Online calculation method for gliding trajectory around no-fly zone based on variant aircraft

Abstract

The invention discloses an online calculation method for a gliding track around a no-fly zone based on a variant aircraft. The method comprises the steps of 1, establishing a trajectory problem model of a variant aircraft around a no-fly zone; step 2, calculating lift resistance change caused by shape change of the aircraft; 3, establishing a track sample database, and dividing a training set and a test set; 4, building a variant aircraft trajectory online calculation network model, and training and storing the model; and 5, inputting a state quantity initial value of the trajectory to obtain a predicted trajectory. The variant aircraft is used as a research object, and the problems of avoidance of the no-fly zone and on-line planning of the trajectory of the variant aircraft are mainly considered. The aerodynamic characteristics of the morphing aircraft are analyzed, a pseudo-spectral method and an artificial intelligence algorithm are combined, the network model is trained, and online optimization solving of the trajectory is achieved.

Description

Technical Field

[0001] The invention belongs to the field of trajectory calculation of variant aircraft, and in particular relates to an online trajectory calculation method for a variant aircraft to avoid a no-fly zone. Background Art

[0002] With the continuous advancement of aviation technology, the limitations of the aerodynamic characteristics of traditional fixed-configuration aircraft are gradually emerging when facing complex and changing flight environments. In order to adapt to the diverse mission requirements in cross-domain flight, variant aircraft came into being. This type of aircraft can adjust its wing shape or plane shape according to flight conditions and mission requirements to achieve optimal flight performance. Variant aircraft represents the development direction of future aircraft, and in this development process, trajectory optimization calculation problems play a vital role.

[0003] At present, there are two main numerical methods for solving trajectories: direct method and indirect method. The indirect method transforms the optimal control problem into a Hamilton boundary value problem based on the Pontryagin minimum principle. When solving, it is necessary to use the Pontryagin minimum principle to derive the first-order necessary conditions for optimal control, and the derivation process is relatively complicated. The direct method adopts the method of discretization first and then optimization, and uses the parameterization method to transform the optimal control problem in continuous space into a nonlinear programming problem. The nonlinear programming problem is solved by numerical methods to obtain the optimal trajectory. The pseudo-spectral method originated from the spectral method for solving differential equations in the early days. This method is consistent with the idea of ​​the collocation method. The collocation points are obtained through a series of complex operations, while the pseudo-spectral method generally selects the roots of orthogonal polynomials as collocation points. Compared with the traditional direct method, the pseudo-spectral method can achieve higher calculation accuracy with fewer parameters, so it is widely used in aircraft trajectory planning problems. However, the solution speed of the pseudo-spectral method still cannot meet the requirements of real-time trajectory calculation.

[0004] In recent years, with the continuous development of artificial intelligence, deep neural networks have successfully predicted the trajectory of aircraft by learning a large amount of data on optimized trajectories and finding the relationship between state quantities and control quantities. Therefore, this paper generates a trajectory sample database based on the pseudo-spectral method and combines it with deep neural networks to realize online trajectory calculation. Carrying out trajectory optimization related work for variable configuration aircraft is of great significance to the development of aviation. Summary of the invention

[0005] In view of the above problems, the purpose of the present invention is to propose an online calculation method based on the gliding trajectory of a variant aircraft around a no-fly zone. For the trajectory optimization problem of a variant aircraft around a no-fly zone, a large number of trajectory samples are generated by the pseudo-spectral method, and a deep neural network model is built and trained, so as to effectively improve the efficiency of trajectory optimization solution and realize online trajectory calculation.

[0006] To achieve the objective of the present invention, a technical solution based on an online calculation method of a gliding trajectory of a morphing aircraft around a no-fly zone comprises the following steps:

[0007] Step 1: Determine the trajectory calculation problem model for the variant aircraft in the gliding phase when circling the no-fly zone. The variant aircraft must reach a certain terminal position while circling the no-fly zone, so the shortest flight time mint f As an optimization goal;

[0008] Step 2: Calculate the values ​​of lift L and drag D that change constantly during flight, and use interpolation to achieve real-time calculation of aerodynamic data;

[0009] Step 3: Establish a sample database of trajectory optimization of variant aircraft around the no-fly zone, use the problem model established in step 1, and use the Radau pseudo-spectral method to solve and generate different optimized trajectories with r0, φ0, θ0;

[0010] Step 4: training and testing the online computing network model for variant aircraft trajectories. The training set and test set constructed in step 3 are used to train and test the online computing network model for variant aircraft trajectories, respectively, to obtain the final network model.

[0011] Step 5: Use the state data of the variant aircraft at the initial moment as the input of the trajectory online network model, and combine it with the Euler integration method to realize the online calculation of the trajectory of the variant aircraft around the no-fly zone.

