A method for rapid calculation of inter-satellite relative positions in a large constellation measurement and control system

By using a unified epoch absolute state pool and a reference star shared RTN coordinate base architecture, combined with near-range CW analytical propagation and far-range two-body numerical differential propagation, a fast and efficient solution for the relative position and velocity of two satellites in a large satellite constellation is achieved. This solves the problems of large computational load and monitoring delay in existing technologies, and improves the stability and efficiency of monitoring.

CN123065484APending Publication Date: 2026-11-10BEIJING TIANLIAN TT&C TECH CO LTD
View PDF 3 Cites 0 Cited by

Patent Information

Application Number
CN202611277413.3
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-21
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing technologies are inefficient in calculating the relative position, relative velocity, and orbital difference parameters of any two satellites within a large satellite constellation, and the computational load is large, making it difficult to meet the requirements for real-time configuration monitoring, resulting in monitoring delays and increased operation and maintenance costs.

Method used

A unified epoch absolute state pool and a shared RTN coordinate base architecture with the reference star are adopted. By constructing a multi-star absolute state pool, and combining near-range CW analytical propagation and far-range two-body numerical differential propagation mechanisms, the relative states of the two stars are quickly calculated. Adaptive filtering updates and anomaly handling mechanisms are used to maintain continuous monitoring output.

Benefits of technology

It improves the efficiency and accuracy of batch output of relative states in large constellation scenarios, reduces the consumption of computing resources, ensures the continuity and accuracy of monitoring, and reduces the workload of manual review.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123065484A_ABST
    Figure CN123065484A_ABST
Patent Text Reader

Abstract

This invention discloses a rapid method for calculating the inter-satellite relative positions of large-scale constellation telemetry and control systems, belonging to the field of aerospace telemetry and control technology. It performs unified epoch alignment on orbital root numbers, GNSS position and velocity, and inter-satellite relative measurements to construct a multi-satellite absolute state pool; based on the reference satellite state, a shared RTN coordinate base is constructed, and a dual-mode approach of near-range C-W analytical propagation and far-range two-body numerical difference is used to complete the relative state propagation of the two satellites; combined with inter-satellite measurements, filtering updates, anomaly removal, and noise adaptive adjustment are performed to output relative positions, phase differences, and configuration offsets, with a multi-dimensional anomaly degradation mechanism to ensure continuous output. This method can significantly improve batch calculation efficiency, reduce computing power consumption, improve calculation stability, and support real-time configuration monitoring of large-scale constellations.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of aerospace telemetry and control technology, specifically relating to a method for rapid calculation of inter-satellite relative positions for large satellite constellations. Background Technology

[0002] With the development of low-Earth orbit communication, remote sensing networking, and commercial aerospace operation and maintenance, the operational mode of satellite constellations has gradually shifted from independent management of a small number of satellites to collaborative operation of multiple satellites. In such telemetry and control scenarios, the system not only needs to obtain the absolute orbital state of a single satellite, but also needs to continuously output the relative position, relative velocity, phase difference, difference in right ascension of the ascending node, and configuration offset between any two satellites under a unified time reference. A common engineering approach is to first calculate the inertial frame position and velocity from the orbital six roots, then project the target satellite's state onto the RTN coordinate system of the reference satellite, or use an approximate relative motion model for short-arc prediction, combined with filtering methods to suppress measurement noise and perturbation disturbances. This type of processing method can meet the needs when the constellation is small, but as the number of satellites increases, the inconsistency of timestamps, sampling frequencies and coordinate references between different data sources will lead to spatiotemporal mismatch in relative state calculations. At the same time, if the mutual calculation of elements, coordinate transformation and filtering updates are repeatedly performed on each pair of satellites, the amount of computation will increase rapidly, making it difficult to meet the real-time configuration monitoring needs of large constellations. This will lead to engineering risks such as increased monitoring latency, delayed alarm criteria, and increased manual review costs on the operation and maintenance side.

[0003] For example, Chinese patent CN202210781070.X discloses a Doppler real-time positioning and velocity measurement method and system for large-scale low-Earth orbit constellations. It calculates the position and velocity of the communication terminal using Doppler observation data, ephemeris information, and inter-satellite single-difference Doppler observation equations, combined with Kalman filtering, least squares, or gradient descent methods. Chinese patent CN202310046556.3 discloses a giant constellation target tracking method based on multi-satellite cooperative passive detection. It achieves the inversion of the external target state through multi-satellite cooperative measurement, random finite set filtering, and consistency term iteration. The common characteristic of the above-mentioned first type of methods is that their state objects are mainly constructed around the ground terminal or the external tracked target, with satellites primarily used as observation platforms. They lack a unified state cache, unified epoch alignment, and shared RTN projection links for arbitrary pairwise relationships within the constellation. Therefore, this type of method struggles to stably and frequently generate large quantities of relative positions, relative velocities, and orbital difference parameters of two satellites in large-scale constellation monitoring scenarios, easily leading to discontinuous updates of configuration monitoring results and affecting the accuracy of subsequent maintenance decisions.

[0004] For example, Chinese patent CN202510244667.4 discloses an autonomous orbit determination method for a large low-Earth orbit constellation based on star sensors and inter-satellite angle measurement. Its focus is on recovering or correcting the orbital state of the entire constellation through inter-satellite measurements and attitude reference information. This type of method is more suitable for overall orbit determination or configuration recovery, but it is not equivalent to the rapid calculation of high-frequency binary-satellite relative quantities for monitoring software. Even if absolute state results are obtained, if subsequent methods still involve repeated element calculations, coordinate transformations, and filtering propagation for each pair of satellites, the repeated calculations will consume significant processing resources as the constellation size increases. This will lead to a decrease in synchronous processing capability under single-process conditions and cause instability in monitoring interface refresh and reduced availability of long-term data assets.

[0005] Furthermore, existing solutions typically assume a single observation type as input, or that the input can be directly incorporated into a unified orbit determination equation. This fails to effectively address inconsistencies in time references, noise models, and state dimensions regarding orbital root numbers, GNSS position and velocity, and inter-satellite relative measurements. Directly applying traditional filtering structures to heterogeneous input scenarios easily leads to frequent switching of observation matrices, unstable propagation of state covariance, and ill-conditioned amplification of angular element values ​​under near-circular orbit conditions. This means that relative state errors do not primarily originate from the dynamic model itself, but rather from the lack of a unified, computable interface between the state representation layer and the filter update layer. This further results in unstable configuration deviation judgments, difficulty in timely identification of abnormal links, and increased costs for subsequent data governance and manual intervention. Summary of the Invention

[0006] The purpose of this invention is to provide a fast method for calculating the relative positions between satellites in a large constellation for telemetry and control, in order to solve the problems in the prior art, such as the state object being biased towards external targets, the separation of absolute orbit determination and relative fast calculation links, and the lack of a unified state expression for heterogeneous inputs, which lead to insufficient efficiency, unstable accuracy, and poor continuous output capability under abnormal conditions in large-scale constellation scenarios.

[0007] To achieve the above objectives, this invention provides a method for rapid calculation of the relative positions between satellites in a large constellation for telemetry and control. The invention provides the following technical solution: The system acquires orbital six-root data inputs, GNSS position and velocity inputs, and inter-satellite relative measurement inputs from multiple satellites. It performs range verification, unified epoch time alignment, and availability marking on each input, and converts available single-satellite inputs into inertial frame position and velocity states to construct a unified epoch multi-satellite absolute state pool. The RTN coordinate base and the transformation matrix from the inertial frame to the RTN coordinate system of the reference star are constructed based on the absolute state of the reference star in the multi-star absolute state pool, and the transformation matrix is ​​reused for all target stars related to the reference star. The relative position and relative velocity of the inertial frame are formed based on the absolute states of the reference star and the target star. The relative position and relative velocity of the inertial frame are projected onto the RTN coordinate system of the reference star to obtain the relative state of the two stars. The relative state of the two stars is propagated according to the orbital parameters of the reference star. Among them, for star pairs that meet the near-circular and near-distance conditions, CW analytical propagation based on the average angular velocity of the reference star is used. For star pairs that exceed the preset distance conditions, two-body numerical difference propagation of the absolute states of the reference star and the target star is used and then written back as the RTN relative state. A relative state observation mapping is constructed based on the currently available inter-satellite relative measurement inputs. The relative state of the two satellites is directly mapped to inter-satellite ranging and relative radial velocity. The propagated relative state of the two satellites is filtered and updated. Abnormal observations are eliminated and observation noise is dynamically adjusted based on the standardized residuals. The system outputs the relative position, relative velocity, phase difference, right ascension difference of the ascending node, and configuration offset of the satellites. In the event of missing input, abnormal reference satellite, abnormal filter residual, or abnormal output jump, it maintains continuous monitoring output through at least one of the following mechanisms: short-term prediction degradation, alternative reference satellite switching, abnormal observation rejection, and result limiting projection.

[0008] Furthermore, the process of acquiring the orbital six-root data input, GNSS position and velocity input, and inter-satellite relative measurement input from multiple satellites, performing range verification, unified epoch time alignment, and availability marking on each input, and converting available single-satellite inputs into inertial frame position and velocity states to construct a unified epoch multi-satellite absolute state pool includes: The values ​​of the semi-major axis, eccentricity, and orbital inclination are verified by inputting the six orbital elements. The position modulus is verified to be within the preset orbital range by inputting the GNSS position and velocity. The relative distance is verified to be positive by inputting the inter-satellite relative measurement. Furthermore, when a line-of-sight unit vector is input, the line-of-sight unit vector is verified to meet the normalization condition. Each input is resampled to a unified solution epoch. Data with a time difference not exceeding a preset time difference is directly used. Data with a time difference within the compensation range is compensated using short arc propagation. Data with a time difference exceeding the compensation range is marked as unavailable for the current period. For satellites with only orbital root data input, the orbital root data is converted into inertial frame position and velocity; for satellites with GNSS position and velocity input, it is directly written into the multi-satellite absolute state pool; for inter-satellite relative measurement input, it is retained as an observation item for subsequent relative state filtering updates.

[0009] Furthermore, the conversion of the orbital six roots into inertial frame position and velocity includes: The off-proximity angle is solved by the mean anterior angle and the eccentricity. When the eccentricity is less than the preset near-circle threshold, the mean anterior angle is directly used as the off-proximity angle. When the iterative solution exceeds the preset convergence condition, it is switched to GNSS position and velocity input or extrapolation of the previous cycle state. The true anomaly angle, orbital radius, and orbital plane position and velocity are calculated based on the deviated anomaly angle, and converted into inertial frame position and velocity vectors through attitude rotation corresponding to the orbital inclination angle, right ascension of the ascending node, and argument of perigee. The semi-major axis is calculated from the position vector and velocity vector of the inertial frame, and the absolute state vector and average angular velocity of each satellite are generated based on the semi-major axis. The absolute state vector is formed by splicing the position vector and velocity vector of the inertial frame.

