there are no high-quality sample points, the constraint threshold is increased by 10%; otherwise, the K-means method is used to obtain the cluster center of high-quality sample points;
[0204] (3) Randomly generate a sample point between the cluster center of high-quality sample points and the sample point with the minimum objective function response value in the sample database of step ②; call the true analysis model to calculate the true model response value of the sample point, and add the sample information to the sample database of step ②; check whether the number of newly added sample points reaches n l , if yes, the expansion mechanism terminates; otherwise, it returns to step (1) and the expansion process continues;
[0205] ④. Use all sample information in the sample database to construct the Kriging proxy model of the objective function and constraint function, and optimize the constructed Kriging proxy model by the constrained differential evolution method to obtain the pseudo optimal solution Call the real analysis model to calculate the real model response value of the pseudo optimal solution, and add the pseudo optimal solution sample information to the sample database;
[0206] ⑤. Solve the sub-optimization problem shown in equation (65) to obtain the sample point x with the maximum probability of comprehensive improvement in the design space PI
[0207]
[0208] Where PI(·) is the probability of objective function improvement, P(gi(·)≤0) is the probability of satisfying the i-th constraint, and d m (X,X P-min ) is the Manhattan distance between the sample point and the optimal comprehensive improvement probability sample point in the current sample database; PI(·), P(gi(·)≤0) and d m (X,X P-min ) are respectively expressed by equation (66), equation (67) and equation (68)
[0209]
[0210] Where Φ(·) is the distribution function of the standard normal distribution, f min is the minimum value of the objective function in the sample database, is the Kriging surrogate model prediction value of the objective function, is the predicted value of the Kriging surrogate model of the i-th constraint function, s(·) is the predicted variance of the Kriging surrogate model of the objective function, s g,i (·) is the predicted variance of the Kriging proxy model of the i-th constraint function; the real analysis model is called to calculate the sample point x with the maximum probability of comprehensive improvement. PI The true model response value, and x PI The sample information is added to the sample database;
[0211] ⑥. Based on the current sample database, construct a key sampling space; the radius of the key sampling space is shown in formula (69)
[0212]
[0213] In the formula, x opt,k is the optimal feasible solution for the kth iteration, x opt,k-1 is the optimal feasible solution for the k-1th iteration, is the sample point with the largest error obtained by the one-by-one verification method when the number of iterations is 1; if x opt,k The optimality of x opt,k-1 If there is improvement, then x opt,k As the focus of the design space center, otherwise x opt,k-1 As a key design space center;
[0214] Randomly generate a sample point x in the key sampling space SSS , call the real analysis model to calculate x SSS The true model response value, and x SSS The sample information is added to the sample database;
[0215] ⑦. Determine whether the number of sample points in the current sample database has reached If it is reached, the optimization is stopped; otherwise, the number of optimization iterations iter is set to iter+1, and go to step 7④ to update the Kriging proxy model of the objective function and the constraint function, and the optimization process continues.
[0216] Step 8: Output the feasible optimal solution in the current sample database as the optimization result of the suborbital vehicle design scheme.
[0217] In order to better demonstrate the effectiveness and engineering practicality of the present invention, the present invention is further explained below by taking the multidisciplinary design optimization problem of a commercial suborbital vehicle as an example with reference to the accompanying drawings and tables.
[0218] Step 9: In this embodiment, the design variable value range is c v,1 ∈[750mm,900mm], c v,2 ∈[2000mm,2200mm],c h,1 ∈[1150mm,1350mm],c h,2 ∈[2100mm,2300mm],L int ∈[2200mm,5000mm],L ext ∈[3800mm,6500mm], b root ∈[10000mm,12300mm], b tip ∈[1200mm,1800mm], χ int∈[50°,60°], χ ext ∈[15°,25°],L h ∈[4230mm,5170mm],W r ∈[1980mm,2420mm],f u ∈[0.018,0.022], f a ∈[0.018,0.022], k α ∈[0.010°s / m,0.015°s / m]. The approximate optimization strategy based on the initial sample expansion mechanism is set as follows: The initial sampling number n doe is 30, the constraint threshold scaling factor η is 0.2, and the initial sample expansion number n l 30, the number of sample points n added in the local search stage a 1, the maximum number of model calls is 150. The optimization convergence curve obtained based on the approximate optimization strategy of the initial sample expansion mechanism and the efficient global optimization method is as follows: Figure 4 The comparison of the suborbital vehicle objective function before and after optimization is shown in Table 1, the design variables are shown in Table 2, and the constraints are shown in Table 3.
