A trajectory replanning method for elliptical rescue orbits under rocket thrust descent failure
By using the radial-based neural network to establish a nonlinear mapping relationship between the fault state and the orbit parameter in the case of rocket thrust drop failure, the problem of adaptive pseudo-spectral method taking time or not converging in online trajectory optimization is solved, and the problem of adaptive pseudo-spectral method is realized Fast and effective trajectory planning and improvement of computing efficiency.
Patent Information
- Application Number
- CN202110388335.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-04-12
- Publication Date
- 2024-11-05
- Estimated Expiration
- 2041-04-12
AI Technical Summary
In the case of rocket thrust drop failure, traditional adaptive pseudo-spectrum method is used for online trajectory optimization, which causes time to take or no convergence, and the contradiction between orbital elements in the objective function causes divergence in search direction, which reduces the Computational efficiency.
Adaptive pseudo-spectral method is used to solve the trajectory optimization problem of elliptical rescue tracks offline, establish a nonlinear mapping relationship between fault state and incoming parameters, and perform data normalization and training through radial-based neural network to provide reasonable initial values, and Migrate it to the actual rocket flight to make decisions online and ultimately transform it into the optimal propellant optimization problem for solving.
It realizes rapid online decision-making of rescue track roots and terminal control variables, provides reasonable initial values, avoids the reduction of calculation efficiency, and improves the efficiency and accuracy of trajectory planning.
Smart Images

Figure CN113189870B_ABST
Abstract
Description
Technical Field
[0001] The invention relates to the technical field of rescue of launch vehicles under failure, and in particular to a trajectory replanning method for an elliptical rescue orbit under a rocket thrust drop failure. Background Art
[0002] As a rocket power unit, the engine is the determining factor for the reliability and safety of the entire rocket flight. In actual flight missions, the failure of the launch vehicle engine and other reasons can easily cause thrust reduction. If the guidance and control scheme under nominal ballistic conditions is continued, it will be difficult to complete the mission, which will lead to the failure of the flight mission.
[0003] In general, trajectory optimization problems are described as optimal control problems, and the adaptive pseudo-spectral method is widely used in traditional solution methods. However, there are some shortcomings in using only the adaptive pseudo-spectral method to solve the online trajectory optimization problem of thrust drop failure. First, improper initial value setting for the online trajectory optimization problem may cause the adaptive pseudo-spectral method to take a long time to solve or not converge. Second, the contradiction between the orbital elements in the objective function will cause the search direction to diverge, or the search step size to decrease, thereby reducing the feasibility of online application. Summary of the invention
[0004] In view of the shortcomings of the adaptive pseudo-spectral method in online trajectory replanning of thrust drop failure, such as long solution time or non-convergence, the present invention provides a trajectory replanning method, which provides reasonable initial values for trajectory planning of online elliptical rescue orbits and avoids the reduction of computational efficiency due to conflicts between variables in the objective function.
[0005] To achieve the above purpose, the technical solution of the present application is: a trajectory replanning method for an elliptical rescue orbit under a rocket thrust drop failure, comprising:
[0006] The secondary flight dynamics equation of the rocket's ascent phase is established in the geocentric inertial coordinate system. The boundary conditions and constraints are set at different fault moments and thrust drop percentages to construct the trajectory optimization problem of the elliptical rescue orbit under various thrust drop faults.
[0007] Adopting an adaptive pseudo-spectral method to solve the trajectory optimization problem of the elliptical rescue orbit offline, a sample set of fault state-orbital parameters is obtained, wherein the input feature of the sample set is the fault state, which includes the moment of thrust failure, the magnitude, position, speed and mass of thrust drop, and the output feature of the sample set is the orbital parameters, which include the orbital semi-major axis, eccentricity, orbital inclination, right ascension of ascending node, and control variables of the terminal;
[0008] The maximum-minimum method is used to normalize the sample data, and all data are normalized to [-1, 1]. The radial basis neural network data center is selected by the orthogonal least squares method, where the radial basis function is the Gaussian basis function. The radial basis neural network is trained offline to establish a nonlinear mapping relationship from the fault state to the orbit entry parameters.
[0009] The radial basis function neural network is transferred to the actual flight of the rocket. The fault status of the flight is used as input, and the radial basis function neural network makes online decisions on the parameters of orbit entry and exit. The rescue orbit elements and terminal control variables determined by the decisions are used to set reasonable initial guess values for trajectory planning. The trajectory optimization problem of the elliptical rescue orbit is converted into an optimal propellant optimization problem by using the rescue orbit elements determined by the decisions. The optimal propellant optimization problem is solved online using the adaptive pseudo-spectral method to obtain the flight trajectory.