[0010] Furthermore, the absolute states of the reference star in the multi-star absolute state pool construct the RTN coordinate base and the transformation matrix from the inertial frame to the RTN coordinate system of the reference star, and reuse the transformation matrix for all target stars associated with the reference star, including: Construct a radial unit vector based on the position vector of the reference star's inertial frame, construct a normal unit vector based on the cross product of the position vector and velocity vector of the reference star's inertial frame, and then construct a tangential unit vector based on the normal unit vector and the radial unit vector. The radial unit vector, tangential unit vector, and normal unit vector are combined into a transformation matrix from the inertial frame to the RTN coordinate system; The transformation matrix is ​​constructed only once for each reference star, and the transformation matrix is ​​reused for the relative state projection of all target stars associated with that reference star; When the reference star is in an abnormal state or the RTN coordinate base degrades, pause the current periodic update that depends on the reference star or switch to an alternative reference star to regenerate the shared RTN coordinate base.

[0011] Furthermore, the propagation of the relative state of the binary stars based on the orbital parameters of the reference star includes: The relative states of the two stars are obtained by RTN projection of the position and velocity differences between the inertial frames of the reference star and the target star, which consist of the relative positions and velocities of the two stars in the RTN. When the relative position of the RTN meets the preset close distance condition, the relative state of the two stars is rapidly propagated using the CW analytical state transition matrix based on the average angular velocity of the reference star and the calculation period. When the relative position of the RTN exceeds the preset distance condition, the same two-body dynamics numerical propagation is performed on the absolute state of the reference star and the absolute state of the target star respectively. Then, the position difference and velocity difference of the inertial frame of the two are recalculated, and the relative state of the RTN after propagation is written back using the RTN coordinate base of the propagated reference star. For a satellite pair that returns from long-range mode to short-range mode, the pair is required to stably meet the short-range condition for several consecutive periods before switching to CW resolution propagation.

[0012] Further, the step of constructing a relative state observation mapping based on currently available inter-satellite relative measurement inputs, directly mapping the binary star relative state to inter-satellite ranging and relative radial velocity measurements, and performing filtering updates on the propagated binary star relative state includes: For each star pair, a relative state filter is established with the relative position and relative velocity of the RTN as the state vectors. The predicted state of the filter is given by the propagated relative state of the RTN. The relative position of the RTN is mapped to an inter-satellite ranging quantity, and the directional projection relationship between the relative position of the RTN and the relative velocity of the RTN is mapped to a relative radial velocity quantity. Two-dimensional observation updates are performed when both distance and velocity measurements are available, and one-dimensional observation updates are performed when only distance measurements are available. When the standardized residual corresponding to the observation residual exceeds the preset threshold, the current observation is judged as an abnormal observation and will not participate in the current period's filter update.

[0013] Furthermore, the step of performing filtering updates on the relative states of the propagated binary stars, and removing outlier observations and dynamically adjusting observation noise based on standardized residuals, includes: A six-dimensional block matrix based on a white noise acceleration model is used as the process noise covariance matrix, so that the position component and the velocity component form a coupled propagation relationship according to the solution period; The filter gain is calculated based on the observation mapping Jacobian matrix at the propagation state, and the updated relative state and the updated covariance matrix are obtained using the filter gain, observation residual and prediction covariance matrix. When the condition number of the observation information matrix exceeds a preset threshold, the corresponding inverse matrix is ​​calculated using a pseudo-inverse. The average of the standardized residuals over the most recent effective periods is calculated. When the average of the standardized residuals continuously exceeds the upward adjustment condition, the observation noise scaling factor is increased. When the downward adjustment condition is continuously met, the observation noise scaling factor is decreased, and the observation noise scaling factor is constrained within a preset upper and lower limit range.

[0014] Furthermore, the output inter-satellite relative positions, relative velocities, phase differences, right ascension differences of ascending nodes, and configuration offsets include: Calculate the inter-satellite relative distance and relative velocity magnitude based on the RTN relative position and RTN relative velocity; The local phase difference is calculated based on the radial and tangential components in the semi-major axis of the reference star and the relative position of the RTN. When the local geometry does not meet the preset conditions, the phase difference is calculated based on the absolute true latitude difference between the reference star and the target star. The difference in right ascension of the ascending node is calculated based on the right ascension of the ascending node of the reference star and the target star. When only position and velocity input are available, the right ascension of the ascending node is calculated inversely from the angular momentum vector and the node vector. Based on the differences between the actual semi-major axis, actual phase, and actual right ascension of the ascending node of each satellite and the pre-stored nominal orbital parameters, a configuration offset is generated, which includes radial scale offset, along-orbit offset, and trans-orbit offset.

[0015] Furthermore, the maintenance of continuous monitoring output through at least one of the following mechanisms includes: short-term prediction degradation, alternative reference star switching, anomaly observation rejection, and result limiting projection. When an input is missing and the duration does not exceed the preset short-term degradation duration, the prediction output for the current period is obtained by extrapolating the relative state of the previous period using the steps described in claim 1, which involve propagating the relative state of the two stars based on the reference star orbital parameters; when the duration exceeds the short-term degradation duration, the star pair is marked as low confidence. When the reference satellite is in an abnormal state, the highest-scoring alternative reference satellite is selected from the candidate satellites in the same orbital plane according to the credibility scoring rules consisting of time consistency, covariance size, geometric proximity and link availability, and the shared RTN coordinate base is regenerated. When the single-cycle jump of the output relative distance or phase difference exceeds the physical upper limit determined according to the preset relative velocity envelope, the output of this cycle will be clipped to the allowable variation range centered on the output of the previous cycle. When the same pair of satellites triggers amplitude-limited projection or the observation noise scaling factor reaches its upper limit for a long period of time, the pair of satellites will be switched to low-confidence monitoring mode.

[0016] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement all the steps of the above-described method for rapid calculation of the relative positions between satellites in a large constellation for telemetry and control.

[0017] This method establishes a processing link combining a unified epoch absolute state pool, a shared RTN coordinate base for reference satellites, bi-satellite relative state propagation, and adaptive filtering updates. The processor first receives orbital root data inputs, GNSS position and velocity inputs, and inter-satellite relative measurement inputs from multiple satellites. It performs range verification, unified epoch time alignment, and availability marking on all types of inputs, and converts available single-satellite inputs into inertial frame position and velocity states, thereby constructing a unified epoch multi-satellite absolute state pool. Subsequently, it constructs the RTN coordinate base and the transformation matrix from the inertial frame to the RTN coordinate system for the reference satellite using the absolute states of the reference satellite in the multi-satellite absolute state pool, and reuses this transformation matrix for all target satellites associated with that reference satellite. Subsequently, the relative position and relative velocity of the inertial frame are formed based on the absolute states of the reference and target stars, and projected onto the RTN coordinate system of the reference star to obtain the relative state of the two stars. The relative state is then propagated according to the orbital parameters of the reference star: for star pairs satisfying the near-circular, close-range condition, analytical relative motion propagation based on the average angular velocity of the reference star is used; for star pairs exceeding the preset distance condition, two-body numerical difference propagation of the absolute states of the reference and target stars is used, followed by writing back the RTN relative state. Then, a relative state observation mapping is constructed based on the currently available inter-star relative measurement inputs, directly mapping the relative state of the two stars to inter-star ranging and relative radial velocity measurements. The propagated relative state of the two stars is then filtered and updated, and abnormal observations are eliminated and observation noise is dynamically adjusted based on the standardized residuals. Finally, the output includes the inter-satellite relative position, relative velocity, phase difference, right ascension difference of ascending node, and configuration offset. In the event of missing input, abnormal reference star, abnormal filter residual, or abnormal output jump, continuous monitoring output is maintained through at least one of the following mechanisms: short-term prediction degradation, alternative reference star switching, abnormal observation rejection, and result limiting projection.

[0018] Step S1, Heterogeneous Input Acquisition and Unified Ephemeris Preprocessing: Acquire the orbital six-root number input, GNSS position and velocity input, and inter-satellite relative measurement input from multiple satellites. Perform range verification, unified epoch time alignment, and availability marking on each input. Convert the available single-satellite inputs into inertial frame position and velocity states to construct a unified epoch multi-satellite absolute state pool.

[0019] Step S2, Multi-Satellite Absolute State Pool Construction: For satellites with only orbital root data input, the orbital root data are converted into inertial frame position and velocity; for satellites with GNSS position and velocity input, the data is directly written into the multi-satellite absolute state pool; for inter-satellite relative measurement input, the data is retained as observation items for subsequent relative state filtering updates. During the orbital root data conversion process, the deviated anomaly angle is first obtained from the mean anomaly angle and eccentricity, then the true anomaly angle, orbital radius, and position and velocity in the orbital plane are determined. Subsequently, the attitude rotation relationship corresponding to the orbital inclination, right ascension of the ascending node, and argument of perigee is combined to obtain the inertial frame position vector and inertial frame velocity vector, and further generate the absolute state vector and mean angular velocity.

[0020] Step S3, Construction of the shared RTN coordinate base for the reference star: Based on the absolute states of the reference stars in the multi-star absolute state pool, a radial unit vector is constructed according to the position vector of the reference star's inertial frame. A normal unit vector is constructed according to the cross product of the position vector and velocity vector of the reference star's inertial frame. Then, a tangential unit vector is constructed from the normal unit vector and the radial unit vector. The radial unit vector, tangential unit vector, and normal unit vector are combined into a transformation matrix from the inertial frame to the RTN coordinate system. This transformation matrix is ​​constructed only once for each reference star and reused for the relative state projection of all target stars associated with that reference star to reduce the computational bloat caused by repeated frame construction. When the reference star state is abnormal or the RTN coordinate base degenerates, the current period update relying on that reference star is paused, or a replacement reference star is used to regenerate the shared RTN coordinate base.

[0021] Step S4, Construction and rapid propagation of the relative state of the two stars: Based on the absolute states of the reference star and the target star, inertial frame position difference vectors and velocity difference vectors are formed and projected onto the reference star's RTN coordinate system to obtain the relative state of the two stars composed of the relative RTN positions and relative RTN velocities. When the relative positions of the two stars meet the preset close-range conditions, the relative state of the two stars is rapidly propagated using the CW analytical state transition relationship according to the average angular velocity of the reference star and the calculation period. When the relative positions of the two stars exceed the preset distance conditions, the same two-body dynamics numerical propagation is performed on the absolute states of the reference star and the target star respectively. Then, the position difference and velocity difference of the two inertial frames are recalculated, and the propagated relative RTN state is written back using the propagated reference star RTN coordinate base. When the far-range mode returns to the close-range mode, the star pair is required to stably meet the close-range conditions for several consecutive periods before switching to analytical propagation to avoid repeated switching near the threshold.

