Fault-tolerant time synchronization method and device for low earth orbit satellite

Abstract

The invention discloses a time synchronization method and device applied between a ground control center and a low-orbit communication satellite, and aims to solve the technical problem that a satellite loses the time synchronization capability due to the failure of a global navigation satellite system (GNSS) component in the prior art. The method comprises the following steps: when a GNSS component works normally, a low-orbit satellite is mainly used for GNSS timing; and when the GNSS component is detected to be invalid or the signal is not stable, the system is automatically switched to a standby calibration mode. In the standby mode, the method utilizes a remote control instruction sent by a ground control center or a ground timestamp attached in a data packet, and combines a time drift prediction model constructed based on historical correction data to perform high-precision correction on a satellite local clock. By supporting a multi-mode time calibration strategy, the fault-tolerant capability and the time synchronization precision of the system under an extreme working condition are remarkably improved, so that the risk caused by time asynchronization in satellite task execution is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of satellite communication, in particular to a method for realizing high-precision time synchronization between a low-orbit satellite and the ground, and is especially suitable for a fault-tolerant time synchronization method that can still maintain reliable synchronization in the case of partial or complete failure of GNSS components. BACKGROUND

[0002] In the field of satellite communication, time synchronization is a key technology to ensure the interpretability and execution accuracy of mission instructions. The traditional satellite time synchronization scheme mainly relies on the GNSS (Global Navigation Satellite System) components carried by the satellite, and compares and calibrates the standard time signals of high-orbit navigation satellites (such as GPS and Beidou) with the same system time obtained by the ground station. In addition, some control systems will attach a timestamp to the sent instructions as an auxiliary calibration means.

[0003] However, in the actual operating environment of a low-orbit satellite, GNSS components may fail or be temporarily unavailable due to various factors such as electromagnetic interference, satellite blocking, or component failure. Once the GNSS signal source is interrupted, the traditional method cannot effectively calibrate the time, causing the satellite local clock to enter an uncontrolled free-running state. In existing technology, the drift rate of common crystal oscillator devices can reach 10ppm. This means that a drift of may occur every minute, and the cumulative error can reach 6ms or more after 10 minutes. This cumulative time error seriously affects the accuracy of satellite remote control operations and the timing of task execution, posing a huge risk to the stable operation of the satellite. Therefore, there is an urgent need for a fault-tolerant time synchronization method to ensure that high-precision ground-sky time consistency can be maintained when GNSS components are unavailable. SUMMARY

[0004] The present application aims to solve the above technical problems and proposes a completely new satellite fault-tolerant time synchronization method and device.

[0005] According to the embodiments of the present application, a low-orbit satellite antenna vehicle-mounted mechanism is provided, characterized by comprising: when the GNSS signal is normally available, calibrating the on-board local clock based on GNSS time observation and recording crystal oscillator drift history data; When the GNSS signal is unavailable, the system switches to a backup calibration mode, uses the timestamp information sent by the ground control center, and combines a time drift prediction model constructed based on the historical drift data to correct the local clock, the time drift prediction model comprising: (1) linear fitting based on multiple historical drift records to predict the drift amount, or (2) the time drift prediction model based on the state space model, a state vector of the state space model includes phase error, frequency deviation, drift rate, temperature coefficient, aging coefficient; the multi-source observation data are fused, predicted and corrected through a filtering algorithm, the multi-source observation data include one-way ground timestamp measurement, and further include two-way time transfer result and GNSS time observation; the model has adaptive clock control function, can dynamically adjust filtering parameters based on online estimation result of the oscillator Allan variance to compensate frequency deviation caused by external temperature, irradiation and aging, and realizes adaptive control of the local clock.

[0006] Preferably, the method further comprises correcting the time drift prediction model by using external excitation quantity, the external excitation quantity includes satellite-borne temperature sensor output, irradiation dosimeter reading and structure stress sensor data.

[0007] Preferably, the filtering algorithm is Kalman filtering, and is adaptive Kalman filtering, and a process noise covariance matrix Q is dynamically adjusted according to the online estimation result of the oscillator Allan variance.

[0008] Preferably, the linear fitting drift prediction method constructs a fitting model by using multiple clock drift data recorded during GNSS normal operation.