[0010] Furthermore, when constructing the trajectory optimization problem of the elliptical rescue orbit under various thrust descent failures:
[0011] Assume that the X1 axis points to the prime meridian direction at the time of launch in the equatorial plane, the Z1 axis is perpendicular to the equatorial plane and points to the North Pole, and the Y1 axis satisfies the right-hand rule. The secondary flight dynamics equation of the rocket's ascent phase is established in the geocentric inertial coordinate system as follows:
[0012]
[0013]
[0014]
[0015] Where r and v are the position and velocity vector of the launch vehicle, respectively. is the first-order derivative of r, is the first-order derivative of v; μ = GM is the gravitational constant of the earth, G is the gravitational constant, M is the mass of the earth; m is the total mass of the rocket, is the first-order derivative of m, I sp is the rocket engine specific impulse; u=[u x ,u y ,u z ] T is the thrust unit vector component of the engine; when the engine fails, the thrust decreases by a ratio of η, and the thrust magnitude is (1-η)T nom , T nom is the nominal thrust of the engine; g0 is the acceleration due to gravity;
[0016] In the case of thrust drop failure, the engine specific impulse remains unchanged, the propellant consumption per second decreases by η, and the total flight time exceeds the nominal flight time. Assuming that the engine thrust drop failure occurs at t0, the launch vehicle needs to optimize the rescue trajectory with the state at t0 as the starting point. Therefore, the starting point equality constraint is expressed as:
[0017] x(t0)=x0 (4)
[0018] Where x0 is the state of the starting point.
[0019] Use m f represents the total mass of the launch vehicle and the payload after the remaining fuel is exhausted, the radius of the earth is R0; the minimum safe orbit altitude is defined as h safe , the terminal quality and height should meet the following requirements:
[0020] m(t f )≥m f ,h safe ≤r(t f )-R0 (5)
[0021] When a thrust drop failure occurs, the elliptical rescue orbit type is used. While adjusting the orbit plane, the orbit height is increased as much as possible. The trajectory optimization problem of the elliptical rescue orbit is described as:
[0022]
[0023] Where J is the optimization target, λ a ,λ e ,λ i ,λ Ω is the weight factor, hp f Corresponding to the perigee height of the orbit, is the perigee height of the target orbit, t f is the terminal moment, a f ,i f ,Ω f are the semi-major axis, inclination, and right ascension of the ascending node of the orbit, They are the semi-major axis, inclination, and right ascension of the ascending node of the target orbit respectively.
[0024] Furthermore, a reasonable initial guess value is set for trajectory planning, specifically: according to the fault state, the radial basis neural network is used to decide the number of terminal orbit elements for rescue and terminal control variables in, are the semi-major axis, eccentricity, inclination, and right ascension of the ascending node of the terminal orbit respectively; in the initial guess, the time vector is The state vector is The control variable vector is The estimated value of flight time is equal to the remaining propellant divided by the seconds consumed; is the thrust unit vector component when the fault occurs; The value determined by the orbital element number of the entry point The angle between the ascending node and the orbital entry point is the geocentric angle Φ k , which is used to approximate the right ascension of the ascending node.
[0025] Furthermore, when the adaptive pseudo-spectral method is used to solve the optimal propellant optimization problem online,
[0026] Since the orbital elements of the rescue orbit are known, the trajectory optimization problem of the elliptical rescue orbit is transformed into the optimal propellant optimization problem:
[0027]
[0028] Among them, δ is a term that avoids the optimal propellant optimization problem from being unsolvable due to the positive error of the semi-major axis of the orbit determined by the radial basis neural network; in order to eliminate the influence of the positive error of the semi-major axis on the online trajectory optimization, the end constraint of the semi-major axis is set to
[0029] The present invention adopts the above technical solution, and can achieve the following technical effects:
[0030] (1) The radial basis function neural network can well fit the nonlinear relationship between “fault state and orbit entry parameters” and can quickly make online decisions on the number of rescue orbit elements and terminal control variables.
[0031] (2) The determined rescue orbit roots and terminal control variables provide reasonable initial values for trajectory planning of the online elliptical rescue orbit.
[0032] (3) The number of rescue orbit elements determined by the decision transforms the trajectory optimization problem of the elliptical rescue orbit into the optimal propellant optimization problem, thus avoiding the reduction of computational efficiency due to the conflict between the variables in the objective function. BRIEF DESCRIPTION OF THE DRAWINGS
[0033] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0034] Figure 1 It is the overall strategy diagram of trajectory optimization in the present invention;
[0035] Figure 2 It is the speed height curve diagram in the present invention;
[0036] Figure 3 It is the altitude curve diagram of the present invention. DETAILED DESCRIPTION
[0037] The embodiments of the present invention are implemented on the premise of the technical solution of the present invention, and detailed implementation methods and specific operation processes are given, but the protection scope of the present invention is not limited to the following embodiments.