[0022] Step S5, Relative State Filter Update: Construct a relative state observation map based on the currently available inter-satellite relative measurement inputs. Establish a relative state filter for each satellite pair, with RTN relative position and RTN relative velocity as state vectors. The filter's predicted state is given by the propagated RTN relative state. Map the RTN relative position to an inter-satellite ranging quantity, and map the directional projection relationship between the RTN relative position and RTN relative velocity to a relative radial velocity quantity. Perform a two-dimensional observation update when both ranging and velocity quantities are available, and a one-dimensional observation update when only ranging quantities are available. During the update process, the filter gain is calculated based on the Jacobian matrix of the observation mapping at the propagation state, and the updated relative state and updated covariance matrix are obtained using the filter gain, observation residuals, and prediction covariance matrix. When the standardized residual corresponding to the observation residual exceeds a preset threshold, the current observation is judged as an abnormal observation and does not participate in the filter update of this period. When the observation information matrix is ​​ill-conditioned, the inverse matrix is ​​calculated using a pseudo-inverse form. At the same time, the average of the standardized residuals of the most recent effective periods is calculated. When the upward adjustment condition is met continuously, the observation noise scaling factor is increased, and when the downward adjustment condition is met continuously, the observation noise scaling factor is decreased. The scaling factor is constrained within a preset upper and lower limit range to adapt to different link quality and measurement fluctuation levels.

[0023] Step S6, Configuration Monitoring and Control Criteria Generation: Calculate the inter-satellite relative distance and relative velocity magnitude based on the RTN relative position and RTN relative velocity; calculate the local phase difference based on the semi-major axis of the reference satellite and the radial and tangential components in the RTN relative position, and if the local geometry does not meet the preset conditions, calculate the phase difference based on the absolute true latitude difference between the reference and target satellites; calculate the right ascension difference of the ascending node based on the right ascension of the ascending node of the reference and target satellites, and if only position and velocity inputs are available, calculate the right ascension of the ascending node from the angular momentum vector and node vector; generate configuration offsets including radial scale offset, along-orbit offset, and trans-orbit offset based on the differences between the actual semi-major axis, actual phase, and actual right ascension of the ascending node of each satellite and the pre-stored nominal orbital parameters, and output configuration monitoring results and control recommendation criteria accordingly.

[0024] Step S7, Abnormal Degradation Handling: When input is missing, reference satellite is abnormal, filter residual is abnormal, long-range mode is abnormal, or output jump is abnormal, the corresponding degradation mechanism is triggered. When input is missing and its duration does not exceed the preset short-term degradation duration, the predicted output for the current period is obtained by extrapolating the posterior relative state of the previous period through a propagation step; when the duration exceeds the short-term degradation duration, the satellite pair is marked as low confidence. When the reference satellite state is abnormal, according to the confidence scoring rules composed of time consistency, covariance scale, geometric proximity, and link availability, the highest-scoring alternative reference satellite is selected from the candidate satellites in the same orbital plane, and a shared RTN coordinate base is regenerated. When the single-period jump of the output relative distance or phase difference exceeds the physical upper limit determined based on the preset relative velocity envelope, the output for the current period is clipped to the allowable variation range centered on the output of the previous period; when the same satellite pair triggers amplitude-limited projection or the observation noise scaling factor reaches the upper limit for a long period in multiple consecutive periods, the satellite pair is switched to low confidence monitoring mode.

[0025] Furthermore, the values ​​of the semi-major axis, eccentricity, and orbital inclination are verified for the six orbital root inputs; the position magnitude is verified to be within the preset orbital range for the GNSS position and velocity inputs; the relative distance is verified to be positive for the inter-satellite relative measurement inputs; and the line-of-sight unit vector satisfies the normalization condition when a line-of-sight unit vector is input.

[0026] Furthermore, each input is resampled to a unified solution epoch. Data with a time difference not exceeding the preset time difference is directly used, data with a time difference within the compensation range is compensated using short arc propagation, and data with a time difference exceeding the compensation range is marked as unavailable for the current period.

[0027] Furthermore, in the process of converting the orbital six elements into the position and velocity of the inertial frame, when the eccentricity is less than the preset near-circular threshold, the mean near point angle is directly used as the deviated near point angle; when the iterative solution of the deviated near point angle exceeds the preset convergence condition, the input is switched to GNSS position and velocity or extrapolation of the previous cycle state, so as to reduce the impact of the instability of the angle element values ​​under near-circular orbit conditions on the subsequent solution link.

[0028] Furthermore, when the reference star is in an abnormal state or the RTN coordinate base degrades, the current periodic update that depends on that reference star can be paused, or a replacement reference star can be used to reconstruct the shared RTN coordinate base in order to maintain the consistency of the local projection reference of the related star pairs of the same reference star.

[0029] Furthermore, for satellite pairs that return from long-range mode to short-range mode, the pair is required to stably meet the short-range condition for several consecutive cycles before switching to CW resolution propagation, thereby suppressing state discontinuities caused by mode jitter near the threshold.

[0030] Furthermore, the relative state filter uses a six-dimensional block matrix based on a white noise acceleration model as the process noise covariance structure, so that the position component and the velocity component form a coupled propagation relationship according to the solution period.

[0031] Furthermore, when the condition number of the observation information matrix exceeds a preset threshold, a pseudo-inverse is used to calculate the corresponding inverse matrix to avoid numerical divergence caused by directly inverting the ill-conditioned matrix.

[0032] Furthermore, the average of the standardized residuals over the most recent effective periods is statistically analyzed. When the scaling factor is exceeded consecutively, the observation noise scaling factor is increased. When the scaling factor is met consecutively, the observation noise scaling factor is decreased. The observation noise scaling factor is then constrained within a preset upper and lower limit range to adapt to different inter-satellite link qualities and measurement stability.

[0033] Furthermore, the configuration offset consists of the difference between the actual semi-major axis, actual phase, and actual right ascension of the ascending node of each satellite and its nominal orbital parameters, which respectively correspond to radial scale offset, along-orbit offset, and across-orbit offset, and are used to generate constellation configuration monitoring parameters.

[0034] Furthermore, when an input is missing and the duration does not exceed the preset short-term degradation duration, the predicted output for the current period is obtained by extrapolating the posterior relative state of the previous period through a propagation step; when the duration exceeds the short-term degradation duration, the star pair is marked as low confidence to distinguish between short-term degradation output and long-term unreliable output.

[0035] Furthermore, when the reference satellite is in an abnormal state, a replacement reference satellite is selected from the candidate satellites in the same orbital plane according to the credibility scoring rules consisting of time consistency, covariance size, geometric proximity and link availability, and a shared RTN coordinate base is regenerated.

[0036] Furthermore, when the single-cycle jump of the output relative distance or phase difference exceeds the physical upper limit determined based on the preset relative velocity envelope, the output of this cycle is clipped to the allowable range of variation; when the same star pair triggers amplitude limiting projection or the observation noise scaling factor reaches the upper limit for a long period of time in multiple consecutive cycles, the star pair is switched to low confidence monitoring mode.

[0037] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement all the steps of the above-described method for rapid calculation of inter-satellite relative positions of a large constellation telemetry and control system.

[0038] This invention employs a unified epoch absolute state pool and a shared RTN coordinate base architecture with reference stars, significantly reducing the computational load of repetitive coordinate transformations, improving the efficiency of batch star pair solutions, reducing computing resource consumption, supporting real-time configuration monitoring of large-scale constellations, and shortening alarm response latency. Through a dual-mode propagation mechanism of near-range CW analytical propagation and far-range two-body numerical difference, combined with a filtering update link directly facing the RTN relative state, dynamic propagation and measurement updates are achieved in a closed loop within the same state space, improving the accuracy and stability of relative state solutions and reducing manual verification workload. Supporting multi-dimensional protection mechanisms such as short-term prediction degradation, alternative reference star switching, anomaly removal, and amplitude-limited projection, this invention effectively improves monitoring continuity under abnormal operating conditions, reduces false alarms and output breakpoints, reduces maintenance and remedial costs, and ensures long-term stable operation of the constellation. Attached Figure Description

[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the embodiments are briefly described below. The drawings only show content related to the method flow of the present invention and are used to assist in understanding the processing link of the present invention, and do not constitute a limitation on the scope of protection of the present invention.

[0040] Figure 1 The flowchart of the overall method in the embodiment of the present invention shows the overall process of heterogeneous input acquisition and unified epoch preprocessing, multi-star absolute state pool construction, reference star shared RTN base construction, binary star relative state construction and fast propagation, relative state filtering update, configuration monitoring and control criterion generation, anomaly degradation processing, and parameter update and version rollback.

[0041] Figure 2 The flowchart of the data acquisition and state construction unit in this embodiment of the invention shows the processing flow of three types of raw input reading, range verification, time alignment, unusable marking, input mode unification, conversion of orbit six roots to position and velocity, absolute state vector generation, and auxiliary orbit parameter calculation.

[0042] Figure 3 The flowchart of the shared RTN base and relative state fast propagation unit in the embodiment of the present invention shows the processing flow of reference star selection, RTN base construction, shared coordinate transformation matrix generation, relative vector calculation, RTN projection, near-range analytical propagation, far-range two-body numerical differential propagation, and monitoring parameter generation.

[0043] Figure 4 The flowchart of the filter update and anomaly degradation unit in this embodiment of the invention shows the process of constructing the relative state observation model, determining available inter-satellite measurements, updating the state, eliminating anomalies, reconstructing observation noise, determining anomaly types, switching alternative reference stars, and output limiting.

[0044] Figure 5 This is a flowchart of the configuration monitoring and online parameter update unit in an embodiment of the present invention. It shows the generation of distance matrix, phase difference matrix, ascending node right ascension difference matrix and configuration offset, phase warning and control suggestion judgment, illumination angle and task auxiliary information calculation, and the process of offline parameter calibration, online evaluation and update and anomaly rollback. Detailed Implementation

[0045] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0046] Example 1: Overall Flowchart of a Fast Calculation Method for Relative Positions Between Satellites in a Large-Scale Constellation Tracking and Control System like Figures 1 to 5 As shown in this embodiment, a method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system is provided. This method includes eight collaborative steps: heterogeneous input acquisition and unified epoch preprocessing, construction of a multi-satellite absolute state pool, construction of a reference satellite shared RTN base, construction and rapid propagation of binary-satellite relative states, relative state filtering and updating, configuration monitoring and control criterion generation, anomaly degradation handling, and parameter updating and version rollback. This method is executed by a processor in a computer device, and the memory is used to store task configuration, parameter dictionary, previous cycle state, covariance matrix, and output buffer.

[0047] In this embodiment, the Earth's gravitational constant is taken as... Earth's rotational angular velocity is taken as The maximum number of satellites that can be calculated in a single operation is set to... In practical engineering applications, this number can be reasonably expanded within a range of no less than several satellites, depending on the processing capacity of a single process and the task configuration; status refresh cycle This cycle can be configured in the sub-second to several-second range according to the telemetry and control refresh frequency and real-time requirements; the standard deviation of GNSS input position noise is taken as... This parameter can be obtained through airborne navigation receiver calibration data or historical playback statistics; the standard deviation of GNSS input velocity noise is taken as... The standard deviation of inter-satellite relative ranging noise is taken as follows: The standard deviation of inter-satellite relative velocity measurement noise is taken as follows: The standard deviation of the process noise reference acceleration is taken as follows: It can be calibrated offline through high-precision orbit simulation and historical propagation residual statistics; the lower limit of the observation noise scaling factor is taken as... Upper limit These upper and lower limits can be configured according to the range of link quality fluctuations; the matrix condition number threshold is set to... The threshold for switching to remote mode is set as follows: In practical engineering applications, this threshold can typically be adjusted based on factors such as track height, configuration spacing, and the applicability range of the linear model. to Reasonable configuration between; short-term degradation duration is set to 10 s, which can be determined based on typical link blockage duration and continuous output requirements; upper limit of relative velocity envelope. Under the near-Earth orbit configuration maintenance condition in this embodiment, This upper limit was determined by calibrating the 95th percentile value with an upward margin under simulation statistics for different track heights and configuration spacing.

