Space-time synchronization system method of heterogeneous giant star seat system

Abstract

The invention discloses a space-time synchronization system method of a heterogeneous giant satellite seat system, which realizes a processing mode of constellation whole network satellite grouping distribution and local each group satellite concentration around the core thought of whole network distribution, local concentration and dynamic node, and ensures the constellation system to run without faults through dynamic main node transfer. The method comprises the following steps: dividing all satellites in a constellation system into a plurality of groups, and appointing a master satellite and slave satellites; link establishment observation is carried out among each group of slave satellites, and distance and time synchronization observation quantity is transmitted to the master satellite through inter-satellite links; each master satellite calculates the position, the speed, the clock error, the frequency offset and the frequency drift of each slave satellite by adopting a centralized method according to the received observation data, and feeds back the position, the speed, the clock error, the frequency offset and the frequency drift to the slave satellites; when the slave satellite does not receive the feedback signal of the master satellite in a synchronization period, triggering a master satellite selection mechanism to realize dynamic master node transfer; all the main satellites are mutually linked for observation, and the optimal estimation values of the state, the clock correction and the clock drift of each main satellite are calculated by adopting a distributed method.

Description

Technical Field

[0001] The present invention relates to giant star constellation technology, and in particular to a space-time synchronization system method for a heterogeneous giant star constellation system. Background Art

[0002] A heterogeneous giant satellite constellation system is a constellation system composed of a large number of satellites of different types or functions, ranging from thousands to tens of thousands. These satellites may have various differences in orbital altitude, communication capabilities, sensor types, computing capabilities, etc. High-precision orbit determination and precise time synchronization are the key to ensuring the efficient operation of satellite constellation systems. However, when the number of satellites increases dramatically, there are many problems and technical difficulties to be studied in achieving time and space synchronization for a giant satellite system composed of tens of thousands of heterogeneous satellites.

[0003] The prior art CN202110997872.X discloses a system and method suitable for joint orbit determination of networked low-orbit satellites. The implementation idea of ​​the method is: first, obtaining observation data and measurement data; second, obtaining the predicted orbital information of each LEO constellation satellite based on the obtained observation data and measurement data; finally, obtaining the orbital information of the LEO constellation satellite based on the obtained measurement data and predicted orbital information.

[0004] This invention only considers low-orbit satellites and cannot be applied to the scenario of heterogeneous giant star constellations. In addition, this method adopts the method of joint processing of on-board distributed autonomous orbit determination and ground data center centralized orbit determination, which is not suitable for scenarios with a large number of satellites. As the number of satellites increases, the orbit determination accuracy of the on-board distributed method will decrease; at the same time, it puts forward additional requirements for ground equipment and cannot achieve fully autonomous orbit determination of the constellation.

[0005] The prior art CN201811509599.6 discloses a distributed autonomous orbit determination method using multi-star-target satellite angular distance fusion, and its implementation idea is: first, establish a multi-star-target satellite angular distance information preprocessing model; second, calculate the right ascension and declination of the observation satellite-target satellite line, and then use the angular distance information combined with the relative distance measurement between the constellation satellites as the observation quantity to establish the measurement equation; finally, combined with the constellation orbit dynamics recursion, a distributed method is used to complete the autonomous orbit determination of the entire satellite constellation.

[0006] This invention uses a distributed method to estimate the navigation parameters of the observation satellite and the target satellite to perform orbit determination for the entire constellation network. It is not suitable for constellation systems with a huge number of satellites. As the number of satellites increases, the orbit determination results obtained by the distributed method have poor accuracy. Summary of the invention

[0007] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method for space-time synchronization of a heterogeneous giant star constellation system, centering on the core idea of ​​"full network distribution, local concentration, and dynamic nodes", which solves the problem of poor orbit determination accuracy of the on-board fully distributed method as the number of satellites increases in the prior art.

[0008] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0009] A method for time-space synchronization of a heterogeneous giant star system is provided, which comprises the steps of:

[0010] S1. Based on the principle of grouping a preset number of adjacent satellites, all satellites in the heterogeneous Giant Star system are divided into several groups. In each group, one satellite is randomly selected as the master satellite, and the rest of the satellites are slave satellites.

[0011] S2. In each group, the slaves establish links with each other for observation, and transmit the calculated distance observations and time synchronization observations to the master through intersatellite links;

[0012] S3. Each master satellite calculates the position, velocity, clock error, frequency deviation and frequency drift of each slave satellite based on the distance observations and time synchronization observations received from all slave satellites using a centralized data processing method, and feeds back to the slave satellites through the intersatellite link;

[0013] S4. In a synchronization cycle, when a slave star fails to receive a feedback signal from the master star, the master star selection mechanism is triggered, and the slave star with the highest priority is selected as the master star to realize dynamic master node transfer;

[0014] S5. All the primary stars establish links with each other for observation, and use distributed data processing methods to calculate the optimal estimate of the state of each primary star and the optimal estimate of the clock error and clock drift.

[0015] The beneficial effects of the present invention are as follows: the present scheme revolves around the core idea of ​​"full network distribution, local concentration, and dynamic nodes", realizes a processing method in which satellites in the entire constellation network are grouped in a distributed manner and each group of satellites is centralized locally, and ensures trouble-free operation of the constellation system through dynamic master node transfer.

[0016] This scheme is aimed at a heterogeneous giant satellite constellation system composed of a large number of satellites of different types or functions. By grouping the heterogeneous giant satellite constellation system, the computational complexity of a single main satellite can be significantly reduced, so that the heterogeneous giant satellite constellation system can achieve time and space synchronization with lower computational complexity.

[0017] This scheme establishes a chain of observations between the slave stars in each group, obtains distance observations and time synchronization observations, and then uses the master star to calculate the data of the slave stars in its group, and then synchronizes the data between all master stars. This method first concentrates locally and then distributes the space-time synchronization across the entire network, which can achieve high-precision spatial position synchronization and time information synchronization of heterogeneous giant constellations.

[0018] During the data synchronization process, this solution takes into account the situation of master star failure. Through the election mechanism, the master star transfer mechanism can be dynamically implemented to ensure that the new centralized master star can take over all management functions to ensure the completion of the space-time synchronization task.

[0019] Furthermore, step S2 further comprises:

[0020] S21. From the mutual link observation between satellites, the two-way inter-satellite ranging observation equation is obtained:

[0021]

[0022] in, are the intersatellite observation quantities between satellites AB and BA respectively; t A ,t B are the nominal launch times of satellite A and B signals respectively; The intersatellite distances between satellites AB and BA at the time of sending and receiving each other; Δ AB , Δ BA are the signal propagation delays between satellites AB and BA respectively; C A , C B are the clock errors of satellites A and B, are the receiving and transmitting delays of satellite A respectively; are the receiving and transmitting delays of satellite B respectively; O AB , O BA are the observation corrections between satellites AB and BA, including antenna phase center correction, relativistic correction and satellite-ground troposphere correction; ε AB , ε BA are the measurement noise between satellites AB and BA respectively;

[0023] S22, preprocessing the two-way inter-satellite ranging observation equation to obtain the inter-satellite distance observation equation and the inter-satellite time synchronization observation equation:

[0024]

[0025] in, and They are distance observation and time synchronization observation respectively; After preprocessing, the distance between satellites AB and BA is 0The instantaneous pseudo-range observation at time; t 0 Theoretical value of the time-star distance; C A , C B are the instantaneous clock differences of satellites A and B respectively; are the residual errors between satellites AB and BA respectively.

