SPACE VEHICLE WITH AN AUTONOMOUS ORBIT CONTROL MODULE AND ANTI-COLLISION MODULE AS WELL AS AUTONOMOUS COLLISION PROTECTION AND STATION PRESERVATION METHOD FOR A SPACE VEHICLE

Description

technical field

[0001] This application relates to a spacecraft equipped with an autonomous computer control system comprising both an autonomous orbit control module, for calculating station-keeping maneuvers aimed at keeping the craft within a mission window, and a collision avoidance module, for calculating station-keeping and avoidance maneuvers aimed at avoiding any collision between the spacecraft and any secondary objects in its orbit or nearby.

[0002] The invention relates more particularly to a satellite operating in low Earth orbit, as is the case with Earth observation satellites. Previous art

[0003] To perform its mission, a satellite, for example in low Earth orbit, must remain within a station-keeping window. Station-keeping operations can be performed automatically, controlled from the ground, or autonomously. In the latter case, the spacecraft has its own computing and orbit control systems, including an Autonomous Orbit Control (AOC) computer module.

[0004] The invention focuses more specifically on reducing the risk of collision for a spacecraft equipped with a COA (Crew Oriented Atomic) system with other non-maneuvering objects in conjunction with the spacecraft's orbit. These collision risks must be managed and can affect the mission.

[0005] The growing population of debris, as a consequence of Kessler syndrome and the detection of smaller debris, poses the challenge of automatic or autonomous management of collision avoidance.

[0006] Collision management requires processing and calculation loops which, today, are handled by a ground segment: Reception of Conjunction Data Messages (known as "CDMs", CDM being the acronym for Conjunction Data Message ) transmitted by an international centralised monitoring organisation such as EUSST (acronym for " EUropean Space Surveillance and Tracking " or JSpoc (acronym for “Joint Space Operation Center”), CDM processing; calculation of collision avoidance maneuvers, remote transmission of said collision avoidance maneuvers to the satellite.

[0007] These processing loops and the exchanges between the ground segment and the spacecraft utilize the limited time intervals during which the satellite can establish communication with the ground station, creating a certain latency in the implementation and execution of avoidance maneuvers. Avoidance maneuvers are generally performed that may later prove unnecessary and degrade the satellite's performance in carrying out its mission. Finally, the coupling between avoidance and the demanding constraints of the mission requires a constant responsiveness effort.

[0008] The thesis of Chiara Maria Paola Rusconi entitled: “ ASTERIA: Integration of Risk Collision Management in Autonomous Orbit Control "Discloses an embedded system and a method for the autonomous management of both station-keeping of the spacecraft and collision avoidance. Such a system still needs improvement."

[0009] The invention aims to provide an embedded system and a method for the autonomous management of both station-keeping of the spacecraft and collision avoidance. This autonomous management is advantageously achieved by taking into account onboard implementation constraints so as not to hinder the execution of tasks necessary for the operability of the spacecraft, while guaranteeing maximum safety. Description of the invention

[0010] To this end, the invention proposes a method for managing collision avoidance and station-keeping of a spacecraft, the spacecraft comprising a propulsion and attitude control system, a navigation system including a GNSS, and telecommunications components for data exchange with a ground segment. The method according to the invention is characterized in that: The spacecraft receives conjunction data messages, hereinafter referred to as CDMs, sent by a ground segment, said CDMs being related to at least one close approach with a secondary object likely to collide with the spacecraft, each of said CDMs describing identification, position, velocity, size and covariance parameters of the secondary object as well as a date of closest passage called the TCA date, at each orbit, at a defined position on the orbit, for example at each ascending node, for example detected by the navigation system (201) and an onboard computing process, an autonomous orbit control module, hereinafter referred to as COA, is activated on board the spacecraft for the establishment of a plan corresponding to the current orbit, called the current safe plan, the current safe plan being in the form of a plan of maneuvers to maintain position at least on a horizon called the risk horizon,The risk horizon includes the current orbit and extends to the orbit containing the nearest TCA date among the TCA dates of the received CDMs. A filtering module on board the spacecraft performs a preliminary filtering of the received CDMs based on geometric and / or temporal criteria to establish a list of risky CDMs. If said preliminary filtering results in at least one risky CDM, a collision risk management module, hereinafter referred to as ACA, is activated on board the spacecraft. This module estimates (on board the spacecraft) a collision risk level, called the onboard risk level, based on onboard navigation data (position, speed, covariance) provided by the GNSS propagated at the TCA date of said risky CDM, and develops a maneuver plan over a control horizon, called the current hold-on and avoidance plan.To satisfy both the requirement to remain within a mission window and the reduction of the risk of collision with the secondary object, the control horizon comprises a predetermined number of orbits, the current orbit being less than the risk horizon, the current hold and avoidance plan being developed as follows: if the previously assessed onboard risk level is less than or equal to a predefined risk threshold, called the onboard risk threshold, the ACA retains the current safe plan as the current hold and avoidance plan; if the previously assessed onboard risk level is greater than the onboard risk threshold, the ACA develops a new maneuver plan on the control horizon from the current safe plan.by removing at least one station-keeping maneuver from said safe current plan and / or by replacing at least one station-keeping maneuver from said safe current plan with one or more additional maneuvers for station-keeping and avoidance, referred to as avoidance maneuvers, the new maneuver plan becoming the station-keeping and avoidance current plan. When the ACA develops this new maneuver plan from the safe current plan, the ACA makes successive modifications starting from the safe current plan, each modification providing a new version of the maneuver plan, and, at each modification made, the ACA reassesses the level of risk on board with the new version of the maneuver plan and based on onboard navigation data provided by the GNSS propagated on the TCA date of said at-risk CDM.

[0011] Note that throughout the patent application, the expression " positioning maneuver" is used to refer to any maneuver calculated by the COA. In contrast, the expression " evasive maneuver " is used to refer to any maneuver calculated by the ACA, although such a maneuver aims to ensure not only avoidance but also remaining within the mission window.

[0012] Note that the risk horizon extends from the current date to the first TCA date associated with the secondary object. It serves to indicate that there is a risk and to define its duration. The control horizon concerns the calculation horizon for avoidance maneuvers, typically a few orbits. This horizon will "slide" with each orbit until it reaches the TCA date. The benefit is to allow for greater robustness in the maneuver calculations. As will be better understood later, it is preferable that this control horizon not be too long, to allow for updates to changes in the secondary object's parameters that could affect the risk level, but also not too short, to avoid systematically recalculating station-keeping and avoidance plans, since several orbits are required to perform an avoidance maneuver, especially if the spacecraft is equipped with a propulsion system providing very low thrust.

[0013] Advantageously, the use of onboard data such as navigation data, updated in real time, known by the satellite and not by the ground segment, presents a significant interest in the approach to avoidance, in particular by allowing to improve the responsiveness of the avoidance and by carrying it out in the last hours, where the uncertainty on the data representative of the CDMs is the lowest.

[0014] Advantageously, since the satellite can use its onboard calculated position, the need to propagate a subsequent calculated position is eliminated, and the accuracy of the calculations is increased. Even more advantageously, the avoidance maneuver sequences implemented by the satellite are more optimal, allowing for the optimization of its nominal mission execution (in terms of expected performance).

