Deep space exploration transfer phase starlight angular distance information optimization, autonomous navigation method and device

Abstract

The embodiment of the present application provides a kind of deep space exploration transfer section starlight angular distance information optimization, autonomous navigation method and device, by the probe with optical sensor, method includes: through optical sensor acquisition deep space image;The deep space image includes multiple stars and at least one planet;According to the deep space image and ephemeris, select the planet closest to the probe from the at least one planet as target planet;To the deep space image collected, construct optical sensor measurement coordinate system, in the optical sensor measurement coordinate system, determine the sum of stellar vectors of each star in the multiple stars and the planet vector of the target planet;According to the sum of stellar vectors of all stars, determine target star combination in all star combinations;According to the sum of stellar vectors of three stars in the target star combination and the planet vector of the target planet, determine the optimal deep space exploration transfer section starlight angular distance information.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of deep space exploration, in particular to a deep space exploration transfer phase starlight angular distance information optimization and autonomous navigation method and device. BACKGROUND

[0002] Autonomous navigation technology has been widely concerned in deep space exploration missions such as small celestial body sampling return and planet landing. These missions require the probe to determine the orbit parameters using its own sensors, thereby avoiding the time delay problem in the navigation process. At present, the probe docking and attachment process is the main application scenario of autonomous navigation. One of the characteristics of the deep space exploration transfer phase is large spatial scale and long target distance, which poses a severe challenge to the action distance of the navigation sensor. The X-pulsar signal receiver is an ideal information source for measuring long-distance target information. However, according to the latest research, the signal error of the X-pulsar will seriously reduce the navigation accuracy, and the signal receiver is too large in volume and weight to be carried by the deep space probe.

[0003] The optical autonomous navigation method has been applied in the transfer phase of several deep space exploration missions, including Deep Space 1 and M-ARGO asteroid exploration missions. In these missions, the probe achieves orbit state estimation by planning an asteroid observation sequence. However, due to the limited accuracy of the asteroid ephemeris, the final navigation accuracy is not high. In the case of good observation conditions, the autonomous navigation accuracy of Deep Space 1 and M-ARGO probes in the transfer phase is only 150 kilometers and 1000 kilometers, respectively.

[0004] In the process of implementing the present application, the applicant found that the prior art at least has the following problems:

[0005] The existing deep space exploration autonomous navigation method has the problems of low navigation accuracy and large navigation parameter calculation amount. SUMMARY

[0006] The embodiment of the present application provides a deep space exploration transfer phase starlight angular distance information optimization and autonomous navigation method and device to solve the problem of low navigation accuracy and large navigation parameter calculation amount of the existing deep space exploration autonomous navigation method.

[0007] To achieve the above purpose, on the one hand, the embodiment of the present application provides a deep space exploration transfer phase starlight angular distance information optimization method, which is adopted by a probe with an optical sensor, and the method comprises:

[0008] acquiring a deep space image through the optical sensor; the deep space image includes a plurality of stars and at least one planet; the number of the plurality of stars is greater than or equal to 3;

[0009] selecting a planet closest to the probe from the at least one planet as a target planet according to the deep space image and ephemeris;

[0010] constructing an optical sensor measurement coordinate system for the collected deep space image, and determining a star vector of each star in the plurality of stars and a planet vector of the target planet in the optical sensor measurement coordinate system; the optical sensor measurement coordinate system coincides with a body coordinate system of the probe;

[0011] determining a target star combination from all star combinations according to the star vectors of all stars; each star combination is composed of any three stars in the plurality of stars, and the sum of inner products between the star vectors of the three stars in the target star combination is the minimum among all star combinations;

[0012] determining optimal deep space probe transfer segment starlight angular distance information according to the star vectors of the three stars in the target star combination and the planet vector of the target planet.

[0013] In a second aspect, an embodiment of the present application provides an autonomous navigation method based on deep space probe transfer segment starlight angular distance information optimization, comprising:

[0014] determining the star vectors of the three stars in the target star combination, the planet vector of a target planet and the starlight angular distance information corresponding to the three stars in the target star combination by using any of the deep space probe transfer segment starlight angular distance information optimization methods described above;

[0015] obtaining solar pressure perturbation and planetary gravity perturbation data through an orbit vibration model;

[0016] inputting the starlight angular distance of each star, the solar pressure perturbation and the planetary gravity perturbation data into an unscented Kalman filter algorithm to determine the running state of the probe.

[0017] In a third aspect, an embodiment of the present application provides a deep space probe transfer segment starlight angular distance information optimization device, which is used by a probe with an optical sensor, and comprises:

[0018] an image acquisition unit configured to acquire a deep space image through the optical sensor; the deep space image comprises a plurality of stars and at least one planet; the number of the plurality of stars is greater than or equal to 3;

[0019] a target planet determination unit configured to select a planet closest to the probe from the at least one planet as a target planet according to the deep space image and ephemeris;

[0020] a vector determination unit configured to construct an optical sensor measurement coordinate system based on the collected deep space image, and determine a star vector of each of the plurality of stars and a planet vector of the target planet in the optical sensor measurement coordinate system, wherein the optical sensor measurement coordinate system coincides with a body coordinate system of the probe;

[0021] a target star combination determination unit configured to determine a target star combination from star vectors of all stars, wherein each star combination is composed of any three stars among all stars, and a sum of inner products between star vectors of three stars in the target star combination is the minimum among all star combinations;

[0022] a starlight angular distance determination unit configured to determine optimal deep space probe transfer segment starlight angular distance information based on star vectors of three stars in the target star combination and the planet vector of the target planet.

[0023] In a fourth aspect, an embodiment of the present application provides an autonomous navigation device based on deep space probe transfer segment starlight angular distance information optimization, comprising:

[0024] The deep space probe transfer segment starlight angular distance information optimization device adopts any one of the deep space probe transfer segment starlight angular distance information optimization methods as described above, so as to determine star vectors of three stars in a target star combination, a planet vector of a target planet, and starlight angular distance information corresponding to the three stars in the target star combination.

[0025] an orbit vibration model unit configured to obtain solar pressure perturbation and planetary gravity perturbation data;

[0026] a running state determination unit configured to input starlight angular distance, solar pressure perturbation and planetary gravity perturbation data of each star into an unscented Kalman filter algorithm, and determine a running state of the probe, wherein the running state comprises three direction position vectors and velocity vectors of the probe.

[0027] The above technical solution has the following beneficial effects: when obtaining starlight angular distance used for determining the motion state of the probe, the number of stars providing starlight angular distance information is limited to three, and a basis for determining the target star combination is given, which significantly reduces the demand for probe operation resources when determining the optimal starlight angular distance, and improves the calculation efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0029] Figure 1 is a flow chart of a deep space exploration transfer segment starlight angular distance information optimization method according to an embodiment of the present application;

[0030] Figure 2 is a flow chart of a deep space exploration transfer segment starlight angular distance information optimization method according to an embodiment of the present application;

[0031] Figure 3 is a structural diagram of a deep space exploration transfer segment starlight angular distance information optimization device according to an embodiment of the present application;

[0032] Figure 4 is a structural diagram of a deep space exploration transfer segment starlight angular distance information optimization device according to an embodiment of the present application;

[0033] Figure 5 is a schematic diagram of a deep space exploration transfer segment autonomous navigation scheme based on starlight angular distance according to an embodiment of the present application;

[0034] Figure 6 is a target measurement principle schematic diagram of an optical sensor measurement coordinate system according to an embodiment of the present application;

[0035] Figure 7 is a starlight angular distance measurement principle schematic diagram according to an embodiment of the present application;

[0036] Figure 8 is a position relationship schematic diagram among a probe, a planet and the sun according to an embodiment of the present application;

[0037] Figure 9 is a tetrahedron schematic diagram composed of three star vectors in a measurement coordinate system according to an embodiment of the present application;

[0038] Figure 10 is a deep space exploration transfer segment autonomous navigation flow schematic diagram according to an embodiment of the present application;

[0039] Figure 11 is a visual star magnitude schematic diagram of the earth;

[0040] Figure 12 is a visual star magnitude schematic diagram of Mars;

[0041] Figure 13 is a visual star magnitude schematic diagram of Jupiter;

[0042] Figure 14 is a schematic diagram of the observability scale under different star combinations, which is one of the embodiments of the present application;

[0043] Figure 15 is a schematic diagram of the observability scale under the combination of star 1 and star 3, which is one of the embodiments of the present application;