[0026] The beneficial effect of the above technical solution is: by preprocessing the two-way inter-satellite ranging observation equation, the inter-satellite distance observation equation related to the spatial position and the inter-satellite time synchronization observation equation related to the clock information can be decoupled, and then the orbit parameters and clock parameters can be solved separately. This technology can reduce the latitude of the unknown number matrix, reduce the amount of calculation, and improve the calculation efficiency.

[0027] Furthermore, the method of calculating the position and velocity of each slave satellite using a centralized data processing method in step S3 includes:

[0028] S31. According to the prior orbit and dynamic parameters, the satellite dynamic equation and its corresponding variational equation are integrated by numerical integration method to obtain the reference orbit and state transfer matrix of the preset epoch; the expression of the satellite dynamic equation is:

[0029]

[0030] in, is the satellite’s acceleration vector; is the initial position vector of the satellite; is the initial velocity vector of the satellite; is the set of satellite dynamic model parameters; f(·) is a nonlinear function;

[0031] S32. Linearize the satellite spacing observation equation using the reference orbit to obtain the linearized observation equation:

[0032]

[0033] in, is the instantaneous star distance after preprocessing; is the instantaneous satellite distance calculated using the reference orbit; ρ AB is the intersatellite distance between satellites A and B; and are the state vectors of satellites A and B, respectively, which include position and velocity; and are the initial position vector, initial velocity vector and dynamic model parameters of satellite A and satellite B; and They are Correction value of and They are Correction value; dD A ,dD B are the intersatellite ranging bias corrections of A and B respectively; is the observation noise;

[0034] S33. According to the linearized observation equation, all the distance observations between the two slave stars in each group are combined to establish a centralized processing parameter estimation equation:

[0035] Y=HX+ε,Y=(ΔP 12 ΔP 1i … ΔP jn ) T

[0036]

[0037] X=(ΔX 1 ΔD 1 ΔX 2 ΔD 2 … ΔX n ΔD n ) T

[0038]

[0039] Where Y is the pseudorange difference observation vector; ΔP 12 , ΔP 1i and ΔP jn are the pseudo-range difference observation values ​​between satellites 1 and 2, satellites 1 and i, and satellites j and n, respectively; T is the transpose; H is the inter-satellite pseudo-range difference observation matrix, which has 2n columns and n(n-1) / 2 rows. Each row in H corresponds to an observation equation, and 1 and -1 in H are the coefficients of the clock error term; ρ 12 , 1i , jn and ρ ij are the inter-satellite pseudoranges from the 1st satellite to the 2nd satellite, the 1st satellite to the ith satellite, the jth satellite to the nth satellite, and the ith satellite to the jth satellite; is the partial derivative; X is the parameter matrix to be estimated; X 1 , X 2 , X i and X n are the parameters to be estimated for satellites 1, 2, i and n, including the position and velocity of the satellite; ΔD 1 , ΔD 2 and ΔD n are the clock corrections of satellites 1, 2 and n respectively; is the position vector of satellite i at the observation time; n is the total number of slave satellites;

[0040] S34, combining all the distance observations between the two slave stars at different times, and using the least squares estimation strategy to obtain the parameter to be estimated X:

[0041] X=(H T PH) ―1 HY

[0042] Among them, P is the weight matrix in the least squares estimation formula.

[0043] The beneficial effects of the above technical solution are as follows: the centralized processing method using the least squares strategy to calculate the position and velocity of each group of satellites can simultaneously process the observation data of all the slave satellites in the group, and can obtain the optimal solution for the state of the group of satellites, thereby improving the accuracy of the overall orbit determination; at the same time, the comprehensive use of all the observation data can improve the reliability of the estimation of the position and velocity parameters of the group of satellites.

[0044] Furthermore, the method of using a centralized data processing method to calculate the clock error, frequency deviation and frequency drift of each slave satellite in step S3 includes:

[0045] A1. Based on the inter-satellite time synchronization observation equation and the satellite reception and transmission delay, the observation equation for estimating the satellite clock error, frequency offset and frequency drift is constructed:

[0046]

[0047] in, is the estimated value of intersatellite time synchronization observation; a A and a B , b A and b B and c A and c B are the clock error, frequency deviation and frequency drift of satellite A and satellite B respectively; t 1 is the observation time after epoch normalization; is the reference epoch;

[0048] A2. Based on the observation equations for estimating satellite clock error, frequency bias and frequency drift, a combined observation equation is constructed:

[0049]

[0050] in, is the time-synchronized observation vector; are the observed values ​​of clock difference, frequency offset and frequency drift information between satellites 1 and 2, satellites 1 and i, satellites j and n, and satellites n-1 and n, respectively; T is transposed; is the design matrix of the observation equation, Δt=t 1―t 0 is the time difference, The 1 or -1 in represents the coefficient of the clock error term; is the combined matrix of clock error, frequency offset and frequency drift of all slave satellites in each group; a 1 ,

[0051] a 2 、a n―1 and a n are the clock errors of satellites 1, 2, n-1 and n respectively; b 1 , b 2 , b n―1 and b n are the frequency deviations of satellites 1, 2, n-1 and n respectively; c 1 、c 2 、c n―1 and c n are the frequency drifts of satellites 1, 2, n-1 and n respectively; A3, using the clock error constraint method, construct the observation vector to eliminate Virtual observation equation for medium rank deficiency:

[0052]

[0053] in, are the prior values ​​of the clock error of satellite j; are the accuracy of satellite j; a j , b j 、c j For The clock parameters of the j-th satellite extracted from ;

[0054] A4. Combined observation vector And the virtual observation equation, we get the combined relationship:

[0055]

[0056] in, is the combined clock information observation; is the combined observation matrix; is the residual error after combination; is a row vector;

[0057] A5. Use the least squares estimation strategy to solve the optimal estimate of the combined relationship:

[0058]

[0059] Among them, the best estimate Contains the clock error, frequency deviation and frequency drift of each slave satellite.

[0060] The beneficial effects of the above technical solution are: by modeling the clock with polynomials, the three parameters of the clock error, frequency deviation and frequency drift of the slave satellite can be solved at the same time. The centralized processing method of each group of satellites using the least squares strategy to solve the clock error, frequency deviation and frequency drift can simultaneously process the observation data of all the slave satellites in the group, and the optimal solution of the clock of the group of satellites can be obtained, thereby improving the accuracy of the overall time synchronization; at the same time, the comprehensive use of all the observation data can improve the reliability of the estimation of the three parameters of the group of satellites.

