High precision relative navigation method based on time varying reference source
By modeling the local state of the reference source with the optimal geometric configuration in the UAV swarm and establishing the correlation of pseudo-observations, the problems of high computational complexity and low navigation accuracy of UAV swarms when GNSS fails are solved, and high-precision navigation under time-varying topology is realized.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- HARBIN ENG UNIV
- Filing Date
- 2026-04-07
- Publication Date
- 2026-06-12
AI Technical Summary
When existing UAV swarms experience GNSS signal failure in complex environments, the relative navigation method based on data links suffers from high computational complexity, low navigation accuracy, and an inability to adapt to time-varying topology changes.
A high-precision relative navigation method based on time-varying reference sources is adopted. This method involves modeling the local state of the reference source with the best geometric configuration in the cluster, setting the initial value of the state vector of the new reference source using broadcast information, establishing correlation through pseudo-observations, and performing information fusion to update the navigation state.
It reduces computational complexity, improves navigation accuracy, and maintains navigation continuity and consistency during formation changes, adapting to time-varying topology changes.
Smart Images
Figure CN122192323A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of UAV swarm collaboration, and in particular relates to a high-precision relative navigation method based on a time-varying reference source. Background Technology
[0002] Unmanned aerial vehicle (UAV) swarm collaboration technology has become a cutting-edge research area due to its enormous potential in collaborative detection, encirclement, and strike missions. The prerequisite for achieving efficient collaboration is providing each node in the swarm with accurate and reliable relative navigation information, that is, determining the relative positions, speeds, and time relationships between swarm members.
[0003] Currently, most cluster systems rely on Global Navigation Satellite Systems (GNSS) to obtain absolute positioning information and indirectly calculate relative relationships by sharing this information. However, in complex environments such as indoors, urban canyons, and dense forests, GNSS signals are easily rendered ineffective or severely interfered with due to obstruction, atmospheric refraction, and multipath effects, significantly reducing the reliability of this method. To address this, the industry has proposed relative navigation schemes based on the combination of data links and autonomous sensors such as Inertial Navigation Systems (INS). These schemes utilize the time of arrival (TOA) of data links to achieve relative ranging and timing, and then fuse information from sensors such as INS, providing continuous relative navigation capabilities for cluster systems in GNSS-denied environments.
[0004] Existing relative navigation methods based on data links are mainly divided into two categories: centralized and distributed. Centralized methods aggregate information from all nodes to a central node for processing, resulting in a heavy computational and communication burden and a risk of single point of failure. If the central node fails, the entire system collapses, making it difficult to meet the robustness and scalability requirements of modern UAV swarms. Distributed methods, on the other hand, distribute the computational tasks among the nodes, making them more practical. One type of distributed method (such as some existing technologies) only models and estimates the state of each node itself. While this method has a low computational cost, it completely ignores the actual correlation between the estimated states of different nodes in the swarm, leading to overconfidence during information fusion—that is, overestimating the certainty of one's own state. Although some scholars have introduced fusion algorithms such as covariance intersection (CI) to alleviate this problem, the fusion results are often too conservative, ultimately sacrificing navigation accuracy.
[0005] Another type of distributed approach, the traditional multi-center relative navigation algorithm, models the global state of all other nodes in the cluster for each node, thus explicitly maintaining the cross-correlation between nodes in the estimation and effectively avoiding the overconfidence problem. However, this global modeling strategy also brings significant drawbacks: as the cluster size increases, the dimensions of the state vector and covariance matrix that each node needs to maintain increase dramatically, leading to a quadratic or even higher increase in computational complexity, resulting in a heavy computational burden. More importantly, when the cluster actually performs a task, its formation needs to be dynamically adjusted according to the task, i.e., time-varying topology. Under time-varying topology, the set of nodes of the reference source also changes accordingly. Existing multi-center relative navigation algorithms do not have a mechanism designed for this dynamic switching. If a node directly switches the modeled reference source during formation change, the estimation results will quickly diverge due to the lack of prior state and correlation information of the new reference source nodes, making it impossible to maintain the continuity and consistency of navigation. In other words, existing methods that can guarantee estimation consistency cannot adapt to time-varying topology, while methods that can adapt to dynamic changes have defects in estimation accuracy or conservatism. Summary of the Invention
[0006] In view of this, the present invention aims to propose a high-precision relative navigation method based on a time-varying reference source, in order to solve the problem that the fusion results of existing methods that can adapt to dynamic changes are too conservative and have low navigation accuracy.