[0009] Another aspect of the application is to provide a low-orbit satellite fault-tolerant time synchronization device, characterized in that it comprises: a GNSS receiving module for receiving standard time signals from navigation satellites; a high-stability local crystal oscillator provides a local clock reference for the satellite; a time drift estimation module for recording clock drift data during GNSS normal operation and constructing a time drift prediction model; a synchronization control unit for switching between a main mode and a backup mode according to the GNSS receiving state, and controlling correction of the local clock; the correction in the backup mode is based on ground timestamp information and the time drift prediction model; the time drift prediction model is a linear fitting model based on historical drift data, or an adaptive clock control model based on a state space model.

[0010] Preferably, the time drift prediction model is a linear fitting model based on historical drift data or an adaptive clock control model based on a state space model.

[0011] Preferably, it further comprises an external excitation acquisition module for acquiring satellite-borne temperature sensor output, irradiation dosimeter reading and structure stress sensor data, and inputting the data to the adaptive clock control model for drift compensation.

[0012] Preferably, the multi-source observation module is further configured to acquire one-way ground timestamp measurement results, two-way time transfer measurement results and GNSS time observation results, and input the results into the filter algorithm to correct the local clock state.

[0013] Preferably, the filter algorithm is an adaptive Kalman filter, which can dynamically adjust the process noise covariance matrix according to the online estimation results of the oscillator Allan variance, so as to maintain high-precision ground-air time synchronization under external environmental change conditions.

[0014] Preferably, when it is detected that the GNSS signal interruption lasts for more than a preset time length or the predicted drift amount exceeds a threshold, the synchronization control unit automatically enters a fault-tolerant recovery process, which includes triggering an alarm notification of the ground station, restarting the GNSS receiving module, and maintaining a time drift prediction model-based time correction maintaining mode during the recovery.

[0015] According to the embodiments of the present specification, the time synchronization with fault tolerance has the ability to combine multi-mode calibration strategies and dynamic time drift prediction models. Specifically, the method does not rely on GNSS time correction alone, but records the calibration deviation data of the local clock when GNSS is working normally; when the GNSS component fails, the system can automatically switch to the standby mode, dynamically build a time drift prediction model using the historical data accumulated in advance, and combine the time stamp sent by the ground to accurately correct the local clock. This intelligent adaptive mechanism ensures high-precision time synchronization in extreme working conditions, significantly improving the reliability and robustness of the system.

[0016] Compared with the prior art, the present application has the following remarkable beneficial effects: 1. High reliability and fault tolerance: The present application can still guarantee the high-precision consistency of ground-air time in the extreme case of GNSS component failure or unavailability, thereby avoiding the system out-of-control caused by GNSS interruption in the traditional scheme, and greatly improving the reliability of satellite operation.

[0017] 2. Multi-mode adaptive calibration: The multi-mode calibration strategy of the present application enables the system to dynamically select the best synchronization scheme according to the GNSS state, rather than using a single fixed calibration method, thereby supporting time synchronization in various complex working conditions.

[0018] 3. Accurate compensation and error compression: By combining the time stamp and the historical drift prediction model, the present application can effectively compensate for the drift of the local clock and control the synchronization error to a very low level. For example, within 120 seconds after GNSS failure, the maximum drift can be controlled to 6ms. By combining the correction of the ground time stamp, the synchronization error can be further compressed to 4ms. This performance is more than 2 times better than the existing fault-tolerant scheme, completely meeting the time error requirements of the remote control system. Requirements.

[0019] 4. Reduce mission risks: By solving the time synchronization problem after GNSS failure, this invention reduces the risks caused by time asynchrony during satellite mission execution, which helps to ensure the accuracy of command execution and extend the satellite's operational life.

[0020] 5. By using AOC+multi-source Kalman fusion, high-precision maintenance is achieved for a longer period of time under GNSS failure. At the same time, it has the ability to actively compensate for external disturbances such as temperature and irradiance, enabling the satellite to maintain millisecond-level synchronization accuracy under complex operating conditions, which is significantly better than existing free-running and linear prediction schemes. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments or related technologies of this specification, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 This specification provides a schematic diagram of the system structure of a satellite synchronization method according to one embodiment. Figure 2 This is a schematic diagram of the primary and backup correction mechanism of a satellite synchronization method provided in one embodiment of this specification. Detailed Implementation

[0023] The technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] It should be noted that the terms "comprising" and "having," and any variations thereof, in the embodiments and drawings of this specification are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or devices.

[0025] Example 1, such as Figure 1 As shown, the method of the present invention is implemented based on the following system structure: Satellite-side components include a GNSS receiver module, a high-stability local crystal oscillator, a time drift estimation module, and a synchronization control unit. The high-stability local crystal oscillator is, for example, a TCXO / OCXO.