[0044] Figure 16 is a schematic diagram of the observability scale under the combination of star 1 and star 2, which is one of the embodiments of the present application;

[0045] Figure 17 is a schematic diagram of the observability scale under the combination of star 2 and star 3, which is one of the embodiments of the present application;

[0046] Figure 18 is a schematic diagram of the iteration process of the simulated annealing particle swarm optimization algorithm, which is one of the embodiments of the present application;

[0047] Figure 19 is a schematic diagram of the state estimation error under different star combinations, which is one of the embodiments of the present application;

[0048] Figure 20 is a schematic diagram of the state estimation error under the optimal star combination, which is one of the embodiments of the present application;

[0049] Figure 21 is a schematic diagram of the state estimation error under the random star combination, which is one of the embodiments of the present application;

[0050] Figure 22 is a schematic diagram of the state estimation error under the random star combination changing with the measurement error, which is one of the embodiments of the present application;

[0051] Figure 23 is a schematic diagram of the state estimation error under the optimal star combination changing with the measurement error, which is one of the embodiments of the present application. DETAILED DESCRIPTION

[0052] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0053] The inventors consider that the time delay and unknown faults of the transfer phase navigation method based on ground stations in the prior art will seriously affect the navigation accuracy of the transfer phase of future long-distance solar system planet exploration missions, and reduce the success rate of subsequent exploration missions to capture, approach and flyby celestial bodies. Therefore, in order to improve the reliability of deep space exploration missions, it is necessary to provide an autonomous navigation method for the transfer phase of deep space exploration. The inventors find that the optical sensor is an ideal information source for measuring long-distance target information. Unlike the X-pulsar signal receiver, the optical sensor can be used to observe a variety of space targets, including stars, planets and planetary satellites, and has a long action distance, low energy consumption and simple structure, which meets the miniaturization trend of future spacecraft. In addition, the development of the gimballed optical sensor also provides convenience for obtaining large-scale space measurement information. Therefore, optical autonomous navigation is the most suitable navigation method for the transfer phase of deep space exploration. In order to overcome the problem of low precision, the starlight angle distance (SAD) composed of the planet vector and the starlight vector gradually becomes an effective observation information. Their ephemeris accuracy is higher under the condition of long-term observation and compilation of ground observation stations. The inventors find that due to the limited computing and storage resources on the satellite, the probe is difficult to process all optical image information on orbit. Therefore, it is necessary to select the most suitable combination of planet vectors and starlight vectors to obtain starlight angle distance information, and then realize the optimal precision autonomous navigation scheme. In order to determine the optimal measurement information combination, the current method usually obtains the optimal measurement information through the traversal method on the basis of observability analysis. However, due to the large number of observable planets and stars, and the navigation system model is a typical nonlinear model, the navigation system will consume a large amount of computing and storage resources when performing observability analysis and traversing measurement information, and cannot be realized autonomously on the satellite. Therefore, the inventors consider that it is necessary to construct a simple and efficient observability index, and simplify the measurement information optimization selection algorithm, so as to select the least optical measurement information on the premise of reducing the on-orbit computing burden, and realize the autonomous navigation of the transfer phase of deep space exploration.

[0054] In a first aspect, the embodiments of the present application provide a deep space exploration transfer phase starlight angle distance information optimization method, which is adopted by a probe with an optical sensor, and the method comprises:

[0055] Step S11, acquiring a deep space image by the optical sensor; the deep space image comprises a plurality of stars and at least one planet; the number of the plurality of stars is greater than or equal to 3;

[0056] Step S12, selecting the planet closest to the probe from the at least one planet as a target planet according to the deep space image and the ephemeris;

[0057] Step S13, constructing an optical sensor measurement coordinate system for the collected deep space image, and determining a star vector of each star in the plurality of stars and a planet vector of the target planet in the optical sensor measurement coordinate system; the optical sensor measurement coordinate system coincides with a body coordinate system of the probe;

[0058] Step S14, determining a target star combination from all star combinations according to the star vectors of all stars; each star combination is composed of any three stars in all stars, and the sum of inner products between the star vectors of the three stars in the target star combination is the minimum value among all star combinations;

[0059] Step S15, determining optimal deep space probe transfer segment starlight angular distance information according to the star vectors of the three stars in the target star combination and the planet vector of the target planet.

[0060] The embodiments of the present application have the following technical effects: when acquiring starlight angular distance for determining the motion state of the probe, the embodiments of the present application limit the number of stars providing starlight angular distance information to three, and provide the basis for determining the target star combination, thereby significantly reducing the demand for probe operation resources when determining the optimal starlight angular distance and improving the calculation efficiency.

[0061] Further, the constructing an optical sensor measurement coordinate system for the collected deep space image, and determining a star vector of each star in the plurality of stars and a planet vector of the target planet in the optical sensor measurement coordinate system comprises:

[0062] The optical sensor measurement coordinate system coincides with a body coordinate system of the probe; the origin of the body coordinate system of the probe is fixed on the center of mass of the probe, the X axis of the body coordinate system of the probe points to the front of the probe, the Y axis of the body coordinate system of the probe points to the right side of the X axis of the body coordinate system of the probe, is perpendicular to the X axis of the body coordinate system of the probe, the Z axis of the body coordinate system of the probe points to the bottom of the probe, is perpendicular to the XY plane of the body coordinate system of the probe, and satisfies the right-hand rule;

[0063] For each star and planet, the following formula is applied respectively to determine the star vector of the star and the planet vector of the planet:

[0064]

[0065] Wherein, f represents the focal length of the optical sensor, (m x ,m y ) represents the coordinates of the target on the imaging plane of the optical sensor, and the target includes each star or planet.

[0066] Further, the target star combination is determined according to the star vectors of all stars in all star combinations, and the target star combination comprises the following steps:

[0067] Initialize a particle swarm, and the particle swarm comprises m particles ζ i ; wherein i = 1, 2,..., m, each particle in the particle swarm represents a potential solution set comprising an initial position and a speed, and each particle corresponds to a star combination, i.e. ζ i = (p i , v i ), wherein p i = {p i1 , p i2 , p i3}, p i1 , p i2 , p i3 are three star numbers in the particle, respectively.

[0068] For each particle ζ i in the particle swarm, a fitness function o(p i ) = |s1s2| + |s1s3| + |s2s3| of each particle ζ i is calculated, s1, s2, s3 are star vectors of three stars in the star combination corresponding to the particle, and f(p i ) = o(p i ) is recorded, the position P i of each particle ζ id is recorded, the fitness f(P id ) = o(P id ) is recorded, the fitness of all m particles ζ i is compared, the position P p of the particle ζ pd corresponding to the minimum fitness is taken as the best position, and the corresponding fitness is the global optimal fitness f(P pd ).

[0069] According to the global optimal fitness f(P pd ), the position of the particle ζ p corresponding to the minimum fitness is P pd , and the initial temperature T0 of the annealing algorithm is calculated as f(P pd ) / ln5.

[0070] F SA (P id ) is a function for calculating the fitness of the annealing algorithm, and the fitness of the annealing algorithm of each particle ζ i at the position P id under the current temperature T is calculated as follows:

[0071]

[0072] where exp() represents a natural exponential function, ζ i represents a current particle, P id represents a position of the current particle;

[0073] An individual optimal position P rd of the rth particle satisfying a preset optimal selection condition is selected by applying the idea of roulette wheel selection algorithm, and is used to replace a global optimal position P pd ;

[0074] P pd is replaced by P rd , and the velocity of each particle is updated as follows by substituting the particle swarm formula:

[0075]

[0076] where k represents an iteration number, c1 and c2 represent learning factors, and are usually taken as 2, r1 and r2 represent two different random numbers between 0 and 1, ω represents a weight, represents a velocity of an individual optimal particle at the k+1th iteration, represents a velocity of an individual optimal particle at the kth iteration, represents an initial position of the particle ζ i ; represents an individual optimal position at the kth iteration;

[0077] The fitness of each particle is calculated again, and the optimal position P rd of each particle and the population optimal position P pd are updated;

[0078] An annealing operation is performed, and the input is a current temperature T, and the output is an initial temperature T0 of the next iteration:

[0079] T0 = δ × T (4)

[0080] where δ ∈ (0, 1) represents an annealing factor;