[0219] Table 1 Comparison of objective functions before and after optimization
[0220]
[0221] Table 2 Comparison of design variables before and after optimization
[0222]
[0223] Table 3 Comparison of constraints before and after optimization
[0224]
[0225]
[0226] As shown in Table 1, the payload mass of the optimized solution is increased by more than 171 kg compared with the initial solution. Figure 5 It can be seen that the optimized configuration has undergone significant changes compared to the initial configuration. The span and wing root chord length are increased by 789.07 mm and 1054.89 mm respectively, thus providing more lift. Figure 6 (b) It can be seen that the Mach number of the optimized solution during the reentry phase is lower than that of the initial solution, thus having a higher lift and lift-to-drag ratio. Figure 6 (e) It can be seen that the maximum dynamic pressure of the optimized solution is reduced by 23.21% compared with the initial solution, thereby reducing the structural design burden of the suborbital vehicle. Figure 6 (f) Figure 6In (g), the maximum heat flux of the optimized solution at the stagnation point and the leading edge of the wing is reduced by 64.85% and 14.95% respectively compared with the initial solution, thereby reducing the risk of ablation faced by the suborbital vehicle during re-entry. In summary, the above optimization results show that the suborbital vehicle carrying performance optimization method considering flight state constraints disclosed in the present invention can obtain a suborbital vehicle design solution with a higher payload and meeting actual engineering needs compared to the initial solution. The above optimization results verify the rationality, effectiveness and engineering practicality of the present invention.
[0227] The specific description above further illustrates the purpose, technical solutions and beneficial effects of the invention in detail. It should be understood that the above is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.
Claims
1. A method for optimizing the carrying performance of a suborbital vehicle considering flight state constraints, characterized in that: The following steps are included: Step 1: Determine the suborbital vehicle launch performance optimization model and determine the initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism; Step 2: Based on the suborbital vehicle design variables, a parametric geometric master model of the suborbital vehicle is established, and the geometric characteristics of the suborbital vehicle are obtained; The geometric characteristics of the suborbital vehicle include reference area, maximum radius of curvature of the head stagnation point, minimum radius of curvature of the head stagnation point, radius of curvature of the wing leading edge, sweep angle of the wing leading edge, total volume, total surface area, wing area and tail area; Step 3: Obtain the lift coefficient and drag coefficient of the suborbital vehicle based on the suborbital vehicle geometric main model and reference area in step 2; Step 4: According to the maximum curvature radius of the head stagnation point of the suborbital vehicle, the minimum curvature radius of the head stagnation point, the curvature radius of the wing leading edge, the sweep angle of the wing leading edge, and the flight conditions in step 2, obtain the heat flux density of the head stagnation point of the suborbital vehicle, the heat flux density of the leading edge of the inner wing section of the wing, and the heat flux density of the leading edge of the outer wing section of the wing; Step 5: Obtain the reentry mass of the suborbital vehicle based on the total volume, total surface area, wing area, and tail area of the suborbital vehicle in step 2; Step 6: Establish a ballistic model of the suborbital vehicle based on the reference area of the suborbital vehicle in step 2, the lift coefficient and drag coefficient in step 3, the reentry mass of the suborbital vehicle, and the flight conditions; Step 7: Based on the suborbital vehicle carrying performance optimization model in step 1, the maximum payload mass of the suborbital vehicle is taken as the optimization goal, and the suborbital vehicle sample is expanded using the initial sample expansion mechanism. The Kriging proxy model of the objective function and the constraint function is constructed using all sample information in the sample database. The Kriging proxy model is used to replace the time-consuming analysis model of the suborbital vehicle, and the constrained differential evolution algorithm is used to solve the pseudo-optimal solution of the proxy model; combined with the comprehensive improvement probability and the key design space method, the proxy model management and update are realized while meeting the design requirements of various disciplines of the suborbital vehicle, and the suborbital vehicle carrying performance optimization process is efficiently guided to converge to the global optimal solution; Step 8: Output the feasible optimal solution in the current sample database as the suborbital vehicle optimization result, that is, to achieve the optimization of the suborbital vehicle carrying performance considering the flight state constraints.