[0026] The GNSS receiver module is used to receive standard time signals from high-orbit navigation satellites; the high-stability local crystal oscillator provides the local clock basis for the satellite; the time drift estimation module is used to predict the clock drift trend based on historical synchronization record data; and the synchronization control unit is responsible for determining the current synchronization mode to be adopted based on the status of the GNSS module and controlling the execution of clock correction.

[0027] Ground-based: Includes the control center master clock and communication module. The control center master clock is synchronized with the high-orbit GNSS time, providing a standard time reference; the communication module includes the current ground timestamp when sending remote control commands or data packets to the satellite. .

[0028] Time synchronization process: After establishing a communication link between the satellite and ground stations and completing initial clock calibration, the system continuously monitors the quality and reception status of the GNSS signal to determine if it is in an ideal synchronization environment. Based on this real-time status assessment, the time synchronization process is designed with a primary mode and a backup mode. Even in primary mode, the system continuously accumulates data for backup mode, ensuring a seamless and accurate switchover in case of emergencies.

[0029] The time synchronization process of this invention is divided into a primary mode and a backup mode based on the operating status of the GNSS component. Even in the primary mode, the system continuously accumulates data for the backup mode, thereby ensuring a seamless and accurate switchover in case of emergencies.

[0030] Step 1: GNSS Time Synchronization (Main Mode) When the GNSS components on the satellite are functioning properly, the system is in main mode. In this mode, the satellite periodically receives standard time from the GNSS signal source. And use this as a reference to calibrate the local clock. .

[0031] Furthermore, during each GNSS calibration, the system accurately records the clock offset data resulting from that calibration. This data is stored in the history of the time drift estimation module, becoming the raw material for building the time drift prediction model.

[0032] At the same time, when the ground control center sends instructions or data packets, it will include the current ground timestamp. When the satellite receives this data packet, it will record the local reception time. By comparison and , the system can estimate the communication propagation delay of the data packet . This can not only be used to assist in correcting the local clock in the main mode (i.e. ), but also provide more accurate and comprehensive basic data for the construction of the subsequent drift model.

[0033] Step 2: Fault-tolerant time synchronization when GNSS fails (backup mode) When the synchronization control unit detects that the GNSS receiving module has failed, such as GNSS signal loss lasting more than the preset 10 seconds, or the GNSS component self-test fails, the system will automatically switch from the main mode to the fault-tolerant backup mode.

[0034] In backup mode, the time synchronization process is as follows: Dynamic drift model construction: the system uses the clock deviation data accumulated in the past N (e.g. N = 10) GNSS correction records accumulated in step 1 to dynamically construct a clock drift trend function. This function can be linearly fitted using the least squares method, represented as .

[0035] Where represents the predicted clock drift since the last GNSS correction, is the time interval, and are the coefficients fitted from historical data.

[0036] Local time correction: when GNSS fails, the system extracts the ground timestamp from the ground data packet and records the reception time . Using the above dynamic drift prediction model, the time interval since the last GNSS correction is calculated.

[0037] The corresponding predicted drift amount .

[0038] Finally, through the correction formula: , the satellite local clock is corrected to keep high precision with the ground time.

[0039] Alarm and recovery: if the predicted drift value calculated by the drift model exceeds the preset allowable error range (e.g. > 5ms), the system will automatically send an alarm packet to the ground station and attempt to restart the GNSS module to restore normal operation.

[0040] Suppose that when GNSS is working normally, the satellite corrects the local clock every 10 seconds, and the recorded drift values are as shown in the following table: Table 1 Local clock drift measurement data in GNSS normal mode By linear fitting the above data, the drift trend function can be obtained as follows: (unit: ms). The coefficients of the fitted function are .

[0041] When the GNSS component fails, if the ground control center sends a data packet after 60 seconds, the time since the last GNSS time correction has passed 60 seconds, and the system uses the predicted drift calculated by the model as follows: Therefore, after receiving the ground timestamp , the system will correct the local clock using .

[0042] The existing scheme relies entirely on the free-running crystal oscillator after the GNSS fails, resulting in uncontrolled time drift and potentially rapid accumulation of errors. For a crystal oscillator with a drift rate of , the clock drift can accumulate up to or even higher within 10 minutes (600 seconds) after the GNSS fails. This error level far exceeds the requirements of most satellite remote control systems for time error .