[0081] It is judged whether a preset termination condition is satisfied, if the preset termination condition is satisfied, the search is stopped, and a return value is output, if the preset termination condition is not satisfied, the function F SA (P id ) is calculated, and the annealing algorithm fitness of the position P i of each particle ζ id at the current temperature T is calculated;

[0082] where the preset termination condition is a judgment condition formula for satisfying a minimum value of an observability measure:

[0083] min(|s1s2|+|s1s3|+|s2s3|)

[0084] s.t.arccos(s w s u )≤F0 w,u=1,2,3,w≠u (5)

[0085] Wherein, s1, s2, s3 are the star vectors of three stars in the target star combination corresponding to the particle; F0 represents the field of view angle of the optical sensor;

[0086] The return value includes the target star combination;

[0087] The roulette selection algorithm is used for selecting the operator, comprising:

[0088] Generating a random number pBet between 0 and 1;

[0089] According to the fitness F SA (P id ) of the annealing algorithm, the cumulative probability P

[0090] According to the cumulative probability, the individual optimal position P rd of the rth particle meeting the preset optimal condition is selected to replace the global optimal position P pd ; the preset optimal condition is: comfit(r-1)<pBet<comfit(r). (7)

[0091] Further, according to the star vectors of the three stars in the target star combination and the planet vector of the target planet, the optimal deep space exploration transfer segment starlight angular distance information is determined, comprising:

[0092] According to the starlight angular distance measurement model of the nth star, the starlight angular distance of the nth star is determined:

[0093] A n,τ =arccos(s0s n,τ )+v n,τ (8)

[0094] Wherein, s n,τ ,n=1,2,3 represents the star vector of the nth star in the target star combination, τ represents the time, s n,τ =(cosα n,τ cosδ n,τ ,cosα n,τ sinδ n,τ ,sinα n,τ ) T ,α n,τ and δ n,τrespectively represent the right ascension and declination of the nth star in the target star combination, A n,τ , n = 1, 2, 3 represent the starlight angular distance composed of the starlight vector of the nth star in the target star combination and the planet vector of the target planet; s0represents the direction vector between the planet and the detector, v n,τ represents the measurement noise, which is subject to Gaussian distribution and satisfies

[0095] According to the measurement model composed of n s starlight angular distance information, the starlight angular distances corresponding to the three stars in the target star combination are determined:

[0096]

[0097] wherein, O represents a zero matrix, z τ represents a measurement information vector, represents n s starlight angular distances, n s represents the number of starlight angular distances, v τ represents the measurement noise, h(x τ ) represents a vector composed of n s starlight angular distances R τ represents the covariance matrix of the measurement noise; the measurement matrix H τ is the partial derivative matrix of h(x τ ) with respect to x τ , and is expressed as:

[0098]

[0099] wherein, the specific expression of H l,τ is:

[0100]

[0101] wherein,

[0102]

[0103]

[0104] wherein, (x, y, z) represents the position vector of the detector, (x0, y0, z0) represents the position vector of the planet, δ l represents the declination of the lth star, and a l represents the right ascension of the lth star.

[0105] In a second aspect, as Figure 2As shown, the embodiment of the present application provides an autonomous navigation method based on deep space exploration transfer phase starlight angular distance information optimization, comprising:

[0106] In step S21, any one of the deep space exploration transfer phase starlight angular distance information optimization methods described above is used to determine the star vector of three stars in the target star combination, the planet vector of one target planet, and the starlight angular distance information corresponding to the three stars in the target star combination.

[0107] In step S22, the solar pressure perturbation and planetary gravity perturbation data are obtained through an orbit vibration model.

[0108] In step S23, the starlight angular distance of each star, the solar pressure perturbation, and the planetary gravity perturbation data are input into the unscented Kalman filter algorithm to determine the running state of the probe.

[0109] In a third aspect, as Figure 3 As shown, the embodiment of the present application provides a deep space exploration transfer phase starlight angular distance information optimization device, which is used by a probe with an optical sensor, and the device comprises:

[0110] An image acquisition unit 300 is configured to acquire a deep space image through the optical sensor; the deep space image includes a plurality of stars and at least one planet; the number of the plurality of stars is greater than or equal to 3;

[0111] A target planet determination unit 301 is configured to select the planet closest to the probe as the target planet from the at least one planet according to the deep space image and ephemeris;

[0112] A vector determination unit 302 is configured to construct an optical sensor measurement coordinate system for the acquired deep space image, and determine the star vector of each star in the plurality of stars and the planet vector of the target planet in the optical sensor measurement coordinate system; the optical sensor measurement coordinate system coincides with the body coordinate system of the probe;

[0113] A target star combination determination unit 303 is configured to determine the target star combination from all star combinations according to the star vectors of all stars; each star combination is composed of any three stars from all stars, and the sum of the inner products between the star vectors of the three stars in the target star combination is the minimum value among all star combinations;

[0114] A starlight angular distance determination unit 304 is configured to determine the optimal deep space exploration transfer phase starlight angular distance information according to the star vectors of the three stars in the target star combination and the planet vector of the target planet.

[0115] Further, the vector determination unit 302 comprises:

[0116] The optical sensor measurement coordinate system coincides with the body coordinate system of the detector; the origin of the body coordinate system of the detector is fixed on the centroid of the detector, the X axis of the body coordinate system of the detector points to the front of the detector, the Y axis of the body coordinate system of the detector points to the right side of the X axis of the body coordinate system of the detector, is perpendicular to the X axis of the body coordinate system of the detector, the Z axis of the body coordinate system of the detector points to the bottom of the detector, is perpendicular to the XY plane of the body coordinate system of the detector, and satisfies the right-hand rule;

[0117] The vector determination module is configured to determine the star vector of the star and the planet vector of the planet by applying the following formula (1) for each star and planet, respectively.

[0118] Further, the target star combination determination unit 303 comprises:

[0119] The particle swarm initialization module is configured to initialize a particle swarm, the particle swarm comprising m particles ζ i ; wherein i = 1, 2,..., m, each particle in the particle swarm represents a potential solution set comprising an initial position and a velocity, and each particle corresponds to a star combination, i.e. i ζ i = (p i , v i ), wherein p i1 = {p i2 , p i3 , p i1}, p i2 , p i3 , and p i are three star numbers in the particle.

[0120] The particle fitness determination module is configured to calculate the fitness function o(p i ) = |s1s2| + |s1s3| + |s2s3| of each particle ζ i , wherein s1, s2, and s3 are the star vectors of the three stars in the star combination corresponding to the particle, and f(p i ) = o(p i ) is recorded, the position P id and the fitness f(P id ) = o(P id ) of each particle ζ i are recorded, the fitness of all m particles ζ p is compared, the position P pd of the particle ζ pd corresponding to the minimum fitness is taken as the best position, and the corresponding fitness is the global optimal fitness f(P pd ).

[0121] The annealing initial temperature determination module is used to determine the global optimal fitness f(P) pd ), where the particle ζ corresponds to the minimum fitness at this point. p Position = P pd The initial temperature T0 of the annealing algorithm is calculated to be f(P). pd ) / ln5;

[0122] Annealing fitness determination module, used to record F SA (P id The function is used to calculate the fitness of the annealing algorithm, given the ζ values ​​of each particle at the current temperature T. i Position P id The fitness calculation of the annealing algorithm is as shown in formula (2);

[0123] The global optimal update module applies the idea of ​​the roulette wheel selection algorithm to select the optimal position P of the r-th particle that meets the preset selection criteria. rd Replace the global optimal position P pd ;

[0124] The particle velocity update module is used to update P pd Replace with P rd Substituting into the particle swarm optimization formula, the velocities of each particle are updated as shown in formula (3):

[0125] The particle position update module is used to recalculate the fitness of each particle and update the optimal position P of each particle. rd and the population's optimal position P pd ;

[0126] The annealing module performs annealing operations. The input is the current temperature, and the output is the initial temperature T0 of the next iteration. The specific formula is formula (4).

[0127] The termination judgment module is used to determine whether the preset termination condition is met. If the preset termination condition is met, the search stops and a return value is output. If the preset termination condition is not met, the annealing fitness determination module is triggered.

[0128] The preset termination condition is the judgment condition formula (5) that satisfies the minimum value of the observability metric;

[0129] The return value includes the target star combination;

[0130] The roulette wheel selection algorithm is used to select operators, including:

[0131] Generate a random number pBet between 0 and 1;

[0132] Based on the fitness F of the annealing algorithm SA (P id), the cumulative probability is calculated by formula (6);

[0133] According to the cumulative probability, the individual optimal position P of the rth particle meeting the preset optimal condition is selected rd Instead of the global optimal position P pd ; the preset optimal condition is formula (7).