[0038] This embodiment provides a trajectory replanning method for an elliptical rescue orbit under a rocket thrust drop failure. Figure 1 The specific implementation steps are as follows:
[0039] S1: Assume that the X1 axis points to the prime meridian direction at the time of launch in the equatorial plane, the Z1 axis is perpendicular to the equatorial plane and points to the North Pole, and the Y1 axis satisfies the right-hand rule. The secondary flight dynamics equation of the rocket's ascent phase is established in the geocentric inertial coordinate system as follows:
[0040]
[0041]
[0042]
[0043] Where r, v are the position and velocity vectors of the launch vehicle, μ = GM is the earth's gravitational constant, m is the total mass of the rocket, and I sp is the rocket engine specific impulse; u=[u x ,u y ,u z ] T is the thrust unit vector component of the engine; when the engine fails, the thrust decreases by a ratio of η, and the thrust magnitude is (1-η)T nom , T nom is the nominal thrust of the engine.
[0044] In the case of thrust drop failure, the engine specific impulse remains unchanged, the propellant consumption per second decreases by η, and the total flight time exceeds the nominal flight time. Assuming that the engine thrust drop failure occurs at t0, the launch vehicle needs to optimize the rescue trajectory with the state at t0 as the starting point. Therefore, the starting point equality constraint is expressed as:
[0045] x(t0)=x0 (4)
[0046] Where x0 is the state of the starting point.
[0047] Use m f represents the total mass of the launch vehicle and the payload after the remaining fuel is exhausted, the radius of the earth is R0; the minimum safe orbit altitude is defined as hsafe , the terminal quality and height should meet the following requirements:
[0048] m(t f )≥m f ,h safe ≤r(t f )-R0 (5)
[0049] When a thrust drop failure occurs, the elliptical rescue orbit type is used. While adjusting the orbit plane, the orbit height should be increased as much as possible. The trajectory optimization problem of the elliptical rescue orbit can be described as:
[0050]
[0051] Where J is the optimization target, λ a ,λ e ,λ i ,λ Ω is the weight factor, hp f Corresponding to the perigee height of the orbit, is the perigee height of the target orbit, t f is the terminal moment, a f ,i f ,Ω f are the semi-major axis, inclination, and right ascension of the ascending node of the orbit, They are the semi-major axis, inclination, and right ascension of the ascending node of the target orbit respectively.
[0052] S2: Adopting the adaptive pseudo-spectral method to solve the trajectory optimization problem of the elliptical rescue orbit offline, a sample set of orbital parameters of the fault state is obtained, wherein the input feature of the sample set is the fault state, and the fault state includes the moment of the thrust failure, the magnitude of the thrust drop, the position, the speed and the mass, and the output feature of the sample set is the orbital parameters, and the orbital parameters include the semi-major axis of the orbit, the eccentricity, the orbit inclination, the right ascension of the ascending node, and the control variables of the terminal.
[0053] S3: In order to eliminate the order of magnitude differences between data in each dimension and avoid large prediction errors caused by large order of magnitude differences between input and output data, the data needs to be normalized. The maximum and minimum method is used to normalize the data, and all data are normalized to between [-1,1]. Considering the numerous computing tasks of the onboard computer, the fewer computing resources occupied by the rescue orbit decision, the better. The orthogonal least squares method is used to select the RBF network data center and establish a nonlinear mapping relationship between "fault state-orbit entry parameters".
[0054] The hidden unit is activated by the basis function. The present invention adopts the Gaussian basis function. The output of the jth hidden layer is:
[0055]
[0056] is the i-th input data, μ j is the basis function center of the hidden node, σ j is the radial basis function expansion speed. The basis function center and expansion speed are determined by the network training. The output layer linearly combines the output of the radial basis function hidden layer to generate the expected output. The output of the kth node of the output layer is:
[0057]
[0058] In the formula, w jk is the weight from the jth hidden layer neuron to the kth output neuron. In order to achieve appropriate approximation accuracy, the following parameters are determined through training: the number of hidden layer neurons, the basis function center of each hidden layer neuron, and the weight of the radial basis function output transferred to the summation layer.