[0043] The present invention achieves high-precision fault-tolerant time synchronization in the case of GNSS failure through the combination of "ground timestamp + dynamic drift model". This method can predict the maximum drift within 120 seconds as (0.05x120)s=6ms, but after combining the actual received ground timestamp for correction, the synchronization error can be further compressed to within 4ms. This performance is superior to existing fault-tolerant schemes and fully meets the stringent requirements of remote control systems for time error of 5ms. The ability to maintain high-precision synchronization even after GNSS failure is a significant technical advantage of the present invention.

[0044] Example Two, based on Example One, this embodiment further proposes a satellite fault-tolerant time synchronization method based on adaptive oscillator control (AOC) and multi-source Kalman fusion to significantly improve time keeping precision and environmental adaptability when GNSS signals are unavailable.

[0045] The on-board time synchronization module of this embodiment establishes the following state space model: Where: : current phase error; : frequency offset; : drift rate; : temperature coefficient, used to reflect the sensitivity of oscillator frequency to temperature change; : aging coefficient, used to reflect the long-term frequency drift trend.

[0046] State transition equation: where, is the state vector, may contain: phase error, frequency offset, drift rate, temperature coefficient or aging coefficient. is the state transition matrix, which describes the relationship between the state variables from the current time to the next time when the system is not disturbed by external input. is the external excitation, including satellite-borne temperature sensor output, dosimeter reading and structural stress sensor data; is the process noise.

[0047] Observation equation: In the above formula, the observation comes from multiple time measurement sources: GNSS time observation (available normally); two-way time transfer results; one-way ground timestamp measurement. In the above formula, is the observation noise, whose covariance is estimated in real time according to the link signal-to-noise ratio, data frame jitter statistics and equipment delay fluctuation. In the above formula, is called the observation matrix in the state space model, which is used to convert the state vector into the form of observable.

[0048] For example, GNSS observation directly gives the phase error and frequency offset : In this way, the Kalman filter will project the observation back to the state space according to to correct the estimate of .

[0049] In the GNSS signal available stage, the system uses GNSS observation to continuously update the state quantity, and simultaneously identifies the temperature coefficient and the aging coefficient The oscillator mathematical model is dynamically corrected to realize long-term precision optimization of adaptive clocking (AOC).

[0050] When GNSS signals are lost, the system automatically switches to the AOC maintenance mode, uses the bidirectional link and the one-way ground timestamp data, and fuses the prediction and correction of the current clock state through multi-source Kalman filtering to realize long-time high-precision maintenance. Since the temperature, irradiation and stress and other external excitation quantities are explicitly introduced into the state model, the algorithm can actively compensate for the drift caused by environmental disturbances, and improve the time stability under conditions of rapid temperature change, irradiation change and the like.

[0051] The multi-source Kalman filtering prediction and correction principle is divided into three steps: S1, prediction: according to the state transition equation and the last step estimation Predict the next state: Indicates the state estimation that has been updated with the current observation value at the current time Indicates the state predicted based on the current time information at the next time

[0052] S2, covariance prediction Wherein is the adaptive process noise matrix, which is obtained by estimating the oscillator noise through Allan variance.

[0053] S3, update When new observation arrives: ; ; ; In the above formula, is the Kalman gain (Kalman Gain), which is a weight matrix used to determine the influence of the prediction value and the observation value on the final estimation. is the error covariance matrix, which represents the uncertainty of the prior state estimation. is the observation matrix, which maps the state vector to the observation space (for example: maps “phase error, frequency deviation” to GNSS observation value). is the observation noise covariance, which can be dynamically adjusted according to the SNR of GNSS and the link jitter. Through this prediction + correction, the filter can fuse the information of bidirectional link + one-way timestamp + temperature / irradiation external excitation.

[0054] ​​ Including phase error, frequency deviation, drift rate, temperature coefficient, aging coefficient; is the current optimal estimate, which integrates the prediction model (AOC adaptive oscillator); and multi-source observation (GNSS, two-way link, ground timestamp). Subsequently used: update the control of the oscillator, short-term prediction maintenance when GNSS is lost.

[0055] In the filter implementation, the adaptive noise covariance adjustment strategy is adopted, and the process noise matrix , etc. are dynamically adjusted according to the online estimation results of the oscillator Allan variance to ensure that the filter can quickly converge and has anti-mutation ability under different working conditions.