[0015] The preliminary filtering of received CDMs and the use of spacecraft navigation data calculated onboard, in real time or near real time, according to the invention, allow for finer filtering of non-risky CDMs, thus reducing onboard computational costs. Furthermore, the calculations of avoidance maneuvers by the ACA benefit from increased accuracy. The calculations can be performed onboard, allowing for greater responsiveness, without the latency between calculations and their execution that is generally characteristic of a ship-to-ground loop. To further reduce onboard computation time, it is also possible to reduce the propagation models used by the ACA, the impact of this reduction remaining acceptable at a time close to the TCA date (i.e., on the control horizon).

[0016] This increased accuracy, along with the use of reduced propagation models, makes it possible to decrease the resources required to develop a maneuver plan that simultaneously maintains position within the mission window and reduces the risk of collision with the secondary object. Similarly, developing the current position-holding and avoidance plan from the current safe-running plan by removing or modifying one or more position-holding maneuvers limits both the calculations required and the impact on the mission, since any avoidance maneuvers calculated by the ACA are scheduled during time slots initially used for position-holding maneuvers—that is, outside of the time slots reserved by the mission.

[0017] Therefore, the spacecraft's resources are sufficient for the development of such a maneuver plan, and the management of the risk of collision and maintaining position can be ensured autonomously by the spacecraft, without intervention from the ground.

[0018] According to one possible feature of the invention, when the ACA develops the new maneuver plan from the current risk-free plan, the ACA successively removes, in reverse chronological order from the TCA date of said risky CDM, the station-keeping maneuvers from the current risk-free plan, each removal leading to a new version of the maneuver plan, and the ACA re-evaluates, at each removal, the level of risk on board with said new version of the maneuver plan.The ACA proceeds in this way as long as the assessed on-board risk level remains above the on-board risk threshold and the number of maneuvers removed is less than a predetermined maximum number of authorized removals, the ACA stopping removals as soon as the assessed on-board risk level is below the on-board risk threshold, the latest new version of the maneuver plan, which has led to obtaining an on-board risk level below the on-board risk threshold, becoming the current plan for maintaining position and avoidance.

[0019] When one or more eliminations of station-keeping maneuvers from the current, risk-free plan sufficiently reduce the risk and allow for the definition of a current station-keeping and avoidance plan without the need to calculate new maneuvers, it is possible that this current station-keeping and avoidance plan may, in practice, lead to a slight deviation from the nominal mission window, but within an expanded mission window. The inventors have demonstrated, however, that this potential window deviation remains acceptable, especially since the ACA's current station-keeping and avoidance plan can be taken into account by the COA during its next activation (at the following orbit).

[0020] According to a possible feature of the invention, if the re-evaluated edge risk level remains above the edge risk threshold after removing a number of maneuvers from the safe running plan equal to the maximum predetermined number of permitted removals, the ACA removes all maneuvers from the safe running plan and calculates a set of avoidance maneuvers on the control horizon, this calculation taking into account both holding and avoidance, said set of avoidance maneuvers becoming the holding and avoidance running plan.

[0021] According to a possible feature of the invention, for the calculation of any avoidance maneuver, the ACA solves a constrained optimization problem with the objective of minimizing a CoPoC risk function and with the constraint of staying within the mission window, the CoPoC function corresponding to a maximum probability of collision within predefined ranges of contraction and expansion of the covariances of the spacecraft and the secondary object at the TCA date.

[0022] According to one possible feature of the invention, the ACA refers the current plan for holding in position and avoidance back to the control system for execution.

[0023] According to one possible feature of the invention, the preliminary filtering of received CDMs includes a temporal filtering step consisting of selecting, from among the received CDMs (or possibly from among the CDMs retained after a geometric filtering step described below), the CDM(s) whose time difference up to the TCA is less than a predetermined number of hours. This predetermined number of hours may be 24 or 48 hours. It is preferably configurable and modifiable remotely.

[0024] According to one possible feature of the invention, the preliminary filtering of the received CDMs includes a geometric filtering step consisting of, for each of the received CDMs (or possibly each of the CDMs retained at the end of the temporal filtering step described above): Calculate the distance between a predicted position of the spacecraft and a predicted position of the secondary object at the TCA date of said CDM, and select the CDM(s) for which the previously calculated distance is less than a predetermined filtering distance. This predetermined filtering distance is, for example, 10 km, 15 km, or 20 km. It is preferably configurable and modifiable from the ground.

[0025] According to one possible feature of the invention, the propagation of onboard navigation data provided by the GNSS is achieved using a propagation model based on a model of the Earth's gravitational potential with a limited number of zonal and tesseral terms, a lunisolar perturbation model, and an atmospheric model that can be parameterized according to solar activity data and drag parameters. The models, data, and parameters used by the propagation model are provided to the spacecraft during its deployment; they are regularly updated by the ground segment, for example, monthly or whenever a significant change, particularly concerning solar activity, is observed. They are systematically sent to the spacecraft along with the CDM data transmission.

[0026] According to one possible feature of the invention, any estimate of the level of risk on board includes: a calculation of the propagation of the orbit and covariance of the spacecraft up to the TCA date of said at-risk CDM, based on the orbit calculated on board the spacecraft provided by the GNSS and the safe maneuver plan (if this is the first assessment of the onboard risk level before the development of a station-keeping and avoidance plan) or the new version of the maneuver plan (if this is a reassessment of the onboard risk level following a modification of the current safe plan during the development of the current station-keeping and avoidance plan), a calculation of the assumed propagation of the orbit and covariance of the secondary object at the TCA date of said at-risk CDM, an adjustment of the TCA date, and a correction of the orbits and covariances of the spacecraft and the secondary object propagated to the adjusted TCA date,The assessment of the onboard risk level is based on the orbits and covariances thus propagated to the adjusted TCA date.

[0027] According to one possible feature of the invention, the spacecraft receives a mission plan from the ground segment. This mission plan defines mission slots reserved for the mission, slots prohibited for the mission and maneuvers to satisfy system constraints of the spacecraft, such as battery recharging, and free slots that can be used for maneuver placement, such as station-holding maneuvers calculated by the COA and the avoidance maneuver(s) calculated by the ACA. In an emergency, mission slots could also be used for maneuvers, in which case the mission would be degraded.

[0028] Preferably, the avoidance maneuver(s) calculated by the ACA are scheduled for free slots in the mission plan prior to a latest avoidance date, with the latest avoidance date preceding the TCA date by a predetermined number of orbits or hours. This predetermined number of orbits or hours could, for example, be between two and four orbits or between two and four hours. It is preferably configurable and modifiable by the ground segment.

[0029] According to one possible feature of the invention, for the development of the current positioning and avoidance plan, the mission window is either a nominal mission window or an extended mission window consistent with the mission. Both the nominal and extended windows are provided by the mission sponsor, and therefore by the ground. The use of an extended window is required when no solution to the constrained optimization problem of calculating an avoidance maneuver using ACA is satisfactory with regard to risk reduction.