[0007] To achieve the above objectives, the present invention adopts the following technical solution: a high-precision relative navigation method based on a time-varying reference source, the method being applied to a single node in a cluster, the method comprising: Based on the states of one or more selected reference sources, the navigation states of the reference sources are modeled to construct local state estimates; The formation change of the response cluster causes the geometry of the reference source for the node to continue modeling to change. As a new reference source to enter the modeling set, the initial value of its state vector is set based on the broadcast information received by the node. The correlation between the state of the node and the state of the new reference source is established based on broadcast information and pseudo-observations; Based on the modeled navigation state, the initial values set, and the established correlations, information fusion is performed to update the node's own navigation state and its relative navigation state to the reference source.
[0008] Furthermore, a preferred approach is proposed, wherein the reference source is a member of the current formation of the cluster with the best geometry.
[0009] Furthermore, a preferred method is proposed, wherein modeling the navigation state of the reference source includes: The navigation error state of the node itself is modeled, and the navigation error state includes: longitude error, latitude error, altitude error, eastward velocity error, northward velocity error, vertical velocity error, three-axis attitude misalignment angle, data link receiver clock phase difference and frequency difference; The navigation state of the selected reference source is modeled, and the navigation state includes: longitude error, latitude error, eastward velocity error, northward velocity error, three-axis attitude misalignment angle, data link receiver clock phase difference and frequency difference.
[0010] Furthermore, a preferred method is proposed in which the broadcast information is exchanged between cluster nodes via a data link. The broadcast information includes the inertial navigation system output information of the sending node, the optimal estimate of its own state, and pseudo-observation information involving other nodes.
[0011] Furthermore, a preferred approach is proposed, wherein the new reference source for entering the modeling set, based on the broadcast information received by the node, sets the initial value of its state vector, including: A node receives broadcast information from a first node, the broadcast information including a first state estimate of the first node for itself and a second state estimate of the first node for the second node; The node calculates the initial value of the state vector of the new reference source based on its own third state estimate of the second node, and in combination with the first state estimate and the second state estimate.
[0012] Furthermore, a preferred approach is proposed, wherein the pseudo-observations used to establish the correlation are constructed based on the difference between the first state estimate and the second state estimate.
[0013] Furthermore, a preferred embodiment is proposed, wherein the pseudo-observation information includes: relative longitude position estimation error relative to the second node, relative latitude position estimation error, relative eastward velocity estimation error, relative northward velocity estimation error, and the variance of the relative position estimation and the variance of the relative velocity estimation.
[0014] Furthermore, a preferred embodiment is proposed, wherein the measurement information utilized in the execution information fusion further includes: Based on the arrival time of the data link signal between the node and other nodes, a relative ranging measurement is obtained; The absolute height is measured based on the altimeter mounted on the node.
[0015] Based on the same inventive concept, the present invention also proposes a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the high-precision relative navigation method based on a time-varying reference source as described in any of the preceding claims.
[0016] Based on the same inventive concept, the present invention also proposes a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the high-precision relative navigation method based on a time-varying reference source as described above.
[0017] Compared with the prior art, the beneficial effects of the present invention are: Unlike existing technologies where each node needs to model the global state of all other nodes in the cluster, this invention proposes that each node only needs to model and estimate the state of a few key nodes with the best geometric configuration in the current formation, i.e., the reference sources. It abandons the inherent approach of treating all nodes equally for global modeling, regardless of their geometric relationships. Instead, it dynamically selects the most important nodes as modeling objects based on navigation accuracy requirements, i.e., the principle of maximizing observability. This directly avoids the problem of the state dimension and computational load blindly increasing with the cluster size. Since each node only needs to model and filter the state of the selected few reference sources, the required state vector dimension and covariance matrix size are significantly reduced. This directly reduces the computational complexity of each node's local processor and also reduces the amount of data required to exchange partial covariance information to maintain global correlation, alleviating the communication load of the cluster data link.
