Reentry vehicle prediction correction guidance method based on neural network feedback correction

Abstract

The invention discloses a reentry vehicle prediction correction guidance method based on neural network feedback correction, and belongs to the technical field of guidance correction. The problem that in the prior art, a traditional reentry vehicle prediction correction guidance method is poor in actual effect and difficult to achieve expectation is solved. The method comprises the following steps: generating an aircraft reentry trajectory under aerodynamic disturbance and an aircraft prediction trajectory without aerodynamic disturbance in a Monte Carlo targeting mode, and calculating to obtain an aerodynamic disturbance identification network data set; constructing an aerodynamic disturbance identification neural network structure, and inputting the aerodynamic disturbance identification network data set into the aerodynamic disturbance identification neural network structure for training to obtain a trained neural network structure; and the trained neural network structure is adopted to correct the aerodynamic model, that is, aerodynamic model correction conditions are set, correction judgment is performed, and aircraft prediction correction guidance is realized. The reentry vehicle prediction correction guidance precision is improved, the calculation efficiency loss is avoided, and the method can be applied to aerodynamic model correction.

Description

Technical Field

[0001] The present invention relates to a reentry vehicle prediction correction guidance method, in particular to a reentry vehicle prediction correction guidance method based on neural network feedback correction, belonging to the technical field of guidance correction. Background Art

[0002] The principle of the reentry vehicle predictive correction guidance algorithm is that the vehicle predicts the terminal state according to the current flight state with a certain guidance instruction, compares it with the expected terminal state bound by the vehicle to obtain the terminal state error, and corrects the guidance instruction through the terminal state error to generate the guidance instruction at each moment to guide the vehicle to the expected terminal state; the predictive correction guidance algorithm does not rely on the nominal trajectory, but only needs to generate the corresponding guidance instruction based on the terminal state error and various path constraints and control quantity constraints. At the same time, it does not have the assumption of small deviation theory, and can adjust the trajectory within a larger range. The predictive correction guidance algorithm has strong adaptability and robustness.

[0003] However, the essence of prediction and correction is to correct the guidance instructions through the error between the predicted terminal state and the expected terminal state. Therefore, the accuracy of the predicted state is directly related to the final guidance accuracy. The flight airspace span of the reentry vehicle is large, and the flight speed range varies widely. Its aerodynamic model inevitably has large deviations. The deviation of the aerodynamic model will cause a large difference between the predicted state and the actual flight state during the guidance process, especially in the dense atmosphere area with lower airspace. Therefore, how to design and improve the prediction and correction algorithm to improve the robustness of the prediction and correction guidance algorithm to aerodynamic model disturbances is a key technology that needs to be solved urgently.

[0004] In the prior art, the control system uses the idea of ​​eliminating deviations based on error feedback. The deviation between the predicted state and the actual flight state of the aircraft is used to identify the aerodynamic disturbance of the aircraft, thereby correcting the model used for aircraft prediction. This can alleviate the problem of poor guidance accuracy caused by inconsistent models to a certain extent, but the actual effect is poor and difficult to achieve expectations.

[0005] In summary, there is a need for a reentry vehicle predictive correction guidance method based on aerodynamic disturbance identification and correction with better effect and improved robustness of the predictive correction guidance algorithm. Summary of the invention

[0006] A brief overview of the present invention is provided below in order to provide a basic understanding of certain aspects of the present invention. It should be understood that this overview is not an exhaustive overview of the present invention. It is not intended to identify key or important parts of the present invention, nor is it intended to limit the scope of the present invention. Its purpose is merely to present certain concepts in a simplified form as a prelude to a more detailed description discussed later.

[0007] In view of this, in order to solve the problem that the traditional re-entry vehicle prediction correction guidance method in the prior art has poor actual effect and is difficult to achieve the expected result, the present invention provides a re-entry vehicle prediction correction guidance method based on neural network feedback correction.

[0008] The technical solution is as follows: A reentry vehicle prediction correction guidance method based on neural network feedback correction specifically includes the following steps:

[0009] S1. Generate the reentry trajectory of the aircraft under aerodynamic disturbance and the predicted trajectory of the aircraft without aerodynamic disturbance by Monte Carlo shooting method, and calculate the aerodynamic disturbance identification network data set;

[0010] S2. constructing an aerodynamic disturbance identification neural network structure, inputting an aerodynamic disturbance identification network data set into the aerodynamic disturbance identification neural network structure for training, and obtaining a trained neural network structure;

[0011] S3. Use the trained neural network structure to correct the aerodynamic model, that is, set the aerodynamic model correction conditions, make correction judgments on the aerodynamic model in each correction command guidance cycle, and realize the aircraft prediction correction guidance.