[0030] This allows for avoidance maneuvers with an acceptable departure from the nominal mission window. Since the current plan maneuvers for station holding and avoidance are better calculated (whether station holding maneuvers calculated by the COA and retained in the final plan or avoidance maneuvers calculated by the ACA) than would be done on the ground without knowledge of real-time onboard navigation data and upcoming maneuvers, the size of the window can be significantly reduced.

[0031] According to one possible feature of the invention, for the verification of the spacecraft's maintenance within the mission window (nominal or extended), the ACA uses a predictive model based on a quadratic evolution of the spacecraft's orbital position, which predictive model is provided to the ACA by the COA with the current risk-free plan, said predictive model being updated by the COA at each COA activation (and therefore at each orbit, for example at the ascending node of the orbit) according to various flight parameters including a possible difference between a theoretical date of passage at the ascending node, provided in the form of ephemerides by the ground, and a date calculated on board of passage at the ascending node of the current orbit, which date calculated on board can be determined by an orbital event determination method using data provided by the GNSS.

[0032] According to one possible feature of the invention, following the development of the current maintenance and avoidance plan, a monitoring process is implemented over the control horizon, in which: The onboard risk level is reassessed at each subsequent COA activation (i.e., at the start of subsequent orbits, for example, at the ascending node of the orbit) with said current station-hold and avoidance plan and with current navigation data provided in real time by the GNSS and propagated at the TCA date of said risky CDM or its adjusted TCA date. If the onboard risk level does not decrease or if the calculated current position onboard the spacecraft diverges from the position predicted by the ACA, the ACA develops a corrected current station-hold and avoidance plan, from the current station-hold and avoidance plan, by removing all future maneuvers on the control horizon and recalculating new avoidance maneuvers for station-hold and avoidance.

[0033] According to one possible feature of the invention, the ACA uses a predictive model based on a quadratic evolution of the spacecraft's orbital position to verify the spacecraft's positioning within the mission window, and the monitoring method further includes a verification of said quadratic model, which triggers, in the event of a divergence observed in the quadratic model, the development by the ACA of the new current plan corresponding to the corrected positioning and avoidance current plan.

[0034] According to one possible feature of the invention, the spacecraft operates in low Earth orbit and the mission window requires maintaining position in orbit and in RAAN (acronym for English Right Ascension of the Ascending Node meaning "right ascension of the ascending node"

[0035] The invention extends to a spacecraft comprising a propulsion and attitude control system, a navigation system including a GNSS, and telecommunications devices for data exchange with a ground segment, characterized in that it is equipped with a COA and an ACA configured to implement the method described above. Brief description of the drawings

[0036] The invention, according to an exemplary embodiment, will be better understood and its advantages will become clearer upon reading the following detailed description, given by way of example and in no way limiting, with reference to the attached drawings in which: [ Fig. 1 ] there figure 1 is a schematic representation of two objects moving towards each other; this figure illustrates the probability that these two objects will collide, via a representation of their respective covariance which reflects the uncertainty that exists concerning the position and velocity of these objects; Fig. 2 ] there figure 2 shows another way to illustrate the probability of two objects colliding, using a combined covariance of the two objects; Fig. 3 ] there figure 3 represents a mission plan in the form of a timeline, which defines various types of slots relating to the use of a spacecraft according to the invention; [ Fig. 4 ] there figure 4 is a graphic representing the trajectory of a spacecraft according to the invention as provided for in the COA of said spacecraft and the trajectory of the same spacecraft as provided for in the ACA of said spacecraft; [ Fig. 5 ] there figure 5 is a schematic representation of a spacecraft according to the invention; [ Fig. 6 ] there figure 6 is a schematic representation of examples of control horizons versus risk horizons.

[0037] Identical elements represented in the aforementioned figures are identified by identical numerical references. Detailed description

[0038] The collision avoidance and station-keeping management method according to the invention applies to a spacecraft comprising (see Fig. 5 ) : a navigation system 201, including in particular computing resources 202 and a GNSS 203 (acronym for English Global Navigation Satellite Systems ) that is to say a satellite geolocation device capable of giving in real time the three-dimensional position and three-dimensional velocity of the spacecraft in an inertial reference frame, a propulsion system 204 and attitude control system 205, in communication with the navigation system 201, communication organs 206 for the exchange of data with a ground segment.

[0039] Throughout the following, for the sake of simplicity, the term " satellite " is used in a non-limiting manner and may refer to a spacecraft according to the invention.

[0040] Note that the expression " ground segment"Ground segment" refers to the ground station(s) responsible for controlling the satellite. Depending on the satellite's orbit, trajectories, mission, etc., as well as the satellite's visibility time to the various existing ground stations, it may be possible to use, for example, several remote ground stations capable of seeing the satellite at different times for satellite control. The "ground segment" then refers to all the ground stations used. When several ground stations are used for satellite control, uplink data signals, such as remote control signals designated by TC, sent from the ground to the satellite, and / or downlink data signals, such as telemetry signals designated by TM, sent from the satellite to the ground, can be distributed among the different ground stations.If the satellite's visibility to existing ground control centers is insufficient given the amount of data to be exchanged with the satellite, or for other reasons related to telecommunications organization, it is possible, for example, to use one or more relay satellites. Communications with relay satellites or with the ground add to other constraints and lead to similar consequences, such as the prohibition of scheduling maneuvers (positioning or avoidance maneuvers) during certain communication windows.

[0041] Typically, the TM signals sent by the satellite to the ground segment include navigation data provided by the satellite's GNSS, such as the satellite's position and velocity at the last up-point node and / or the satellite's current position and velocity at the time of the TM communication.

[0042] Typically, the TC signals sent by the ground segment to the satellite include conjunction messages called CDMs, relating to secondary objects that may be on the satellite's trajectory.

[0043] TC signals can also include, for example, mission-related data, in particular a mission plan 100 (see Fig.3 ) which defines, in time and / or position on orbit, on the one hand recovery points 101, mission slots 102 reserved for the mission and during which no maneuver is theoretically permitted, and on the other hand slots 103 available for the placement of maneuvers.

[0044] TC signals are regularly sent by the ground segment, for example approximately every four orbits or every six hours, when the satellite is visible to the ground control center(s). CDMs received by the ground segment are, for example, transmitted to the satellite.

[0045] Like TC remote controls, TM telemetry is sent by the satellite to the ground segment on a regular basis, preferably at each visibility by the ground segment.

[0046] In a CDM, a conjunction is defined, for example, by one or more of the following parameters: a primary object, here the satellite, a secondary object, the primary object and the secondary object each having a covariance of uncertainty on its position and its speed, and a TCA date, that is to say a closest passage date (or date of closest passage) which corresponds to the moment when the two objects are supposed to be closest to each other.

[0047] The CDM could include all the information listed previously. The CDM could also be simplified by sending only a portion of the information, omitting data that can be calculated on board.