[0018] Existing multi-center modeling methods, when changing formations, suffer from estimation divergence due to a lack of prior information about the new object if the modeling object is directly switched. This invention, however, does not rely on external absolute information but is based on a key geometric fact: the relative position vector between any two nodes is objectively unique and independent of the observer. Utilizing this principle, a node indirectly derives the initial state value of the new reference source relative to itself by receiving its own state estimate of another common node broadcast by the new reference source node, combined with its existing estimate of that common node. This process is completed entirely within a distributed framework using existing information, without the need for central node intervention or additional initialization measurements. To correctly integrate information within the filtering framework, a cross-correlation between the node's own state estimate and the new reference source's state estimate must be established. This invention constructs the aforementioned state difference relationship used to calculate the initial value into a pseudo-measure that can be incorporated into the filtering update process. The observation value of this pseudo-measure originates from the broadcast data of the new reference source node, and its noise covariance is carefully designed and calculated from the covariance sub-blocks contained in the broadcast information. By updating this pseudo-measurement with standard filtering, the required cross-correlation can be established in principle and quickly, ensuring the consistency and continuity of the filtered estimation.
[0019] By introducing an initial state setting method for new reference sources and a rapid construction strategy based on pseudo-measurement correlation, this invention enables cluster nodes to smoothly and continuously transition to tracking and estimating new reference sources during formation changes and dynamic switching of reference source sets. Simulation results show that after a 1300-second formation change, the method of this invention (MRNF-TVTN) maintains stable navigation accuracy, while the traditional multi-center method (MRNF) shows a significant performance degradation after the switch. This demonstrates that this invention overcomes the inherent limitation of existing global modeling methods in adapting to topology changes.
[0020] This invention inherits the advantages of multi-center modeling, which maintains the state correlation between nodes and avoids overconfidence in estimation. Furthermore, since the reference source is optimally selected based on the geometric configuration, modeling it maximizes the utilization of high-precision measurement information. Simulation results show that, under the same time-varying topology scenario, the method of this invention (MRNF-TVTN) achieves a relative position root mean square error of 10.60 meters, significantly outperforming the distributed method (44.07 meters) that only models its own state, and the high-precision dual-grid method HDGF (31.80 meters) using a specific fusion strategy. It is also far superior to the multi-center method MRNF (25.92 meters) that does not handle topology changes. Attached Figure Description
[0021] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart of the high-precision relative navigation method based on a time-varying reference source described in this invention; Figure 2 This is the simulated trajectory diagram described in this invention; Figure 3 This is an RMSE curve of the relative positioning error for all the methods described in this invention. Detailed Implementation
[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other, and the described embodiments are only some embodiments of the present invention, not all embodiments.
[0023] Implementation Method 1, see [link] Figure 1 This embodiment describes a high-precision relative navigation method based on a time-varying reference source. This embodiment addresses the problem that existing methods for adapting to dynamic changes often result in overly conservative fusion results and low navigation accuracy. The method is applied to a single node in a cluster and includes: Based on the states of one or more selected reference sources, the navigation states of the reference sources are modeled to construct local state estimates; The formation change of the response cluster causes the geometry of the reference source for the node to continue modeling to change. As a new reference source to enter the modeling set, the initial value of its state vector is set based on the broadcast information received by the node. The correlation between the state of the node and the state of the new reference source is established based on broadcast information and pseudo-observations; Based on the modeled navigation state, the initial values set, and the established correlations, information fusion is performed to update the node's own navigation state and its relative navigation state to the reference source.
[0024] In this embodiment, the reference source is the member with the best geometry in the current formation of the cluster.
[0025] In this implementation, the reference source is explicitly defined as the member with the optimal geometric configuration. This is not a simple noun restriction, but rather an optimization and specific implementation guideline for the selective modeling principle. The aim is to ensure that the node selects to model the node that contributes the most to improving its relative positioning observability, thereby maximizing the navigation accuracy gains from limited computing resources.
[0026] In this embodiment, modeling the navigation state of the reference source includes: The navigation error state of the node itself is modeled, and the navigation error state includes: longitude error, latitude error, altitude error, eastward velocity error, northward velocity error, vertical velocity error, three-axis attitude misalignment angle, data link receiver clock phase difference and frequency difference; The navigation state of the selected reference source is modeled, and the navigation state includes: longitude error, latitude error, eastward velocity error, northward velocity error, three-axis attitude misalignment angle, data link receiver clock phase difference and frequency difference.
[0027] This implementation method ensures that the core information of relative navigation, namely relative horizontal position, speed, and time, can be estimated, while further simplifying the state dimensions to be estimated and reducing unnecessary calculations.