[0012] Further, in said S1, the aerodynamic disturbance identification network data set includes aerodynamic disturbance coefficient deviation and state quantity deviation;

[0013] The aircraft height h, aircraft speed v, flight path angle γ and aircraft mechanical energy e are selected as the predicted aircraft state. The actual flight state is (h1, v1, γ1, e1), where h1 is the actual aircraft height, v1 is the actual aircraft speed, γ1 is the actual flight path angle, and e1 is the actual aircraft mechanical energy. The state deviation (δh, δv, δγ, δe) between the predicted aircraft state and the actual flight state is calculated;

[0014] The state quantity deviation (δh, δv, δγ, δe) is expressed as:

[0015]

[0016] Among them, δh is the height state quantity deviation, δv is the speed state quantity deviation, δγ is the path angle state quantity deviation, and δe is the mechanical energy state quantity deviation;

[0017] The mechanical energy e of the aircraft is expressed as:

[0018]

[0019] Among them, r is the current distance from the center of the earth of the aircraft;

[0020] Select the aerodynamic drag coefficient disturbance deviation δC dAs the aerodynamic disturbance coefficient deviation.

[0021] Furthermore, in S2, the aerodynamic disturbance identification neural network structure includes an input layer, two hidden layers and an output layer, and the aerodynamic disturbance identification neural network is trained using L2 regularization to avoid overfitting, and the state quantity deviation is used as input, and the aerodynamic disturbance coefficient deviation is used as output;

[0022] L2 regularization L2 It is expressed as:

[0023]

[0024] L data =δC d -Φ(δh,δv,δγ,δe)

[0025]

[0026] Among them, L data is the deviation δC between the neural network prediction value Φ(δh,δv,δγ,δe) and the aerodynamic disturbance coefficient d The error is λ, λ is the regularization parameter, is the square of the L2 norm of the neural network parameter set w, w i are neural network parameters;

[0027] The input layer includes 4 nodes, each hidden layer includes 10 nodes, and both hidden layers use Relu activation function;

[0028] The Relu activation function f(x) is expressed as:

[0029] f(x)=max(0,x)

[0030] Among them, max is a limiting function, and x represents the output data of the input layer or the first hidden layer.

[0031] Further, in S3, the correction judgment process is represented as follows: if the aerodynamic model correction condition is met, the aerodynamic disturbance evaluation result is obtained through the neural network structure according to the current state deviation to correct the aerodynamic model, and the corrected model is used to perform the prediction correction guidance algorithm; if the aerodynamic model correction condition is not met, the current aerodynamic model is continued to be used to perform the prediction correction guidance algorithm;

[0032] The prediction correction guidance algorithm is a guidance mode with separated longitudinal and transverse profiles. The longitudinal profile guidance adopts the dual correction logic of angle of attack and roll angle. The longitudinal profile adopts the roll angle correction strategy based on range and the angle of attack profile correction strategy based on altitude. The lateral profile adopts the roll angle corridor flipping strategy.

[0033] The prediction-correction guidance algorithm is expressed as:

[0034] minJ(Δs togo )→σ cmd

[0035] minJ(h)→(α max ,α min )

[0036]

[0037] Among them, J(·) is the objective function, Δs togo is the flight distance to be flown, σ cmd is the roll angle command, Δψ is the heading angle deviation, Δψ threshold is the heading angle deviation threshold, sign(·) is the sign function used to take the positive or negative sign of the variable, is the roll angle command of the kth guidance cycle, is the roll angle command of the k-1th guidance cycle, α max is the maximum angle of attack, α min is the minimum angle of attack.

[0038] The beneficial effects of the present invention are as follows: the present invention constructs a set of neural networks, whose input and output layers can meet the dimensional requirements of aircraft state deviation and aerodynamic deviation, generates the reentry trajectory of the aircraft under aerodynamic deviation in a Monte Carlo shooting manner, and records the predicted trajectory of the corresponding aircraft under the prediction and correction guidance instructions, obtains the aircraft state deviation by disturbing the reentry trajectory and the predicted trajectory, thereby generating a data set for training the neural network to train and verify the network, and realizes aerodynamic identification with a small amount of online calculation through offline neural network training; the present invention uses the trained neural network for the prediction and correction algorithm, sets the aerodynamic model correction conditions, and realizes the aerodynamic model correction only under the condition of severe aerodynamic disturbance through the correction conditions, and when the uncorrected prediction and correction guidance algorithm can meet the guidance accuracy, no additional aerodynamic model correction is performed, thereby avoiding the loss of calculation efficiency and further improving the correction accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0040] Figure 1 The figure is a flow chart of the predictive correction guidance method for the reentry vehicle based on the neural network aerodynamic evaluation;