[0048] As illustrated in the Fig. 1, the probability of collision between a primary object 1 moving at a speed vp and a secondary object 2 moving at a speed vs can be based on the intersection, at the TCA date, of the ellipsoid 11 representing the covariance of the primary object 1 at the position of said primary object at the TCA date, and of the ellipsoid 21 representing the covariance of the secondary object 2 at the position of the secondary object at the TCA date.

[0049] Alternatively, as illustrated in the Fig. 2 The probability of collision can be based on the intersection, at the TCA date, of, on the one hand, the combined ellipsoid 30 reduced to the position of the secondary (if not primary) object at the TCA date, and on the other hand, the HBR section 40 reduced to the position of the primary (respectively secondary) object at the TCA date, the HBR section designating a sphere having a diameter equal to the sum of the characteristic dimensions of the two objects (HBR being the acronym for the English Hard Body Radius ) .

[0050] Throughout this detailed description, it is assumed that the CDMs received by the satellite concern at least one secondary object. In the event that several problematic secondary objects are identified by the ground segment and CDMs concerning different secondary objects are consequently received by the satellite, the method according to the invention, described for a single secondary object, would be repeated for each of the secondary objects.

[0051] Note that, statistically, a CDM can be received by the ground segment up to seven days before its TCA date, and that the ground segment generally receives from the international centralized monitoring organizations EUSST and JSpoc, for each identified secondary object, a CDM every six to eight hours, which corresponds to four to five orbits for the satellite in the case of a satellite evolving in low Earth orbit.

[0052] In addition to the navigation system and the propulsion and attitude control system mentioned above, the satellite according to the invention comprises: an autonomous orbit control module or COA 207, whose role is to calculate station-keeping maneuvers to comply with the mission (the latter being, for example, an Earth observation mission), a collision risk management module or ACA, referenced 208, whose role is to calculate collision risks and, if necessary, to propose avoidance strategies.

[0053] For example, the COA and the ACA work collaboratively and in a synchronized manner. Indeed, maneuvers initiated by the COA could, for instance, create a risk of collision, and to avoid such an effect, the COA is advantageously coordinated with the ACA. Similarly, an avoidance strategy proposed by the ACA could, for example, conflict with the mission requirement, and to avoid such an effect, the ACA is advantageously coordinated with the COA.

[0054] According to the invention, the COA is activated for example at each orbit, at the passage through the ascending node, and the ACA is also activated at each orbit, after the COA has completed its calculations, which allows the ACA to take into account the station-keeping maneuvers calculated by the COA.

[0055] As an example, the ACA could be activated each time a new event is deemed likely to change the risk value. Events affecting the risk calculation include, for example: the receipt of a new CDM (whether it is an update of a known risk, i.e. a CDM relating to a secondary object which has already been the subject of a previous CDM, or the appearance of a new risk i.e. a CDM relating to a new secondary object which has not been the subject of any CDM until now), the planning of a maneuver, either by the COA, or by the ACA.

[0056] On the other hand, as the TCA date approaches, the uncertainty, calculated from real-time onboard navigation data and propagated to the TCA, decreases. Therefore, even in the absence of new data, activating the ACA can be advantageous for updating a previously assessed risk and verifying, in particular, whether that risk has disappeared (i.e., whether the associated risk level has fallen below the onboard risk threshold). Thus, it seems beneficial to call the ACA periodically. Activating the ACA and updating the risk can also be used to verify whether the risk has deteriorated, for example, by exceeding an onboard risk threshold. Furthermore, managing the spacecraft's computing resources is a concern that leads to limiting the activation of the ACA.

[0057] Thus, activating the ACA once per orbit after the COA has been activated (at the ascending node) and has completed its calculations proves, for example, to be a good compromise.

[0058] Activating the COA (Collision Orientation Assessment) on the current orbit, for example at the ascending node, allows the establishment of a plan, called the current safe orbit plan, which "corresponds" to the current orbit in that it was established at the beginning of that orbit, and which defines station-keeping maneuvers (for mission execution) that do not take into account potential collision risks. This current safe orbit plan is, for example, established over a calculation horizon, called the risk horizon, which extends to the nearest TCA (Terminal Control Area) date for a given secondary object. The duration of the risk horizon, that is, the question of how long before said TCA date the ACA must be activated to develop an avoidance strategy for said secondary object, is determined by several criteria.

[0059] To determine the upper bound of the risk horizon, two opposing criteria are considered. On the one hand, it is pointless to anticipate risk too far in advance because propagating uncertainties too far ahead impairs the accuracy of estimates (the later the better with respect to the uncertainty of the risk). Conversely, making an early avoidance decision makes it possible to limit the scope of the avoidance maneuver by optimizing its placement across a wider range of maneuvering windows and / or to limit the impact of the avoidance on mission planning.

[0060] Furthermore, the satellite's maneuverability implies a limit date (and therefore a lower bound for the risk horizon), known as the latest avoidance date, beyond which it is no longer possible to maneuver in order to completely eliminate the risk of collision. This latest avoidance date is defined in particular by one or more of the parameters below: the thrust acceleration capacity of the satellite's propulsion system, the mission, which imposes maneuver slots and mission slots (mission slots during which the craft must remain stable within the mission window), the thrust configuration which includes in particular the preheating of the nozzles or the attitude rally, - the onboard operational process which dictates the time required to start a maneuver, the consideration of anomalies which would prevent the avoidance strategy from being executed.

[0061] The latest avoidance date can be estimated, for example, as 2 to 4 orbits or 2 to 4 hours before the TCA date of the at-risk CDM.

[0062] The risk horizon is, for example, set to one day (24 hours) prior to the TCA date of the relevant risk CDM. Alternatively, the risk horizon can be set to 48 hours. Furthermore, the risk horizon can be configured remotely from the ground and therefore modified by sending a corresponding remote control (TC) signal.

[0063] According to the invention, the CDMs received by the spacecraft undergo preliminary filtering on board.

[0064] The EU-SST data received by the ground segment is, for example, pre-filtered by the ground segment before being sent to the satellite. This is done both to detect potential SST errors (duplicates, conjunctions of auto-collisions) and to limit the amount of data sent on board. However, this ground-based pre-filtering does not take into account either the satellite's navigation data or upcoming station-keeping maneuvers. Such pre-filtering by the ground segment remains very crude, particularly to avoid the risk of discarding a conjunction that, once recalculated with knowledge of the station-keeping maneuvers and the navigation data calculated in real time on board, would be larger than expected.

[0065] It is therefore advantageous to reduce the number of CDMs that the ACA will process at each orbit from among the CDMs received by the satellite, as provided for in the invention with the execution of a preliminary filtering on board, based on temporal and / or geometric criteria.

[0066] This preliminary filtering may, for example, include a time-based filtering step, which consists of removing CDMs with TCA dates that are too far in the future. This time-based filtering may involve removing CDMs whose time difference until the TCA exceeds a predetermined number of hours, a number that can be configurable or fixed, for example, at 36 or 48 hours. The time-based filtering can thus consider an interval longer than the risk horizon (for example, if the risk horizon is 24 hours and the time-based filtering interval is 48 hours). This allows for consideration of potential TC communication problems between the ground segment and the satellite, and / or potential communication problems between international centralized monitoring organizations and the ground segment, and / or potential failures of the ground segment or international centralized monitoring organizations.