[0061] Furthermore, the method of calculating the optimal estimated value of the state of each primary star using the distributed data processing method includes:

[0062] S51, the extended Kalman filter state transfer equation is:

[0063] x k+1 =f(x k ,p,t)+w

[0064] Among them, x k+1 is the real state vector of the primary star at time k+1; x k is the true state vector of the primary star at time k; p is the covariance matrix; t is the time of primary star motion; w is the system process noise; f(·) is a nonlinear function;

[0065] S52, the time update calculation formula of the extended Kalman filter is:

[0066]

[0067] in, is the estimated value of the primary star state predicted at time k+1; is the current state vector of the primary star at time k; Φ is the state transfer matrix, Find the partial derivative of the f(·) function with respect to the true state vector of the primary star at time k, is the prediction error covariance matrix at time k+1, is the error covariance matrix at time k, Φ T is the transpose of the state transfer matrix, Q is the system process noise covariance matrix;

[0068] S53, the observation update calculation formula of the extended Kalman filter is used to obtain the optimal estimated value of the state of each primary star:

[0069]

[0070] in, is the optimal estimate of the state of the primary star at time k+1, H a is the observation matrix of the extended Kalman filter, H a TFor H a Find the transpose; R is the observation noise covariance matrix; is the inter-satellite residual Δρ of all the primary stars that can be linked in the current epoch ij The vector formed, is the optimal estimation error covariance matrix at time k+1.

[0071] The beneficial effects of the above technical solution are: the extended Kalman filter algorithm is used for distributed processing to solve the optimal estimated value of the state between several master satellites in the whole network, and the calculation tasks can be distributed to multiple master satellites, reducing the calculation burden of each master satellite and improving the calculation efficiency; at the same time, the distributed algorithm can reduce processing delays, improve the overall calculation speed of the system, and meet the needs of real-time orbit determination.

[0072] Furthermore, the residual error Δρ between satellites ij The methods for obtaining include:

[0073] S531. Construct the intersatellite two-way observation equation between the primary satellites i and j as follows:

[0074]

[0075] Among them, ρ ij , ij is the pseudo-range observation quantity of primary star i receiving primary star j and primary star j receiving primary star i; c is the speed of light; t i ,t j are the clock errors of primary stars i and j respectively; ε ij , ε ij is the observation noise from primary star i to primary star j and from primary star j to primary star i; is the theoretical inter-satellite distance between the main stars i and j, expressed as:

[0076]

[0077] Among them, x i 、x j are the position components of the primary stars i and j in the x-axis direction; y i ,y j are the position components of the primary stars i and j in the y-axis direction; z i 、z j are the position components of primary stars i and j in the z-axis direction respectively;

[0078] S532. Use the reference orbit linearization to obtain the linear observation equation without satellite clock error:

[0079]

[0080] Among them, ρ is the linear observation without satellite clock error; is the approximate satellite-to-earth distance calculated using the reference orbit; are the correction values ​​of the positions of the primary stars i and j relative to the reference orbit; L m is the star spacing unit vector, which is in the form of ε′ is the observation noise;

[0081] S533. Based on the linear observation equation without satellite clock error, the Kalman filter observation equation is obtained as follows:

[0082]

[0083] Among them, Δρ ij is the residual error of the inter-satellite distance.

[0084] The beneficial effect of the above technical solution is that by processing the inter-satellite two-way observation equation, a linearized observation equation can be obtained, and the expression form of the inter-satellite distance residual related to the Kalman filter algorithm can be obtained, which is convenient for introducing into the extended Kalman filter to calculate the parameters to be estimated.

[0085] Furthermore, the method of calculating the optimal estimated value of the clock error and clock drift of each primary star by using the distributed data processing method includes:

[0086] B1. Construct a linear model containing two states to represent the satellite clock error:

[0087]

[0088] Among them, b k ,d k is the clock error and clock drift of the primary star at time k; b k+1 ,d k+1 is the clock error and clock drift of the primary star at time k+1; dt is the sampling interval; ω b ,ω d They are white FM noise and random walk FM noise respectively;

[0089] B2. The time update calculation formula of the extended Kalman filter is:

[0090]

[0091] in, is the satellite clock estimate predicted at time k, including clock error and clock drift; φ k,k―1 The state transition matrix solved for the clock parameters; φ T k,k―1 For φ k,k―1 Find the transpose; is the satellite clock parameter at time k-1; is the clock error covariance matrix predicted at time k; is the error covariance matrix at time k-1; Q k―1 is the covariance matrix of white FM noise and random walk FM noise at time k-1;

[0092] B3. The observation update calculation formula of the extended Kalman filter is:

[0093]

[0094] Among them, K k is the Kalman gain matrix at time k, is the observation matrix, For Find the transpose; R k is the clock error covariance matrix at time k; (·) ―1 To find the inverse operation; is the optimal estimate of the satellite clock error and clock drift at time k; Z k―1 is the true measured value of the satellite clock error and clock drift at time k-1; is the optimal estimation error covariance matrix at time k, and I is the unit matrix.

[0095] The beneficial effects of the above technical solution are: the extended Kalman filter algorithm is used for distributed processing to solve the optimal estimated values ​​of the clock difference and clock drift between several master stars in the whole network, and the calculation tasks can be distributed to multiple master stars, reducing the calculation burden of each master star and improving the calculation efficiency; at the same time, the distributed algorithm can reduce the time synchronization delay, quickly respond to clock changes, and meet the needs of implementing time synchronization.

[0096] Furthermore, the master star selection mechanism is triggered to select the slave star with the highest priority as the master star. The method for realizing dynamic master node transfer includes:

[0097] S41, modify each slave star in the group to be a candidate state;

[0098] S42, each candidate state slave star sends its own campaign information to other candidate state slave stars around it to form a campaign information set; the campaign information includes communication capability, computing capability, orbit error and clock error;

[0099] S43, the slave star initiating the election is regarded as the first election star, any slave star receiving the election information is regarded as the second election star, and the election information of the second election star is added to the election information set;

[0100] S44, according to the election information of the first election star and the second election star, selecting the one with a higher priority as the new first election star;

[0101] S45, judging whether there is a candidate slave star that has not participated in the comparison in the group, if so, proceeding to step S46, otherwise, taking the first candidate star selected most recently as the master star, and proceeding to step S47;

[0102] S46, the first candidate star sends its own election information to other candidate slave stars around it, takes any slave star that receives the election information as the second candidate star, and stores the election information of the first candidate star and the second candidate star in a new election information set, and returns to step S44;

[0103] S47, the selected master satellite broadcasts the information of its successful selection to all slave satellites in the group, and other slave satellites confirm and synchronize their status.

[0104] The beneficial effects of the above technical solution are: the main star selection mechanism ensures that when the original main star fails, the main star transfer can be realized dynamically, ensuring that the new centralized main star can take over all management functions to ensure the completion of the space-time synchronization task; and this technology can select a new main star through direct comparison of election information, and there will be no election failure.