[0041] Figure 2 This is a schematic diagram of the neural network structure for aerodynamic disturbance identification;

[0042] Figure 3A schematic diagram of an embodiment of a prediction correction guidance method;

[0043] Figure 4 A schematic diagram of an embodiment of a flow chart of a longitudinal profile roll angle correction;

[0044] Figure 5 Schematic diagram of the effect of different aerodynamic disturbances on the altitude and velocity profile of the aircraft;

[0045] Figure 6 A schematic flow chart of an embodiment of a reentry vehicle prediction and correction guidance method based on neural network aerodynamic evaluation;

[0046] Figure 7 The change of loss function in aerodynamic model training is shown as a schematic diagram;

[0047] Figure 8 This is a schematic diagram of the aircraft guidance altitude velocity profile result;

[0048] Fig. 9 This is a schematic diagram of the aircraft guidance altitude results;

[0049] Fig.10 This is a schematic diagram of the aircraft guidance speed results;

[0050] Fig.11 This is a schematic diagram of the aircraft guidance altitude range results;

[0051] Fig.12 This is a schematic diagram of the flipping result of the navigation angle corridor of the aircraft;

[0052] Fig.13 This is a schematic diagram of the correction results of the aircraft guidance angle of attack;

[0053] Fig.14 This is a schematic diagram of the aircraft guidance roll angle correction results. DETAILED DESCRIPTION

[0054] In order to make the technical solutions and advantages of the embodiments of the present invention more clearly understood, the exemplary embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than an exhaustive list of all the embodiments. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict.

[0055] refer to Figure 1-Figure 14 The present embodiment is described in detail, a reentry vehicle prediction correction guidance method based on neural network feedback correction, specifically comprising the following steps:

[0056] S1. Generate the reentry trajectory of the aircraft under aerodynamic disturbance and the predicted trajectory of the aircraft without aerodynamic disturbance by Monte Carlo shooting method, and calculate the aerodynamic disturbance identification network data set;

[0057] S2. constructing an aerodynamic disturbance identification neural network structure, inputting an aerodynamic disturbance identification network data set into the aerodynamic disturbance identification neural network structure for training, and obtaining a trained neural network structure;

[0058] S3. Use the trained neural network structure to correct the aerodynamic model, that is, set the aerodynamic model correction conditions, make correction judgments on the aerodynamic model in each correction command guidance cycle, and realize the aircraft prediction correction guidance;

[0059] Further, in said S1, the aerodynamic disturbance identification network data set includes aerodynamic disturbance coefficient deviation and state quantity deviation;

[0060] The aircraft height h, aircraft speed v, flight path angle γ and aircraft mechanical energy e are selected as the predicted aircraft state. The actual flight state is (h1, v1, γ1, e1), where h1 is the actual aircraft height, v1 is the actual aircraft speed, γ1 is the actual flight path angle, and e1 is the actual aircraft mechanical energy. The state deviation (δh, δv, δγ, δe) between the predicted aircraft state and the actual flight state is calculated;

[0061] The state quantity deviation (δh, δv, δγ, δe) is expressed as:

[0062]

[0063] Among them, δh is the height state quantity deviation, δv is the speed state quantity deviation, δγ is the path angle state quantity deviation, and δe is the mechanical energy state quantity deviation;

[0064] The mechanical energy e of the aircraft is expressed as:

[0065]

[0066] Among them, r is the current distance from the center of the earth of the aircraft;

[0067] Select the aerodynamic drag coefficient disturbance deviation δC d As the aerodynamic disturbance coefficient deviation;

[0068] Specifically, in this embodiment, the pseudo code of step S1 is described as follows:

[0069] Begin Monte Carlo Shooting

[0070] Define the number of target shots num;

[0071] Select the initial value of the aircraft state and the typical control profile;

[0072] for i in num:

[0073] Integrate and generate flight trajectories under aerodynamic disturbances;

[0074] Integrate and generate the flight trajectory without aerodynamic disturbances;

[0075] Calculate the terminal state deviation (δh, δv, δγ, δe) and the aerodynamic pull coefficient δC d

[0076] end

[0077] The terminal state deviation (δh, δv, δγ, δe) is the state quantity deviation, and the aerodynamic pull coefficient δC d That is, the disturbance deviation of the aerodynamic drag coefficient.