[0067] Alternatively, or in combination with other methods, preliminary filtering may include, for example, a geometric filtering step, which consists of removing CDMs (Crew-Diffused Materials) for which the distance between the predicted satellite position and the assumed position of the secondary object at the TCA date exceeds a predetermined filtering distance. This filtering distance is, for example, 20 km, 15 km, or 10 km. It is advantageously configurable, and therefore modifiable from the ground by sending a remote control signal.

[0068] For the purposes of this geometric filtering step, the predicted position of the satellite at the TCA date is calculated by the satellite's onboard filtering module, based on the satellite's position and velocity at the ascending node of the current orbit, provided by the onboard GNSS, with high accuracy. Furthermore, the assumed position of the secondary object at the TCA date is that provided by the CDM. Finally, to keep the preliminary filtering process fast and computationally efficient, the TCA date considered is, for example, the one provided by the CDM, without any adjustment (unlike what might be done when calculating a risk level or an avoidance maneuver, where the use of an adjusted TCA date is preferred, as explained later).

[0069] Only one of the two filtering steps defined above (temporal or geometric filtering) can be executed.

[0070] Alternatively, and preferably, both filtering steps are performed, in either order. The second filtering step then considers only the CDMs retained after the first filtering step.

[0071] According to one implementation example, if several CDMs (Conjunction Data Modules) have been received for the same secondary object and remain after the two temporal and geometric filtering steps, an additional preliminary filtering step could consist of selecting only the last CDM received by the satellite from among the CDMs of the same conjunction with this secondary object that remain after the previous filtering steps. In particular, several different conjunctions may exist for the same object due to cyclical aftershocks on multiple orbits.

[0072] If, after the preliminary screening carried out on board, at least one CDM remains, it means that a risk of collision with the corresponding secondary object has been identified. The CDM(s) retained after the preliminary screening are referred to as high-risk CDMs.

[0073] The ACA will then calculate the risk on at least one of the filtered CDMs. At least the risk for the closest CDM in time is calculated.

[0074] In the event of an identified collision risk, the ACA estimates a risk level representative of the risk of collision between the satellite and a secondary object at a TCA date, for each of the selected risk CDMs or for at least one risk CDM. This risk level is called the onboard risk level because it is calculated by the satellite's computing resources (and not by the ground segment) and because it takes into account navigation data calculated onboard the satellite and station-keeping maneuvers calculated onboard the satellite.

[0075] If a collision risk is identified, the ACA is instructed, based on a risk level, to establish a maneuver plan over a control horizon to simultaneously maintain the aircraft within the mission window and reduce the risk of collision with the secondary object. This plan is called " current plan for maintaining position and avoidance the term fluent The preceding expression refers to the fact that the said plan is established during the current orbit for a horizon (the control horizon) that includes this orbit. The control horizon includes a predetermined number of orbits including the current orbit, this number being, for example, between 2 and 4, preferably equal to 3.

[0076] The ACA's risk level calculation relies on finding the maximum collision probability (CoPoC function) while considering uncertainties. A contraction / dilation process on the covariances is used to account for unmodeled uncertainties in the dynamics, navigation, and orbit determination of the secondary object.

[0077] To reduce computation time, the probability of collision for given covariances (PoC function, whose CoPoC is the maximum over predetermined ranges of dilated / contracted covariances for the primary and secondary objects) is evaluated by an analytical expansion. CoPoC = max Kp , Ks PoC Kp Ks P c t 1 t 2 = 1 2 π 3 / 2 det Σ R ¯ 0 ∫ T c t 1 t 2 exp − 1 2 r → − μ R ¯ 0 T Σ R ¯ 0 − 1 r → − μ R ¯ 0 d r → .

[0078] Where ∑ R represents the sum of the covariances and where µ R represents the relative position vector between the primary and the secondary.

[0079] In particular, it is possible to approximate this integral using a finite sum.

[0080] The initial state (position, velocity) and covariance (excluding expansion / contraction) are provided by the GNSS. To achieve the lowest possible initial covariance for the satellite, it is preferable, for example, to equip the satellite with a high-quality GNSS receiver capable of measuring, at a frequency of 1 Hz, the satellite's position and velocity with standard deviations of σP = 1 m and σV = 0.003 m / s, respectively. A navigation filter is preferably associated with this GNSS receiver to filter out measurement noise.

[0081] The calculation of the risk level by the ACA is based on the propagation, up to the TCA, of the state (position, velocity) and the dilated / contracted covariance of the satellite taking into consideration, on the one hand, the position and velocity calculated on board the satellite at the time of the activation of the ACA provided by the GNSS of the satellite, and on the other hand, the risk-free maneuver plan of the COA.

[0082] Preferably, the propagation calculation is refined, for example, by an adjustment (of a few seconds) of the TCA date.

[0083] This adjustment could, for example, consist of shifting the TCA date provided by the CDM based on the relative position and velocity of the two objects. In other words, the adjusted TCA date can be calculated using the formula TCA ajustée = TCA duCDM + dt , with dt = dot(dr, dv) / norm(dv) 2< , where: dot: dot product dr: relative position vector spacecraft / secondary object calculated from the position of the spacecraft and the position of the secondary, at the TCA date of the CDM. dv: relative velocity vector spacecraft / secondary object, calculated from the position of the spacecraft and the position of the secondary, at the TCA date of the CDM.

[0084] The propagation model used is, for example, a simplified model based on: A terrestrial potential model; for example, a 6x6 terrestrial potential model (simplified terrestrial potential with 6 zonals and 6 tesserales), an atmospheric model, a lunisolar perturbation model, solar activity data, and atmospheric drag parameters, which help avoid propagation and prediction errors due to drag during periods of high solar activity, particularly for satellites in low Earth orbit. Drag can be accounted for using an average ballistic coefficient, depending on solar activity.

[0085] The remaining uncertainties are, for example, managed by the expansion / contraction of the covariance of the satellite and the secondary object.

[0086] The models, data, and parameters of the propagation model are stored by the satellite. These models, data, and parameters can then be updated on the ground and sent back to the satellite.

[0087] It should be noted that the propagation model is specific to each satellite and depends, in particular, on its altitude. For example, atmospheric drag plays a significant role in first-order propagation for a satellite in low Earth orbit, whereas it becomes almost negligible compared to the uncertainty of the navigation solution for high-altitude missions. Thus, in the case of a satellite in low Earth orbit, incorporating atmospheric drag into the propagation model allows for the onboard determination of the satellite's propagated state at the TCA date provided by the CDM or at the adjusted TCA date, with high accuracy.

[0088] In theory, only a complete dynamics model can provide adequate accuracy for risk assessment. However, using such a complete model requires resources (in terms of computing power) greater than those available from satellites and would necessitate implementing propagation calculations on the ground.