[0134] Further, the starlight angular distance determination unit 304 is configured to:

[0135] The starlight angular distance of the ith star is determined according to the starlight angular distance measurement model of the ith star by formula (8).

[0136] According to the n s starlight angular distance information, the starlight angular distances corresponding to the three stars in the target star combination are determined by formula (9) to formula (17).

[0137] In a fourth aspect, as shown in Figure 4 , the embodiment of the present application provides an autonomous navigation device based on deep space exploration transfer segment starlight angular distance information optimization, comprising:

[0138] The deep space exploration transfer segment starlight angular distance information optimization device 400 adopts any one of the deep space exploration transfer segment starlight angular distance information optimization methods as described above, so as to determine the star vector of the three stars in the target star combination, the planet vector of the target planet and the starlight angular distance information corresponding to the three stars in the target star combination.

[0139] The orbit vibration model unit 401 is used to obtain solar pressure perturbation and planetary gravity perturbation data.

[0140] The running state determination unit 402 is used to input the starlight angular distance of each star, the solar pressure perturbation and the planetary gravity perturbation data into the unscented Kalman filtering algorithm, and determine the running state of the probe; the running state includes the three direction position vectors and the velocity vectors of the probe.

[0141] The above technical solutions of the embodiments of the present application will be described in detail in combination with specific application examples, and the technical details not introduced in the implementation process can refer to the related description in the foregoing.

[0142] Deep space probe transfer phase plays a vital role in planetary exploration mission. Optical sensor is an ideal information source for autonomous navigation of deep space probe transfer phase. In order to achieve optimal autonomous navigation under the condition of severe resource constraints on the spacecraft, the starlight angular distance is selected as the navigation information source. The minimum number of stars required by the navigation system is determined by the observability matrix analysis. Then, combined with the starlight angular distance navigation principle, the observable ability scalar calculation metric is proposed, which overcomes the disadvantage of large amount of calculation of traditional observability analysis method. The influence of different starlight angular distances on navigation accuracy can be analyzed by using the metric. On this basis, the optimal navigation method is combined with the optimal starlight angular distance selection algorithm to improve the autonomous navigation performance of deep space probe transfer phase.

[0143] 1. Deep space probe transfer phase autonomous navigation scheme based on starlight angular distance

[0144] 1.1 Transfer phase autonomous navigation system framework

[0145] The deep space probe transfer phase autonomous navigation scheme based on starlight angular distance information is shown in Figure 5 . In Figure 5 , the navigation system state model mainly consists of the gravitational perturbation of the sun and the planet and the solar pressure perturbation model. The measurement model is mainly composed of the starlight angular distance information measurement principle, which is calculated by the starlight vector and the planet vector measured by the optical sensor. Due to the large number of observable planets and stars, the navigation system needs to analyze the observability in combination with the perturbation model and the measurement model, and select the optimal measurement information as the navigation measurement information source according to the analysis result. Finally, based on the measurement information and the system state model, the position state estimation result of the probe is obtained by the Unscented Kalman Filter (UKF) algorithm.

[0146] 1.2 Navigation system model

[0147] The following describes the definition of the coordinate system related to the navigation system.

[0148] The sun inertial coordinate system is an inertial coordinate system with the sun as the center and fixed relative to the X axis along the intersection direction of the ecliptic and the solar equatorial plane. The Z axis points to the vertical direction of the solar equator, and the Y axis forms a right-handed coordinate system with the X axis and the Z axis. In the embodiment of the present application, the sun inertial coordinate system is selected as the navigation coordinate system.

[0149] The body coordinate system has its origin fixed at the center of mass of the spacecraft. The X axis points to the front of the spacecraft. The Y axis points to the right of the X axis and is perpendicular to the X axis. The Z axis points to the bottom of the spacecraft and is perpendicular to the XY plane and satisfies the right-hand rule.

[0150] The embodiment of the present application assumes that the optical sensor measurement coordinate system coincides with the body coordinate system, and the target measurement principle is as shown in Figure 6The target direction vector is represented as equation (1).

[0151] 1.2.1 System State Model

[0152] The system state model is constructed based on the on-orbit dynamics model of the probe, including the gravitational perturbation of the sun and planets and the solar radiation pressure perturbation. The dynamics model is represented as equation (18):

[0153]

[0154] where and represent the position and velocity of the probe in the navigation coordinate, respectively, t represents time, and δ(t) represents process noise (used to describe the unmodeled dynamics model), represents the derivative of χ with respect to time t, and f(x, t) is represented as:

[0155]

[0156] where u s represents the gravitational constant of the sun, N = 8 represents the number of main planets in the solar system, μ i represents the gravitational constant of the i-th planet, (x i , y i , z i ) represents the position of the i-th planet in the navigation coordinate system, (x si , y si , z si ) represents the relative position of the probe and the i-th planet, represents the distance between the sun and the probe, represents the distance between the sun and the i-th planet, represents the distance between the probe and the i-th planet, η represents the shadow coefficient, P SR represents the solar radiation pressure under 1 light year, C R represents the surface reflection coefficient of the probe, A R represents the cross-sectional area of the spacecraft perpendicular to the direction of the sun, and m represents the mass of the probe. AU 2 represents the square of 1 astronomical unit. Further, the discrete form of f(x, t) is represented as equation (20):

[0157]

[0158] where T represents the time interval. The discrete form of the system state model is represented as:

[0159] x k = f(x k-1 ) + w k (21)

[0160] where w k represents the discrete form of state noise δ(t), which is subject to Gaussian distribution and satisfies E(w k ) = 0,

[0161] 1.2.2 Measurement Model

[0162] The measurement model of the navigation system is mainly constructed according to the principle of starlight angular distance measurement. Since the planets run slowly and their sizes are larger relative to the planets' satellites, it is relatively low in difficulty for the optical sensor to acquire and process images. Therefore, in order to improve the measurement accuracy of starlight angular distance, the center of mass direction vector of each planet in the eight planets of the solar system is selected as the planet vector of starlight angular distance. The measurement principle is shown in Figure 7 In Figure 7 , t and r represent the position vectors of the planet and the probe in the navigation coordinate system, respectively. s represents the relative position vector of the planet and the probe in the navigation coordinate system. s i , i = 1, 2, 3 represent the starlight vector of the i th star, which is specifically represented as: s i = (cosα i cosδ i , cosα i sinδ i , sinα i ) T , where α i and δ i represent the right ascension and declination of the i th star, respectively. A i , i = 1, 2, 3 represent the starlight angular distance composed of the starlight vector of the i th star and the planet vector. The measurement model of the i th starlight angular distance is represented by formula (8).

[0163] The measurement model composed of n s starlight angular distance information is represented by formula (9) to formula (17).

[0164] The star direction information can be obtained through ephemeris. However, since the number of stars is extremely large, it is necessary to determine which star or which several stars are selected as measurement information according to the analysis results of star magnitude and observability. Since the position information of near celestial bodies is more accurate, the optical image information obtained is also more accurate, so the planet closest to the probe is selected as the planet vector measurement information source of starlight angular distance.

[0165] 2. Observability Analysis

[0166] Observability is the key to obtain high-precision navigation results. Based on the observability analysis results, the available quantity information at different time can be determined, and the performance of autonomous navigation system at different time can be estimated. Therefore, the observability analysis process is the key step to select the optimal measurement information combination and improve the navigation performance.

[0167] 2.1 Observability matrix

[0168] The observability matrix can be used to determine the minimum number of star angular distance measurement data required to achieve autonomous navigation. The observability matrix is defined as:

[0169]

[0170] where n represents the dimension of the system state. When the rank of the matrix O k is n, the system satisfies the observability condition.

[0171] When there is only 1 star angular distance information, the observability matrix (22) after the simplest row transformation is obtained:

[0172]

[0173] Obviously, the rank of the observability matrix is 5, which is not full rank.

[0174] When there are 2 star angular distance information, the observability matrix (22) after the simplest row transformation is obtained:

[0175]

[0176] where

[0177]

[0178] Obviously, the rank of the observability matrix is 5, which is not full rank.

[0179] When there are 3 star angular distance information, the observability matrix (22) after the simplest row transformation is obtained:

[0180]

[0181] Obviously, the observability matrix is column full rank at this time, and the system satisfies the observability condition. Three star angular distances are composed of three star vectors and one planet vector. Therefore, when there are at least 3 stars and 1 planet, the probe can achieve autonomous navigation.