[0048] Step S1: Heterogeneous Input Acquisition and Unified Ephemeral Preprocessing S101. Read three types of raw input. The processor reads the task configuration from memory and receives three types of raw input from the measurement and control interface.

[0049] The first type of input consists of six root numbers representing the orbits of each satellite: , in, Indicates the first The semi-major axis of a satellite, measured in meters; This represents the eccentricity, which is a dimensionless quantity. Indicates the track inclination angle, in radians; The right ascension of the ascending node is expressed in radians. This indicates the argument of perigee, expressed in radians. This represents the angle at the point of approach, expressed in radians. This represents the timestamp corresponding to the set of roots, in seconds.

[0050] The second type of input is GNSS position and velocity input: , in, Indicates the first The position vector of the satellite in the J2000 inertial frame, in meters; This represents the satellite's velocity vector in the J2000 inertial frame, expressed in meters per second. This represents the GNSS status timestamp, in seconds.

[0051] The third type of input is inter-satellite relative measurement input: , in, Indicates satellite To satellite The relative distance is measured in meters. This indicates a relative radial velocity measurement, in meters per second. This represents the unit vector of the line of sight from the reference star to the target star; This represents the inter-satellite measurement timestamp, in seconds.

[0052] In engineering implementation, the inter-satellite relative measurement input is a higher-level concept and can adopt various lower-level combinations. Preferably, it can include an input form containing only inter-satellite ranging, a combined input form containing inter-satellite ranging and relative radial velocity, and a combined input form containing inter-satellite ranging or ranging and velocity and an accompanying line-of-sight unit vector. For the input form containing only inter-satellite ranging, the subsequent filtering stage performs one-dimensional observation updates; for the input form containing both ranging and relative radial velocity, the subsequent filtering stage performs two-dimensional observation updates; for the input form with an accompanying line-of-sight unit vector, the line-of-sight unit vector can be used for measurement validity verification, link geometric consistency checks, or as auxiliary geometric information for relative radial velocity mapping, but it does not change the processing mainline of this invention, which uses the RTN relative state as the core state space.

[0053] S102, Execution Scope Verification and Time Alignment The processor performs a physical range check on the six track element inputs, including checking if the semi-major axis is greater than zero, if the eccentricity is between zero and one, and if the track inclination is between zero and... Within the range. For the GNSS position and velocity input, check whether the position vector magnitude falls within a reasonable range for low Earth orbit. In this embodiment, the position magnitude is preferably checked within... to The range can be given based on historical orbit statistics or mission orbit design parameters according to the target orbit altitude. For inter-satellite relative measurement inputs, check whether the relative distance is positive; when the input contains a line-of-sight unit vector, also check whether its magnitude satisfies the normalization condition.

[0054] Subsequently, the processor resamples all types of inputs to a unified solution epoch. The difference between the original data timestamp and the unified solution epoch should not exceed [a certain value]. In this embodiment, the nearest neighbor data is directly used; the time difference threshold can be configured within a smaller or larger range depending on the interface clock synchronization accuracy. For time differences within... to In this embodiment, data between time differences is compensated using short-arc propagation; the compensation range can be calibrated offline based on the orbit propagation error envelope and data refresh interval. For time differences exceeding [a certain threshold], [the following applies]. Data that is not available in the current period is marked as unavailable to prevent outdated data from being directly processed in the current period and causing spatiotemporal mismatch.

[0055] S103, Unified Input Mode For satellites with only orbital root data input, the processor converts the orbital root data into inertial frame position and velocity states. For satellites with GNSS position and velocity input, the processor directly writes them into the multi-satellite absolute state pool. For inter-satellite relative measurement input, the processor does not mistakenly treat it as an absolute orbital state, but retains it as an observation item for subsequent relative state filtering updates. After this step, data from different sources are mapped to absolute state input links and relative observation input links at a unified epoch, allowing subsequent steps to be executed on a unified data organization structure.

[0056] Step S2: Construct a multi-star absolute state pool S201. Convert the six fundamental numbers of the orbit into position and velocity in the inertial frame. When the satellite only has six orbital roots as input, the processor first solves the Kepler equations to obtain the anomalous angle from the mean anomaly and the eccentricity. The corresponding basic orbital relationship is: , in, Indicates the first The anomalous angle of a satellite, in radians; Indicates the eccentricity; This represents the mean anterior angle, expressed in radians. This formula belongs to the Kepler equations in basic orbital dynamics and is used to convert the time phase information in the element input into orbital geometric position parameters. In this embodiment, the processor uses Newton's iteration to solve the problem. Initial value is taken The preferred iteration termination threshold is... The convergence threshold can be set according to the requirements of double-precision floating-point accuracy and orbit solution stability.

[0057] When the eccentricity is less than the near-circular threshold, to reduce the unnecessary amplification of numerical errors during iteration in near-circular orbits, the processor directly uses the mean near-point angle as the eccentricity near-point angle. In this embodiment, the near-circular threshold is preferably set to... This threshold can be determined offline through near-circular orbit simulation error statistics. When convergence is not achieved after more than 20 iterations, an abnormal degradation mechanism is triggered, switching to GNSS position and velocity input or extrapolation of the previous cycle's state to avoid the ill-conditioned roots continuously affecting subsequent system establishment and propagation.

[0058] After obtaining the partial anomaly angle, the processor calculates the true anomaly angle, orbital radius, and orbital plane position and velocity based on well-known orbital mechanics relationships. Then, using the attitude rotation relationship defined by the orbital inclination, right ascension of the ascending node, and argument of perigee, it transforms the orbital plane state into the inertial frame, thereby obtaining the inertial frame position vector. and velocity vector Since this rotation is a standard coordinate transformation process, the formulas will not be elaborated here.

[0059] To unify the methods for obtaining auxiliary orbital parameters under both element-based paths and GNSS paths, the processor also inversely calculates the semi-major axis from the position and velocity of the inertial frame based on the specific mechanical energy relationship: , in, Indicates the first The semi-major axis of a satellite, measured in meters; Represents position vector The L2 norm, i.e., the orbital radius, is expressed in meters; Represents the velocity vector The L2 norm, measured in meters per second; This represents the Earth's gravitational constant, expressed in cubic meters per square second. This formula pertains to the orbital energy relationship between two bodies and is used to recover orbital scale parameters when the position and velocity are known, providing a unified parameter input for subsequent calculations of average angular velocity and characterization of phase difference.

[0060] S202, Generate the absolute state vector The processor constructs a uniform absolute state vector for each satellite: , in, Indicates the first Within the first solution cycle, the first The absolute state vector of each satellite; This represents the satellite's position vector at a unified epoch, in meters; This represents the satellite's velocity vector at a unified epoch, expressed in meters per second. This represents a six-dimensional real space. This state vector organizes the single-star position and velocity into a unified data structure for subsequent shared RTN system establishment, relative difference, and propagation.

[0061] When a satellite is missing either its position or velocity in the current cycle, the processor does not generate a new absolute state vector. Instead, it uses the filtered state from the previous cycle for a one-step propagation and marks the satellite as the predicted state. This mechanism allows the system to maintain continuous processing even when the input is short-term incomplete.

[0062] S203, Calculate auxiliary track parameters The processor calculates the average angular velocity based on the semi-major axis of each satellite, which is then used as input for subsequent CW propagation. , in, Indicates the first The average angular velocity of the satellite, measured in radians per second; Represents the Earth's gravitational constant; This represents the semi-major axis, in meters. This formula pertains to the calculation of average motion parameters for near-circular orbits, and is used to convert absolute orbital scale information into local relative dynamic propagation parameters.

[0063] When it is necessary to calculate the latitude, longitude, and local time of the nadir point, the processor first converts the J2000 inertial frame position to the Earth-fixed frame based on the Greenwich Mean Time angle corresponding to the current epoch. Then, it obtains the longitude and latitude from the Earth-fixed frame position components and further derives the local time. This part mainly serves as an auxiliary monitoring information output and does not change the main calculation link for inter-satellite relative positions in this invention.

[0064] When the element input is unavailable and only GNSS status is available, the processor first calculates the semi-major axis from the position and velocity, and then calculates the average angular velocity. When the semi-major axis obtained from the inversion is not within the reasonable range of the target orbit, the current satellite is not used as a reference satellite, but only as a target satellite to participate in passive calculation, so as to reduce the amplified impact of abnormal absolute state on the local coordinate base.

[0065] Step S3: Construct a reference satellite shared RTN coordinate base S301. Construct RTN basis vectors The processor uses each available satellite as a candidate reference star and constructs a local orthogonal basis for the RTN based on its unified absolute state. The construction relationship is as follows: , in, Indicates the first The radial unit vector of the reference star, with its direction pointing from the Earth's center to the reference star; It represents the normal unit vector, whose direction is consistent with the orbital angular momentum; This represents a tangential unit vector, located in the orbital plane and orthogonal to the radial direction; Represents the position vector of the reference star; Represents the velocity vector of the reference star; symbol Represents the cross product of vectors; This represents the L2 norm. This relationship belongs to the local orbital coordinate system construction formula, used to project the relative state of the subsequent target star onto a local processing scale consistent with the orbital geometry of the reference star.

[0066] when If the value is too small, it indicates an abnormal angular momentum vector magnitude, which may be caused by input errors, numerical degradation, or state abnormalities. In this case, the processor does not generate the RTN basis for the reference star. In this embodiment, the degradation criterion is preferably selected as follows: This threshold can be determined offline through numerical stability testing. When the orthogonality error of the three vectors exceeds a preset small threshold, the processor uses Gram-Schmidt orthogonalization to reconstruct the tangential and normal vectors to ensure the orthogonality of the transformation matrix.

[0067] S302. Generate the transformation matrix from the inertial frame to the RTN coordinate system. The processor combines the three sets of unit vectors into a transformation matrix: , in, Indicates from the inertial frame to the first Rotation matrix of the reference star's RTN coordinate system; , and These represent the transposes of three sets of unit vectors, used to construct the row vectors of a matrix. This matrix is ​​used to project the position difference vector and velocity difference vector in the inertial frame onto the radial, orbital, and normal directions of the reference star.

[0068] If any unit vector is missing, the matrix will not be generated, and all star pairs that depend on the matrix will pause updates in this cycle, retaining only the results from the previous cycle or switching to alternative reference star paths.

[0069] S303, Shared Multiplexing RTN Base Unlike the approach of constructing local coordinate systems for each pair of stars separately, in this embodiment, the processor constructs an RTN transformation matrix only once for each reference star and reuses it for the relative state projection of all target stars associated with that reference star. In this way, within the same period, multiple star pairs associated with the same reference star adopt the same epoch, the same local basis, and the same numerical specification, thereby reducing redundant coordinate construction and suppressing local spatiotemporal inconsistencies caused by independent pair-by-pair coordinate system construction.