[0105] Furthermore, step S44 further includes:

[0106] There are four situations when determining the priority of the first campaign star and the second campaign star:

[0107] First, the orbit error of the first candidate star is smaller than the orbit error of the second candidate star, and the clock error of the first candidate star is smaller than the clock error of the second candidate star;

[0108] Second, the orbit error of the first candidate star is smaller than the orbit error of the second candidate star, and the clock error of the second candidate star is smaller than the clock error of the first candidate star;

[0109] Third, the orbit error of the second candidate star is smaller than that of the first candidate star, and the clock error of the first candidate star is smaller than that of the second candidate star;

[0110] Fourth, the orbit error of the second candidate satellite is smaller than that of the first candidate satellite, and the clock error of the second candidate satellite is smaller than that of the first candidate satellite;

[0111] When the first, second and third conditions are met, the first campaign star is considered to have a higher priority; when the fourth condition is met, the second campaign star is considered to have a higher priority.

[0112] Furthermore, the method for calculating the preset number includes:

[0113] Starting from 1, the number of main stars is selected in ascending order, and the complexity of the selected number of main stars is calculated using the complexity model. The expression of the complexity model is:

[0114] O = (3*N / k) 3

[0115] Among them, O is the complexity; N is the total number of satellites in the entire Giant Star constellation network; k is the number of main satellites;

[0116] When the difference between the complexities corresponding to two adjacent numbers of primary stars selected is greater than a preset threshold, the larger value of the two numbers of primary stars is selected as the preset number.

[0117] The beneficial effect of the above technical solution is: this solution deduces that the computational complexity is only related to the complexity of the matrix inversion operation through the process of solving the clock error. Therefore, when selecting the main satellite, a model is constructed based on the computational complexity and the total number of satellites in the entire network and the number of main satellites. The optimal number of main satellites can be quickly selected through the computational complexity corresponding to two adjacent values, thereby ensuring the efficiency of space-time synchronization. BRIEF DESCRIPTION OF THE DRAWINGS

[0118] Figure 1 Flow chart of the space-time synchronization system method for the heterogeneous giant star system.

[0119] Figure 2 This is a simulation diagram of the single-star computing complexity (taking N=10000 as an example). DETAILED DESCRIPTION

[0120] The specific implementation modes of the present invention are described below so that those skilled in the art can understand the present invention. However, it should be clear that the present invention is not limited to the scope of the specific implementation modes. For those of ordinary skill in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the attached claims, these changes are obvious, and all inventions and creations utilizing the concept of the present invention are protected.

[0121] refer to Figure 1 , Figure 1 A method for time-space synchronization of a heterogeneous giant star system is shown; Figure 1 As shown, the method S includes steps S1 to S5.

[0122] In step S1, all satellites in the heterogeneous Giant Star system are divided into several groups based on the principle that a preset number of adjacent satellites are grouped together. In each group, one satellite is randomly selected as the master satellite, and the remaining satellites are slave satellites.

[0123] In one embodiment of the present invention, the method for calculating the preset number includes:

[0124] Starting from 1, the number of main stars is selected in ascending order, and the complexity of the selected number of main stars is calculated using the complexity model. The expression of the complexity model is:

[0125] O=(3*N / k z ) 3

[0126] Among them, O is the complexity; N is the total number of satellites in the Giant Star constellation network; k z is the number of main stars;

[0127] When the difference between the complexities corresponding to two adjacent numbers of primary stars selected is greater than a preset threshold, the larger value of the two numbers of primary stars is selected as the preset number.

[0128] like Figure 2 As shown in the figure, the curve depicts the change in the computational complexity of each main satellite as the number of main satellites gradually increases. Here, the vertical axis data is logarithmically processed. Figure 2 The simulation results show that in the centralized processing method, as the number of primary satellites increases, when the total number of satellites in the Giant Star system is 10,000, when k z =100, the computational complexity of a single primary star will be significantly reduced.

[0129] To this end, in this scheme, the heterogeneous Giant Star constellation system is grouped according to the principle of dividing every 100 adjacent satellites into a group, so that time and space synchronization can be achieved with lower computational complexity.

[0130] In step S2, in each group, the slave satellites establish links with each other for observation, and transmit the calculated distance observations and time synchronization observations to the master satellite through the intersatellite link.

[0131] During implementation, the preferred step S2 of this solution further includes:

[0132] S21. From the mutual link observation between satellites, the two-way inter-satellite ranging observation equation is obtained:

[0133]

[0134] in, are the intersatellite observation quantities between satellites AB and BA respectively; t A ,t B are the nominal launch times of satellite A and B signals respectively; The intersatellite distances between satellites AB and BA at the time of sending and receiving each other; Δ AB , Δ BA are the signal propagation delays between satellites AB and BA respectively; C A , C B are the clock errors of satellites A and B, are the receiving and transmitting delays of satellite A respectively; are the receiving and transmitting delays of satellite B respectively; O AB , OBA are the observation corrections between satellites AB and BA, including antenna phase center correction, relativistic correction and satellite-ground troposphere correction; ε AB , ε BA are the measurement noise between satellites AB and BA respectively;

[0135] S22, preprocessing the two-way inter-satellite ranging observation equation to obtain the inter-satellite distance observation equation and the inter-satellite time synchronization observation equation:

[0136]

[0137] in, and They are distance observation and time synchronization observation respectively; After preprocessing, the distance between satellites AB and BA is 0 The instantaneous pseudo-range observation at time; t 0 Theoretical value of the time-star distance; C A , C B are the instantaneous clock differences of satellites A and B respectively; are the residual errors between satellites AB and BA respectively.

[0138] In step S3, each master satellite calculates the position, velocity, clock error, frequency deviation and frequency drift of each slave satellite based on the received distance observations and time synchronization observations of all slave satellites using a centralized data processing method, and feeds back to the slave satellites through the inter-satellite link.

[0139] In one embodiment of the present invention, the method of calculating the position and velocity of each slave satellite using a centralized data processing method in step S3 includes:

[0140] S31. According to the prior orbit and dynamic parameters, the satellite dynamic equation and its corresponding variational equation are integrated by numerical integration method to obtain the reference orbit and state transfer matrix of the preset epoch; the expression of the satellite dynamic equation is:

[0141]

[0142] in, is the satellite’s acceleration vector; is the initial position vector of the satellite; is the initial velocity vector of the satellite; is the set of satellite dynamic model parameters; f(·) is a nonlinear function;

[0143] S32. Linearize the satellite spacing observation equation using the reference orbit to obtain the linearized observation equation:

[0144]

[0145] in, is the instantaneous star distance after preprocessing; is the instantaneous satellite distance calculated using the reference orbit; ρ AB is the intersatellite distance between satellites A and B; and are the state vectors of satellites A and B, respectively, which include position and velocity; and are the initial position vector, initial velocity vector and dynamic model parameters of satellite A and satellite B; and They are Correction value of and They are Correction value; dD A ,dD B are the intersatellite ranging bias corrections of A and B respectively; is the observation noise;