[0089] In the method according to the invention, the propagation calculations performed by the ACA take into account the position and velocity calculated on board the satellite, as well as the GNSS covariance, which describes only the uncertainty in the onboard navigation solution and is therefore lower than the covariance generally considered by the ground segment, which must take into account other sources of uncertainty. Furthermore, the propagation calculations performed by the ACA also consider the maneuvers planned by the COA before the TCA date (of the CDM or adjusted). In particular, for these two reasons, the satellite's state at the TCA date (of the CDM or adjusted) can advantageously be determined onboard with greater accuracy than that which would be obtained by the ground segment.Indeed, on the one hand the ground segment does not know the precise real-time orbit calculated on board the satellite nor the maneuvers planned by the COA and on the other hand the ground segment must use a greater covariance than that calculated on board the satellite.

[0090] Therefore, a simplified propagation model, which takes into account, for example (in the case of a satellite in low orbit) only a terrestrial potential model, lunisolar perturbation data, an atmospheric model and drag parameters, becomes acceptable, and the satellite's computing resources are sufficient to support on-board risk assessment and management.

[0091] A fixed-step RK4 integration algorithm (for example, on the order of 60 seconds) can be used. The simplification of calculations is, for example, offset by the increased accuracy provided by taking into account the position and velocity calculated in real time on board the satellite, the GNSS covariance, and the COA maneuvers.

[0092] For the calculation of the CoPoC, the presumed state (position, velocity) of the secondary object at the TCA date or the adjusted TCA date is that provided by the at-risk CDM. Similar to the satellite covariance, the covariance of the secondary object at the TCA date or the adjusted TCA date is considered within predefined expansion / contraction ranges.

[0093] The maximum probability of collision obtained in these ranges of expansion / contraction of the covariances of the satellite and the secondary object corresponds to the level of risk on board.

[0094] This level of risk on board is then compared to a predefined onboard risk threshold. Advantageously, the onboard risk threshold can, for example, be greater than a " soil risk threshold "which would be appropriate to use if the risk level were calculated by the ground segment without considering, in particular, the orbit calculated in real time on board the satellite or the future station-keeping maneuvers up to the TCA date of the at-risk CDM or up to the adjusted TCA date. The onboard risk threshold could, for example, be chosen between 1.10⁻⁴ and 5.10⁻⁴."

[0095] If the risk level assessed by the ACA is less than or equal to the edge risk threshold, the ACA retains the current risk-free plan as the current plan for maintaining position and avoidance.

[0096] If the level of risk on board assessed by the ACA is higher than the on-board risk threshold, the ACA develops a new maneuver plan from the current risk-free plan, in order to guarantee the avoidance of the secondary object.

[0097] When the ACA develops a new maneuver plan, it initially proceeds, for example, by removing station-keeping maneuvers from the current, non-risk plan developed by the COA, as explained below. Indeed, the station-keeping maneuvers planned by the COA have an impact on the risk of collision; that is, they can advantageously reduce it or, conversely, increase it. The ACA therefore first determines whether removing one or more of these maneuvers can sufficiently reduce the risk without exceeding the extended mission window.

[0098] For example, the ACA initially proceeds by successive cancellations and verifies, with each cancellation, whether the risk level has decreased to the point of falling below the onboard risk threshold. If not, an avoidance maneuver can be implemented instead of a station-holding maneuver or during an available time slot.

[0099] For example, the ACA first removes the maneuver that precedes the adjusted TCA date and is closest to that adjusted TCA date. The plan thus obtained by removing the last maneuver before the adjusted TCA date defines a new version of the maneuver plan.

[0100] For example, the ACA then re-estimates the onboard risk level as it did previously, using GNSS navigation data propagated up to the adjusted TCA date, but with the new version of the maneuvering plan instead of the current risk-free plan. It compares the newly estimated risk level to the onboard risk threshold.

[0101] If the newly estimated risk level is, for example, less than or equal to the ship's risk threshold, the ACA adopts the new maneuver plan as the current hold and avoidance plan. It is possible that the resulting hold and avoidance plan could lead to a departure from the mission window. However, this plan will be corrected by the COA at the next call, so that any potential departure from the window remains limited to an expanded mission window. This strategy nevertheless allows for a shift in the hold maneuvers, creating separation from the secondary object to be avoided.

[0102] If the newly estimated level of risk on board is higher than the on-board risk threshold, the ACA will, for example, develop yet another new version of the maneuver plan by continuing to modify the current risk-free plan already modified by removing the penultimate station-keeping maneuver planned before the adjusted TCA date.

[0103] For example, it reassesses the level of onboard risk with this new version of the maneuver plan and compares the level of risk obtained with the onboard risk threshold.

[0104] If the reassessed risk level is less than or equal to the onboard risk threshold, the ACA adopts, for example, the new version of the maneuver plan as the current plan for maintaining position and avoidance.

[0105] Conversely, if the reassessed risk level is still above the edge risk threshold, the ACA continues, for example, its modification of the current risk-free plan by further removing the following station-keeping maneuver in the reverse chronological order.

[0106] The number of maneuvers removed is limited, for example, to allow us to stay within a wider mission window.

[0107] For example, ACA can be programmed to only suppress a maneuver if it is an in-orbital-plane maneuver aimed at correcting an error in the satellite's Position on Orbit, and not an out-of-plane maneuver aimed at correcting an error in RAAN or an emergency maneuver aimed at avoiding an imminent window exit or bringing the satellite back into the mission window after a window exit.

[0108] For example, the ACA proceeds with successive deletions, as described previously, until the onboard risk level, reassessed with the latest version of the maneuver plan, falls below the onboard risk threshold, while avoiding, for example, deleting more than a predetermined number of station-keeping maneuvers. This number is, for example, between 1 and 3, preferably equal to 2. Limiting the number of deletions prevents, for example, an excessive exit from the operating window, thus halting a potentially unsuccessful deletion strategy and switching to a new avoidance strategy. If the ACA is configured to retain off-orbital-plane maneuvers and emergency maneuvers and only delete non-urgent maneuvers within the orbital plane, the ACA can, for example, be authorized to delete, within the current off-risk plan, all non-urgent maneuvers planned for the orbital plane.An urgent maneuver is defined as a PSO (position on orbit) error correction maneuver triggered when a mission window exit is imminent or already in effect.

[0109] If, after removing the predetermined number of station-keeping maneuvers from the current safe plan (or all non-urgent maneuvers planned in the orbital plan by the safe plan), the level of risk on board is still not sufficiently reduced, the method according to the invention provides for example that the ACA calculates one or more avoidance maneuvers in order to guarantee avoidance without going out of the mission window.

[0110] In this case, for example, it is more efficient in terms of total calculation time and management of the satellite's computing resources, for the ACA to remove, for example, all station-keeping maneuvers from the current non-risk plan of the COA and to develop a complete maneuver plan respecting both the mission and the avoidance of the secondary object.

[0111] Each calculated avoidance maneuver is, for example, planned as a replacement for a position-maintaining maneuver from the out-of-risk plan initially established by the COA.

[0112] The avoidance maneuver(s) are, for example, the solution to the problem of maintaining position and avoiding the secondary object. The calculation of an avoidance maneuver is based, for example, on the generic formulation of an optimization problem minimizing the CoPoC (which is also used to calculate the risk) while taking into account window holding constraints, particularly window holding in orbit.

