CN-121979698-A - Strong robust hierarchical distributed fusion method for remote sensing satellite

CN121979698ACN 121979698 ACN121979698 ACN 121979698ACN-121979698-A

Abstract

The invention discloses a strong robust hierarchical distributed fusion method for remote sensing satellites, which comprises the steps of determining a hierarchical management organization structure of satellite clusters and performing role division on the satellites; and determining the responsibility of the platform based on the role configuration, and processing the data step by the task execution node, the local fusion center node and the system fusion center node and synchronizing the data to the backup node. The invention realizes the purpose that the failure of part of nodes of the fusion system does not affect the whole function of the system by multi-star data backup, role automatic switching and task autonomous migration, and the tasks of the failure nodes can be automatically migrated to the backup nodes.

Inventors

CAI JIANWEI
LUO QIFENG
HUI XIAOHONG
Song Chenying
DONG SHAOPENG
LIN BOXUAN
ZHAO WEN
LI HANG
LUO RONGZHENG
LI XIAOJUAN
ZHANG HONGJIE
CAO JINGLONG

Assignees

中国空间技术研究院

Dates

Publication Date: 20260505
Application Date: 20251212

Claims (10)

1. A strong robust hierarchical distributed fusion method facing remote sensing satellites is characterized by comprising the following steps: The method comprises the steps of S1, determining a satellite cluster layered management organization structure and carrying out role division on satellites, wherein the satellite cluster layered management organization structure comprises a system fusion center layer, a local fusion center layer and a sensor layer, the roles comprise a system fusion center layer role, a local fusion center layer role and a sensor layer role, the system fusion center layer role comprises a system fusion center node and a system fusion center backup node, the local fusion center layer role comprises a local fusion center node and a local fusion center backup node, and the sensor layer role comprises a task execution node; S2, the task execution node processes the sensor acquired data and transmits a processing result to the local fusion center node; the local fusion center node receives the processing results of the task execution nodes and synchronizes to the local fusion center backup node; the local fusion center node carries out fusion processing on processing results of the task execution nodes and transmits the processing results to the system fusion center node; the system fusion center node receives the processing result of the local fusion center node and synchronizes to the system fusion center backup node; The system fusion center node carries out fusion processing on the processing result of the local fusion center node to obtain target information; When the local fusion center node or the system fusion center node fails, the local fusion center node or the system fusion center node is respectively switched into a local fusion center backup node or a system fusion center backup node.
2. The method for strongly robust hierarchical distributed fusion for remote sensing satellites according to claim 1, wherein after the satellites are subjected to role division, a role configuration list is uniformly managed by a system fusion center node and is distributed to subordinate nodes in an event-triggered mode, and after the role configuration list is changed, the system fusion center node distributes changed role information to subordinate local fusion center nodes, local fusion center backup nodes and task execution nodes; The role configuration list also contains attribution relations of roles of different levels, the task execution node transmits data to a local fusion center node specified by the attribution relations in the role configuration list, and the local fusion center node transmits the data to a system fusion center node specified by the attribution relations in the role configuration list.
3. The method of claim 1, wherein the task execution node processes the sensor collected data by using the method of the present invention comprises: The task execution node preprocesses the acquired data of the sensor to obtain a standard remote sensing image, and then extracts target information in the standard remote sensing image, wherein the target information comprises imaging time, target position, target category or motion state information; and the task execution node transmits data to the local fusion center node according to a time-triggered rule.
4. The method for strongly robust hierarchical distributed fusion for remote sensing satellites according to claim 1, wherein the process results of the plurality of task execution nodes are fused by the local fusion center node, the local fusion center node receives target information of the plurality of task execution nodes, spatial registration and time registration are performed on the target information, the same targets acquired by different satellites and different times are associated, and locally optimal target identity and state information are acquired through an optimal estimation algorithm.
5. The method for carrying out fusion processing on processing results of local fusion center nodes by using a system fusion center node according to claim 1 is characterized in that the processing results of the local fusion center node comprise a locally optimal target identity and state information, track association matching is realized based on the locally optimal target identity and state information, the locally estimated same target information is associated together, and then estimation fusion is carried out to obtain the globally optimal target identity and state information.
6. The method of claim 1, wherein when the local fusion center node synchronizes the processing result to the local fusion center backup node or when the system fusion center node synchronizes the processing result of the local fusion center node to the system fusion center backup node, the data synchronization adopts a time-triggered strategy, and uses a unique identifier, a time stamp and a version number to track the evolution history of the data.
7. The method for strongly robust hierarchical distributed fusion for remote sensing satellites according to claim 1, the method is characterized by monitoring the health state of the nodes by the following method: The system fusion center node monitors the health states of all the nodes, and meanwhile, the system fusion center backup node receives the heartbeat information of the system fusion center node, and the system fusion center backup node monitors the health states of the system fusion center nodes; when the task execution node fails, executing the task by the new task execution node; when the local fusion center node fails, the local fusion center backup node is switched to be the local fusion center node autonomously; when the system fusion center node fails, the autonomous switching system fusion center backup node is the system fusion center node; the system fusion center backup node or the local fusion center backup node fails, and a new backup node is selected.
8. The method for the robust hierarchical distributed fusion for remote sensing satellites according to claim 7, wherein the system fusion center node receives that the health state of heartbeat information of the task execution node is abnormal or that the heartbeat information of the task execution node is not received after timeout, and considers that the task execution node fails; When a new task executing node executes a task, updating the global role configuration list, sending all the nodes, executing the new task executing node according to the role rule after receiving the global role configuration list, and sending the target information to an upper local fusion center node.
9. The method for strongly robust hierarchical distributed fusion for remote sensing satellites according to claim 7, wherein the system fusion center node receives that the health state in the heartbeat information of the local fusion center node is abnormal or that the heartbeat information of the local fusion center node is not received over time, and considers that the local fusion center node fails; when the backup node of the local fusion center is switched to the local fusion center node autonomously, the system fusion center node modifies a global role configuration list maintained by the node, the role of the original local fusion center node is modified to be offline abnormal, the role of the backup node of the original local fusion center is modified to be the local fusion center node, a new local fusion center backup node is selected, the global role configuration list is distributed to all nodes, the role information of the role configuration file of the node is modified to be the local fusion center node after the backup node of the original local fusion center receives the new role configuration list, the operation is executed according to the rule of the local fusion center node, the task execution node receives the global role configuration list and then sends the processing result to the new local fusion center node, and the new local fusion center node receives the target information of the task execution node, carries out fusion processing and sends the processing result to the system fusion center node.
10. The method for strongly robust hierarchical distributed fusion for remote sensing satellites according to claim 7, wherein when the system fusion backup central node monitors that the health state of the system fusion central node is abnormal, the system fusion central node is considered to be invalid; When the autonomous switching system fusion center backup node is a system fusion center node, the role configuration list managed by the node is modified, the role of the original system fusion center node is modified to be offline, the role of the original system fusion center backup node is modified to be the system fusion center node, a new system fusion center backup node is selected, the global role configuration list is distributed to all nodes, the local fusion center node, the local fusion center backup node and the task execution node receive the distributed global role configuration list, heartbeat information is sent to the new system fusion center node, and the new system fusion center backup node receives the global role configuration list and executes operations according to role rules.