[0078] Furthermore, in S2, the aerodynamic disturbance identification neural network structure includes an input layer, two hidden layers and an output layer, and the aerodynamic disturbance identification neural network is trained using L2 regularization to avoid overfitting, and the state quantity deviation is used as input, and the aerodynamic disturbance coefficient deviation is used as output;

[0079] L2 regularization L2 It is expressed as:

[0080]

[0081] L data =δC d -Φ(δh,δv,δγ,δe)

[0082]

[0083] Among them, L data is the deviation δC between the neural network prediction value Φ(δh,δv,δγ,δe) and the aerodynamic disturbance coefficient d The error is λ, λ is the regularization parameter, is the square of the L2 norm of the neural network parameter set w, w i are neural network parameters;

[0084] The input layer includes 4 nodes, each hidden layer includes 10 nodes, both hidden layers use Relu activation function, and the output layer includes 1 node;

[0085] The Relu activation function f(x) is expressed as:

[0086] f(x)=max(0,x)

[0087] Among them, max is a limiting function, and x represents the output data of the input layer or the first hidden layer;

[0088] refer to Figure 2 , Input Layer is the input layer, Represents 4 nodes, Hidden Layer is the hidden layer, Represents 10 nodes, Output Layer is the output layer, Represents 1 node;

[0089] Further, in S3, the correction judgment process is represented as follows: if the aerodynamic model correction condition is met, the aerodynamic disturbance evaluation result is obtained through the neural network structure according to the current state deviation to correct the aerodynamic model, and the corrected model is used to perform the prediction correction guidance algorithm; if the aerodynamic model correction condition is not met, the current aerodynamic model is continued to be used to perform the prediction correction guidance algorithm;

[0090] The prediction correction guidance algorithm is a guidance mode with separated longitudinal and transverse profiles. The longitudinal profile guidance adopts the dual correction logic of angle of attack and roll angle. The longitudinal profile adopts the roll angle correction strategy based on range and the angle of attack profile correction strategy based on altitude. The lateral profile adopts the roll angle corridor flipping strategy.

[0091] The prediction-correction guidance algorithm is expressed as:

[0092] minJ(Δs togo )→σ cmd

[0093] minJ(h)→(α max ,α min )

[0094]

[0095] Among them, J(·) is the objective function, Δs togo is the flight distance to be flown, σ cmd is the roll angle command, Δψ is the heading angle deviation, Δψ threshold is the heading angle deviation threshold, sign(·) is the sign function used to take the positive or negative sign of the variable, is the roll angle command of the kth guidance cycle, is the roll angle command of the k-1th guidance cycle, α max is the maximum angle of attack, α min is the minimum angle of attack;

[0096] Specifically, the aerodynamic model correction condition may be an altitude condition, a speed condition or a range condition, the purpose of which is to ensure the stability of the aircraft guidance method by making the aerodynamic correction only occur in certain flight sections, while reducing the amount of calculation;

[0097] refer to Figure 4 and Figure 6 The present invention can effectively identify the aerodynamic disturbance of the aircraft in a complex disturbance environment, so that the aircraft can correct the aerodynamic model error of the aircraft in real time during the prediction and correction guidance process, thereby ensuring the guidance accuracy. The current state of the aircraft (h, θ, φ, v, γ, ψ), θ is the current geocentric longitude of the aircraft, φ is the current geocentric latitude of the aircraft, ψ is the flight heading angle, and the predicted terminal state (h fe ,θ fe ,φ fe ,v fe ,γ fe ,ψ fe ), where h fe is the predicted height, θ fe To predict the geocentric longitude, φ fe To predict the geocentric latitude, v fe is the predicted speed, γ fe is the predicted path angle, ψ fe To predict the heading angle, the desired terminal state (h c ,θ c ,φ c ,v c ,γ c ,ψ c ), where h c is the desired height, θ c is the desired geocentric longitude, φ c is the desired geocentric latitude, v c is the expected speed, γ c is the desired path angle, ψ fe is the desired heading angle;

[0098] refer to Figure 8-Figure 14 In the initial state, the initial height H0 = 50, in km, the initial speed V0 = 2500, in m / s, the initial path angle γ0 = -1°, in the terminal state, the output height H f =20, unit is km, output speed V f =1500, in m / s, aircraft range Range = 680, in km;

[0099] Among them, Velocity is the speed in m / s, Height is the altitude in km, Time is the time in s, Range to Go is the range to be flown, Aziumuth Deviation is the heading angle deviation, Alpha Angle is the angle of attack, and Correct Bank Angle is the corrected bank angle.