[0059] S4: The rescue trajectory decision model is migrated and applied online, with the actual flight fault status as input, and the radial basis neural network is used to decide the terminal trajectory elements of the rescue. and terminal control variables in, are the semi-major axis, eccentricity, inclination, and right ascension of the ascending node of the terminal orbit respectively; in the initial guess, the time vector is The state vector is The control variable vector is The estimated value of flight time is equal to the remaining propellant divided by the seconds consumed; is the thrust unit vector component when the fault occurs; The value determined by the orbital element number of the entry point The angle between the ascending node and the orbital entry point is the geocentric angle Φ k , which is used to approximate the right ascension of the ascending node.
[0060] Since the orbital elements of the rescue orbit are known, the trajectory optimization problem of the elliptical rescue orbit can be transformed into the optimal propellant optimization problem
[0061]
[0062] Among them, δ is a term to avoid the optimal propellant optimization problem from being unsolvable due to the positive error of the semi-major axis of the orbit determined by the neural network. In order to eliminate the influence of the positive error of the semi-major axis on the online trajectory optimization, the end constraint of the semi-major axis is set to
[0063] Example 1
[0064] The entire second-stage flight phase of the launch vehicle is taken as the research object, and the parameters are from the literature Z. Song, C. Wang, Q. Gong, Joint dynamic optimization of the target orbit and flight trajectory of a launch vehicle based on state-triggered indices, Acta Astronaut., 174 (2020) 82-93. The matching points used by the adaptive pseudo-spectral method are the zero points of the orthogonal Legendre polynomial. The state distribution of the fault established in the sample set is based on the occurrence time of 0 to 378s, with a step size of 0.5s; the thrust drop size is 13% to 50%, with a step size of 0.25%, and 5330 elliptical rescue orbit type samples. In the sample set, 90% of the data are randomly selected as the training set, and the remaining 10% of the data are used as the test set. The diffusion factor of the radial basis neural network training is 0.5, and the number of hidden layer neurons trained in the end is 300. The absolute error of the semi-major axis decision is in the range of [-1.5,1.5] km, and the absolute error of the eccentricity is in the range of [-2×10 -4 ,2×10 -4 ] interval, the absolute error of the orbital inclination is within [-2×10 -3 ,6×10 -3 ]°, the absolute error of the right ascension of the ascending node is [-5×10 -3 ,1×10 -2 Compared with the circular orbit rescue type, the semi-major axis decision error of the elliptical orbit rescue type is larger, because the nonlinear relationship between the fault state and the semi-major axis in the elliptical orbit rescue type is more complicated.
[0065] The trajectory based on the orbital parameter decision method of the present invention is consistent with the trajectory obtained by using only the adaptive pseudo-spectral method, such as Figure 2 As shown in Figure 2, the velocity curve obtained by the proposed method coincides with that obtained by the adaptive pseudo-spectral method. The flight trajectory is adjusted after 147.5s to extend the flight time and ensure that the payload enters the elliptical rescue orbit. Figure 3 It can be seen that the orbital insertion height of the method of the present invention is slightly lower than the orbital insertion height of the trajectory obtained by only using the adaptive pseudo-spectral method. This is due to the introduction of the δ term. The calculation time of the online trajectory replanning method proposed in the present invention is 98.8% less than that of only using the adaptive pseudo-spectral method. The improvement in calculation efficiency is mainly due to the decision of the orbital insertion parameters. On the one hand, the determined orbital insertion parameters provide a suitable reference value for the setting of the initial value; on the other hand, the objective function is simplified, which improves the search efficiency of the optimal solution.
[0066] The embodiments of the present invention have better practicability and are not intended to limit the present invention in any form. The technical features or combinations of technical features described in the embodiments of the present invention should not be considered isolated, and they can be combined with each other to achieve better technical effects. The scope of the preferred embodiments of the present invention may also include other implementations, and this should be understood by those skilled in the art of the invention embodiments.