2. The method according to claim 1, characterized in that: The specific implementation method of step one is as follows: Determine the initial parameters of the suborbital vehicle launch performance optimization model and the approximate optimization strategy based on the initial sample expansion mechanism; Suborbital vehicle payload mass m pl The maximum is the objective function, and the take-off mass m is considered takeoff , axial overload N a , Normal overload N n , reentry end velocity V end , reentry range R, lift-to-drag ratio at the end of the reentry phase C l / C d , flight pressure Q, reentry end velocity inclination Θ touch , stagnation point heat flux q stag and the heat flux density q at the leading edge of the wing wing Constraints are set to establish the suborbital vehicle carrying performance optimization model, as shown in formula (1): find X=[c v,1 ,c v,2 ,c h,1 ,c h,2 ,L int ,L ext ,b root ,b tip ,χ int ,χ ext ,L h ,W r ,f u ,f a ,k α ] min f(X)=-m pl Where X represents the design variables, including the coordinates of the control point 1 in the vertical direction of the aircraft head contour c v,1 , the coordinates of the control point 2 in the vertical direction of the aircraft head contour c v,2 , horizontal direction control point 1 coordinate c of the aircraft head contour h,1 , the coordinates of the control point 2 in the vertical direction of the aircraft head contour c h,2 、Inner wing section half span L int 、Outer wing section half span L ext , wing root chord length b root , wing tip chord length b tip , inner wing section sweep angle χ int , outer wing section sweep angle χ ext , head length L h , rear body width W r , Unexpected fuel mass ratio f u , additional fuel mass ratio f a and the angle of attack decrease rate k α As the design variable; X LB With X UB denote the lower and upper bounds of the design variables respectively; The initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism include the initial sampling number n doe , constraint threshold scaling factor η, initial sample expansion number n l , the number of new sample points n in the local search stage a And the maximum number of model calls 3. The method according to claim 1, characterized in that: The specific implementation method of step 2 is as follows: Based on the NURBS method shown in formula (2), the suborbital vehicle head, fuselage, wings, rear body and tail are parametrically modeled to obtain the parametric main geometric model: Where, d i is the curve control point, R i,p (u) is a rational basis function defined on u∈[0,1], ω i is the weight coefficient, N i,p (u) is the i-th p-order B-spline basis function, n represents the number of nodes; N i,p (u) can be expressed by equation (3) and equation (4); when the node u and the order p are determined, the B-spline basis function can be uniquely determined Where N i,p (u) is the i-th p-order B-spline basis function, u i is the i-th node; The parameterized shapes of the head, fuselage, wing, rear body and tail are generated by equations (2)-(4); the geometric features are measured and used as input to obtain the lift coefficient, drag coefficient, head stagnation point heat flux density, wing inner wing section leading edge heat flux density, wing outer wing section leading edge heat flux density, reentry mass and ballistic model; the geometric features include reference area, head stagnation point maximum curvature radius, head stagnation point minimum curvature radius, wing leading edge curvature radius, wing leading edge sweep angle, total volume, total surface area, wing area and tail area.