[0146] S33. According to the linearized observation equation, all the distance observations between the two slave stars in each group are combined to establish a centralized processing parameter estimation equation:

[0147] Y=HX+ε,Y=(ΔP 12 ΔP 1i … ΔP jn ) T

[0148]

[0149] X=(ΔX 1 ΔD 1 ΔX 2 ΔD 2 … ΔX n ΔD n ) T

[0150]

[0151] Where Y is the pseudorange difference observation vector; ΔP 12 , ΔP 1i and ΔP jn are the pseudo-range difference observation values ​​between satellites 1 and 2, satellites 1 and i, and satellites j and n, respectively; T is the transpose; H is the inter-satellite pseudo-range difference observation matrix, which has 2n columns and n(n-1) / 2 rows. Each row in H corresponds to an observation equation, and 1 and -1 in H are the coefficients of the clock error term; ρ 12 ,1i , jn and ρ ij are the inter-satellite pseudoranges from the 1st satellite to the 2nd satellite, the 1st satellite to the ith satellite, the jth satellite to the nth satellite, and the ith satellite to the jth satellite; is the partial derivative; X is the parameter matrix to be estimated; X 1 , X 2 , X i and X n are the parameters to be estimated for satellites 1, 2, i and n, including the position and velocity of the satellite; ΔD 1 , ΔD 2 and ΔD n are the clock corrections of satellites 1, 2 and n respectively; is the position vector of satellite i at the observation time; n is the total number of slave satellites;

[0152] The function of the matrix H is to establish a linear relationship between the observed values ​​and the parameters to be estimated. To facilitate the understanding of H, we take three satellites as an example (there will be three groups of observations), and the expression of H is:

[0153]

[0154] Among them, the first line is the observation equation between satellites 1 and 2; the second line is the observation equation between satellites 1 and 3, and the third line is the observation equation between satellites 2 and 3.

[0155] Taking 4 satellites as an example (there will be 6 sets of observations), the expression of H is:

[0156]

[0157] Among them, the first line is the observation equation between satellites 1 and 2; the second line is the observation equation between satellites 1 and 3, and the third line is the observation equation between satellites 1 and 4; the fourth line is the observation equation between satellites 2 and 3, the fifth line is the observation equation between satellites 2 and 4, and the sixth line is the observation equation between satellites 3 and 4.

[0158] In H, there are only 4 non-zero elements in each row (2 partial derivatives, 2 ±1), and the rest of the elements are 0; the determination of positive and negative is explained: the determination of ± is related to the pseudo-range differential observation ΔP in Y ij The definition of ΔP is related to: ij It represents the pseudo-range differential observation value from satellite i to satellite j. The clock difference coefficient of satellite i with the number written in front is 1, and the clock difference coefficient of satellite j with the number written in the back is -1.

[0159] Based on the matrix H corresponding to the above 3 satellites and 4 satellites, it can be known that when there are n satellites, the rows of the matrix H are, from top to bottom, the observation equations of satellite 1 and satellites 2 to n, the observation equations of satellite 2 and satellites 3 to n, the observation equations of satellite 3 and satellites 4 to n, ..., the observation equations of satellite n-1 and satellite n.

[0160] S34, combining all the distance observations between the two slave stars at different times, and using the least squares estimation strategy to obtain the parameter to be estimated X:

[0161] X=(H T PH) ―1 HY

[0162] Among them, P is the weight matrix in the least squares estimation formula.

[0163] During implementation, the method of using a centralized data processing method to calculate the clock error, frequency deviation and frequency drift of each slave satellite in step S3 of this solution preferably includes:

[0164] A1. Based on the inter-satellite time synchronization observation equation and the satellite reception and transmission delay, the observation equation for estimating the satellite clock error, frequency offset and frequency drift is constructed:

[0165]

[0166] in, is the estimated value of intersatellite time synchronization observation; a A and a B , b A and b B and c A and c B are the clock error, frequency deviation and frequency drift of satellite A and satellite B respectively; Δ 1 is the observation time after epoch normalization; is the reference epoch;

[0167] A2. Based on the observation equations for estimating satellite clock error, frequency bias and frequency drift, a combined observation equation is constructed:

[0168]

[0169]

[0170] in, is the time-synchronized observation vector; are the observed values ​​of clock difference, frequency offset and frequency drift information between satellites 1 and 2, satellites 1 and i, satellites j and n, and satellites n-1 and n, respectively; T is transposed; is the design matrix of the observation equation, Δt=t 1 ―t 0 is the time difference, The 1 or -1 in the equation represents the coefficient of the clock error term, Δt, Δt 2 are the coefficients of the clock error, frequency bias and frequency drift of the corresponding satellite in the observation equation respectively; is the combined matrix of clock error, frequency offset and frequency drift of all slave satellites in each group; a 1 、a 2 、a n―1 and a n are the clock errors of satellites 1, 2, n-1 and n respectively; b 1 , b 2 , b n―1 and b n are the frequency deviations of satellites 1, 2, n-1 and n respectively; c 1 、c 2 、c n―1 and c n are the frequency drifts of satellites 1, 2, n-1 and n respectively;

[0171] for Matrix, taking 4 satellites as an example, its expression is:

[0172]

[0173] exist In the equation, columns 1 to 3 correspond to the clock error, frequency deviation, and frequency drift coefficients of satellite 1, columns 4 to 6 correspond to the clock error, frequency deviation, and frequency drift coefficients of satellite 2, columns 7 to 9 correspond to the clock error, frequency deviation, and frequency drift coefficients of satellite 3, columns 10 to 12 correspond to the clock error, frequency deviation, and frequency drift coefficients of satellite 4, and so on. When there are n satellites, then The matrix consists of 3n columns in total, and each column consists of n(n-1) / 2 rows.

[0174] exist In the equation, the first line is the observation equation between satellites 1 and 2, the second line is the observation equation between satellites 1 and 3, and so on. The definition logic of the positive and negative signs in is the same as the definition logic of the matrix H in S33. The first satellite is positive and the second satellite is negative.

[0175] A3. Use the clock error constraint method to construct the observation matrix to eliminate the observation Virtual observation equation for medium rank deficiency:

[0176]

[0177] in, are the prior values ​​of the clock error of satellite j; are the accuracy of satellite j; a j , b j 、c j For The clock parameters of the j-th satellite extracted from ;

[0178] A4. Combined observation vector And the virtual observation equation, we get the combined relationship:

[0179]

[0180] in, is the combined clock information observation; is the combined observation matrix; is the residual error after combination; is a row vector, in which the three consecutive elements corresponding to the extracted j-th satellite are 1, and the remaining elements are 0.

[0181] is used from Extract the clock parameters of the jth satellite (a j ; b j ;c j ), The three consecutive 1s in correspond to the positions of the three clock parameters to be extracted (i.e., the positions of the corresponding j-th satellite parameters), and the other positions are all 0.