[0070] Step S4: Construction of binary star relative states and rapid propagation S401, Constructing the relative vectors of the inertial frame. For any reference star and target star The processor first constructs the position difference vector and velocity difference vector in the inertial frame: , in, Indicates the first The position difference vector between the target star and the reference star for each period, in meters; This represents the velocity difference vector, with units of meters per second. , These represent the position vector and velocity vector of the target star during that period, respectively; , These represent the position and velocity vectors of the reference satellite during that period, respectively. This step transforms the absolute state relationship between the two satellites into relative state primitives that can be further projected.

[0071] When the absolute state of either the reference star or the target star is missing, the star pair only retains the result from the previous cycle and marks it as not updated.

[0072] S402, Project to RTN coordinate system The processor projects the relative vector of the inertial frame onto the RTN coordinate system of the reference star to obtain the relative state of the two stars: , in, Indicates the first Periodic star pairs The relative state vector of the RTN; Represents the relative position in the RTN coordinate system, where Radial relative position components, For the relative position components along the track, These are the normal relative position components, all in meters; Represents the relative velocity components in the RTN coordinate system, all in meters per second; This is the transformation matrix from the inertial frame to the reference star RTN frame.

[0073] When the magnitude of any relative position after projection exceeds the applicable range of near-range resolving propagation, the processor marks the star pair as a far-range mode. In this embodiment, it is preferable to use whether the relative distance exceeds... This threshold can be obtained through orbital simulation scanning as a criterion.

[0074] S403, CW parsing and propagation are performed in close-range mode. For star pairs that meet the near-circular proximity condition, the processor uses a CW analytical relationship driven by the average angular velocity of the reference star to rapidly propagate the relative states of the two stars. The overall prediction relationship is as follows: , in, Indicates star pair Predicting the relative state of the RTN in the next cycle; Indicates the average angular velocity of the reference star and solution cycle A defined CW analytical state transition matrix; This represents the average angular velocity of the reference star; This indicates the state refresh cycle. This propagation relationship belongs to the local relative motion model of near-circular orbits, which is used to perform fast state updates directly in the local relative coordinate system without repeatedly performing absolute orbit integration for each star pair.

[0075] For ease of implementation, this embodiment provides analytical prediction expressions for the position and velocity components:

[0076] as well as ,

[0077] in, , , These represent the three relative position components of the RTN in the current cycle; , , These represent the three relative velocity components of the RTN in the current cycle; , , Indicates the predicted position component for the next period; , , This indicates the predicted velocity component for the next cycle; It is a dimensionless phase advance quantity.

[0078] when When the value is very small, directly calculating the sine and cosine may introduce floating-point cancellation errors. In this embodiment, when... At this time, the processor uses a first-order approximation for stabilization. This lower limit can be determined based on numerical accuracy tests. This process only affects computational stability and does not change the physical meaning of propagation.

[0079] S404, Calculate monitoring output parameters The processor calculates the inter-satellite relative distance and relative velocity magnitude based on the relative positions and velocities of the two RTNs. Their physical meanings are the degree of separation between the binary stars and the intensity of their relative motion, respectively. In this embodiment, the relative distance and relative velocity are calculated using the Euclidean norm.

[0080] The processor further calculates the local phase difference based on the radial and tangential components in the semi-major axis of the reference star and the relative position of the RTN: ,in, This represents the local phase difference between the target star and the reference star, expressed in radians. This represents the arctangent function for the four quadrants, used to maintain the correct angle quadrant; Indicates the relative position component along the track, in meters; Represents the radial relative position component, in meters; The semi-major axis of the reference star is represented in meters. This relationship utilizes a local geometric approximation to project the relative positions of the binary stars into an in-orbit phase shift, suitable for configuration maintenance monitoring.

[0081] When the local geometry does not satisfy the above approximation conditions, for example The processor now defines the phase difference based on the absolute true latitude difference between the reference and target stars. The true latitude definition relationship is as follows: , The phase difference is then returned to the standard angle range: , in, , These represent the true latitudes of the reference star and the target star, respectively, in radians; , These represent the arguments at perigee; , They represent the true anterior angles, respectively. This indicates that the angle is turned back to the interval. The angle normalization function. This alternative path is used to maintain phase difference output continuity when local geometric approximations are unreliable.

[0082] The difference in right ascension of the ascending node is calculated using the angular reversal method: , in, The difference in right ascension between the ascending nodes of the target star and the reference star is expressed in radians. , These represent the right ascension of the ascending node of the reference star and the target star, respectively. If only position and velocity are input, the processor first calculates the right ascension of the ascending node from the angular momentum vector and the node vector, and then performs the above difference operation.

[0083] The configuration offset is generated based on pre-stored nominal orbital parameters. The processor calculates the difference between the actual and nominal semi-major axis, the difference between the actual and nominal phase, and the difference between the actual and nominal right ascension of the ascending node, and then forms the configuration offset. , , , , in, Indicates the first The semi-major axis offset of a satellite, in meters; This indicates the pre-stored nominal semi-major axis, in meters; Indicates phase offset, in radians; Indicates the nominal true latitude; Indicates the right ascension offset of the ascending node, in radians; Indicates the nominal right ascension of the ascending node; The configuration offset vector has a first component representing radial scale offset, a second component representing arc length offset along the track, and a third component representing arc length offset across the track. This represents the orbital inclination angle. This vector uniformly expresses configuration deviations as a three-dimensional quantity that can be directly used for monitoring and criterion generation.

[0084] When the orbital inclination angle satisfies When the value is too small, i.e., close to the equatorial orbital degradation condition, the trans-orbital offset component output is zero and marked as right ascension difference degradation, in order to avoid numerical instability at small inclination angles.

[0085] S405, Perform two-body numerical differential propagation in long-distance mode. When the star pair is marked as long-range mode, the processor does not directly propagate the relative states of the RTN. Instead, it performs the same two-body dynamics numerical propagation on the absolute states of the reference star and the target star respectively. The basic dynamic equations are: , in, This represents the first derivative of the position vector with respect to time, i.e., the velocity vector; This represents the first derivative of the velocity vector with respect to time, i.e., gravitational acceleration; This represents the current position vector of the satellite; This represents the current velocity vector of the satellite; Represents the Earth's gravitational constant; This represents the orbital radius. This equation belongs to the standard two-body dynamics model, used to maintain the physical consistency of propagation beyond the local linear approximation range.

[0086] In this embodiment, the integrator step size is taken as .when During propagation, a 10-step sub-step is preferred; this setting can be selected based on numerical stability and processor load balancing. The integrator uses a fixed-step fourth-order Runge-Kutta numerical integral. After propagation, the processor recalculates the position and velocity differences between the inertial frames of the reference and target satellites, and uses the propagated reference satellite RTN coordinate base to write back the RTN relative state, thus maintaining consistency with the data structure output from the near-range mode.

[0087] When any substep in the long-range mode propagation exhibits an abnormally small orbital radius or significant orbital energy divergence, the processor stops the numerical integration of the current cycle, performs an extrapolation using the absolute state of the previous cycle, and outputs a low-confidence flag to ensure continuous output results even in abnormal situations. For satellite pairs returning from long-range mode to short-range mode, this embodiment requires that the satellite pair satisfy the hysteresis distance condition for three consecutive cycles before switching to analytical propagation; this number of consecutive cycles can be offline calibrated based on mode jitter statistics.

[0088] Step S5: Relative State Filter Update S501. Establish a relative state filter and construct the observation function. The processor establishes a six-dimensional relative state filter for each satellite pair. The state vector directly uses the relative position and relative velocity of the RTN, instead of filtering each satellite independently and then subtracting the results. The state prediction result is given in step S403 or step S405.

[0089] The process noise covariance matrix adopts a discrete white noise acceleration model: , in, Indicates the first Periodic star pairs The process noise covariance matrix; This represents the standard deviation of the equivalent unmodeled relative acceleration, in meters per square second. Indicates the calculation period, in seconds; This represents a three-dimensional identity matrix. The upper left block of the matrix reflects the accumulation of position noise, the upper right and lower left blocks reflect the coupling between position and velocity, and the lower right block reflects the accumulation of velocity noise, thus ensuring that the position and velocity components maintain a consistent propagation relationship in discrete time.

[0090] In this embodiment, the initial covariance matrix is ​​preferably in diagonal form, with a standard deviation of 50 m for the position component and 0.5 m / s for the velocity component. This initial value can be calibrated offline based on the initial orbit determination quality or historical orbit statistics.

[0091] When available observations are inter-satellite ranging and velocity measurement, the observation vector is defined as: , Its observation function is defined as: , in, Indicates the first The observation vector for each period; This indicates a relative distance measurement, with the unit being meters; This indicates a relative radial velocity measurement, in meters per second. This represents a nonlinear observation function that maps the relative states of a six-dimensional RTN to observables. , , Represents the relative position component of the RTN, in meters; , , Represents the relative velocity component of RTN, in meters per second; This represents a smoothing term used to avoid a denominator of zero; in this embodiment, it is preferably selected as... This value can be set according to the floating-point precision requirements. The first dimension of the observation function output is the relative distance, and the second dimension output is the relative velocity projection along the line of sight.

[0092] In engineering implementation, when the inter-satellite relative measurement input only includes inter-satellite ranging, the processor uses only the first dimension of the observation function to construct a one-dimensional observation model and perform a one-dimensional update; when the inter-satellite relative measurement input includes both inter-satellite ranging and relative radial velocity, the processor uses the above-mentioned two-dimensional observation model to perform a two-dimensional update; when the inter-satellite relative measurement input also includes a line-of-sight unit vector, the line-of-sight unit vector can be used to verify the directional consistency of the measured relative radial velocity, or to determine whether the link geometry meets the current cycle update requirements.

[0093] When the relative distance is too small, the radial velocity denominator may become unstable. In this embodiment, when the distance is less than 1 m, the processor replaces the radial velocity observation denominator with 1 m to ensure continuous observation mapping. This lower limit can be determined based on inter-satellite safety spacing and numerical stability requirements. If only ranging is measured and not velocity, only the first-dimensional observation is used to perform a one-dimensional update.

[0094] S502, Perform Extended Kalman Update The processor uses Extended Kalman Update to obtain the posterior relative state and posterior covariance. The update relation is: , , , in, Indicates the first Periodic star pairs The Kalman gain matrix; Represents the predicted covariance matrix; This represents the first-order Jacobian matrix of the observation function at the predicted state; Represents the observation noise covariance matrix; Indicates the predicted relative state; This represents the updated posterior relative state; This represents the updated posterior covariance matrix; This represents the identity matrix. The first equation is used to calculate the trade-off coefficient between observations and predictions, the second equation is used to correct the state using the observation residuals, and the third equation uses the Joseph form to update the covariance to improve numerical stability.

[0095] When the condition number of the observation information matrix exceeds a threshold, direct inversion may amplify numerical errors. In this embodiment, when the condition number exceeds... At this time, the processor uses a pseudo-inverse instead of direct inversion. This threshold can be determined offline through ill-conditioned matrix simulation testing.