[0056] Ground testing and on-orbit simulation results show that: within 120 seconds of GNSS failure, The synchronization error is not more than 1.5 milliseconds; within 900 seconds of GNSS failure, The synchronization error is not more than 3 milliseconds; under the conditions of temperature ±10℃ mutation and irradiation dose change of 2krad, the time drift suppression rate is increased by more than 20% compared with the uncompensated scheme.

[0057] In order to verify the effectiveness of the adaptive clocking (AOC) + multi-source Kalman fusion method, a low-orbit satellite equipped with a 10MHz OCXO (constant temperature crystal oscillator) is selected as the test object, and the frequency stability of the crystal oscillator at room temperature is / ℃ (within 1 second), the temperature coefficient is about , and the aging rate is about / day. Assume that the initial local clock phase error ; when GNSS fails, the temperature slowly rises from 25℃ to 35℃, and the irradiation dose increases ; the two-way link delay jitter is , and the one-way uplink timestamp accuracy is ; the initial covariance of the Kalman filter is set to: The state space vector is defined as: The relative frequency deviation caused by temperature is: ; The frequency deviation caused by aging during Δt is: Substitute: Therefore, the total relative frequency deviation is: .

[0058] The resulting accumulated phase error is Within 900 seconds For process noise, the following approximation is obtained based on the 1-second Allan deviation: bidirectional link within 900s ≈4.5ns.

[0059] For observation noise, bidirectional link One-way stamp: Compared to local oscillator prediction (ns level), the observed noise is approximately 10 ohms higher. 6 times.

[0060] During the GNSS failure phase, the phase deviation given by the Kalman filter prediction step is on the order of nanoseconds, while the observation noise is on the order of milliseconds. Therefore, during the filter update process, the weights are automatically tilted towards the local oscillator prediction to ensure the stability of the filter output. As shown in the calculation above, at time 900s: Predicted value After the update ≈46ns, uncertainty ≈4.5ns (3σ≈13.5ns). In this embodiment, the AOC+ multi-source Kalman fusion method calculates the error, using temperature and irradiation as external excitation inputs. Real-time correction of drift rate through filtering During the T=900-second hold period, the bidirectional link and unidirectional timestamp observations update the state every 60 seconds; adaptive adjustment. This technical solution performs real-time correction of the state vector to compensate for drift, thereby maintaining time synchronization accuracy under changing environmental conditions. Multi-source Kalman filtering is used to process temperature sensor readings. As external stimulus input Predict the corresponding frequency drift Each time an observation is updated (every 60 seconds), the filter is corrected based on the GNSS or ground timestamp. The residual = prediction drift - actual drift, obtained from optimal fusion of the filters. Substituting these values ​​into the aforementioned multi-source Kalman filter prediction and correction principle formula, the filter is adjusted... and The residual caused by temperature abrupt changes can thus be effectively suppressed.

[0061] As a comparative example, if uncompensated free operation is adopted: crystal oscillator drift = temperature drift + aging drift, then: For observation noise, bidirectional link Total frequency deviation ≈5.104×10 -11 s / s (i.e. 51 ps / s), corresponding to a time error growth rate of about: Cumulative drift over T = 900 s: ≈45.9 ns (which would be smaller if noise accumulation were ignored, but in practice would be amplified).

[0062] A linear fit correction is made using historical drift data, assuming a fit residual of σ ≈ ±3 ms; due to temperature jumps and irradiation, the time error over 900 s is ≈3.8 ms (measured simulation result).

[0063] Technical effect comparison. Compare the error calculation of Example 1, Example 2, and the prior art (using a 10 MHz OCXO, temperature variation ±10°C, bidirectional link update period 60 seconds): Table 1 Performance comparison table of time synchronization methods under GNSS failure conditions As can be seen from the table, under the same holding time, the 3σ synchronization error of the present embodiment is reduced by about 39.5% compared to Example 1, and more than 98% compared to a free-running crystal oscillator, and still maintains a synchronization accuracy of <3 ms under temperature variation and irradiation interference.

[0064] If the traditional linear extrapolation method (only is shown) is used, when the temperature and irradiation change simultaneously, it is easy to appear under-compensation or over-correction, so that the time error within 900 s can reach the order of microseconds. By the AOC + multi-source Kalman filtering method described in the present embodiment, the time holding error can be controlled within tens of nanoseconds under the same conditions, and the accuracy is improved by about 10 4 times compared to the traditional method.

