Multi-target clustering and optimal observation path planning method and system

Abstract

The application provides a multi-target clustering and optimal observation path planning method and system, which comprises the following steps: collecting targets to form an observation target set U; extracting the target with the highest priority in sequence according to the priority, and clustering other targets around the target with the highest priority until no target with the highest priority can be extracted, and then clustering the remaining targets according to the principle of the least number after clustering; screening out the best path observation node according to the distribution of the targets after clustering near the subsatellite point of the space-based remote sensor; based on the best path observation node, screening out the targets satisfying the condition of not consuming the space-based remote sensor maneuvering resource forward and backward respectively according to the over-the-top time, and forming an optimal observation target set U b ; adding the remaining targets into the set U b one by one, and selecting the target consuming the least space-based remote sensor maneuvering resource and satisfying the maneuvering resource to add into the set U b , and circulating until the space-based remote sensor maneuvering resource is exhausted. The application has obvious convenience and engineering applicability.

Description

Technical Field

[0001] This invention relates to the field of on-orbit observation path planning, and more specifically, to a multi-objective clustering and optimal observation path planning method and system. Background Technology

[0002] Dynamic observation path planning has always been a challenging area in space-based remote sensor mission planning due to its uncertainty and variability. The main disturbances in space-based remote sensor observation path planning include changes in satellite resource status and the insertion of new targets. Furthermore, current research primarily considers paths from a single target source, with limited coverage of path planning problems involving multiple targets.

[0003] Therefore, in the case of multiple targets operating concurrently, how to quickly plan an optimal observation path using the limited resources of space-based remote sensors has become an urgent problem to be solved.

[0004] The literature (Wang Zhiyong. Research on Multi-view SAR Target Recognition and Autonomous Planning Method of Observation Path [D]. University of Electronic Science and Technology of China, 2022) proposes an autonomous planning model of observation path based on reinforcement learning, which is used to solve the optimal solution of a multi-view SAR target recognition system under specific task requirements to satisfy multiple objective functions. However, it does not mention the innovative points of this invention, such as multi-target clustering and finding the optimal observation path with the best observation node.

[0005] The literature (Zhang Yaoyuan, Yang Hongwei, Yuan Ronggang, et al. Hierarchical clustering planning for satellite multi-target repeated observation tasks [J]. China Space Science and Technology, 2023, 43(1):29-43) proposes a hierarchical clustering-based method for planning satellite multi-target repeated observation tasks. This method resolves the conflict between repeated observation meta-tasks and initial observation meta-tasks, allowing for repeated observations of target points without reducing the target point observation completion rate. Compared to this invention, it does not mention innovative aspects such as finding the optimal observation path using the best observation node.

[0006] The literature (Shang Xijie, Feng Yang, Lin Xiaoyong, et al. Discussion on group task planning methods for imaging satellite networking [J]. Digital Technology and Application, 2023, 41(11):87-90) proposes to use a greedy algorithm to perform preliminary planning for sub-tasks to obtain a better initial solution, and then use a genetic algorithm to optimize the preliminary planning results to obtain a better task planning scheme. Compared with this invention, it does not mention the innovative points of this invention, such as multi-objective clustering and finding the optimal observation path with the best observation node.

[0007] The literature (Zhang Meiyan, Cai Wenyu. Underwater multi-target detection path planning based on multi-AUV task cooperation [J]. Journal of Sensor Technology, 2018, 31(7): 1101-1107) proposed using a genetic algorithm to heuristically solve this NP-Complete problem, and designed a fitness function that considers the total cruise path and the number of targets visited to improve the energy consumption balance among multiple AUVs, thus realizing optimized collaborative detection of multiple underwater targets by multiple AUVs. Compared with this invention, it does not mention innovative points such as multi-target clustering and finding the optimal observation path with the best observation node.

[0008] The literature (Zhao Yuxin, Du Denghui, Cheng Xiaohui, et al. A method for observation path planning of marine mobile observation network based on reinforcement learning [J]. Journal of Intelligent Systems, 2022, 17(1): 192-200) proposes a method for solving discrete and continuous action design problems in path planning using reinforcement learning algorithms, and conducts single-platform and multi-platform experiments using DQN and DDPG algorithms respectively. Compared with this invention, it does not demonstrate the advantages of multi-objective clustering, optimal observation path, speed and simplicity. Summary of the Invention

[0009] To address the shortcomings of existing technologies, the purpose of this invention is to provide a multi-objective clustering and optimal observation path planning method and system.

[0010] A multi-objective clustering and optimal observation path planning method provided by the present invention includes: Step S1: Collect targets to form an observation target set U; Step S2: Extract the highest priority target from the observation target set U in sequence according to priority. Cluster the other targets with the highest priority target as the center until there is no highest priority target to extract. Then, cluster the remaining targets according to the principle of minimizing the number of targets after clustering to obtain multiple clustered targets. Step S3: Based on the distribution of clustered targets near the nadir point of the space-based remote sensor, select the best path observation nodes; Step S4: Based on the optimal path observation nodes, select targets that do not consume space-based remote sensor maneuvering resources by prioritizing and retrieving them from the top of the sky, forming the optimal observation target set U. b ; Step S5: Add the remaining targets to set U one by one. b Select the target that consumes the least maneuvering resources from the space-based remote sensor and meets the maneuvering resource requirements, and add it to set U. b Repeat this process until the space-based remote sensor's maneuvering resources are exhausted.

[0011] Furthermore, in step S1, the target includes hotspot area information on the ground or targets acquired by other space-based remote sensors; The targets in the observation target set U are arranged sequentially according to the observation time.