[0182] 2.2 Observability measure

[0183] The rank of the observability matrix provides the minimum number of stars required for autonomous navigation. In order to select the optimal star combination, it is necessary to quantify the autonomous navigation system's ability to observe, i.e. the observability metric. According to Figure 7 , the position relationship between the probe, the planet and the sun is shown as Figure 8 . Where t = k t t0, r = k r r0, s = k s s0. k t , k r and k s represent the length of t, r and s respectively. t, k t and t0are obtained from the planet ephemeris. θ represents the angle between vectors t and s. φ is the complementary angle of the angle between vectors s and r. s0represents the direction vector from the planet to the probe in the navigation coordinate system, which is expressed as:

[0184]

[0185] where R represents the rotation matrix from the measurement coordinate system to the navigation coordinate system, represents the rotation matrix from the probe body coordinate system to the navigation coordinate system, represents the rotation matrix from the measurement coordinate system to the body coordinate system, which is assumed to be the identity matrix in the embodiments of the invention. l0represents the direction vector from the planet to the probe in the measurement coordinate system, which can be obtained directly from the measurement sensor. Therefore, r0is expressed as:

[0186]

[0187] Further:

[0188] θ = arccos (t0s0) (29)

[0189] φ = arccos (-r0s) (30)

[0190] According to the sine theorem, the probe position in the navigation coordinate system is expressed as:

[0191]

[0192] According to equation (31), the determination accuracy of r0depends on the rotation matrix and the vector l0. Since l0depends on the nearest planet, it is important to quantify the determination accuracy of the rotation matrix for evaluating the navigation accuracy. represents the determined attitude matrix. According to the above analysis, the starlight angular distance autonomous navigation can be equivalently regarded as determining the position of the probe in the navigation coordinate system through the attitude determination result of the planet and starlight vectors with known azimuth information. According to the above analysis, the starlight angular distance autonomous navigation accuracy can be equivalently represented by the starlight attitude determination accuracy.

[0193] Theorem 1: The determination accuracy of is represented as:

[0194]

[0195] wherein represents the determination error of represents the determination error covariance of

[0196] Proof: Let l1, l2 and l3 represent the starlight vectors in the measurement coordinate system, then:

[0197]

[0198] wherein v represents the measurement noise, which is equal to the measurement noise in formula (9). According to the least square method, the estimation result of is represented as:

[0199]

[0200] The estimation error is represented as:

[0201]

[0202] then,

[0203] Theorem 2. When the three starlight vectors are perpendicular to each other, the attitude matrix achieves the optimal determination accuracy.

[0204] Proof: According to Theorem 1, the determination accuracy of is converted to:

[0205]

[0206] wherein (·) * represents the adjugate matrix.

[0207] In formula (36), R is the optical sensor design parameter. Therefore, when the optimal accuracy is achieved, that is, reaches the minimum value, at this time, det(L T L) reaches the maximum value and tr[(L T L)​​* [Reaching the minimum value.]

[0208] On the one hand, for det(L) T L), such as Figure 9 As shown, its size is actually equal to the volume of a tetrahedron composed of three stellar vectors. This is because the calculated volume of the tetrahedron is:

[0209]

[0210] The volume of a tetrahedron can also be calculated using the following formula:

[0211]

[0212] in α, β, γ represent the angles between any two stellar vectors, and ||l1||, ||l2||, ||l3|| represent the magnitudes of the three stellar vectors, all equal to 1. It is clear that when V... s When the maximum value is reached, we have:

[0213] sinp=sin(p-α)=sin(p-β)=sin(p-γ) (39)

[0214] Then we have:

[0215]

[0216] On the other hand, tr[(L T L) * ]=λ1+λ2+λ3, where λ1, λ2, λ3 represent the matrix (L) T L) * The three eigenvalues. Additionally, When tr[(L) T L) * When ]=λ1+λ2+λ3 reaches its minimum value, λ1=λ2=λ3. When When , the stellar vector matrix L is an orthogonal matrix. Then λ1=λ2=λ3=1. This means that when At that time, det(L T L) reaches its maximum value and tr[(L) T L) * [Reaching the minimum value.]

[0217] In summary, when the starlight vectors of the three stars are perpendicular to each other, the attitude matrix... The accuracy is determined to be optimal.

[0218] According to Theorem 1 and Theorem 2, the closer the inner product of star vectors is to 0, the higher the state estimation accuracy is. When the sum of inner products of three stars two by two is equal to 0, the state estimation accuracy is optimal, therefore, the observability measure of starlight angular distance autonomous navigation system is represented as:

[0219] O = |s1s2| + |s1s3| + |s2s3| (41)

[0220] Therefore, the optimal star combination can be selected based on the observability measure.

[0221] 3. Optimal navigation scheme based on preferred algorithm

[0222] According to the foregoing observability analysis, the observability measure o can be used to evaluate the autonomous navigation accuracy. The minimum value of o will be used as the optimization target to select the optimal star combination, and then optimize the autonomous navigation accuracy. In this section, the navigation star combination selection scheme of the observability measure will be given, and the simulated annealing particle swarm algorithm is used to select the optimal star combination.

[0223] 3.1 Starlight angular distance preferred model based on observability measure

[0224] The star vector l in the measurement coordinate system i (i = 1, 2, 3) can be obtained by processing the image information obtained by the optical sensor. The observability measure o can be obtained by calculating the inner product of two star vectors. Since the starlight vector is 3, the calculation process of o greatly reduces the computational complexity compared to the traditional observability evaluation method involving complex matrix calculation. Therefore, the optimal star combination can be obtained under the condition that o is minimum. The specific optimization model is represented as formula (5) as the constraint condition.

[0225] 3.2 Simulated annealing particle swarm optimization algorithm

[0226] The simulated annealing algorithm is a global optimization algorithm based on the mechanism of metal annealing. In the case of sufficient computing resources and sufficient computing time, this algorithm can converge to the global optimal solution with probability. However, the limited processing resources of the probe limit the effectiveness of the algorithm. The particle swarm algorithm has good stability in optimal solution search and global convergence, but its search domain is small and it is easy to fall into a local optimal solution.

[0227] Considering the advantages and disadvantages of the two algorithms, the simulated annealing and particle swarm algorithm are combined in the optimal navigation scheme to optimize the star measurement information combination. The specific optimization selection process is the optimal star combination optimization scheme based on the simulated annealing particle swarm algorithm.

[0228] Parameters to be initialized: weight parameter ω, learning factors c1 and c2, annealing speed δ, evolution generation number M of particle swarm, and particle number m; output result: when the observability measure o reaches the minimum value, the global optimal position of the particle is output.

[0229] The specific procedure is as follows:

[0230] Step 1, initializing the particle swarm ζ i , i = 1, 2,..., m, wherein each particle represents a potential solution set (i.e., a star combination) containing an initial position and speed;

[0231] Step 2, calculating the fitness function O(p i ) of each particle ζ i , denoted as f(p i ), recording the position P id of each particle (the i-th particle), the global optimal position P pd , the fitness f(P id ), and the global optimal fitness f(P pd ).

[0232] Step 3, calculating the initial temperature f(P pd ) > 0 of the annealing algorithm according to the global optimal fitness f(P pd ).

[0233] Step 4, recording the function F SA (P id ) as a function for calculating the fitness of the annealing algorithm, wherein the annealing algorithm fitness of each P id at the current temperature T is calculated according to formula (2).

[0234] Step 5, applying the idea of roulette selection algorithm to select one from the individual optimal position P id to replace the global optimal position P pd , denoted as P rd . The roulette selection method is a kind of selection operator, also known as the proportional selection method, and the basic idea is to select the individuals with high fitness from the population by using the probability method, and the higher the fitness of each individual, the higher the probability of being selected. In the embodiment of the present application, the algorithm idea is as follows:

[0235] pBet = rand(), (pBet is a random number between 0 and 1)

[0236] According to formula (6), the cumulative probability is calculated according to the annealing algorithm fitness F SA (P id ).