[0065] The linear fitting method used in Example 1 has limited accuracy in long-time GNSS failure and complex environments such as temperature and irradiation, but it is still applicable in relatively stable environments, short-time GNSS failure, or limited computing resources, and can provide millisecond-level time holding capability with relatively low implementation complexity; while Example 2 can still compress the synchronization error to within milliseconds when the external environment changes significantly through adaptive clocking and multi-source Kalman fusion, significantly improving the robustness and accuracy of the system. The two methods complement each other and meet the layered application needs under different task conditions.

[0066] Those skilled in the art can understand that the drawings are only schematic diagrams of an embodiment, and the modules or processes in the drawings are not necessarily necessary to implement the present application.

[0067] Those skilled in the art can understand that the modules in the device in the embodiments can be distributed in the device in the embodiments as described in the embodiments, or can be changed to be located in one or more devices different from the embodiments. The modules in the above embodiments can be combined into one module, or can be further split into multiple sub-modules.

[0068] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A low earth orbit satellite fault-tolerant time synchronization method, characterized by, Comprise: When GNSS signal is normally available, calibrate the on-board local clock based on GNSS time observation, and record the history data of crystal oscillator drift; When GNSS signal is not available, the system switches to the backup calibration mode, uses the time stamp information sent by the ground control center, and combines the time drift prediction model constructed based on the history drift data to correct the local clock, the time drift prediction model comprises: (1) Linear fitting based on multiple history drift records to predict the drift amount, or (2) Time drift prediction model based on state space model, the state vector of the state space model includes phase error, frequency deviation, drift rate, temperature coefficient, aging coefficient; Through filtering algorithm, multiple source observation data are fused, predicted and corrected, the multiple source observation data include one-way ground time stamp measurement, and also include two-way time transfer result and GNSS time observation; The model has adaptive clock control function, can dynamically adjust the filtering parameters based on the online estimation results of the oscillator Allan variance to compensate the frequency deviation caused by external temperature, irradiation and aging, and realizes adaptive control of the local clock.

2. The method of claim 1, wherein, The method further comprises correcting the time drift prediction model by using external excitation amount, the external excitation amount includes on-board temperature sensor output, radiation dosimeter reading and structure stress sensor data.

3. The method of claim 1, wherein, The filtering algorithm is Kalman filtering, and is adaptive Kalman filtering, and the process noise covariance matrix Q is dynamically adjusted according to the online estimation results of the oscillator Allan variance.

4. The method of claim 1, wherein, The linear fitting drift prediction method uses multiple clock drift data recorded during normal GNSS operation to construct a fitting model.

5. A low earth orbit satellite fault-tolerant time synchronization apparatus, characterized by, Comprise: GNSS receiving module, for receiving standard time signal from navigation satellite; High-stability local crystal oscillator, providing local clock reference for satellite; Time drift estimation module, for recording clock drift data and constructing time drift prediction model during normal GNSS operation; Synchronization control unit, for switching between main mode and backup mode according to GNSS receiving state, and controlling correction of local clock; The correction in the backup mode is based on ground time stamp information and the time drift prediction model; The time drift prediction model is a linear fitting model based on history drift data, or an adaptive clock control model based on state space model.

6. The apparatus of claim 5, wherein, The time drift prediction model is a linear fitting model based on history drift data or an adaptive clock control model based on state space model.

7. The apparatus of claim 6, wherein, Further comprising an external excitation acquisition module for acquiring on-board temperature sensor output, radiation dosimeter reading and structure stress sensor data, and inputting the data into the adaptive clock control model for drift compensation.

8. The apparatus of claim 7, wherein, Further comprising a multiple source observation module for acquiring one-way ground time stamp measurement result, two-way time transfer measurement result and GNSS time observation result, and inputting the results into filtering algorithm to fuse and correct the state of the local clock.

9. The apparatus of claim 8, wherein, The filter algorithm is an adaptive Kalman filter, which can dynamically adjust the process noise covariance matrix according to the online estimation result of the oscillator Allan variance, so as to maintain high-precision ground-air time synchronization under the condition of external environmental changes.

10. The device of any one of claims 5 to 9, wherein, When it is detected that the GNSS signal interruption lasts for more than a preset time length or the predicted drift amount exceeds a threshold, the synchronization control unit automatically enters a fault-tolerant recovery process, which includes triggering an alarm notification of the ground station, restarting the GNSS receiving module, and maintaining a time drift prediction model-based time correction maintaining mode during the recovery.

Images