[0096] Anomalies are identified using standardized residuals. The corresponding criterion is: , in, Represents standardized residual statistics; Represents the observation residual vector; Let represent the residual covariance matrix. When When the preset threshold is exceeded, the current observation is judged as an abnormal observation and will not participate in the current period's filter update. In this embodiment, the threshold is preferably 9, which can be determined based on the chi-square threshold or historical false alarm rate statistics; the same criteria for the corresponding dimension are used when updating the one-dimensional ranging.

[0097] S503, Perform dynamic reconstruction of observation noise. The processor first defines the basic observation noise matrix: During online operation, an observation noise scaling factor is introduced. And order: , in, Indicates star pair The basic observation noise covariance matrix; Indicates the relative standard deviation of the distance measurement; Indicates the relative standard deviation of speed measurement; Indicates the first Observation noise scaling factor for each period; This represents the observation noise covariance matrix actually used for the current filter update. This scaling factor is used to adaptively adjust the observation confidence based on the level of the most recent measurement residuals.

[0098] Let the mean of the standardized residuals for the most recent effective periods be: , in, This represents the mean of the standardized residuals within the window; This indicates the length of the statistical window. In this embodiment, The window length can be configured between 3 and 10 cycles, depending on the duration of link fluctuations and response speed requirements.

[0099] If the mean residual value is large for several consecutive periods, it indicates that the current measurement fluctuation is higher than the basic noise assumption, and the observation noise scaling factor should be increased. , If the mean residual value is small for several consecutive periods, it indicates that the current measurement is relatively stable, and the observation noise scaling factor can be appropriately reduced. , in, This represents the observation noise scaling factor for the next period; To increase the proportion; To lower the percentage; and These represent the upper and lower limits of the scaling factor, respectively. The scaling ratios can be calibrated offline through a trade-off experiment between residual convergence speed and false alarm rate. In this embodiment, if the scaling factor of a certain satellite pair remains close to the upper limit for a long period within a 24-hour statistical window, the link is marked as a low-quality measurement source, and subsequent degradation processing is triggered first.

[0100] After this processing, the filter can automatically adjust the observation confidence level according to the changes in inter-satellite link quality, so that the same model can output a more stable relative state under different link conditions.

[0101] Step S6: Configuration monitoring and control criterion generation S601, Generate configuration monitoring parameter matrix The processor outputs the phase difference matrix, range matrix, and ascending node right ascension difference matrix based on the posterior relative states of each satellite pair, and generates the configuration offset of each satellite relative to its virtual nominal position. The virtual nominal position is generated from the Walker constellation nominal orbit parameters in the mission configuration. For each satellite with a determined orbital plane number and in-plane number, the processor obtains the nominal semi-major axis, nominal phase, and nominal ascending node right ascension, and then combines them with the configuration offset vector from step S404 to form the final output.

[0102] S602, Perform phase maintenance judgment The processor performs configuration monitoring status determination based on the deviation of each satellite from the nominal phase. In this embodiment, the phase maintenance warning threshold is preferably... The preferred trigger threshold is... This threshold can be set offline based on constellation configuration maintenance strategy, propulsion capability, and mission tolerance. When the absolute value of the phase deviation is below the warning threshold, a normal monitoring status is output; when it is between the warning threshold and the control threshold, a warning status is output; when it reaches or exceeds the control threshold, an orbital ascent or descent suggestion is generated.

[0103] Simultaneously, the processor verifies whether the resolvable margin of the semi-major axis control meets the execution conditions. The criterion is that sufficient margin must be maintained between the target semi-major axis adjustment and the comprehensive error estimate. In this embodiment, the resolvable margin threshold is preferably 10 m, which can be comprehensively calibrated based on the orbit determination accuracy, actuator discrete error, and environmental perturbation error. When the margin is insufficient, only a continued observation suggestion is output, without directly generating an execution-level control suggestion.

[0104] S603, Output Task Auxiliary Information The processor can also calculate illumination angle and task visibility auxiliary information for subsequent task planning, imaging window arrangement or link resource scheduling, but this information does not change the state update logic of the main solution link for relative position in this invention.

[0105] Step S7: Abnormal Degradation Handling S701, Determine the type of abnormality In each cycle, the processor determines whether there are any anomalies such as missing input, abnormal reference satellite RTN base construction, abnormal filtering residuals, abnormal long-range mode, or abnormal output jumps. The result of this determination directly determines whether to use normal updates, degradation prediction, alternative reference satellite switching, or a limited projection path for the current cycle.

[0106] S702, Short-term Predictive Degradation When input loss occurs and its duration does not exceed the short-term degradation duration, the processor extrapolates the predicted output for the current period using the posterior relative state of the previous period through a propagation step. In this embodiment, the short-term degradation duration is preferably 10 s, but in actual engineering, it can be reasonably configured within the range of 5 s to 30 s based on the duration of link obstruction. When the duration exceeds the specified duration, the satellite pair is marked as low confidence to indicate to the upper-layer system that the output mainly relies on model prediction rather than real-time observation.

[0107] S703, Alternate Reference Star Switching When the reference satellite is in an abnormal state, the processor calculates a reliability score among candidate satellites in the same orbital plane and selects the one with the highest score as the replacement reference satellite. The scoring relationship is as follows: , in, Indicates candidate satellites The overall credibility score; , , , These represent the weights for time consistency, covariance size, geometric proximity, and link availability, respectively. , , , These represent the scores for the corresponding sub-items.

[0108] In this embodiment, the weights are preferably selected as follows: , , , The weight is determined by ensuring a high continuous monitoring output rate and a low false handover rate after switching using an offline validation set. Among the sub-scoring components, the time consistency score decreases as the deviation between the candidate star's state timestamp and the current epoch increases; the covariance score decreases as the candidate's absolute state covariance trace increases; the geometric proximity score decreases as the phase deviation between the candidate star and the original reference star increases; and the link availability score is determined based on the coverage of available measurement links between the candidate star and the target constellation. If the overall scores are tied, the candidate with the larger covariance score is prioritized, followed by the candidate with the closer geometric proximity, and finally, a stable ranking can be performed by satellite number.

[0109] This alternative mechanism allows the shared RTN base to be rebuilt even when the reference star is faulty, reducing the large-area star pair update interruption caused by single-point reference failure.

[0110] S704, Perform output limiting projection When the single-cycle jump in relative distance or phase difference exceeds the physical limit, the processor performs a limiting projection on the current output. Taking relative distance as an example, if the distance difference between the current cycle and the previous cycle exceeds the displacement limit corresponding to the allowable velocity envelope, the output of the current cycle is clipped to the allowable variation range centered on the output of the previous cycle. This allowable variation range is jointly determined by the relative velocity limit and the calculation cycle.

[0111] In this embodiment, This upper limit is obtained by adding approximately 20% margin upwards to the 95th percentile of the relative velocity statistics for different configurations under low-Earth orbit maintenance conditions. If the result from the previous cycle is not available, the jump limit is not executed. If the limit is applied consecutively more than 5 times, the system will forcibly switch the satellite pair to a low-confidence mode and prompt for manual review to prevent abnormal states from being smoothly hidden for a long time.

[0112] Step S8: Parameter Update and Version Rollback S801, offline calibration In the offline phase, a high-precision orbit simulator is used to generate training and validation samples. These samples are then combined with available fragments from historical telemetry and control playback data or semi-physical integration data to calibrate the process noise matrix, observation noise matrix, and long-range mode switching threshold, rather than training a black-box model. The samples are preferably designed to cover operating conditions such as normal measurements, different packet loss rates, continuous link interruptions, and long-range satellite pair propagation to ensure the parameters are applicable to various operational scenarios.

[0113] S802, Online Update During the online phase, the distribution of various residuals is statistically analyzed every 24 hours. If the new parameter version reduces the mean relative position error and does not increase the 95th percentile error on the validation set, while the false alarm rate of phase exceeding the threshold does not increase and the effective continuous monitoring output rate does not decrease, then the noise parameter version is updated; otherwise, the old version is retained. This update strategy ensures that parameter adaptation does not sacrifice continuous output capability and criterion reliability.

[0114] S803, Abnormal Rollback If, after an online update, the error increases by more than a preset percentage for one consecutive hour, or the 95th percentile error increases by more than a preset percentage, or the false alarm rate significantly increases due to phase exceeding the threshold, the system will automatically roll back to the previous stable parameter version. In this embodiment, a rollback is triggered when the error increases by more than 15%, the 95th percentile error increases by more than 10%, or the false alarm rate increases by more than 2 percentage points. These percentages can be determined based on long-term operational stability requirements. This mechanism is used to suppress the accumulation of systematic deviations caused by online parameter drift.

[0115] Example 2: Single-process multi-satellite synchronous fast solution for 10 low-Earth orbit Walker satellites This embodiment describes the operation of the invention in a specific engineering scenario for a monitoring server in the telemetry, tracking, and command (TT&C) center of a commercial low-Earth orbit (LEO) communication constellation. The server processor can be a multi-core CPU, a digital signal processor (DSP), or other processing platforms with vector computing capabilities. In this embodiment, a 16-core CPU with a clock speed of 3.2 GHz and 64 GB of memory is preferred. In actual engineering applications, the processing platform configuration can be adjusted according to the constellation size, concurrent computation load, and real-time requirements of the task, and does not constitute a necessary limitation on the implementation of the invention. The number of satellites processed in a single operation is fixed at 10, numbered from Sat-1 to Sat-10. The sampling period and output period are both... In practical engineering applications, this cycle can also be configured between sub-second and several-second intervals depending on the refresh rate of the telemetry and control interface. GNSS position velocity per... Updated once, inter-satellite ranging / velocity measurement every Update once, the optimal number of six orbitals per orbital is determined. The update is performed once to correct long-term drift; in other mission scenarios, the update interval of the six track elements can also be adjusted within the range of several seconds to several minutes according to the refresh rate of the telemetry and control interface, the track maintenance rhythm and the computing power of the equipment.

[0116] In the current epoch Sat-1 was selected as the reference satellite. The inertial frame position and velocity inputs for Sat-1 were... , The position and velocity inputs of the inertial frame of Sat-2 are: , This set of data corresponds to a near-circular, close-range relative flight scenario.

[0117] The processor first executes step S301 to construct the RTN basis for Sat-1. From the current input, the reference star position is along the first coordinate axis of the inertial frame, and the velocity is along the second coordinate axis. Therefore, the RTN basis is aligned with the inertial frame direction in this period, resulting in an identity matrix as the transformation matrix from the inertial frame to the RTN. The processor then executes steps S401 and S402 to obtain the initial RTN state of Sat-2 relative to Sat-1: radial relative position approximately 12 m, orbital relative position approximately 104.8 m, and normal relative position approximately 8.2 m; radial relative velocity approximately -0.12 m / s, orbital relative velocity approximately -0.03 m / s, and normal relative velocity approximately 0.015 m / s. From this, the relative distance is calculated to be approximately 105.8 m, and the relative velocity magnitude is approximately 0.1246 m / s.