[0028] In this embodiment, the broadcast information is exchanged between cluster nodes via a data link. The broadcast information includes the inertial navigation system output information of the sending node, the optimal estimate of its own state, and pseudo-observation information involving other nodes.
[0029] In this embodiment, the step of setting the initial value of the state vector of a new reference source for entering the modeling set, based on the broadcast information received by the node, includes: A node receives broadcast information from a first node, the broadcast information including a first state estimate of the first node for itself and a second state estimate of the first node for the second node; The node calculates the initial value of the state vector of the new reference source based on its own third state estimate of the second node, and in combination with the first state estimate and the second state estimate.
[0030] By leveraging the geometric fact that the relative position vector is objectively unique, the algorithm indirectly calculates the value of another common node by receiving the estimation information of the new reference source and combining it with its own existing estimation of the common node, thus laying the foundation for subsequent filtering convergence, and with low communication overhead.
[0031] Furthermore, the pseudo-observations used to establish the correlation are constructed based on the difference between the first state estimate and the second state estimate.
[0032] In this embodiment, the state difference relationship used to calculate the initial value is constructed as a pseudo-measurement, and a noise covariance matrix calculated based on broadcast covariance information is designed for this pseudo-measurement. The abstract problem of establishing cross-correlation is transformed into a standard Kalman filter measurement update problem, thereby quickly completing the correlation establishment and ensuring the consistency of state estimation.
[0033] Furthermore, the pseudo-observation information includes: relative longitude position estimation error relative to the second node, relative latitude position estimation error, relative eastward velocity estimation error, relative northward velocity estimation error, and the variance of the relative position estimation and the variance of the relative velocity estimation.
[0034] In this embodiment, the measurement information utilized in the execution information fusion also includes: Based on the arrival time of the data link signal between the node and other nodes, a relative ranging measurement is obtained; The absolute height is measured based on the altimeter mounted on the node.
[0035] Implementation Method 2, see below Figure 2 and Figure 3 This embodiment describes a complete implementation of the high-precision relative navigation method based on a time-varying reference source described in Embodiment 1, including: Based on the states of one or more selected reference sources, the navigation states of the reference sources are modeled to construct local state estimates; Taking node i as an example, its navigation state x i It consists of its own navigation state and the navigation states of other nodes being modeled, and its specific form is as follows:
[0036] Here, This represents the navigation error state of node i, while This represents the state of the j-th node (excluding node i) out of the n-1 nodes modeled by node i, where 1 ≤ j ≤ n-1. It also represents the navigation error state of node i. Includes: Strapdown Inertial Navigation System (SINS) longitude error, latitude error, and altitude error. SINS eastward velocity error, northward velocity error, and vertical velocity error SINS three-axis attitude misalignment angle Data link receiver clock phase error and frequency error .
[0037] Its dimension is 11, and its expression is:
[0038] The state of the j-th node among n nodes modeled by node i Includes the following: SINS longitude and latitude errors SINS eastward velocity error and northward velocity error SINS system three-axis attitude deviation angle ; and data link receiver clock phase error and frequency error . The dimension is 9, which is represented as: The state propagation equation is expressed as follows:
[0039] in and These represent the states at time steps k and k-1, respectively. and This represents the state transition matrix and noise driving matrix at time step k; It follows a Gaussian distribution. ~ N(0, Q) system white noise.
[0040] In a drone swarm, each node exchanges information with other nodes via data links and calculates the relative distance based on the time of arrival (TOA) of the transmitted messages. Assume node i receives a broadcast message from node j and obtains the distance measurement from the TOA. The corresponding data link relative measurement model can be expressed as:
[0041] in( , , ) T and( , , ) T These represent the positions of node i and node j in the Earth coordinate system, respectively. and This represents the data link clock error between node i and node j; It is relative ranging noise.
[0042] In a dynamic network environment, the reference source changes with the formation. For a new reference source to be modeled, the current node needs to set initial values for its state estimation. Assuming the cluster needs to switch reference source sets, the nodes... New benchmark source required Modeling (where) , (Represents the new set of benchmark sources). Based on the following fact: by nodes Modeling nodes The true position and node corresponding to the error state Modeling nodes The vector formed by the true positions corresponding to the error states is equal to the node. The true location and node corresponding to the error state of the self-modeling Modeling nodes The vector formed by the true positions corresponding to the error states. From this, we can obtain: (1) in, and Representing nodes respectively and nodes Pure inertial navigation output; , For nodes before formation change The original set of reference sources; Represents a node Modeling nodes The state vector (including longitude error, latitude error, eastward velocity error, and northward velocity error); and Meaning and Similarly, only the superscript and subscript have different meanings; Represents a node The self-modeling state. Simplifying equation (1) yields: (2) Therefore, node For the new benchmark source The initial values for modeling can be obtained as follows: (3) in, and Through nodes Data link message retrieval for broadcast, By node Obtained from its own error state.