[0237] According to the cumulative probability, the individual optimal position Prd Instead of global optimal position P pd . Condition is formula (7) ;

[0238] Step 6, replace P pd with P rd , substitute particle swarm formula (3), update the speed of each particle;

[0239] Step 7, recalculate the fitness of each particle, update the optimal position P rd of each particle and the population optimal position P pd

[0240] Step 8, annealing operation is performed by formula (4) ;

[0241] Step 9, judge whether the termination condition set in advance is met, if yes, stop searching, output the return value, if not, go to step 4

[0242] Output: output the result, output the global optimal solution of the particle swarm meeting the problem constraint condition.

[0243] According to the above steps, the particle swarm that makes the observability measure optimal, i.e. the star combination, is obtained.

[0244] On the basis of the above measurement information optimization, the deep space exploration transfer segment autonomous navigation scheme based on starlight angular distance information optimization is obtained as shown in Figure 10 .

[0245] 4. Simulation verification

[0246] The performance analysis of the starlight angular distance autonomous optical navigation method will be given. The scene is selected as the transfer segment of the Jupiter exploration task, and the simulation parameters are shown in Table 1.

[0247] Item Value Unit Initial position (5.31,13.76,-0.01)*1010 m Initial velocity (-14.04,35.86,0.42)*103 m / s Simulation time 1.8 Years Initial position error (100010001000) m Initial velocity error (101010) m / s Sensor sensitivity threshold 6.5 Star magnitude Optical sensor resolution 4096*4096 Pixels Optical sensor focal length 170 mm Measurement noise 0.1 Pixels Sampling period 12 Hours

[0248] Table 1 Simulation conditions

[0249] 4.1 Description of simulation scenario

[0250] In the simulation, the star ephemeris is obtained from the Yabu star catalog. Considering that the starlight angular distance information is obtained from 3 stars and 1 central planet, the number of visible stars needs to be determined according to different central stars. From the perspective of the whole detection process of the probe, the Earth, Mars and Jupiter are the three closest central planets that can be observed by the probe. Moreover, the large scale of the three planets makes their image information and centroid information easy to obtain and process. According to the introduction of starlight angular distance information in the previous section, the planet closest to the probe is selected as the central planet of the starlight angular distance information. The 3 stars are selected according to the optimization algorithm to achieve the optimization of navigation accuracy.

[0251] After the simulation scenario is constructed, the visibility of the planets and stars is determined by the optical sensor sensitivity threshold and the apparent magnitude of the targets. The apparent magnitudes of the three planets are shown in Figure 11 , Figure 12 and Figure 13 .

[0252] According to Figure 11 , Figure 12 and Figure 13 , the central planet can be different at different times. It is noted that there are much more than 3 stars that can be observed at any time. Therefore, the star angular distance measurement information can be optimized to achieve orbit determination according to the observability metric and the simulated annealing particle swarm algorithm.

[0253] 4.2 Star angular distance information optimization selection comparison and analysis

[0254] In order to prove the effectiveness of the method proposed by the embodiment of the application, a Monte Carlo simulation is performed in the scenario mentioned above.

[0255] The observability of different star combinations is shown in Figure 14 . In the simulation, according to the optical sensor sensitivity threshold and the Hubble Guide Star Catalog, 3596 stars can be observed. The coordinate axes in the figure represent different stars. The observability under different three-star combination conditions is reflected by the color of the block in the figure. When the color tends to be closer to blue, the observability metric is smaller, and the navigation accuracy is higher.

[0256] According to Figure 14 , in order to better represent the change of the system observability under different star combinations, Figure 15 , Figure 16 and Figure 17 three images are given. In the three images, it is assumed that two of the three stars are fixed, and the number of each fixed star is given in the figure, Figure 15 is star 1 No. 798 and star 3 No. 1841, Figure 16 is star 1 No. 798 and star 2 No. 2803, Figure 17 is star 3 No. 1841 and star 2 No. 2803. The images reflect the change of the system observability scale with the third star under the condition that two stars are fixed.

[0257] According to Figure 14 , Figure 15 , Figure 16 and Figure 17 , it can be seen that the system observability changes very complicatedly with the star combination, and it is difficult to obtain the optimal star combination through an analytical relationship. For a probe with severely limited on-board resources, it is difficult to select the optimal star combination to make the system observability optimal through the exhaustive method.

[0258] To handle the above problem, the simulated annealing particle swarm optimization algorithm is used to select the optimal star combination. The iteration process of the algorithm is shown in Figure 18 .

[0259] Based on the selection result of the preferred algorithm, the spacecraft obtains the position state estimation result under the optimal star combination and the random star combination. The simulation is performed 1000 times, and the spacecraft position state estimation error is shown in Figure 19 .

[0260] The evaluation estimation error of different star combinations is shown in Figure 20 and Figure 21 , and part of the curve is shown in the enlarged view of the figure. The root mean square error result of the probe position estimation result is shown in Table 2.

[0261] X direction Y direction Z direction Total error Optimal star combination 750.46 2564.93 3174.31 4149.49 Random star combination 8690.96 13023.4 2153.27 15804.38

[0262] Table 2 Root mean square error of probe position estimation

[0263] According to Figure 20 , Figure 21 , the results of Table 2, the navigation accuracy of the optimal star combination is higher than that of the random star combination in most cases. These results prove the effectiveness of the method.

[0264] 4.3 Effect of measurement noise

[0265] Measurement noise often affects navigation accuracy based on optical image processing accuracy. The state estimation result of the embodiment of the application is given under the condition that the measurement noise changes from 0.1 pixel to 1.0 pixel, as shown in Figure 22 and Figure 23 . 1000 simulations are performed under each measurement noise value condition, and the average value of the state estimation error is obtained. It is obvious from Figure 22 that the state estimation accuracy changes nonlinearly with the measurement noise, and the accuracy under the optimal star combination is generally higher than that under the random star combination. In addition, it is obvious that the state estimation accuracy decreases as the measurement noise increases. This is because the starlight angular distance is composed of the “probe-planet” vector and the star vector, and the navigation accuracy mainly depends on the extraction accuracy of the above two vectors. If the navigation accuracy requirement is higher, the image processing algorithm accuracy needs to be higher, which will also lead to higher requirements for the complexity and efficiency of the image processing algorithm. Therefore, the navigation accuracy and calculation efficiency need to be balanced in actual engineering practice.

[0266] It is obvious that the position estimation accuracy changes differently in different directions with the noise. This is related to the orbit state of the probe. According to Figure 22 and Figure 23The positions of planets in the solar system indirectly provide distance information for the state estimation accuracy of the probe, and the relative position information of the planets and the probe is mainly reflected in the position estimation result of the probe. The angle between the vector in different directions and the coordinate system is different for the position estimation result in different directions. Therefore, if the state estimation accuracy in a certain direction of the probe is to be improved, a suitable to-be-observed star needs to be selected in the process of designing the orbit of the probe.

[0267] 6. Conclusion

[0268] The embodiment of the present application provides a deep space probe transfer segment autonomous navigation method based on starlight angular distance information optimization. Due to limited computing resources, it is difficult to select optimal measurement information from all observable stars. For a nonlinear autonomous navigation system, the calculation process of the traditional measurement information optimization selection method based on observability analysis is very complex. In order to overcome this problem, the embodiment of the present application combines the starlight angular distance positioning principle to propose a scalar calculation-based observability measurement scale. Based on this observability measurement, a navigation scheme is proposed to select the optimal measurement information by using the simulated annealing particle algorithm. The effectiveness of the above method is verified by simulation, and the influence of measurement noise on navigation accuracy is also analyzed based on the simulation results.

[0269] It should be understood that the specific order or hierarchy of steps in the processes disclosed is an example. Based upon design preferences, it should be understood that the specific order or hierarchy of steps in the processes can be re-arranged while remaining within the scope of the present disclosure. The accompanying method claims present elements of the various steps in a sample order, and as such claims should not be construed as necessarily limited to the particular opening of closing sequence.

[0270] In the above detailed description, various features are grouped together in single embodiments for the purpose of streamlining the disclosure. This disclosed approach is not to be interpreted as reflecting an intention that the embodiments of the claimed subject matter require more features than are explicitly recited in each claim. On the contrary, as indicators of the true breadth of each claim, the claims appended hereto should be construed to include all embodiments falling within the scope of the claims, and to encompass all features available under the principles of patent law. Accordingly, the claims as they are expressed below are hereby expressly incorporated into this detailed description, with each claim acting as a separate embodiment of the application.

[0271] The disclosed embodiments have been described above. Those skilled in the art will readily devise their own modifications of the embodiments without departing from the spirit and scope of the disclosure. Accordingly, the disclosure is not intended to be limited to the embodiments described herein, but is to be accorded the full scope that the principles and novel features disclosed herein support.