Description

Strong robust hierarchical distributed fusion method for remote sensing satellite Technical Field The invention belongs to the technical field of remote sensing satellite data processing, and particularly relates to a strong-robustness hierarchical distributed fusion method for multi-source information of a remote sensing satellite. Background With the development of science and technology, the field of remote sensing satellites is rapidly developed towards the increasing application demands of the fields of military, industry, environmental protection, disaster reduction and the like, the quantity of the transmitted remote sensing satellites is increased, the orbit height is increased, indexes such as satellite breadth and resolution are increased, the carried load covers visible light, infrared, microwaves and the like, the load type comprises the spectrum ranges of photoelectricity, radar, SAR, laser and the like, the spectrum quantity spans from a few to thousands of spectrum ranges, the serious problems of large data quantity and information overload are faced, and how to effectively utilize multi-source massive remote sensing data becomes the current hot spot problem. The data fusion system can fuse the remote sensing data of different sensors to realize the complementary integration of multiple sensing information. The redundancy among information is measured through multi-source remote sensing data, the generating capacity of the system is enhanced, the coverage area of space and time is expanded through multi-overlapping coverage, and the reliability of results is improved through collaborative observation. Common converged system architectures include both centralized and distributed types. The centralized fusion architecture transmits the original remote sensing data of all satellites to a fusion center for processing and fusion. Although the fusion center can utilize the original measurement data of all satellite platforms, no information is lost, and the fusion result is optimal. However, due to the large amount of original measurement data, the on-board bandwidth is limited, the data transmission time is long, and the timeliness is not high. Each satellite platform of the distributed fusion architecture has its own processor for preprocessing the original, and then transmits the intermediate result to the central node for fusion processing. The method has low requirement on channel capacity, strong system vitality and wide application in engineering. However, once the central node is damaged, the fusion system is at risk of reduced performance or even failure. Disclosure of Invention The invention aims to overcome the defects, and provides a strong robust hierarchical distributed fusion method for remote sensing satellites, which solves the technical problems that once a fusion center of a traditional distributed fusion system is damaged, the efficiency of the fusion system is reduced and even fails. The invention provides a high-robustness hierarchical distributed fusion method for remote sensing satellites, which realizes the purpose that partial node failure of a fusion system does not influence the overall function of the system by multi-star data backup, role automatic switching and task autonomous migration, and the tasks of failure nodes can be automatically migrated to backup nodes. In order to achieve the above purpose, the present invention provides the following technical solutions: a strong robust hierarchical distributed fusion method facing remote sensing satellites comprises the following steps: The method comprises the steps of S1, determining a satellite cluster layered management organization structure and carrying out role division on satellites, wherein the satellite cluster layered management organization structure comprises a system fusion center layer, a local fusion center layer and a sensor layer, the roles comprise a system fusion center layer role, a local fusion center layer role and a sensor layer role, the system fusion center layer role comprises a system fusion center node and a system fusion center backup node, the local fusion center layer role comprises a local fusion center node and a local fusion center backup node, and the sensor layer role comprises a task execution node; S2, the task execution node processes the sensor acquired data and transmits a processing result to the local fusion center node; the local fusion center node receives the processing results of the task execution nodes and synchronizes to the local fusion center backup node; the local fusion center node carries out fusion processing on processing results of the task execution nodes and transmits the processing results to the system fusion center node; the system fusion center node receives the processing result of the local fusion center node and synchronizes to the system fusion center backup node; The system fusion center node carries out fusion processing on the processing result of t