[0043] To address the issue of modeling a new reference source in dynamic networking scenarios for relative navigation algorithms, this implementation proposes a method for setting the initial value of the new reference source. Furthermore, establishing the cross-correlation relationship between the new reference source and the current node is a crucial aspect that must be considered in algorithm design. Failure to establish this cross-correlation relationship often leads to inconsistent estimations, causing the current node's estimate of the new reference source to diverge.
[0044] During formation reconfiguration, establishing cross-correlation with the new baseline source is a crucial consideration. Failure to establish such cross-correlation often leads to inconsistencies in estimation, ultimately causing the current node's estimate of the new baseline source to diverge. The cross-correlation with the new baseline source is established through corresponding measurement updates using a standard filtering method (Extended Kalman Filter), where the measurements used to construct this correlation are called pseudo-observations. . Node A new reference source is needed. Taking the scenario as an example, the measurement equation is given by equation (3), which can be simplified to obtain: (4) in, Information is obtained through data link broadcasting. It is relative ranging noise, and satisfies Pseudo-observation noise covariance matrix as follows: (5) in, Latitude and These are the radii of curvature of the meridian and the gyrus, respectively. and The location and velocity variance are shared via data link broadcast. and The specific expression is defined as follows: (6) The superscript indicates that the node is a node. Specific sub-blocks extracted from the covariance matrix. Specifically, Indicates from node Extracted from its own covariance matrix corresponding to the node The submatrix of positional states; and and Indicates from node Nodes extracted from the covariance matrix With nodes The cross-covariance submatrix between positional states. From node Extracted from the covariance matrix corresponding to the nodes Submatrix of position state. Related terms for velocity state. , , and The definition is similar to that of position state; the extracted state is simply the position replaced by velocity.
[0045] In relative navigation, the data link is used not only to acquire Time of Arrival (TOA) information but also to transmit necessary data packets. The information contained in these packets is used for information fusion processing. Table 1 summarizes the content of the shared data packets transmitted via the data link, mainly including node self-information and pseudo-measurement information. Node self-information is used to update altimeter and TOA measurements, while pseudo-measurement information is used for setting initial values for new modeling members and constructing cross-correlation relationships.
[0046] Table 1
[0047] This embodiment evaluates the relative navigation performance of a multi-UAV swarm operating in a time-varying topology environment under conditions of complete GNSS loss through simulation experiments. Figure 2 As shown, the simulated trajectory uses a serpentine high-maneuver pattern with six UAVs, and the total duration is 1600 seconds. Inter-node communication between UAVs is achieved through a broadcast mechanism based on Time Division Multiple Access (TDMA).
[0048] The results obtained by different methods were compared in the following ways: (1) MRNF TVTN: The high-precision relative navigation method based on time-varying reference source proposed in this embodiment is applicable to time-varying topology scenarios.
[0049] (2) MRNF: A relative navigation method based on multicenter modeling, which is not applicable to time-varying topology scenarios.
[0050] (3) DRN: This method uses the propagation equation of the absolute navigation state error of autonomous vehicles as the state model, and each UAV only models its own state.
[0051] (4) HDGF: This method uses a high-precision dual-grid error propagation model to construct a state model and selectively integrates received data link information through a specific fusion strategy.
[0052] Simulation curves as follows Figure 3As shown in Table 2, the corresponding ARMSE values are also shown. In the time-varying topology scenario, the cluster reference source set is updated every 1300 seconds. By comparing MRNF-TVTN and the MRNF algorithm, it is found that after 1300 seconds, the MRNF-TVTN algorithm, which uses a dedicated initial value setting method and a newly added reference source correlation construction mechanism, significantly outperforms the MRNF algorithm, which only modifies the model without additional processing. Furthermore, in the overall simulation trajectory, the MRNF-TVTN algorithm proposed in this embodiment outperforms the distributed methods DRN and HDGF in both applicability and accuracy. Due to its utilization of more measurement data and accurate maintenance of cross-correlation, it exhibits a more significant advantage, ultimately achieving higher accuracy than DRN and HDGF.