[0272] The above description includes examples of one or more embodiments. Of course, not all possible combinations of components or methods described above will be employed to make or use the embodiments nor will all of

[0273] Those skilled in the art will further appreciate that the various illustrative logical blocks, modules, and steps described in connection with the embodiments disclosed herein can be implemented as electronic hardware, computer software, or combinations of both. To clearly illustrate this interchangeability of hardware and software, various illustrative components, blocks, modules, and steps have been described above generally in terms of their functionality. Whether such functionality is implemented as hardware or software depends upon the particular application and design constraints imposed on the overall system. Skilled artisans can implement the described functionality in varying ways for each particular application, but such implementation decisions should not be interpreted as causing a departure from the scope of the present embodiments.

[0274] The above description is only specific implementation of the present application, and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A deep space probe transfer phase star angle distance information optimization method, characterized in that, The method is adopted by a probe with an optical sensor, and the method comprises: collecting a deep space image by the optical sensor; the deep space image comprises a plurality of stars and at least one planet; the number of the plurality of stars is greater than or equal to 3; selecting a planet closest to the probe from the at least one planet as a target planet according to the deep space image and ephemeris; constructing an optical sensor measurement coordinate system for the collected deep space image, and determining a star vector of each star in the plurality of stars and a planet vector of the target planet in the optical sensor measurement coordinate system; the optical sensor measurement coordinate system coincides with a body coordinate system of the probe; determining a target star combination in all star combinations according to the star vectors of all stars; each star combination is composed of any three stars in all stars, the sum of inner products between the star vectors of the three stars in the target star combination is the minimum value in all star combinations, the inner product of the star vectors is closer to 0, the state estimation accuracy is higher, and an analog annealing particle swarm algorithm is used to select the optimal star combination; determining optimal deep space probe transfer segment starlight angular distance information according to the star vectors of the three stars in the target star combination and the planet vector of the target planet.

2. The deep space probe transfer phase starshot angular distance information optimization method of claim 1, wherein, The method for constructing the optical sensor measurement coordinate system for the collected deep space image and determining the star vector of each star in the plurality of stars and the planet vector of the target planet comprises: the optical sensor measurement coordinate system coincides with the body coordinate system of the probe; the origin of the body coordinate system of the probe is fixed on the center of mass of the probe, the X-axis of the body coordinate system of the probe points to the front of the probe, the Y-axis of the body coordinate system of the probe points to the right side of the X-axis of the body coordinate system of the probe, is perpendicular to the X-axis of the body coordinate system of the probe, the Z-axis of the body coordinate system of the probe points to the bottom of the probe, is perpendicular to the XY plane of the body coordinate system of the probe, and satisfies the right-hand rule; for each star and planet, the star vector of the star and the planet vector of the planet are determined by applying the following formula respectively: where f represents the focal length of the optical sensor, (m x ,m y ) represents the coordinates of the target on the imaging plane of the optical sensor, said target comprising each star or planet.

3. The deep space probe transfer phase starshot angular distance information optimization method of claim 1, wherein, The method for determining the target star combination in all star combinations according to the star vectors of all stars comprises: Initialize a particle swarm, the particle swarm including m particles ζ i ; wherein i = 1, 2, …, m, each particle in the particle swarm represents a potential solution set including an initial position and a velocity, each particle corresponds to a star combination, namely ζ i = (p i , v i ), wherein p i = {p i1 , p i2 , p i3}, p i1 , p i2 , p i3 are three star numbers in the particle respectively; For each particle ζ in the particle swarm i Calculate ζ for each particle i fitness function o(p) i )=|s1s2|+|s1s3|+|s2s3|, where s1, s2, and s3 are the stellar vectors of the three stars in the stellar combination corresponding to the particle, denoted as f()=o(), and record the ζ of each particle. i The position is P id Fitness f(P) id )=o(P id Compare all m particles ζ i The fitness is determined by taking the particle ζ with the minimum fitness. p Position P pd The optimal position corresponds to the global optimal fitness f(P). pd ); Based on the global optimal fitness f(P) pd ), where the particle ζ corresponds to the minimum fitness at this point. p Position = P pd The initial temperature T0 of the annealing algorithm is calculated to be f(P). pd ) / ln5; F SA (P id ) is a function that calculates the fitness of the annealing algorithm for the position P i of each particle ζ id at the current temperature T, and is calculated as follows: where exp() denotes the natural exponential function, ζ i denotes the current particle, P id denotes the position of the current particle; The idea of roulette wheel selection algorithm is applied, and the individual optimal position P of the rth particle meeting the preset optimal selection condition is selected rd Instead of the global optimal position P pd ; P pd is replaced by P rd , the particle swarm formula is substituted, and the velocity of each particle is updated as follows: where k denotes the iteration number, ci and c2 denote learning factors, ri and r2 denote two different random numbers between 0 and 1, and ω denotes a weight, denotes the velocity of the individual optimal particle at the k+1 iteration, denotes the velocity of the individual optimal particle at the k iteration, denotes the initial position of the particle ζ i denotes the individual optimal position at the k iteration;​ Recalculate the fitness of each particle, update the optimal position P of each particle rd and the population optimal position P pd ; performing an annealing operation, inputting a current temperature T, and outputting an initial temperature T0 of the next iteration: T0 = δ * T, wherein δ ∈ (0, 1) represents an annealing factor; determining whether the preset termination condition is met, if the preset termination condition is met, stopping the search, outputting a return value, and if the preset termination condition is not met, turning to the step of recording the position of each particle SA (P id ) is a function of calculating the annealing algorithm fitness of each particle ζ i at the position P id of the particle ζ at the current temperature T. wherein the preset termination condition is a judgment condition formula for satisfying the minimum value of the observability measure: min (|s1s2| + |s1s3| + |s2s3|) s.t. arccos(s w s u )≤ F0w, u = 1,2,3, w≠ u wherein s1, s2, and s3 are star vectors of the three stars in the star combination corresponding to the particle; F0 represents the field of view angle of the optical sensor; the return value comprises the target star combination; the roulette selection algorithm is used for selecting an operator, comprising: generating a random number pBet between 0 and 1; According to the annealing algorithm fitness F SA (P id ), the cumulative probability According to the cumulative probability, an individual optimal position P of the rth particle satisfying a preset optimal condition is selected rd Instead of the global optimal position P pd ; and the preset optimal condition is: comfit(r-1) 4. The deep space probe transfer phase starshot angular distance information optimization method of claim 1, wherein, The method for determining the optimal deep space probe transfer segment starlight angular distance information according to the star vectors of the three stars in the target star combination and the planet vector of the target planet comprises: determining the starlight angular distance of the nth star according to a starlight angular distance measurement model of the nth star: A n,τ = arccos(s0s n,τ )+ v n,τ , where s n,τ ,n = 1,2,3 represents the star vector of the nth star in the target star combination, τ represents the time, s n,τ =(cosα n,τ cosδ n,τ ,cosα n,τ sinδ n,τ ,sinα n,τ ) T ,α n,τ and δ n,τ represent the right ascension and declination of the nth star in the target star combination, A n,τ ,n = 1,2,3 represents the starlight angular distance composed of the starlight vector of the nth star in the target star combination and the planet vector of the target planet; s0 represents the direction vector between the planet and the probe, v n,τ represents the measurement noise, which is subject to Gaussian distribution and satisfies E(v n,τ ) = 0, According to n s a measurement model constituted by n pieces of starlight angular distance information to determine the starlight angular distances corresponding to three stars in a target star combination: where E(v τ ) = O, O represents a zero matrix, z τ represents a measurement information vector, represents n s starlight angular distances, n s represents the number of starlight angular distances, v τ represents a measurement noise, h(x τ ) represents a vector of n s starlight angular distances R τ represents a covariance matrix of the measurement noise; the measurement matrix H τ is the partial derivative matrix of h(x τ ) with respect to x τ is expressed as: wherein H l,τ The specific expression form is: wherein, where (x, y, z) represents a position vector of the probe, (x0, y0, z0) represents a position vector of the planet, δ l represents the declination of the lth star, a l represents the right ascension of the lth star.