[0113] For example, one can search for the thrust direction in the orbital plane that minimizes the CoPoC the most, using the entire maneuver window. Far from the TCA date, the maneuver is generally primarily tangential. Very close to the TCA date, it can take a specific direction, with a radial component.

[0114] If this problem proves insoluble, priority is given, for example, to avoidance. A wider window compatible with the mission and provided, for example, by ground control, can then be used initially. This wider window can also be used in case of degraded operation, particularly during a GNSS unavailability.

[0115] As explained above, the propagation calculations performed by the ACA may show a discrepancy between the TCA date provided by the CDM and the TCA date predicted by these propagation calculations. Therefore, as for example when calculating a risk level, the ACA performs an avoidance maneuver calculation based on the position, velocity, and covariance of the satellite and the secondary object propagated at the adjusted TCA date (as defined above). This advantageously increases the accuracy of the calculations and reduces the number of avoidance maneuvers required to prevent any risk of collision.

[0116] Once developed by the ACA, the current station-keeping and avoidance plan is, for example, transmitted to the navigation system for execution. Each maneuver within this plan is defined by the duration and direction of thrust to be applied by the propulsion system at a given date or orbital position.

[0117] Thus, the satellite is, for example, capable of estimating its attitude, for example using STR sensors (from the English Star Trackers), and on the other hand, to control its attitude, for example using reaction wheels, in order to execute each maneuver in the corresponding thrust direction. The satellite stores, for example, a synchronous onboard code responsible for preparing commands (preheating, thrust activation, etc.) for the propulsion system. The propulsion system may, for example, include one or more electric thrusters and / or one or more plasma thrusters or, more generally, any low-thrust propulsion system.

[0118] Note that the development of the current workstation maintenance and avoidance plan by the ACA is, for example, advantageously an asynchronous function. This allows more computation time to be allocated to the ACA.

[0119] There Fig. 4This illustrates an example of the effect, on the satellite's trajectory, of replacing the COA's safe current flight plan with a station-keeping and avoidance current flight plan calculated by the ACA. The satellite's mission window is marked by lines 301 and 302. The satellite's trajectory calculated by the COA, i.e., the satellite's trajectory resulting from the implementation of the safe current flight plan, corresponds to the dashed line 12. This trajectory crosses the ellipsoid 21 representing the covariance of a piece of debris 2 (secondary object), at a TCA date (tCA) provided by a CDM for said debris. This means that a risk of collision exists between the satellite and debris 2, probably with a level of risk sufficient for an avoidance procedure to be implemented by the ACA.

[0120] Following the example of the figure 4After calculation, the risk level assessed by the ACA for this debris 2 being indeed higher than the onboard risk threshold, the ACA develops a current station-keeping and avoidance plan. The trajectory of the satellite resulting from this new plan is represented by the solid line 13. For the development of this plan, the last station-keeping maneuver 14 scheduled before the TCA date by the COA's current risk-free plan is removed by the ACA. In the present case (example of the Fig. 4This deletion reduces the risk but results in an immediate exit from the mission window {301-302}. The development of the current positioning and avoidance plan by the ACA continues by deleting the previous (or following in reverse chronological order) positioning maneuver 15, which guarantees the reduction of risk but ultimately leads to an exit from the mission window.The ACA having removed the two station-keeping maneuvers provided for in the current risk-free plan over the risk horizon from t 0 to t TCA without managing to define a plan which satisfies both the mission and the avoidance, it recalculates avoidance maneuvers 16, 17 allowing to respect both the station-keeping in the mission window and the avoidance, that is to say defining a trajectory 13 of the satellite which on the one hand remains in the mission window and on the other hand is sufficiently distant, at the date TCA (t TCA ) or at an adjusted date TCA, from the ellipsoid 21 representing the covariance of the secondary object.

[0121] Preferably, the method according to the invention further comprises the implementation by the COA / ACA pair of a monitoring function on the (sliding) control horizon. The control horizon comprises a predetermined number of orbits, including the current orbit. This number of orbits is, for example, between 2 and 4, for example equal to 3, as illustrated in the figure 6 .

[0122] There figure 6This shows examples of risk horizons and control horizons. The control horizon HC1 is initially composed of an integer number of orbits starting at the current activation. The hold and avoidance strategy is calculated over the entire risk period. The maneuvers within the control horizon are fixed. With each successive activation, the control horizon decreases by one orbit. When the control horizon reaches its limit, a new control horizon HC2 is defined. The new maneuvers calculated at activation will also be fixed within this control horizon. The pre-calculated maneuvers over a control period do not change unless a significant deviation from the prediction is observed, in which case a complete recalculation of the maneuver plan is performed.

[0123] The validity of the maneuvers in the current station-keeping and avoidance plan is verified as they are executed. If there is a discrepancy between the state (position, velocity) calculated in real time on board the satellite by the GNSS and the state predicted by the plan, the current station-keeping and avoidance plan is updated and / or modified to ensure mission execution and avoidance of the secondary object. Monitoring also covers, for example, avoidance and station-keeping performance, with the risk being regularly reassessed (at each orbit) and the station-keeping within the mission window being regularly checked.

Claims

1. A method for managing collision avoidance and station-keeping of a spacecraft (1), the spacecraft comprising a propulsion (204) and attitude control (205) system, a navigation system (201) including a GNSS (203), and telecommunication members (206) for exchanging data with a ground segment, comprising the following steps: - the spacecraft receives conjunction data messages, hereinafter noted CDM, sent by the ground segment, said CDM being related to a close approach with at least one secondary object liable to collide with the spacecraft, each of said CDMs describing parameters of identification, position, speed, size and covariance of the secondary object as well as a time of closest approach date called TCA date: - at each orbit, at a defined position on the orbit, activating, on board the spacecraft, an autonomous orbit control module (207), hereinafter referred to as COA, for the establishment of a plan corresponding to the current orbit, referred to as a current risk-free plan, the current risk-free plan being in the form of a station-keeping maneuver plan (14, 15) at least over a risk horizon, the risk horizon comprising the current orbit and extending to the orbit containing the closest TCA date among the TCA dates of the CDMs received, - a filtering module on board the spacecraft, performs preliminary filtering of the CDMs received according to geometric and / or time criteria, to establish a list of CDMs at risk, - in the case where the preliminary filtering results in at least one CDM at risk, activating, on board the spacecraft, a collision risk management module (208), hereinafter referred to as ACA, which - estimates a level of collision risk, referred to as on-board risk level, based on on-board navigation data provided by the GNSS (203) propagated at the TCA date of said CDM at risk, - and develops a maneuver plan on a control horizon, referred to as current station-keeping and avoidance plan, to ensure both continued operation within a mission window (301-302) and reduction of the risk of collision with the secondary object, the control horizon comprising a predetermined number of orbits, including the current orbit, and being lower than the risk horizon, the current station-keeping and avoidance plan being developed as follows: - if the previously evaluated on-board risk level is less than or equal to a predefined risk threshold, referred to as the on-board risk threshold, the ACA maintains the risk-free plan as the current plan for station-keeping and avoidance, - if the previously evaluated on-board risk level is greater than the on-board risk threshold, the ACA develops a new maneuver plan from the current risk-free plan, by eliminating at least one station-keeping maneuver from said current risk-free plan and / or replacing at least one station-keeping maneuver from said current risk-free plan with an additional maneuver (16, 17) for remaining within the mission window and avoidance, referred to as an avoidance maneuver, with the new maneuver plan becoming the current station-keeping and avoidance plan, and in that when the ACA develops the new maneuver plan from the current risk-free plan, the ACA (208) makes successive modifications based on the current risk-free plan, each modification providing a new version of the maneuver plan, and, for each of the modifications made, the ACA reevaluates the on-board risk level using the new version of the maneuver plan and based on on-board navigation data provided by the GNSS propagated on the TCA date of said CDM at risk.