4. The method according to claim 1, characterized in that: The specific implementation method of step three is as follows: ①. Based on the parametric main geometric model and computational domain in step 2, divide the unstructured grid; and add local refinement processing at the head stationary point and wing; Add boundary layer grids at the head and wings, and the first layer grid height is shown in formula (5): Where, the dimensionless wall distance y is + ≈1, L r is the reference length of the aircraft, Re is the Reynolds number, which is expressed by formula (6): In the formula, ρ ∞ is the atmospheric density, v ∞ is the incoming flow velocity in flight conditions, μ is the atmospheric viscosity; After local refinement and adding boundary layer mesh, mesh A is obtained; ②. Based on the reference area of the suborbital vehicle measured in step 2 and the grid A in ①, the lift coefficient and drag coefficient of the suborbital vehicle are solved by the CFD method; the density-based solver and energy equation are used in the CFD solution, and the incoming flow is set as the pressure far-field boundary condition; the convective upwind splitting format is used for spatial discretization, and the Green-Gauss node method is used to calculate the gradient; before starting the solution, a 3-level FMG initialization is used to improve the solution convergence speed.
5. The method according to claim 1, characterized in that: The specific implementation method of step 4 is as follows: ①. Calculate the stagnation point gas sensible enthalpy and gas wall enthalpy based on equations (7) and (8) In the formula, c p is the air specific heat ratio, T ∞ is the incoming flow temperature, R is the ideal gas constant, W is the atmospheric constant, T w is the wall temperature; ②. Based on formula (9), calculate the stagnation point heat flux value In the formula, r min is the minimum head stagnation point curvature radius, r max is the maximum head stagnation point curvature radius, ρ sea is the sea level atmospheric density, ρ ∞ is the incoming flow density; ③. Based on equations (10) and (11), calculate the heat flux density values of the leading edges of the inner and outer wing sections: In the formula, q sp is the heat flow of a sphere of equal radius, χ int is the leading edge sweep angle of the inner wing section, χ ext is the sweep angle of the leading edge of the outer wing section, n a is the correction index, α is the angle of attack in the flight condition; q sp With n a The specific value of is given by formula (12) and formula (13) In the formula, r l is the radius of curvature of the wing leading edge, h 300K is the enthalpy of air at a temperature of 300K, and χ is the wing sweep angle.
6. The method according to claim 1, characterized in that: The specific implementation method of step five is as follows: ①. Based on formula (14), calculate the dry weight of the suborbital vehicle: m dry =m ext +m p +m tps +m pl +m uc +m es +m of (14) In the formula, m ext is the external structure mass, m p To propel the system mass, m tps is the mass of the thermal protection system, m uc is the landing gear mass, m es is the mass of the onboard electronic system, m of is the mass of other liquids; where m ext It is expressed by formula (15) In the formula, m fu is the fuselage mass, m wing is the wing mass, m vt is the tail mass, w fu is the fuselage unit mass factor, w wing is the wing unit mass factor, w vt is the tail unit mass factor, S fu is the fuselage surface area, S wing is the wing surface area, S vt is the tail surface area, g0 is the gravitational acceleration at sea level; Propulsion system mass m p , thermal protection system mass m tps 、Fuel mass m fuel and other liquid mass m of It is expressed by equation (16)-equation (19) m p =n p ×w p ×S in (16) m tps =w tps ×S wing (17) m fuel =ρ fuel ×V tank ×k fuel (18) m of =m fuel ×k of (19) Where n p is the number of engines, w p is the engine unit mass factor, S in is the engine intake area, w tps is the unit quality factor of the thermal protection system, ρ fuel is the fuel density, V tank is the volume of the fuel tank, k fuel is the fuel loading factor, k of is the fuel-other liquid ratio coefficient; ② Solve the linear equations in equation (20) to obtain the suborbital vehicle takeoff mass m takeoff and dry weight m dry ③. Calculate the reentry mass of the suborbital spacecraft based on formula (21) m entry =m dry +(f u +f a )·m fuel (21 ) 。 7. The method according to claim 1, characterized in that: The specific implementation method of step six is as follows: ①. Based on formula (22), determine the reentry angle of attack of the suborbital vehicle: Where v(t) is the flight speed; α max is the initial reentry angle of attack, α min is the final reentry angle, k α is the rate of decrease of angle of attack, v max is the maximum speed before the start of pneumatic deceleration, t1 is the time required to reach the maximum speed, and t2 is the time it takes for the speed change to be Δv; Δv is expressed by equation (23) ②. Solve the partial differential equations in equation (24) to obtain the velocity, velocity inclination, heading angle, latitude, longitude and distance from the center of the earth during the reentry of the suborbital spacecraft Where V is the flight speed, θ is the velocity inclination, σ is the yaw angle, r is the distance from the center of the earth, φ is the latitude, λ is the longitude, m is the reentry mass of the spacecraft, g r is the component of gravity acceleration in the direction of the center of the earth, ω e is the angular velocity of the Earth's rotation, g ω is the component of gravity acceleration in the direction of the earth's rotation, ν is the inclination angle of the suborbital vehicle, L(v,α) is the lift, and D(v,α) is the drag.