Claims
1. A trajectory replanning method for an elliptical rescue orbit under a rocket thrust drop failure, characterized in that: include: The secondary flight dynamics equation of the rocket's ascent phase is established in the geocentric inertial coordinate system. The boundary conditions and constraints are set at different fault moments and thrust drop percentages to construct the trajectory optimization problem of the elliptical rescue orbit under various thrust drop faults: Assume that the X1 axis points to the prime meridian direction at the time of launch in the equatorial plane, the Z1 axis is perpendicular to the equatorial plane and points to the North Pole, and the Y1 axis satisfies the right-hand rule. The secondary flight dynamics equation of the rocket's ascent phase is established in the geocentric inertial coordinate system as follows: Where r and v are the position and velocity vector of the launch vehicle, respectively. is the first-order derivative of r, is the first-order derivative of v; μ = GM is the gravitational constant of the earth, G is the gravitational constant, M is the mass of the earth; m is the total mass of the rocket, is the first-order derivative of m, I sp is the rocket engine specific impulse; u=[u x ,u y ,u z ] T is the thrust unit vector component of the engine; when the engine fails, the thrust decreases by a ratio of η, and the thrust magnitude is (1-η)T nom , T nom is the nominal thrust of the engine; g0 is the acceleration due to gravity; In the case of thrust drop failure, the engine specific impulse remains unchanged, the propellant consumption per second decreases by h, and the total flight time exceeds the nominal flight time. Assuming that the engine thrust drop failure occurs at t0, the launch vehicle needs to optimize the rescue trajectory with the state at t0 as the starting point. Therefore, the starting point equality constraint is expressed as: x(t0)=x0 (4) In the formula, x0 is the state of the starting point; Use m f represents the total mass of the launch vehicle and the payload after the remaining fuel is exhausted, the radius of the earth is R0; the minimum safe orbit altitude is defined as h safe , the terminal quality and height should meet the following requirements: m(t f )≥m f , h safe ≤ r (t f )-R0 (5) When a thrust drop failure occurs, the elliptical rescue orbit type is used. While adjusting the orbit plane, the orbit height is increased as much as possible. The trajectory optimization problem of the elliptical rescue orbit is described as: Where J is the optimization target, λ hp ,λ a ,λ i ,λ Ω is the weight factor, hp f Corresponding to the perigee height of the orbit, is the perigee height of the target orbit, t f is the terminal moment, a f ,i f ,Ω f are the semi-major axis, inclination, and right ascension of the ascending node of the orbit, are the semi-major axis, inclination, and right ascension of the ascending node of the target orbit respectively; Adopting an adaptive pseudo-spectral method to solve the trajectory optimization problem of the elliptical rescue orbit offline, a sample set of fault state-orbital parameters is obtained, wherein the input feature of the sample set is the fault state, which includes the moment of thrust failure, the magnitude, position, speed and mass of thrust drop, and the output feature of the sample set is the orbital parameters, which include the orbital semi-major axis, eccentricity, orbital inclination, right ascension of ascending node, and control variables of the terminal; The maximum-minimum method is used to normalize the sample data, and all data are normalized to [-1, 1]. The radial basis neural network data center is selected by the orthogonal least squares method, where the radial basis function is the Gaussian basis function. The radial basis neural network is trained offline to establish a nonlinear mapping relationship from the fault state to the orbit entry parameters. The radial basis function neural network is transferred to the actual flight of the rocket. The fault status of the flight is used as input, and the radial basis function neural network makes online decisions on the parameters of entry and exit of the orbit. The rescue orbit elements and terminal control variables obtained by the decision are used to set reasonable initial guess values for trajectory planning. The trajectory optimization problem of the elliptical rescue orbit is converted into the optimal propellant optimization problem by using the rescue orbit elements obtained by the decision. The optimal propellant optimization problem is solved online by using the adaptive pseudo-spectral method to obtain the flight trajectory. Set a reasonable initial guess value for trajectory planning, specifically: according to the fault state, use the radial basis neural network to decide the number of terminal orbit elements for rescue and terminal control variables in, are the semi-major axis, eccentricity, inclination, and right ascension of the ascending node of the terminal orbit respectively; in the initial guess, the time vector is The state vector is The control variable vector is t0 is the time when the engine thrust reduction failure occurs, and x0 is the state at the starting point; The estimated value of flight time is equal to the remaining propellant divided by the seconds consumed; is the thrust unit vector component when the fault occurs; The value determined by the orbital element number of the entry point The angle between the ascending node and the orbital entry point is the geocentric angle Φ k , used to approximate the right ascension of the ascending node; When the adaptive pseudo-spectral method is used to solve the optimal propellant optimization problem online, since the orbital elements of the rescue orbit are known, the trajectory optimization problem of the elliptical rescue orbit is transformed into the optimal propellant optimization problem: Among them, m f represents the total mass of the launch vehicle and the payload after the remaining fuel is exhausted, the radius of the earth is R0, and the minimum safe orbit altitude is defined as h safe ,u=[u x ,u y ,u z ] T is the thrust unit vector component of the engine; δ is a term that avoids the optimal propellant optimization problem from being unsolvable due to the positive error of the semi-major axis of the orbit determined by the radial basis neural network; in order to eliminate the influence of the positive error of the semi-major axis on the online trajectory optimization, the end constraint of the semi-major axis is set to