2. The management method according to claim 1, wherein, when the ACA develops the new maneuver plan from the current risk-free plan, the ACA (208) successively eliminates the station-keeping maneuvers from the current risk-free plan in reverse chronological order from the TCA date of said CDM at risk, each elimination leading to a new version of the maneuver plan, and the ACA reevaluates, with each elimination, the on-board risk level using said new version of the maneuver plan, as long as the evaluated on-board risk level remains above the on-board risk threshold and the number of maneuvers eliminated is less than a predetermined maximum number of authorized eliminations, the ACA stopping the eliminations as soon as the evaluated on-board risk level is below the on-board risk threshold, the most recent version of the maneuver plan, which resulted in an on-board risk level below the on-board risk threshold, becoming the current station-keeping and avoidance plan.

3. The management method according to claim 2, wherein, if the reevaluated on-board risk level remains greater than the on-board risk threshold after eliminating a number of maneuvers of the current risk-free plan equal to the predetermined maximum number of eliminations authorized, the ACA eliminates all maneuvers of the current risk-free plan and calculates a set of avoidance maneuvers on the control horizon, this calculation taking both station-keeping and avoidance into account, said set of avoidance maneuvers becoming the current station-keeping and avoidance plan.

4. The method according to one of claims 1 to 3, wherein, for the calculation of any avoidance maneuver, the ACA solves a constrained optimization problem with the objective of minimizing a CoPoC risk function and the constraint of remaining within the mission window, the CoPoC function corresponding to a maximum probability of collision in predefined ranges of contraction and expansion of the covariances of the spacecraft and the secondary object at the TCA date.

5. The method according to one of claims 1 to 4, wherein the ACA sends the current station-keeping and avoidance plan back to the control system for execution.

6. The method according to one of claims 1 to 5, wherein the preliminary filtering of the CDMs comprises a time-filtering step consisting in selecting, from the CDMs received, the CDM(s) whose time difference to the TCA date is less than a predetermined number of hours.

7. The method according to one of claims 1 to 6, wherein, the preliminary filtering of the CDMs received comprises a geometric filtering step consisting in, for each of the CDMs received or each of the CDMs retained at the end of the time filtering step: - calculating a distance between a forecast position of the spacecraft and an assumed position of the secondary object on the TCA date of said CDM, - selecting the CDM(s) for which the previously calculated distance is less than a predetermined filtering distance.

8. The method according to one of claims 1 to 7, wherein the propagation of the on-board navigation data provided by the GNSS is carried out with a propagation model based on a terrestrial gravitational potential model, a lunar-solar disturbance model and an atmospheric model integrating solar activity parameters and drag parameters provided by the ground segment.

9. The method according to one of claims 1 to 8, wherein any estimation of the on-board risk level comprises: - calculating the propagation of the orbit and the covariance of the spacecraft up to the TCA date of said CDM at risk, according to the orbit of the spacecraft provided by the GNSS and the risk-free maneuver plan or the new version of the maneuver plan, - calculating the propagation of the orbit and the covariance of the secondary object at the assumed TCA date of said CDM at risk, - adjusting the TCA date, and correcting the orbits and covariances of the spacecraft and the secondary object propagated to the adjusted TCA date, - evaluating the on-board risk level is carried out on the basis of the orbits and covariances thus propagated on the adjusted TCA date.

10. The method according to one of claims 1 to 9, wherein the spacecraft receives a mission plan (100) from the ground segment, which mission plan defines mission slots (102) reserved for the mission and free slots (103) that can be used for scheduling maneuvers, the station-keeping maneuvers calculated by the COA and the avoidance maneuver(s) calculated by the ACA being forecast on free slots of the mission plan.

11. The method according to claim 10, wherein the avoidance maneuver(s) calculated by the ACA are forecast on free slots of the mission plan preceding an avoidance date at the latest, the avoidance date at the latest preceding the TCA date by a predetermined number of orbits or hours.

12. The method according to one of claims 1 to 11, wherein, for the development of the current station-keeping and avoidance plan, the mission window is a nominal mission window or an extended mission window, the nominal mission window or the extended mission window being predetermined consistent with the mission.

13. The method according to one of claims 1 to 12, wherein to verify that the spacecraft remains within the mission window, the ACA uses a predictive model based on a quadratic course of the spacecraft's position on the orbit, which predictive model is provided to the ACA by the COA with the risk-free maneuver plan of the current orbit, said predictive model being updated by the COA at each activation of the COA during each orbit (for example, at each ascending node of the orbit) according to various flight parameters, including a difference between a theoretical date of passage at the ascending node, provided in the form of ephemeris by the ground, and an on-board calculated date of passage at the ascending node of the current orbit.

14. The method according to one of claims 1 to 13, wherein, following the development of the current station-keeping and avoidance plan, a monitoring method is implemented over the control horizon, wherein: - the on-board risk level is reevaluated at each subsequent activation of the COA with said current station-keeping and avoidance plan and with current navigation data provided in real time by the GNSS and propagated on the TCA date of said CDM at risk, - if the on-board risk level does not decrease or if the spacecraft's current position calculated on board deviates from a position predicted by the ACA, the ACA develops a corrected current station-keeping and avoidance plan, from the current station-keeping and avoidance plan, by eliminating all upcoming maneuvers on the control horizon and by recalculating new avoidance maneuvers for keeping within the mission window and avoidance.

15. The method according to claim 14, wherein the ACA uses a predictive model based on a quadratic course of the spacecraft's position on the orbit to verify that the spacecraft remains within the mission window, and wherein the monitoring method further comprises verifying said quadratic model, which triggers the ACA to develop the current corrected station-keeping and avoidance plan in the event of a detected divergence in the quadratic model.

16. The method according to one of claims 1 to 15, wherein the spacecraft maneuvers in low orbit and wherein the mission window requires keeping in Position on Orbit and in RAAN.

17. A spacecraft (1) comprising a propulsion (201) and attitude control system, a navigation system (202) including a GNSS, telecommunication members (203) for exchanging data (TC, TM) with a ground segment, characterized in that it comprises an autonomous orbit control module (204), referred to as COA, and a collision risk management module (205), referred to as ACA, which COA and ACA are configured to implement a method for managing collision avoidance and station-keeping of the spacecraft according to any one of claims 1 to 16.