[0118] The processor approximates the semi-major axis based on Sat-1 inverse calculation. The average angular velocity is obtained from step S203. .exist Under these conditions, the processor executes step S403, using CW analytical propagation to obtain the predicted relative position for the next cycle as approximately: radial 11.88 m, orbital 104.77 m, and normal 8.215 m; the corresponding predicted velocity components are approximately: radial -0.119 m / s, orbital -0.031 m / s, and normal 0.015 m / s. This prediction result indicates that, under the current close-range configuration, local relative motion can propagate rapidly within the shared RTN basis of the reference satellite without requiring separate integration of the absolute orbits of the two satellites.

[0119] During this period, inter-satellite measurements from both Sat-1 and Sat-2 were received simultaneously, with a measured distance of 105.2 m and a measured relative radial velocity of -0.056 m / s. The processor executes step S501, inputting the predicted relative state into the observation function to obtain the predicted observation values: a predicted distance of approximately 105.76 m and a predicted radial velocity of approximately -0.041 m / s. Therefore, the distance residual for this period is approximately -0.56 m, and the velocity residual is approximately -0.015 m / s, both within the statistical threshold. The processor then executes step S502 to perform an extended Kalman update. After the update, the posterior RTN position of Sat-2 relative to Sat-1 is corrected to: radial approximately 11.93 m, orbital approximately 104.24 m, and normal approximately 8.19 m; the posterior velocity is corrected to: radial approximately -0.118 m / s, orbital approximately -0.032 m / s, and normal approximately 0.015 m / s.

[0120] The processor further executes step S404, calculating the local phase difference based on the posterior RTN state. The phase difference, approximately [value missing], can be obtained from the orbital position and the semi-major axis of the reference star. That is, about If we calculate the right ascension of the ascending nodes of Sat-1 and Sat-2 from the roots, they are approximately 1.2040 rad and 1.2041 rad respectively. Then the difference in right ascension between the ascending nodes is approximately... That is, about If the nominal parameters of Sat-2 satisfy the nominal semi-major axis as... The nominal true latitude is smaller than the current actual true latitude. The nominal right ascension of the ascending node is smaller than the current actual value. The configuration offset generated by the processor is approximately: 12 m radial offset, 68.78 m along the track, and 0.49 m across the track. Since this phase difference is much smaller than the phase warning threshold, Sat-2 does not trigger a warning or control suggestion compared to Sat-1.

[0121] Within the same cycle, the processor constructs 10 absolute state vectors and 10 reference satellite RTN bases for 10 satellites, and outputs the relative state results of 90 directed satellite pairs. Compared with pairwise independent system construction and pairwise independent filtering, this processing method reuses the local coordinate base of the reference satellites and the unified state pool, which is beneficial for balancing computational speed and result consistency under single-process conditions.

[0122] This embodiment uses a high-precision two-body plus J2 perturbation simulator to generate the true orbit values ​​of 10 low-Earth orbit Walker satellites for 6 consecutive hours, with an output period of [missing information]. The simulation was further enhanced by using historical telemetry and control playback interface timing data to verify time alignment and caching logic. Under normal operating conditions, the standard deviation of GNSS position noise is 8 m, the standard deviation of velocity noise is 0.03 m / s, the standard deviation of inter-satellite ranging noise is 2 m, and the standard deviation of velocity noise is 0.01 m / s. The control scheme uses independent calculation for each pair plus fixed noise extended Kalman filtering, while the present invention uses a unified state pool, shared RTN base, and relative state extended Kalman filtering. Under the statistical criteria of fixed thread binding, CPU usage not exceeding 70%, and repeated simulations 100 times, in a 10-satellite scenario, the average single-cycle time of the control scheme is 46.8 ms, and the 95th percentile time is 49.6 ms, while the average time of the present invention is 13.9 ms, and the 95th percentile time is 15.1 ms; the root mean square error of the relative position of the two satellites in the control scheme is... The solution of the present invention is as follows: The root mean square error of the relative velocity of the control scheme is... The solution of the present invention is as follows: The above results demonstrate that, with a scale of 10 satellites, this invention can stably support single-process multi-satellite synchronous rapid output and improve relative status accuracy. It can effectively reduce the computing power configuration cost of the constellation telemetry and control system, increase the configuration monitoring refresh frequency and alarm response speed, reduce the workload of manual data verification, and significantly improve the telemetry, control and maintenance efficiency of large constellations.

[0123] Subsequently, based on the above calculation and processing results, the processor continues to execute the subsequent steps as described in Embodiment 1, such as configuration monitoring criterion generation, abnormal degradation processing, and parameter version maintenance. The specific execution logic is the same as in Embodiment 1, and will not be repeated here.

[0124] Example 3: Continuous output monitoring of configurations with link interruption and measurement packet loss This embodiment illustrates the working method of the present invention in maintaining continuous monitoring output under abnormal operating conditions. It still uses 10 satellites and an output period of... The Walker constellation scenario. Between 1200 s and 1210 s, the inter-satellite ranging / velocity measurement links between Sat-4 and Sat-5 were completely interrupted, with a random 5% packet loss. Traditional solutions in this situation typically skip updates or output null values, which can easily cause breakpoints in the configuration monitoring interface and lead to misjudgments of phase maintenance criteria.

[0125] arrive At this time, the posterior state of Sat-4 relative to Sat-5 in the previous cycle is as follows: the relative position of the RTN is approximately -18.4 m radially, 2650.0 m along the orbit, and -6.1 m normally; the relative velocity of the RTN is approximately 0.09 m / s radially, -1.73 m / s along the orbit, and 0.01 m / s normally. Since there are no new inter-satellite measurement inputs in this cycle, the processor determines an input missing anomaly in step S701 and then executes step S702, performing CW extrapolation based solely on the posterior state of the previous cycle and the average angular velocity of the reference satellite to obtain the prediction result for this cycle: the relative position of the RTN is approximately -18.31 m radially, 2648.27 m along the orbit, and -6.09 m normally; the relative velocity of the RTN is approximately 0.091 m / s radially, -1.731 m / s along the orbit, and 0.010 m / s normally. Since the missing duration has not exceeded 10 s, the system continues to output this prediction value and adds a prediction state confidence flag. The processor then calculates the phase difference, obtaining approximately The reading is still below the warning threshold, so control recommendations will not be triggered falsely.

[0126] The link was restored at 1211 s, but the measured distance in the first frame abruptly increased to 2900 m. This link restoration time is merely an example of an abnormal scenario; in other engineering tasks, the link restoration time can occur at different epochs depending on the occlusion geometry, telemetry and control scheduling, and interface restoration timing. The processor executes step S704, performing a jump check based on the previous cycle result and the allowed relative velocity envelope, and finds that the distance jump in this cycle exceeds... The corresponding physical upper limit is not used directly for the frame measurement; instead, it is identified as an abnormal observation and subjected to amplitude limiting. Normal measurement resumes at 1212 seconds, and the filter re-enters normal update mode. This way, abnormal measurements in the early stages of link recovery do not directly cause non-physical spikes in the monitoring curve.

[0127] If the Sat-4 reference satellite status is simultaneously abnormal within the same anomaly window, the processor executes step S703 to calculate the score of the alternative reference satellite in the candidate set within the same orbital plane. For example, if candidate satellite Sat-6 satisfies a time consistency score of 1, a covariance score of 0.92, a geometric proximity score of 0.84, and a link availability score of 1, its overall score is approximately 0.936. If its score is higher than other candidate satellites, Sat-6 is selected as the alternative reference satellite, and a shared RTN basis is regenerated. Subsequently, satellite pairs relying on the original reference satellite can quickly switch to the new local coordinate base path, avoiding large-scale pauses in updates.

[0128] In this embodiment, the 10-second short-term degradation duration is set based on the configuration monitoring continuity requirements and the typical link blockage duration; the abnormal measurement threshold adopts the standardized residual statistical threshold; the output jump limit is set based on the engineering velocity envelope of the low-Earth orbit relative flight; and the alternative reference star rating weight is selected through an offline validation set to achieve a combination that results in a high continuous monitoring output rate and a low false handover rate after the switch. All the above parameters are derived from simulation statistics, historical telemetry and control playback, and mission operation experience.

[0129] The simulation object was the same as in Example 2, and the test lasted for 6 hours. During this time, three packet loss rates (1%, 5%, and 10%) were randomly injected every hour, and several satellite pairs were subjected to continuous 10-second link interruptions. The control scheme still used independent pair-by-pair calculation with fixed-noise extended Kalman filtering, while the present invention's scheme used a unified state pool, a shared RTN base, relative state extended Kalman filtering, and degenerate continuous output. Under the statistical caliber of 100 repeated simulations, the effective continuous monitoring output rate was 96.7% for the control scheme and 99.3% for the present invention's scheme at 1% packet loss; 91.2% for the control scheme and 98.4% for the present invention's scheme at 5% packet loss; and 84.5% for the control scheme and 96.8% for the present invention's scheme at 10% packet loss. Under the condition of 5% packet loss and continuous 10-second link interruption, the false alarm rate for phase over-threshold was 7.8% for the control scheme and 2.1% for the present invention. The results show that the present invention can improve the continuity of configuration monitoring and the reliability of criteria under abnormal operating conditions, effectively reduce the probability of false alarms caused by link interruption and measurement disturbance, reduce the workload of manual re-judgment by operation and maintenance personnel, and improve the continuous availability and operational stability of constellation monitoring system in complex space environment.

[0130] Subsequently, based on the above calculation and processing results, the processor continues to execute the subsequent steps as described in Embodiment 1, such as restoring filter updates, refreshing the configuration monitoring matrix, maintaining low confidence markers, and performing parameter statistics. The specific execution logic is the same as in Embodiment 1, and will not be repeated here.

[0131] Example 4: Electronic Equipment This embodiment provides an electronic device for implementing the aforementioned method for rapid calculation of inter-satellite relative positions in a large-scale constellation telemetry and control system. The electronic device includes a memory, a processor, and a communication interface.

[0132] The memory stores computer programs, task configuration parameters, historical state cache, covariance matrix, nominal orbital parameters, and residual statistical data. The processor calls and executes the computer programs. The communication interface receives orbital root data input, GNSS position and velocity input, and inter-satellite relative measurement input, and outputs results such as inter-satellite relative position, relative velocity, phase difference, ascending node right ascension difference, configuration offset, and confidence level to an external telemetry and control system.

[0133] When the processor executes the computer program, it performs the following functions: acquires the orbital root data inputs, GNSS position and velocity inputs, and inter-satellite relative measurement inputs from multiple satellites; performs range verification, unified epoch time alignment, and availability marking on each input; and constructs a unified epoch multi-satellite absolute state pool; constructs a shared RTN coordinate base based on the absolute state of the reference satellite; constructs the relative state of two satellites and performs near-range analytical propagation or far-range two-body numerical differential propagation; performs relative state filtering and updating based on inter-satellite ranging and velocity measurements; outputs configuration monitoring parameters and performs short-term prediction degradation, alternative reference satellite switching, abnormal observation elimination, and result limiting projection under abnormal operating conditions.