[0182] A5. Use the least squares estimation strategy to solve the optimal estimate of the combined relationship:

[0183]

[0184] Among them, the best estimate Contains the clock error, frequency deviation and frequency drift of each slave satellite.

[0185] In step S4, within a synchronization cycle, when a slave star fails to receive a feedback signal from the master star, the master star selection mechanism is triggered, and the slave star with the highest priority is selected as the master star to realize dynamic master node transfer.

[0186] During implementation, this solution will first trigger the master star selection mechanism and select the slave star with the highest priority as the master star. The methods for implementing dynamic master node transfer include:

[0187] S41, modify each slave star in the group to be a candidate state;

[0188] S42, each candidate state slave star sends its own campaign information to other candidate state slave stars around it to form a campaign information set; the campaign information includes communication capability, computing capability, orbit error and clock error;

[0189] S43, the slave star initiating the election is regarded as the first election star, any slave star receiving the election information is regarded as the second election star, and the election information of the second election star is added to the election information set;

[0190] S44, according to the election information of the first election star and the second election star, selecting the one with a higher priority as the new first election star;

[0191] S45, judging whether there is a candidate slave star that has not participated in the comparison in the group, if so, proceeding to step S46, otherwise, taking the first candidate star selected most recently as the master star, and proceeding to step S47;

[0192] S46, the first candidate star sends its own election information to other candidate slave stars around it, takes any slave star that receives the election information as the second candidate star, and stores the election information of the first candidate star and the second candidate star in a new election information set, and returns to step S44;

[0193] S47, the selected master satellite broadcasts the information of its successful selection to all slave satellites in the group, and other slave satellites confirm and synchronize their status.

[0194] In this plan, there are four situations when determining the priority of the first campaign star and the second campaign star:

[0195] First, the orbit error of the first candidate star is smaller than the orbit error of the second candidate star, and the clock error of the first candidate star is smaller than the clock error of the second candidate star;

[0196] Second, the orbit error of the first candidate star is smaller than the orbit error of the second candidate star, and the clock error of the second candidate star is smaller than the clock error of the first candidate star;

[0197] Third, the orbit error of the second candidate star is smaller than that of the first candidate star, and the clock error of the first candidate star is smaller than that of the second candidate star;

[0198] Fourth, the orbit error of the second candidate satellite is smaller than that of the first candidate satellite, and the clock error of the second candidate satellite is smaller than that of the first candidate satellite;

[0199] When the first, second and third conditions are met, the first campaign star is considered to have a higher priority; when the fourth condition is met, the second campaign star is considered to have a higher priority.

[0200] In step S5, all the primary satellites establish links with each other for observation, and use a distributed data processing method to calculate the optimal estimate of the state of each primary satellite and the optimal estimate of the clock error and clock drift.

[0201] In one embodiment of the present invention, a method for calculating the optimal estimated value of the state of each primary star using a distributed data processing method includes:

[0202] S51, the extended Kalman filter state transfer equation is:

[0203] xk+1 =f(x k ,p,t)+w

[0204] Among them, x k+1 is the real state vector of the primary star at time k+1; x k is the true state vector of the primary star at time k; p is the covariance matrix; t is the time of primary star motion; w is the system process noise; f(·) is a nonlinear function;

[0205] S52, the time update calculation formula of the extended Kalman filter is:

[0206]

[0207] in, is the estimated value of the primary star state predicted at time k+1; is the current state vector of the primary star at time k; Φ is the state transfer matrix, Find the partial derivative of the f(·) function with respect to the true state vector of the primary star at time k, is the prediction error covariance matrix at time k+1, is the error covariance matrix at time k, Φ T is the transpose of the state transfer matrix, Q is the system process noise covariance matrix;

[0208] S53, the observation update calculation formula of the extended Kalman filter is used to obtain the optimal estimated value of the state of each primary star:

[0209]

[0210] in, is the optimal estimate of the state of the primary star at time k+1, H a is the observation matrix of the extended Kalman filter, H a T For H a Find the transpose; R is the observation noise covariance matrix; is the inter-satellite residual Δρ of all the primary stars that can be linked in the current epoch ij The vector formed, is the optimal estimation error covariance matrix at time k+1.

[0211] When implemented, the residual error Δρ of the preferred star spacing of this scheme is ij The methods for obtaining include:

[0212] S531. Construct the intersatellite two-way observation equation between the primary satellites i and j as follows:

[0213]

[0214] Among them, ρij , ji is the pseudo-range observation quantity of primary star i receiving primary star j and primary star j receiving primary star i; c is the speed of light; t i ,t j are the clock errors of primary stars i and j respectively; ε ij , ε ij is the observation noise from primary star i to primary star j and from primary star j to primary star i; is the theoretical inter-satellite distance between the main stars i and j, expressed as:

[0215]

[0216] Among them, x i 、x j are the position components of the primary stars i and j in the x-axis direction; y i ,y j are the position components of the primary stars i and j in the y-axis direction; z i 、z j are the position components of primary stars i and j in the z-axis direction respectively;

[0217] S532. Use the reference orbit linearization to obtain the linear observation equation without satellite clock error:

[0218]

[0219] Among them, ρ is the linear observation without satellite clock error; is the approximate satellite-to-earth distance calculated using the reference orbit; are the correction values ​​of the positions of the primary stars i and j relative to the reference orbit; L m is the star spacing unit vector, which is in the form of ε′ is the observation noise;

[0220] S533. Based on the linear observation equation without satellite clock error, the Kalman filter observation equation is obtained as follows:

[0221]

[0222] Among them, Δρ ij is the residual error of the inter-satellite distance.

[0223] The method of calculating the optimal estimated value of the clock error and clock drift of each primary star using the distributed data processing method includes:

[0224] B1. Construct a linear model containing two states to represent the satellite clock error:

[0225]

[0226] Among them, b k ,d kis the clock error and clock drift of the primary star at time k; b k+1 ,d k+1 is the clock error and clock drift of the primary star at time k+1; dt is the sampling interval; ω b ,ω d They are white FM noise and random walk FM noise respectively;

[0227] B2. The time update calculation formula of the extended Kalman filter is:

[0228]

[0229] in, is the satellite clock estimate predicted at time k, including clock error and clock drift; φ k,k―1 The state transition matrix solved for the clock parameters; φ T k,k―1 For φ k,k―1 Find the transpose; is the satellite clock parameter at time k-1; is the clock error covariance matrix predicted at time k; is the error covariance matrix at time k-1; Q k―1 is the covariance matrix of white FM noise and random walk FM noise at time k-1;

[0230] B3. The observation update calculation formula of the extended Kalman filter is:

[0231]

[0232] Among them, K k is the Kalman gain matrix at time k, is the observation matrix, For Find the transpose; R k is the clock error covariance matrix at time k; (·) -1 To find the inverse operation; is the optimal estimate of the satellite clock error and clock drift at time k; Z k―1 is the true measured value of the satellite clock error and clock drift at time k-1; is the optimal estimation error covariance matrix at time k, and I is the unit matrix.