[0012] The present invention is further improved in that: in step 1, the actual situation of the morphing aircraft is deeply analyzed, and the deformation of the aircraft is included in the research scope as a key control parameter. At the same time, the constraints of the no-fly zone are taken into consideration to ensure that the flight path planning of the morphing aircraft is more in line with the actual operating environment.

[0013] The present invention is further improved in that: in the step 2, special attention is paid to the analysis of the variant aircraft. Through this strategy, the aerodynamic characteristics of the aircraft, especially the lift-drag characteristics, can be optimized with the adjustment of the deformation amount. This deformation capability significantly enhances the performance of the variant aircraft. At the same time, through the precise trajectory optimization algorithm, the optimization accuracy of the flight trajectory is further improved.

[0014] A further improvement of the present invention is that in step 4, the trajectory data is learned by constructing a deep neural network model, which improves the computational efficiency of trajectory optimization and ensures that the aircraft can make rapid adjustments to maintain flight safety when facing changing flight conditions.

[0015] The beneficial effects of the present invention are as follows: the present invention takes the variant aircraft as the research object, focusing on the obstacle avoidance capability of the variant aircraft and the online optimization calculation of the trajectory. In order to improve the efficiency of trajectory planning, the present invention adopts deep neural network technology to optimize the trajectory solution process and realize real-time trajectory calculation. By adjusting the shape of the aircraft, its aerodynamic characteristics can be optimized, thereby improving the flight trajectory. In the context of increasingly complex flight missions, improving the computational efficiency of trajectory optimization can not only improve the flexibility of mission execution, but also enhance the safety and response speed of the aircraft, which is crucial to the success of the flight mission. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the process of the present invention;

[0017] Figure 2 It is a schematic diagram of the state quantity of trajectory samples;

[0018] Figure 3 Schematic diagram of online trajectory planning architecture; DETAILED DESCRIPTION

[0019] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0020] The process of the present invention is as follows Figure 1 As shown, this example provides an online calculation method based on the gliding trajectory of a variant aircraft around a no-fly zone, including the following steps:

[0021] Step 1: This invention takes a variable-sweep aircraft as an example to analyze the trajectory optimization problem of the aircraft's gliding phase. Assuming that the earth is a uniform sphere, the aircraft is a point mass flying without power, and that the elastic deformation of the aircraft structure is not considered during the flight, the motion model of the variable-sweep aircraft is as follows:

[0022]

[0023] In the formula: r represents the distance from the center of the earth; Φ, θ represent longitude and latitude respectively, which are used to determine the position of the waverider; v represents velocity; γ represents track angle; ψ represents heading angle; σ represents roll angle; g represents gravitational acceleration; L, D represent lift acceleration and drag acceleration respectively.

[0024] In addition, considering the situations that the aircraft may encounter during actual flight, the no-fly zone is introduced as a constraint into the trajectory optimization problem. The no-fly zone constraint can be expressed as:

[0025] d(ζ(φ,θ),Ι)-R r <0

[0026] Where: The projection of the optimal trajectory on the horizontal plane is Ι(φ,θ), ζ(φ,θ) is the center of the no-fly zone, and R r The height of the no-fly zone, which can be infinitely high.

[0027] Step 2: The shape of the aircraft directly affects the aerodynamic coefficient, thereby changing the lift and drag characteristics of the aircraft. The calculation expression of lift and drag is as follows:

[0028]

[0029] Where: S ref represents the aerodynamic reference area of ​​the variant aircraft; q d represents dynamic pressure, q d =0.5ρv 2 ; ρ is the atmospheric density; m represents the mass of the aircraft; lift and drag coefficient C L ,C D It is a function related to the angle of attack α, Mach number Ma, and sweep angle Λ.

[0030] Step 3: Use the Radau pseudo-spectral method to solve the trajectory and establish a trajectory sample database. The core idea is: first, discretize the state and control variables at a series of LGR nodes, and then use Lagrange interpolation polynomials to approximate the state and control variables of the aircraft. In this way, the differential operations in the state equation and the integral operations in the performance function are converted into algebraic operations. Finally, the optimal control problem is converted into a nonlinear programming problem, which is easy to solve. Since different optimized trajectories need to be generated for network training, the orbiting range is added to the position information at the initial and terminal moments, and finally 500 different optimized trajectories are generated. The generated trajectory data is as follows: Figure 2 shown.