[0134] In one optional implementation, the electronic device can be deployed on a ground-based telemetry and control center server, an edge telemetry and control node, or an on-orbit processing platform. Specifically, the processor can be a general-purpose central processing unit, a digital signal processor, or a dedicated processor with vector computing capabilities; the memory can be random access memory, read-only memory, flash memory, or a combination thereof; and the communication interface can be an Ethernet interface, a serial bus interface, or an inter-satellite / ground-satellite telemetry and control data interface. None of these different hardware configurations change the core processing logic of this invention, which uses a unified state pool and a shared RTN base for rapid inter-satellite relative position calculation.

[0135] The embodiments of the present invention have been described in detail above. Those skilled in the art can make reasonable adjustments to the parameter configuration and execution logic according to the constellation size, orbit configuration and measurement and control accuracy requirements without departing from the basic principles of orbit mechanics and filtering of the present invention. Such equivalent substitutions and adaptive adjustments all fall within the protection scope defined by the claims of the present invention.

Claims

1. A method for rapid calculation of the relative positions between satellites in a large constellation for telemetry and control, characterized in that, include: The system acquires orbital six-root data inputs, GNSS position and velocity inputs, and inter-satellite relative measurement inputs from multiple satellites. It performs range verification, unified epoch time alignment, and availability marking on each input, and converts available single-satellite inputs into inertial frame position and velocity states to construct a unified epoch multi-satellite absolute state pool. The RTN coordinate base and the transformation matrix from the inertial frame to the RTN coordinate system of the reference star are constructed based on the absolute state of the reference star in the multi-star absolute state pool, and the transformation matrix is ​​reused for all target stars related to the reference star. The relative position and relative velocity of the inertial frame are formed based on the absolute states of the reference star and the target star. The relative position and relative velocity of the inertial frame are projected onto the RTN coordinate system of the reference star to obtain the relative state of the two stars. The relative state of the two stars is propagated according to the orbital parameters of the reference star. For star pairs that meet the near-circular and near-distance conditions, CW analytical propagation based on the average angular velocity of the reference star is used. For star pairs that exceed the preset distance conditions, two-body numerical difference propagation of the absolute states of the reference star and the target star is used and then written back as the RTN relative state. A relative state observation mapping is constructed based on the currently available inter-satellite relative measurement inputs. The relative state of the two satellites is directly mapped to inter-satellite ranging and relative radial velocity. The propagated relative state of the two satellites is filtered and updated. Abnormal observations are eliminated and observation noise is dynamically adjusted based on the standardized residuals. The system outputs the relative position, relative velocity, phase difference, right ascension difference of the ascending node, and configuration offset of the satellites. In the event of missing input, abnormal reference satellite, abnormal filter residual, or abnormal output jump, it maintains continuous monitoring output through at least one of the following mechanisms: short-term prediction degradation, alternative reference satellite switching, abnormal observation rejection, and result limiting projection.

2. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 1, characterized in that, The process involves acquiring orbital root data inputs, GNSS position and velocity inputs, and inter-satellite relative measurement inputs from multiple satellites; performing range verification, unified epoch time alignment, and availability marking on each input; converting available single-satellite inputs into inertial frame position and velocity states; and constructing a unified epoch multi-satellite absolute state pool, including: The values ​​of the semi-major axis, eccentricity, and orbital inclination are verified by inputting the six orbital elements. The position modulus is verified to be within the preset orbital range by inputting the GNSS position and velocity. The relative distance is verified to be positive by inputting the inter-satellite relative measurement. Furthermore, when a line-of-sight unit vector is input, the line-of-sight unit vector is verified to meet the normalization condition. Each input is resampled to a unified solution epoch. Data with a time difference not exceeding a preset time difference is directly used. Data with a time difference within the compensation range is compensated using short arc propagation. Data with a time difference exceeding the compensation range is marked as unavailable for the current period. For satellites with only orbital root data input, the orbital root data is converted into inertial frame position and velocity; for satellites with GNSS position and velocity input, it is directly written into the multi-satellite absolute state pool; for inter-satellite relative measurement input, it is retained as an observation item for subsequent relative state filtering updates.

3. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 2, characterized in that, The process of converting the six orbital roots into inertial frame position and velocity includes: The off-proximity angle is solved by the mean anterior angle and the eccentricity. When the eccentricity is less than the preset near-circle threshold, the mean anterior angle is directly used as the off-proximity angle. When the iterative solution exceeds the preset convergence condition, it is switched to GNSS position and velocity input or extrapolation of the previous cycle state. The true anomaly angle, orbital radius, and orbital plane position and velocity are calculated based on the deviated anomaly angle, and converted into inertial frame position and velocity vectors through attitude rotation corresponding to the orbital inclination angle, right ascension of the ascending node, and argument of perigee. The semi-major axis is calculated from the position vector and velocity vector of the inertial frame, and the absolute state vector and average angular velocity of each satellite are generated based on the semi-major axis. The absolute state vector is formed by splicing the position vector and velocity vector of the inertial frame.

4. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 1, characterized in that, The step of constructing the RTN coordinate base and the transformation matrix from the inertial frame to the RTN coordinate system of the reference star based on the absolute state of the reference star in the multi-star absolute state pool, and then reusing the transformation matrix for all target stars associated with the reference star, includes: Construct a radial unit vector based on the position vector of the reference star's inertial frame, construct a normal unit vector based on the cross product of the position vector and velocity vector of the reference star's inertial frame, and then construct a tangential unit vector based on the normal unit vector and the radial unit vector. The radial unit vector, tangential unit vector, and normal unit vector are combined into a transformation matrix from the inertial frame to the RTN coordinate system; The transformation matrix is ​​constructed only once for each reference star, and the transformation matrix is ​​reused for the relative state projection of all target stars associated with that reference star; When the reference star is in an abnormal state or the RTN coordinate base degrades, pause the current periodic update that depends on the reference star or switch to an alternative reference star to regenerate the shared RTN coordinate base.

5. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 4, characterized in that, The propagation of the relative state of the binary stars based on the orbital parameters of the reference star includes: The relative states of the two stars are obtained by RTN projection of the position and velocity differences between the inertial frames of the reference star and the target star, which consist of the relative positions and velocities of the two stars in the RTN. When the relative position of the RTN meets the preset close distance condition, the relative state of the two stars is rapidly propagated using the CW analytical state transition matrix based on the average angular velocity of the reference star and the calculation period. When the relative position of the RTN exceeds the preset distance condition, the same two-body dynamics numerical propagation is performed on the absolute state of the reference star and the absolute state of the target star respectively. Then, the position difference and velocity difference of the inertial frame of the two are recalculated, and the relative state of the RTN after propagation is written back using the RTN coordinate base of the propagated reference star. For a satellite pair that returns from long-range mode to short-range mode, the pair is required to stably meet the short-range condition for multiple consecutive periods before switching to CW resolution propagation.

6. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 1, characterized in that, The step of constructing a relative state observation mapping based on currently available inter-satellite relative measurement inputs, directly mapping the binary star relative state to inter-satellite ranging and relative radial velocity measurements, and performing filtering updates on the propagated binary star relative state includes: For each star pair, a relative state filter is established with the relative position and relative velocity of the RTN as the state vectors. The predicted state of the filter is given by the propagated relative state of the RTN. The relative position of the RTN is mapped to an inter-satellite ranging quantity, and the directional projection relationship between the relative position of the RTN and the relative velocity of the RTN is mapped to a relative radial velocity quantity. Two-dimensional observation updates are performed when both distance and velocity measurements are available, and one-dimensional observation updates are performed when only distance measurements are available. When the standardized residual corresponding to the observation residual exceeds the preset threshold, the current observation is judged as an abnormal observation and will not participate in the current period's filter update.

7. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 6, characterized in that, The process of filtering and updating the relative states of the two stars after propagation, and removing outlier observations and dynamically adjusting observation noise based on standardized residuals, includes: A six-dimensional block matrix based on a white noise acceleration model is used as the process noise covariance matrix, so that the position component and the velocity component form a coupled propagation relationship according to the solution period; The filter gain is calculated based on the observation mapping Jacobian matrix at the propagation state, and the updated relative state and the updated covariance matrix are obtained using the filter gain, observation residual and prediction covariance matrix. When the condition number of the observation information matrix exceeds a preset threshold, the corresponding inverse matrix is ​​calculated using a pseudo-inverse. The average of the standardized residuals over the most recent effective periods is calculated. When the average of the standardized residuals continuously exceeds the upward adjustment condition, the observation noise scaling factor is increased. When the downward adjustment condition is continuously met, the observation noise scaling factor is decreased, and the observation noise scaling factor is constrained within a preset upper and lower limit range.

8. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 1, characterized in that, The output inter-satellite relative positions, relative velocities, phase differences, right ascension differences of ascending nodes, and configuration offsets include: Calculate the inter-satellite relative distance and relative velocity magnitude based on the RTN relative position and RTN relative velocity; The local phase difference is calculated based on the radial and tangential components in the semi-major axis of the reference star and the relative position of the RTN. When the local geometry does not meet the preset conditions, the phase difference is calculated based on the absolute true latitude difference between the reference star and the target star. The difference in right ascension of the ascending node is calculated based on the right ascension of the ascending node of the reference star and the target star. When only position and velocity input are available, the right ascension of the ascending node is calculated inversely from the angular momentum vector and the node vector. Based on the differences between the actual semi-major axis, actual phase, and actual right ascension of the ascending node of each satellite and the pre-stored nominal orbital parameters, a configuration offset is generated, which includes radial scale offset, along-orbit offset, and trans-orbit offset.

9. The method for rapid calculation of inter-satellite relative positions in a large constellation tracking and control system according to claim 1, characterized in that, Maintaining continuous monitoring output through at least one of the following mechanisms: short-term prediction degradation, alternative reference star switching, anomaly observation rejection, and result limiting projection, includes: When an input is missing and the duration does not exceed the preset short-term degradation duration, the prediction output for the current period is obtained by extrapolating the relative state of the previous period using the steps described in claim 1, which involve propagating the relative state of the two stars based on the reference star orbital parameters; when the duration exceeds the short-term degradation duration, the star pair is marked as low confidence. When the reference satellite is in an abnormal state, the highest-scoring alternative reference satellite is selected from the candidate satellites in the same orbital plane according to the credibility scoring rules consisting of time consistency, covariance size, geometric proximity and link availability, and the shared RTN coordinate base is regenerated. When the single-cycle jump of the output relative distance or phase difference exceeds the physical upper limit determined according to the preset relative velocity envelope, the output of this cycle will be clipped to the allowable variation range centered on the output of the previous cycle. When the same pair of satellites triggers amplitude-limited projection or the observation noise scaling factor reaches its upper limit for a long period of time, the pair of satellites will be switched to low-confidence monitoring mode.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for rapid calculation of inter-satellite relative positions of a large constellation telemetry and control system according to any one of claims 1 to 9.

Citation Information

Patent Citations

  • A Giant Constellation Target Tracking Method Based on Multi-Star Cooperative Passive Detection

    CN115790575B

  • Doppler real-time positioning and speed measuring method and system for large-scale low-orbit constellation

    CN115902982A

  • Autonomous orbit determination method for low earth orbit mega constellation based on star sensor and inter-star angle measurement

    CN119935158B