[0233] In summary, this scheme can achieve high-precision satellite constellation spatial position synchronization and time information synchronization by grouping heterogeneous giant star constellation systems and then combining local concentration with full-network distributed space-time synchronization.

Claims

1. A method for time-space synchronization of a heterogeneous giant star system, characterized in that: Includes steps: S1. Based on the principle of grouping a preset number of adjacent satellites, all satellites in the heterogeneous Giant Star system are divided into several groups. In each group, one satellite is randomly selected as the master satellite, and the rest of the satellites are slave satellites. S2. In each group, the slaves establish links with each other for observation, and transmit the calculated distance observations and time synchronization observations to the master through intersatellite links; S3. Each master satellite calculates the position, velocity, clock error, frequency deviation and frequency drift of each slave satellite based on the distance observations and time synchronization observations received from all slave satellites using a centralized data processing method, and feeds back to the slave satellites through the intersatellite link; S4. In a synchronization cycle, when a slave star fails to receive a feedback signal from the master star, the master star selection mechanism is triggered, and the slave star with the highest priority is selected as the master star to realize dynamic master node transfer; S5. All the primary stars establish links with each other for observation, and use distributed data processing methods to calculate the optimal estimate of the state of each primary star and the optimal estimate of the clock error and clock drift.

2. The method for time-space synchronization of a heterogeneous giant star system according to claim 1, characterized in that: Step S2 further comprises: S21. From the mutual link observation between satellites, the two-way inter-satellite ranging observation equation is obtained: in, are the intersatellite observation quantities between satellites AB and BA respectively; t A ,t B are the nominal launch times of satellite A and B signals respectively; The intersatellite distances between satellites AB and BA at the time of sending and receiving each other; Δ AB , Δ BA are the signal propagation delays between satellites AB and BA respectively; C A , C B are the clock errors of satellites A and B, are the receiving and transmitting delays of satellite A respectively; are the receiving and transmitting delays of satellite B respectively; O AB , O BA are the observation corrections between satellites AB and BA, including antenna phase center correction, relativistic correction and satellite-ground troposphere correction; ε AB , ε BA are the measurement noise between satellites AB and BA respectively; S22, preprocessing the two-way inter-satellite ranging observation equation to obtain the inter-satellite distance observation equation and the inter-satellite time synchronization observation equation: in, and They are distance observation and time synchronization observation respectively; is the instantaneous pseudo-range observation between satellites AB and BA at time t0 after preprocessing; is the theoretical value of the inter-satellite distance at time t0; C A , C B are the instantaneous clock differences of satellites A and B respectively; are the residual errors between satellites AB and BA respectively.

3. The method for time-space synchronization of a heterogeneous giant star system according to claim 2, characterized in that: The method of using a centralized data processing method to calculate the position and speed of each slave satellite in step S3 includes: S31. According to the prior orbit and dynamic parameters, the satellite dynamic equation and its corresponding variational equation are integrated by numerical integration method to obtain the reference orbit and state transfer matrix of the preset epoch; S32. Linearize the satellite spacing observation equation using the reference orbit to obtain the linearized observation equation: in, is the instantaneous star distance after preprocessing; is the instantaneous satellite distance calculated using the reference orbit; ρ AB is the intersatellite distance between satellites A and B; and are the state vectors of satellites A and B, including position and velocity; and are the initial position vector, initial velocity vector and dynamic model parameters of satellite A and satellite B; and They are Correction value of and They are Correction value; dD A ,dD B are the intersatellite ranging bias corrections of A and B respectively; is the observation noise; S33. According to the linearized observation equation, all the distance observations between the two slave stars in each group are combined to establish a centralized processing parameter estimation equation: Y=HX+ε,Y=(ΔP 12 ΔP 1i … ΔP jn ) T X=(ΔX1 ΔD1 ΔX2 ΔD2 … ΔX n ΔD n ) T Where Y is the pseudorange difference observation vector; ΔP 12 , ΔP 1i and ΔP jn are the pseudo-range difference observation values ​​between satellites 1 and 2, satellites 1 and i, and satellites j and n, respectively; T is the transpose; H is the inter-satellite pseudo-range difference observation matrix, which has 2n columns and n(n-1) / 2 rows. Each row in H corresponds to an observation equation, and 1 and -1 in H are the coefficients of the clock error term; ρ 12 , 1i , jn and ρ ij are the inter-satellite pseudoranges from the 1st satellite to the 2nd satellite, the 1st satellite to the ith satellite, the jth satellite to the nth satellite, and the ith satellite to the jth satellite; is the partial derivative; X is the parameter matrix to be estimated; X1, X2, X i and X n are the parameters to be estimated for satellites 1, 2, i and n, including the position and velocity of the satellites; ΔD1, ΔD2 and ΔD n are the clock corrections of satellites 1, 2 and n respectively; is the position vector of satellite i at the observation time; n is the total number of slave satellites; S34, combining all the distance observations between the two slave stars at different times, and using the least squares estimation strategy to obtain the parameter to be estimated X: X =(H T OH) ―1 HY Among them, P is the weight matrix in the least squares estimation formula.

4. The method for time-space synchronization of a heterogeneous giant star system according to claim 2, characterized in that: The method of using a centralized data processing method to calculate the clock error, frequency deviation and frequency drift of each slave satellite in step S3 includes: A1. Based on the inter-satellite time synchronization observation equation and the satellite reception and transmission delay, the observation equation for estimating the satellite clock error, frequency offset and frequency drift is constructed: in, is the estimated value of intersatellite time synchronization observation; a A and a B , b A and b B and c A and c B are the clock error, frequency bias and frequency drift of satellite A and satellite B respectively; Δ1 is the observation time after epoch normalization; is the reference epoch; A2. Based on the observation equations for estimating satellite clock error, frequency bias and frequency drift, a combined observation equation is constructed: in, is the time-synchronized observation vector; are the observed values ​​of clock difference, frequency offset and frequency drift information between satellites 1 and 2, satellites 1 and i, satellites j and n, and satellites n-1 and n, respectively; T is transposed; is the design matrix of the observation equation, Δt=t1―t0 is the time difference, The 1 or -1 in represents the coefficient of the clock error term; is the combined matrix of clock error, frequency deviation and frequency drift of all slave satellites in each group; a1, a2, a n-1 and a n are the clock errors of satellites 1, 2, n-1 and n respectively; b1, b2, b n―1 and b n are the frequency deviations of satellites 1, 2, n-1 and n respectively; c1, c2, c n―1 and c n are the frequency drifts of satellites 1, 2, n-1 and n respectively; A3. Use the clock error constraint method to construct the observation vector to eliminate Virtual observation equation for medium rank deficiency: in, are the prior values ​​of the clock error of satellite j; are the accuracy of satellite j; a j , b j 、c j For The clock parameters of the j-th satellite extracted from ; A4. Combined observation vector And the virtual observation equation, we get the combined relationship: in, is the combined clock information observation; is the combined observation matrix; is the residual error after combination; is a row vector; A5. Use the least squares estimation strategy to solve the optimal estimate of the combined relationship: Among them, the best estimate Contains the clock error, frequency deviation and frequency drift of each slave satellite.