8. The method according to claim 1, characterized in that: The specific implementation method of step seven is as follows: ①. Determine the initial parameters of the approximate optimization strategy based on the initial sample expansion mechanism, including the initial sampling number n doe , constraint threshold scaling factor η, initial sample expansion number n l , the number of new sample points n in the local search stage a And the maximum number of model calls ②Use Latin hypersquare experimental design method to construct n in the design space doe Initial sample points are taken and the real analysis model is called to calculate the model response value at the initial sample point; all sample points and response values are added to the sample database; Set the optimization iteration number iter to 1; ③. Based on the initial sample expansion mechanism, obtain n l Sample points; (1) Based on the sample database in step ②, establish a random forest classifier as shown in formula (25) Where n RF The number of sample points for constructing the random forest classifier, is the classification value of the i-th sample point, c i is the constraint violation degree of the i-th sample point, is the normalized value of the jth constraint function, is the constraint threshold of the jth constraint function, P thresh is the constraint violation threshold; P thresh Expressed as P thresh =c i,min +η·(c i,max -c i,min ) (26) In the formula, c i,min is the minimum constraint violation degree, c i,max is the maximum constraint violation degree, η is the constraint threshold scaling factor; Generate a large number of simple sample points in the design space, and classify the simple sample points by the constructed random forest classifier; if the classification value is 1, the simple sample point is recorded as a high-quality sample point; (2) Count the number of high-quality sample points; if there are no high-quality sample points, the constraint threshold is increased by 10%; otherwise, the K-means method is used to obtain the cluster center of high-quality sample points; (3) Randomly generate a sample point between the cluster center of high-quality sample points and the sample point with the minimum objective function response value in the sample database of step ②; call the true analysis model to calculate the true model response value of the sample point, and add the sample information to the sample database of step ②; check whether the number of newly added sample points reaches n l , if yes, the expansion mechanism terminates; otherwise, it returns to step (1) and the expansion process continues; ④. Use all sample information in the sample database to construct the Kriging proxy model of the objective function and constraint function, and optimize the constructed Kriging proxy model by the constrained differential evolution method to obtain the pseudo optimal solution Call the real analysis model to calculate the real model response value of the pseudo optimal solution, and add the pseudo optimal solution sample information to the sample database; ⑤. Solve the sub-optimization problem shown in equation (27) to obtain the sample point x with the maximum probability of comprehensive improvement in the design space PI find X=[c v,1 ,c v,2 ,c h,1 ,c h,2 ,L int ,L ext ,b root ,b tip ,χ int ,χ ext ,L h ,W r ,f u ,f a ,k α ] s.t.X lb ≤X≤X ub Where PI(·) is the probability of objective function improvement, P(g i (·)≤0) is the probability of satisfying the i-th constraint, d m (X,X P-min ) is the Manhattan distance between the sample point and the optimal comprehensive improvement probability sample point in the current sample database; PI(·), P(g i (·)≤0) and d m (X,X P-min ) are respectively expressed by equation (28), equation (29), and equation (30) Where Φ(·) is the distribution function of the standard normal distribution, f min is the minimum value of the objective function in the sample database, is the Kriging surrogate model prediction value of the objective function, is the predicted value of the Kriging surrogate model of the i-th constraint function, s(·) is the predicted variance of the Kriging surrogate model of the objective function, s g,i (·) is the predicted variance of the Kriging