[0100] Although the present invention has been described according to a limited number of embodiments, it will be apparent to those skilled in the art, with the benefit of the above description, that other embodiments may be envisioned within the scope of the invention thus described. In addition, it should be noted that the language used in this specification is selected primarily for readability and teaching purposes, rather than for explaining or defining the subject matter of the present invention. Therefore, many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the appended claims. The disclosure of the present invention is illustrative, not restrictive, with respect to the scope of the present invention, which is defined by the appended claims.

Claims

1. A reentry vehicle prediction correction guidance method based on neural network feedback correction, characterized in that: The following steps are involved: S1. Generate the reentry trajectory of the aircraft under aerodynamic disturbance and the predicted trajectory of the aircraft without aerodynamic disturbance by Monte Carlo shooting method, and calculate the aerodynamic disturbance identification network data set; S2. constructing an aerodynamic disturbance identification neural network structure, inputting an aerodynamic disturbance identification network data set into the aerodynamic disturbance identification neural network structure for training, and obtaining a trained neural network structure; S3. Use the trained neural network structure to correct the aerodynamic model, that is, set the aerodynamic model correction conditions, make correction judgments on the aerodynamic model in each correction command guidance cycle, and realize the aircraft prediction correction guidance.

2. A reentry vehicle prediction correction guidance method based on neural network feedback correction according to claim 1, characterized in that: In said S1, the aerodynamic disturbance identification network data set includes aerodynamic disturbance coefficient deviation and state quantity deviation; The aircraft height h, aircraft speed v, flight path angle γ and aircraft mechanical energy e are selected as the predicted aircraft state. The actual flight state is (h1, v1, γ1, e1), where h1 is the actual aircraft height, v1 is the actual aircraft speed, γ1 is the actual flight path angle, and e1 is the actual aircraft mechanical energy. The state deviation (δh, δv, δγ, δe) between the predicted aircraft state and the actual flight state is calculated; The state quantity deviation (δh, δv, δγ, δe) is expressed as: Among them, δh is the height state quantity deviation, δv is the speed state quantity deviation, δγ is the path angle state quantity deviation, and δe is the mechanical energy state quantity deviation; The mechanical energy e of the aircraft is expressed as: Among them, r is the current distance from the center of the earth of the aircraft; Select the aerodynamic drag coefficient disturbance deviation δC d As the aerodynamic disturbance coefficient deviation.

3. A reentry vehicle prediction correction guidance method based on neural network feedback correction according to claim 2, characterized in that: In S2, the aerodynamic disturbance identification neural network structure includes an input layer, two hidden layers and an output layer, and the aerodynamic disturbance identification neural network is trained using L2 regularization to avoid overfitting, and the state quantity deviation is used as input, and the aerodynamic disturbance coefficient deviation is used as output; L2 regularization L2 It is expressed as: L data =δC d -Φ(δh,δv,δγ,δe) Among them, L data is the deviation δC between the neural network prediction value Φ(δh,δv,δγ,δe) and the aerodynamic disturbance coefficient d The error is λ, λ is the regularization parameter, is the square of the L2 norm of the neural network parameter set w, w i are neural network parameters; The input layer includes 4 nodes, each hidden layer includes 10 nodes, both hidden layers use Relu activation function, and the output layer includes 1 node; The Relu activation function f(x) is expressed as: f(x)=max(0,x) Among them, max is a limiting function, and x represents the output data of the input layer or the first hidden layer.

4. A reentry vehicle prediction correction guidance method based on neural network feedback correction according to claim 3, characterized in that: In S3, the correction judgment process is represented as follows: if the aerodynamic model correction condition is met, the aerodynamic disturbance evaluation result is obtained through the neural network structure according to the current state deviation to correct the aerodynamic model, and the corrected model is used to perform the prediction correction guidance algorithm; if the aerodynamic model correction condition is not met, the current aerodynamic model is continued to be used to perform the prediction correction guidance algorithm; The prediction correction guidance algorithm is a guidance mode with separated longitudinal and transverse profiles. The longitudinal profile guidance adopts the dual correction logic of angle of attack and roll angle. The longitudinal profile adopts the roll angle correction strategy based on range and the angle of attack profile correction strategy based on altitude. The lateral profile adopts the roll angle corridor flipping strategy. The prediction-correction guidance algorithm is expressed as: minJ(Δs togo )→σ cmd minJ(h)→(α max ,a min ) Among them, J(·) is the objective function, Δs togo is the flight distance to be flown, σ cmd is the roll angle command, Δψ is the heading angle deviation, Δψ threshold is the heading angle deviation threshold, sign(·) is the sign function used to take the positive or negative sign of the variable, is the roll angle command of the kth guidance cycle, is the roll angle command of the k-1th guidance cycle, α max is the maximum angle of attack, α min is the minimum angle of attack.

Images