5. The method for time-space synchronization of a heterogeneous giant star system according to claim 1, characterized in that: The method of calculating the optimal estimated value of the state of each primary star using a distributed data processing method includes: S51, the extended Kalman filter state transfer equation is: x k+1 =f(x k ,p,t)+w Among them, x k+1 is the real state vector of the primary star at time k+1; x k is the true state vector of the primary star at time k; p is the covariance matrix; t is the time of primary star motion; w is the system process noise; f(·) is a nonlinear function; S52, the time update calculation formula of the extended Kalman filter is: in, is the estimated value of the primary star state predicted at time k+1; is the current state vector of the primary star at time k; Φ is the state transfer matrix, Find the partial derivative of the function f(·) with respect to the real state vector of the primary star at time k, is the prediction error covariance matrix at time k+1, is the error covariance matrix at time k, Φ T is the transpose of the state transfer matrix, Q is the system process noise covariance matrix; S53, the observation update calculation formula of the extended Kalman filter is used to obtain the optimal estimated value of the state of each primary star: in, is the optimal estimate of the state of the primary star at time k+1, H a is the observation matrix of the extended Kalman filter, H a T For H a Find the transpose; R is the observation noise covariance matrix; is the inter-satellite residual Δρ of all the primary stars that can be linked in the current epoch ij The vector formed, is the optimal estimation error covariance matrix at time k+1.

6. The method for time-space synchronization of a heterogeneous giant star system according to claim 5, characterized in that: The residual error of the inter-satellite distance Δρ ij The methods for obtaining include: S531. Construct the intersatellite two-way observation equation between the primary satellites i and j as follows: Among them, ρ ij , ji is the pseudo-range observation quantity of primary star i receiving primary star j and primary star j receiving primary star i; c is the speed of light; t i ,t j are the clock errors of primary stars i and j respectively; ε ij , ε ij is the observation noise from primary star i to primary star j and from primary star j to primary star i; is the theoretical inter-satellite distance between the main stars i and j, expressed as: Among them, x i 、x j are the position components of the primary stars i and j in the x-axis direction; y j ,y j are the position components of the primary stars i and j in the y-axis direction; z i 、z j are the position components of the primary stars i and j in the z-axis direction respectively; S532. Use the reference orbit linearization to obtain the linear observation equation without satellite clock error: Among them, ρ is the linear observation without satellite clock error; is the approximate satellite-to-earth distance calculated using the reference orbit; are the correction values ​​of the positions of the primary stars i and j relative to the reference orbit; L m is the star spacing unit vector, which is in the form of ε′ is the observation noise; S533. Based on the linear observation equation without satellite clock error, the Kalman filter observation equation is obtained as follows: Among them, Δρ ij is the residual error of the inter-satellite distance.

7. The method for time-space synchronization of a heterogeneous giant star system according to claim 1, characterized in that: The method of calculating the optimal estimated value of the clock error and clock drift of each primary star using the distributed data processing method includes: B1. Construct a linear model containing two states to represent the satellite clock error: Among them, b k ,d k is the clock error and clock drift of the primary star at time k; b k+1 ,d k+1 is the clock error and clock drift of the primary star at time k+1; dt is the sampling interval; ω b ,ω d They are white FM noise and random walk FM noise respectively; B2. The time update calculation formula of the extended Kalman filter is: in, is the satellite clock estimate predicted at time k, including clock error and clock drift; φ k,k―1 The state transition matrix solved for the clock parameters; φ T k,k―1 For φ k,k―1 Find the transpose; is the satellite clock parameter at time k-1; is the clock error covariance matrix predicted at time k; is the error covariance matrix at time k-1; Q k―1 is the covariance matrix of white FM noise and random walk FM noise at time k-1; B3. The observation update calculation formula of the extended Kalman filter is: Among them, K k is the Kalman gain matrix at time k, is the observation matrix, For Find the transpose; R k is the clock error covariance matrix at time k; (·) ―1 To find the inverse operation; is the optimal estimate of the satellite clock error and clock drift at time k; Z k―1 is the true measured value of the satellite clock error and clock drift at time k-1; is the optimal estimation error covariance matrix at time k, and I is the unit matrix.

8. The method for time-space synchronization of a heterogeneous giant star system according to claim 1, characterized in that: Trigger the master star selection mechanism and select the slave star with the highest priority as the master star. The methods for realizing dynamic master node transfer include: S41, modify each slave star in the group to be a candidate state; S42, each candidate state slave star sends its own campaign information to other candidate state slave stars around it to form a campaign information set; the campaign information includes communication capability, computing capability, orbit error and clock error; S43, the slave star initiating the election is regarded as the first election star, any slave star receiving the election information is regarded as the second election star, and the election information of the second election star is added to the election information set; S44, according to the election information of the first election star and the second election star, selecting the one with a higher priority as the new first election star; S45, judging whether there is a candidate slave star that has not participated in the comparison in the group, if so, proceeding to step S46, otherwise, taking the first candidate star selected most recently as the master star, and proceeding to step S47; S46, the first candidate star sends its own election information to other candidate slave stars around it, takes any slave star that receives the election information as the second candidate star, and stores the election information of the first candidate star and the second candidate star in a new election information set, and returns to step S44; S47, the selected master satellite broadcasts the information of its successful selection to all slave satellites in the group, and other slave satellites confirm and synchronize their status.

9. The method for time-space synchronization of a heterogeneous giant star system according to claim 8, characterized in that: Step S44 further comprises: There are four situations when determining the priority of the first campaign star and the second campaign star: First, the orbit error of the first candidate star is smaller than the orbit error of the second candidate star, and the clock error of the first candidate star is smaller than the clock error of the second candidate star; Second, the orbit error of the first candidate star is smaller than the orbit error of the second candidate star, and the clock error of the second candidate star is smaller than the clock error of the first candidate star; Third, the orbit error of the second candidate star is smaller than that of the first candidate star, and the clock error of the first candidate star is smaller than that of the second candidate star; Fourth, the orbit error of the second candidate satellite is smaller than that of the first candidate satellite, and the clock error of the second candidate satellite is smaller than that of the first candidate satellite; When the first, second and third conditions are met, the first campaign star is considered to have a higher priority; when the fourth condition is met, the second campaign star is considered to have a higher priority.

10. The method for time-space synchronization of a heterogeneous giant star system according to any one of claims 1 to 9, characterized in that: The method for calculating the preset number includes: Starting from 1, the number of main stars is selected in ascending order, and the complexity of the selected number of main stars is calculated using the complexity model. The expression of the complexity model is: O=(3*N / k z ) 3 Among them, O is the complexity; N is the total number of satellites in the Giant Star constellation network; k z is the number of main stars; When the difference between the complexities corresponding to two adjacent numbers of primary stars selected is greater than a preset threshold, the larger value of the two numbers of primary stars is selected as the preset number.

Images