[0053] Table 2
[0054] Implementation Method 3: This implementation method proposes a computer device, including a memory and a processor. The memory stores a computer program. When the processor runs the computer program stored in the memory, the processor executes the high-precision relative navigation method based on a time-varying reference source as described in any one of Implementation Methods 1 to 2.
[0055] Implementation Method 4: This implementation method proposes a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the high-precision relative navigation method based on a time-varying reference source as described in any one of Implementation Methods 1 to 2.
[0056] Those skilled in the art will understand that embodiments of this disclosure can be provided as methods, systems, or computer program products. Therefore, this disclosure can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this disclosure can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0057] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0058] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0059] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this disclosure and not to limit its protection scope. Although this disclosure has been described in detail with reference to the above embodiments, those skilled in the art should understand that after reading this disclosure, they can still make various changes, modifications or equivalent substitutions to the specific implementation of the invention, but these changes, modifications or equivalent substitutions are all within the protection scope of the pending claims.
Claims
1. A high-precision relative navigation method based on a time-varying reference source, characterized in that, The method is applied to a single node in a cluster, and the method includes: Based on the states of one or more selected reference sources, the navigation states of the reference sources are modeled to construct local state estimates; The formation change of the response cluster causes the geometry of the reference source for the node to continue modeling to change. As a new reference source to enter the modeling set, the initial value of its state vector is set based on the broadcast information received by the node. The correlation between the state of the node and the state of the new reference source is established based on broadcast information and pseudo-observations; Based on the modeled navigation state, the initial values set, and the established correlations, information fusion is performed to update the node's own navigation state and its relative navigation state to the reference source.
2. The high-precision relative navigation method based on a time-varying reference source according to claim 1, characterized in that, The reference source is the member with the best geometry in the current formation of the cluster.
3. The high-precision relative navigation method based on a time-varying reference source according to claim 1, characterized in that, The modeling of the navigation state of the reference source includes: The navigation error state of the node itself is modeled, and the navigation error state includes: longitude error, latitude error, altitude error, eastward velocity error, northward velocity error, vertical velocity error, three-axis attitude misalignment angle, data link receiver clock phase difference and frequency difference; The navigation state of the selected reference source is modeled, and the navigation state includes: longitude error, latitude error, eastward velocity error, northward velocity error, three-axis attitude misalignment angle, data link receiver clock phase difference and frequency difference.
4. The high-precision relative navigation method based on a time-varying reference source according to claim 1, characterized in that, The broadcast information is exchanged between cluster nodes via a data link. The broadcast information includes the inertial navigation system output information of the sending node, the optimal estimate of its own state, and pseudo-observation information involving other nodes.
5. The high-precision relative navigation method based on a time-varying reference source according to claim 1, characterized in that, The new baseline source for entering the modeling set, based on the broadcast information received by the node, sets the initial value of its state vector, including: A node receives broadcast information from a first node, the broadcast information including a first state estimate of the first node for itself and a second state estimate of the first node for the second node; The node calculates the initial value of the state vector of the new reference source based on its own third state estimate of the second node, and in combination with the first state estimate and the second state estimate.
6. The high-precision relative navigation method based on a time-varying reference source according to claim 5, characterized in that, The pseudo-observations used to establish the correlation are constructed based on the difference between the first state estimate and the second state estimate.
7. The high-precision relative navigation method based on a time-varying reference source according to claim 6, characterized in that, The pseudo-observation information includes: relative longitude position estimation error relative to the second node, relative latitude position estimation error, relative eastward velocity estimation error, relative northward velocity estimation error, and the variance of the relative position estimation and the variance of the relative velocity estimation.
8. The high-precision relative navigation method based on a time-varying reference source according to claim 1, characterized in that, The measurement information utilized in the execution information fusion also includes: Based on the arrival time of the data link signal between the node and other nodes, a relative ranging measurement is obtained; The absolute height is measured based on the altimeter mounted on the node.
9. A computer device, characterized in that: It includes a memory and a processor, wherein the memory stores a computer program, and when the processor runs the computer program stored in the memory, the processor executes the high-precision relative navigation method based on a time-varying reference source according to any one of claims 1-8.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the high-precision relative navigation method based on a time-varying reference source as described in any one of claims 1-8.