proxy model of the i-th constraint function; the real analysis model is called to calculate the sample point x with the maximum probability of comprehensive improvement. PI The true model response value, and x PI The sample information is added to the sample database; ⑥. Based on the current sample database, construct a key sampling space; the radius of the key sampling space is shown in formula (31) In the formula, x opt,k is the optimal feasible solution for the kth iteration, x opt,k-1 is the optimal feasible solution for the k-1th iteration, is the sample point with the largest error obtained by the one-by-one verification method when the number of iterations is 1; if x opt,k The optimality of x opt,k-1 If there is improvement, then x opt,k As the focus of the design space center, otherwise x opt,k-1 As a key design space center; Randomly generate a sample point x in the key sampling space SSS , call the real analysis model to calculate x SSS The true model response value, and x SSS The sample information is added to the sample database; ⑦. Determine whether the number of sample points in the current sample database has reached If it is reached, the optimization is stopped; otherwise, the number of optimization iterations iter is set to iter+1, and go to step 7④ to update the Kriging proxy model of the objective function and the constraint function, and the optimization process continues.
9. The method according to claim 8, characterized in that: The objective function Kriging proxy model in step 7 is: payload mass m pl Kriging proxy model; the constraint function Kriging proxy model in step 7 is: takeoff mass m takeoff Kriging proxy model, axial overload N a Kriging proxy model, normal overload N n Kriging proxy model, reentry end velocity V end The Kriging proxy model of the reentry range R, the Kriging proxy model of the reentry end lift-to-drag ratio C l / C d Kriging proxy model of the flight pressure Q, Kriging proxy model of the flight pressure Q, and the velocity inclination angle Θ at the end of the reentry phase touch Kriging proxy model, head stagnation point heat flux density q stag The Kriging proxy model and the heat flux density q at the leading edge of the wing wing Kriging surrogate model.
10. The method of claim 1, 2, 3, 4, 5, 6, 7, 8 or 9, wherein: The method further includes step nine: applying the suborbital vehicle carrying performance optimization method considering flight state constraints described in steps one to eight to optimize the carrying performance of the suborbital vehicle, thereby improving the performance and design efficiency of the suborbital vehicle; According to the suborbital vehicle optimization results obtained in step eight, the suborbital vehicle is made to maximize its carrying performance under the constraints of takeoff mass, maximum axial overload, maximum normal overload, reentry stage terminal speed, reentry stage range, reentry stage terminal lift-to-drag ratio, maximum flight operating pressure, reentry stage terminal speed inclination, maximum stagnation point heat flux density, and maximum wing leading edge heat flux density. The suborbital vehicle carrying performance optimization includes the fields of single-stage horizontal take-off and landing suborbital vehicle carrying performance optimization, two-stage horizontal take-off and landing suborbital vehicle carrying performance optimization, and vertical take-off and landing suborbital vehicle carrying performance optimization.
Citation Information
Patent Citations
aircraft approximate optimization method based on a filter and an adaptive Kriging model
CN109918809A
Variant aircraft aerodynamic optimization method based on improved position vector expectation improvement degree
CN112329140A
Method and system for designing transonic airfoil profile of sweepback wing of aircraft
CN117910150A
Guided rocket projectile inner and outer trajectory joint design method adopting coupling iterative optimization
CN117932793A
Co-optimization of blunt body shapes for moving vehicles
US8725470B1
Cited By
Rocket aerodynamic configuration optimization method considering carrying capacity
CN120373207A
A rocket aerodynamic shape optimization method considering carrying capacity
CN120373207B