5. An autonomous navigation method based on deep space probe transfer phase star angle distance information optimization, characterized in that, comprising: determining the star vector of each star, the planet vector of the target planet and the starlight angular distance information of the three stars in the target star combination by using the preferred method of deep space exploration transfer phase starlight angular distance information according to any one of claims 1-4; obtaining the solar pressure perturbation and planetary gravity perturbation data through an orbit vibration model; inputting the starlight angular distance of each star, the solar pressure perturbation and the planetary gravity perturbation data into an unscented Kalman filter algorithm to determine the running state of the probe.

6. A deep space probe transfer phase starshot angular distance information optimization apparatus, characterized in that, adopted by a probe with an optical sensor, the device comprising: an image acquisition unit for acquiring a deep space image through the optical sensor; the deep space image includes a plurality of stars and at least one planet; the number of the plurality of stars is greater than or equal to 3; a target planet determination unit for selecting the planet closest to the probe as the target planet from the at least one planet according to the deep space image and ephemeris; a vector determination unit for constructing an optical sensor measurement coordinate system for the acquired deep space image, and determining the star vector of each star in the plurality of stars and the planet vector of the target planet in the optical sensor measurement coordinate system; the optical sensor measurement coordinate system coincides with the body coordinate system of the probe; a target star combination determination unit for determining the target star combination from all star combinations according to the star vector of each star; each star combination is composed of any three stars from all stars, and the sum of the inner products between the star vectors of the three stars in the target star combination is the minimum value among all star combinations; the closer the inner product of the star vector is to 0, the higher the state estimation accuracy is; an analog particle swarm optimization algorithm is used to select the optimal star combination; a starlight angular distance determination unit for determining the optimal deep space exploration transfer phase starlight angular distance information according to the star vector of the three stars in the target star combination and the planet vector of the target planet.

7. The deep space probe transfer phase starshot angular distance information optimization apparatus of claim 6, wherein, The vector determination unit comprises: The optical sensor measurement coordinate system coincides with the body coordinate system of the probe; the origin of the body coordinate system of the probe is fixed on the center of mass of the probe, the X-axis of the body coordinate system of the probe points to the front of the probe, the Y-axis of the body coordinate system of the probe points to the right side of the X-axis of the body coordinate system of the probe, perpendicular to the X-axis of the body coordinate system of the probe, the Z-axis of the body coordinate system of the probe points to the bottom of the probe, perpendicular to the XY plane of the body coordinate system of the probe, and satisfies the right-hand rule; a vector determination module for determining the star vector of each star and the planet vector of the planet by applying the following formula to each star and planet respectively: where f represents the focal length of the optical sensor, (m x ,m y ) represents the coordinates of the target on the imaging plane of the optical sensor, said target comprising each star or planet.

8. The deep space probe transfer phase starshot angular distance information optimization apparatus of claim 6, wherein, The target star combination determination unit comprises: A particle swarm initialization module is configured to initialize a particle swarm, the particle swarm including m particles ζ i ; wherein i = 1, 2, …, m, each particle in the particle swarm represents a potential solution set including an initial position and velocity, each particle corresponds to a star combination, i.e. ζ i = (p i , v i ), wherein p i = {p i1 , p i2 , p i3}, p i1 , p i2 , p i3 are three star numbers in the particle respectively; a particle fitness determination module, configured to calculate a fitness function o(p i ) of each particle ζ i in the particle group, where s1, s2 and s3 are star vectors of three stars in the star combination corresponding to the particle, and f(p i ) = o(p i ) is recorded as the fitness of the particle ζ i , and the position of the particle ζ i is recorded as P id , and the fitness f(P id ) = o(P id ) is recorded, the fitness of all m particles ζ i is compared, the position P p of the particle ζ pd corresponding to the minimum fitness is taken as the best position, and the corresponding fitness is the global optimal fitness f(P pd ). An annealing initial temperature determining module is configured to determine an annealing initial temperature T0according to a global optimal fitness f(P pd ) corresponding to the particle ζ p with the smallest fitness at this time pd , and calculate the annealing initial temperature T0= f(P pd ) / ln5. Annealing fitness determination module, used to record F SA (P id The function is used to calculate the fitness of the annealing algorithm, given the ζ values ​​of each particle at the current temperature T. i Position P id The fitness calculation for the annealing algorithm is as follows: where exp() denotes the natural exponential function, ζ i denotes the current particle, P id denotes the position of the current particle; A global optimal updating module is configured to apply the idea of roulette selection algorithm to select an individual optimal position P of the rth particle that meets a preset optimal selection condition rd instead of the global optimal position P pd ; a particle velocity update module for updating the velocity of each particle in the particle group according to the following equation: pd replaced by P rd , and substituting into the particle group formula, the velocity of each particle is updated as follows: where k denotes the iteration number, ci and c2 denote learning factors, ri and r2 denote two different random numbers between 0 and 1, and ω denotes a weight, denotes the velocity of the individual optimal particle at the k+1 iteration, denotes the velocity of the individual optimal particle at the k iteration, denotes the initial position of the particle ζ i denotes the individual optimal position at the k iteration;​ a particle position updating module for recalculating the fitness of each particle and updating the optimal position P of each particle rd and the population optimal position P pd ; an annealing module for annealing operation, inputting the current temperature T and outputting the initial temperature T0 of the next iteration: T0 = δ * T, wherein δ ∈ (0, 1) represents the annealing factor; A termination judging module is configured to judge whether a preset termination condition is met. If the preset termination condition is met, the search is stopped and a return value is output. If the preset termination condition is not met, the annealing fitness determining module is triggered. The preset termination condition is a judging condition formula that the observability measure reaches a minimum value: min(|s1s2|+|s1s3|+|s2s3|) s.t. arccos(s w s u )≤ F0w, u = 1,2,3, w≠ u Wherein, s1, s2, s3 are star vectors of three stars in a target star combination corresponding to a particle; F0 represents a field of view angle of an optical sensor; The return value includes the target star combination; A roulette selection algorithm is used to select an operator, including: generating a random number pBet between 0 and 1; According to the annealing algorithm fitness F SA (P id ), the cumulative probability According to the cumulative probability, an individual optimal position P of the rth particle satisfying a preset optimal condition is selected rd Instead of the global optimal position P pd ; and the preset optimal condition is: comfit(r-1) 9. The deep space probe transfer phase starshot angular distance information optimization apparatus of claim 6, wherein, A starlight angular distance determining unit is configured to: determine the starlight angular distance of the nth star according to a starlight angular distance measurement model of the nth star: A n,τ = arccos(s0s n,τ )+ v n,τ , where s n,τ ,n = 1,2,3 represents the star vector of the nth star in the target star combination, τ represents the time, s n,τ = (cosα n,τ cosδ n,τ ,cosα n,τ sinδ n,τ ,sinα n,τ ) T ,α n,τ and δ n,τ represent the right ascension and declination of the nth star in the target star combination, A n,τ ,n = 1,2,3 represents the starlight angular distance composed of the starlight vector of the nth star in the target star combination and the planet vector of the target planet; s0 represents the direction vector between the planet and the probe, v n,τ represents the measurement noise, which is subject to Gaussian distribution and satisfies E(v n,τ ) = 0, According to n s a measurement model constituted by n pieces of starlight angular distance information determines the starlight angular distances corresponding to three stars in a target star combination: where E(v τ ) = 0, O represents a zero matrix, z τ represents a measurement information vector, represents n s starlight angular distances, n s represents the number of starlight angular distances, v τ represents measurement noise, h(x τ ) represents a vector composed of n s starlight angular distances, and represents a covariance matrix of measurement noise; a measurement matrix H τ is a partial derivative matrix of h(x τ ) with respect to x τ is represented as: wherein H l,τ is a specific expression of: wherein where (x, y, z) represents a position vector of the probe, (x0, y0, z0) represents a position vector of the planet, δ l represents the declination of the lth star, a l represents the right ascension of the lth star.

10. An autonomous navigation device based on deep space probe transfer phase star angle distance information, characterized in that, including: A deep space exploration transfer phase starlight angular distance information optimization device using the deep space exploration transfer phase starlight angular distance information optimization method according to any one of claims 1-4 is used to determine star vectors of three stars in a target star combination, a planet vector of a target planet, and starlight angular distance information corresponding to the three stars in the target star combination; An orbit vibration model unit is configured to obtain solar pressure perturbation and planetary gravity perturbation data; A running state determining unit is configured to input the starlight angular distance of each star, the solar pressure perturbation and the planetary gravity perturbation data into an unscented Kalman filtering algorithm to determine a running state of the probe; The running state includes three direction position vectors and velocity vectors of the probe.

Images