[0031] Step 4: Training and testing of optimized trajectory data. The state quantity is used as the input data of the neural network, and the control quantity is used as the output data of the neural network. During the model training process, the Adam optimizer is used, with the initial learning rate set to 0.001, the number of iterations set to 300, the sample size in each training batch set to 256, and the sigmoid activation function is used for training. This training strategy ensures that the model fully learns the data features while ensuring the prediction accuracy of the trajectory.

[0032] Step 5: Input the initial value of the state quantity of a trajectory, use the trained and saved network model to predict the control quantity at the current moment, substitute the control quantity and state quantity at the current moment into the motion equation for integral inverse solution, obtain the state quantity at the next moment, iterate the cycle in sequence, and finally predict the entire trajectory to achieve online calculation of the trajectory. Figure 3 It is a specific process design diagram of trajectory planning, which details the entire process of online trajectory planning calculation.

[0033] The processor used in the simulation experiment is Core(TM) i7-12700CPU@2.3GHz. The simulation results of the present invention show that the variable-sweep aircraft is used for trajectory optimization calculation. Under the same constraints, the gliding time is shortened by 18.873 seconds to reach the specified position, which greatly improves the flight performance of the aircraft. In addition, the pseudo-spectral method is used to solve the trajectory optimization problem of the variant aircraft to avoid the no-fly zone in 68 seconds, while the deep neural network model is used for trajectory optimization in only 1 second, which greatly improves the efficiency of trajectory planning.

Claims

1. An online calculation method based on the gliding trajectory of a morphing aircraft around a no-fly zone, characterized in that: The steps include: Step 1: Determine the trajectory calculation problem model for the variant aircraft when gliding around the no-fly zone. The variant aircraft must reach the determined terminal position while circling the no-fly zone, so the shortest flight time min t f As an optimization goal; Step 2: Calculate the values ​​of lift l and drag D that change constantly during flight, and realize real-time calculation of aerodynamic data through interpolation method; Step 3: Establish a sample database of trajectory optimization of variant aircraft around the no-fly zone, use the problem model established in step 1, and use the Radau pseudo-spectral method to solve and generate different optimized trajectories with r0, φ0, θ0; Step 4: training and testing the online computing network model for variant aircraft trajectories. The training set and test set constructed in step 3 are used to train and test the online computing network model for variant aircraft trajectories, respectively, to obtain the final network model. Step 5: Use the state data of the variant aircraft at the initial moment as the input of the trajectory online network model, and combine it with the Euler integration method to realize the online calculation of the trajectory of the variant aircraft around the no-fly zone.

2. The online calculation method of the glide trajectory of a morphing aircraft around a no-fly zone as claimed in claim 1, characterized in that: The step 1 is to establish a trajectory calculation problem model for the variant aircraft in the gliding phase around the no-fly zone, specifically including: Determine the optimization goal as the shortest flight time min t f ; Determine the deformation amount of the morphing aircraft, usually the sweep angle Λ or the span χ; A no-fly zone is introduced as a constraint into the problem.

3. The online calculation method of the glide trajectory of a morphing aircraft around a no-fly zone as claimed in claim 1, characterized in that: In the step 2, the lift L and drag D of the variant aircraft are related to the Mach number Ma, the angle of attack α, and the deformation of the aircraft. The existing aerodynamic data will be interpolated to obtain the values ​​of the lift L and drag D at each moment.

4. The online calculation method based on the gliding trajectory of a morphing aircraft around a no-fly zone as claimed in claim 1, characterized in that: The step 3 of establishing a trajectory sample database of variant aircraft around the no-fly zone specifically includes: The Radau pseudospectral method is used to solve the trajectory of the variant aircraft around the no-fly zone; Adding rotation to the initial position of the variant vehicle to generate a large number of different trajectories, the rotation range is as follows: In the formula, r0 represents the initial altitude of the variant aircraft, Δr represents the disturbance value added to the altitude; φ0 represents the longitude of the variant aircraft at the initial moment, Δφ represents the disturbance value added to the longitude; θ0 represents the longitude of the variant aircraft at the initial moment, Δθ represents the disturbance value added to the longitude.

5. The online calculation method based on the gliding trajectory of a morphing aircraft around a no-fly zone as claimed in claim 1, characterized in that: In step 4, a network model for online calculation of the trajectory of a variant aircraft is constructed, and the model is verified in step 5, which specifically includes: The six state variables are used as the network input, and the three control variables are used as the network output. A deep neural network with six hidden layers is constructed, and the number of neurons in the hidden layers is [64, 64, 64, 32, 32, 32] in sequence. The network model adopts As an activation function, train and save the network model; The state of the variant aircraft at the initial moment is used as the input of the trajectory online network model, and combined with the Euler integration method, the online calculation of the trajectory of the variant aircraft around the no-fly zone is realized.

Images