[0012] Furthermore, in step S2, targets within the circle with the highest observation priority as the center and half the width of the land surface observed by the space-based remote sensor as the radius are clustered into one observation target.

[0013] Furthermore, in step S3, the optimal path observation node is the target that consumes the least maneuvering resources in the observation path. In the vertical orbit direction, the target is divided into two sets, U+ and U-, with the nadir point trajectory as the dividing line. Based on the principle that the set has a large number of targets and the roll angle variance of the space-based remote sensor observation targets within the set is small, the priority observation set and the candidate set are determined. In the priority observation set, the target observed by the space-based remote sensor whose roll angle is closest to the average roll angle of the priority observation set is determined as the optimal path observation node.

[0014] Furthermore, in step S4, targets that do not consume space-based remote sensor maneuvering resources include: the space-based remote sensor having maneuvered into position before observing the next target of the target.

[0015] A multi-objective clustering and optimal observation path planning system provided by the present invention includes: Module M1: Collects targets to form an observation target set U; Module M2: Extract the highest priority target from the observation target set U in sequence according to priority. Other targets are clustered with the highest priority target as the center until no highest priority target can be extracted. The remaining targets are then clustered according to the principle of minimizing the number of targets after clustering, resulting in multiple clustered targets. Module M3: Based on the distribution of clustered targets near the nadir point of the space-based remote sensor, select the best path observation nodes; Module M4: Based on the optimal path observation nodes, targets that do not consume space-based remote sensor maneuvering resources are selected forward and backward according to the overpass time, forming the optimal observation target set U. b ; Module M5: Add the remaining targets to set U one by one. b Select the target that consumes the least maneuvering resources from the space-based remote sensor and meets the maneuvering resource requirements, and add it to set U. b Repeat this process until the space-based remote sensor's maneuvering resources are exhausted.

[0016] Furthermore, in module M1, the targets include hotspot information on the ground or targets acquired by other space-based remote sensors; The targets in the observation target set U are arranged sequentially according to the observation time.

[0017] Furthermore, in module M2, targets within the circle centered on the highest priority target and with a radius equal to half the width of the land surface observed by the space-based remote sensor are clustered into a single observation target.

[0018] Furthermore, in module M3, the optimal path observation node is the target that consumes the least maneuvering resources in the observation path. In the vertical track direction, the target is divided into two sets, U+ and U-, with the nadir point trajectory as the dividing line. Based on the principle that the set has a large number of targets and the roll angle variance of the space-based remote sensor observation targets within the set is small, the priority observation set and the candidate set are determined. In the priority observation set, the target observed by the space-based remote sensor whose roll angle is closest to the average roll angle of the priority observation set is determined as the optimal path observation node.

[0019] Furthermore, in module M4, targets that do not consume space-based remote sensor maneuvering resources include those where the space-based remote sensor has already maneuvered into position before observing the next target.

[0020] Compared with the prior art, the present invention has the following beneficial effects: Based on constraints such as imaging swath width and maneuverability of space-based remote sensors, this invention proposes a multi-target clustering and optimal observation path planning method, which solves the technical problem of selecting observation paths for multi-target imaging of space-based remote sensors. Compared with existing path planning methods, it has significant convenience and engineering applicability, and is of great significance for realizing observation path planning for space-based remote sensors or multi-space-based remote sensor networks. Attached Figure Description

[0021] Other features, objects, and advantages of the present invention will become more apparent from the following detailed description of non-limiting embodiments with reference to the accompanying drawings: Figure 1 A schematic diagram of the multi-objective clustering and optimal observation path planning process; Figure 2 A schematic diagram illustrating an application scenario for a space-based remote sensor observing ground targets while in orbit. Figure 3 This is the result of clustering ground targets. Detailed Implementation

[0022] The present invention will now be described in detail with reference to specific embodiments. These embodiments will help those skilled in the art to further understand the present invention, but do not limit the invention in any way. It should be noted that those skilled in the art can make several changes and improvements without departing from the concept of the present invention. These all fall within the protection scope of the present invention.

[0023] like Figure 1 As shown, the multi-objective clustering and optimal observation path planning method of the present invention includes the following steps: Step S1: Collect targets to form an observation target set U.

[0024] The targets are those obtained from ground-based hotspot information or other space-based remote sensors, such as forest fires, vegetation changes, and calibration areas. The target set U should be arranged sequentially according to the observation time.

[0025] Step S2: Based on priority, extract the highest-priority targets sequentially from the observation target set U. Other targets are then clustered around the highest-priority target until no higher-priority target can be extracted. The remaining targets are then clustered according to the principle of minimizing the number of targets in each cluster, resulting in multiple clustered targets. Specifically, using the highest-priority target as the center and half the width of the land surface observed by the space-based remote sensor as the radius, targets within this circle are clustered into one observation target. The priority can be user-defined, with more urgent tasks having higher priority, such as natural disaster relief and ecological environment damage monitoring.

[0026] Step S3: Based on the distribution of clustered targets near the ground point of the space-based remote sensor, select the best path observation nodes.

[0027] The optimal path observation node is generally the target that consumes the least maneuvering resources along the observation path. In the vertical orbit direction, the target is divided into U-shaped sections using the nadir trajectory as the dividing line. + and U - Two sets are used to determine the priority observation set and the candidate set based on the principle that the set has a large number of targets and the roll angle variance of the space-based remote sensor observation targets within the set is small. In the priority observation set, the space-based remote sensor observation target whose roll angle is closest to the average roll angle of the set is determined as the optimal path observation node.

[0028] Step S4: Based on the optimal path observation nodes, select targets that do not consume space-based remote sensor maneuvering resources by prioritizing and retrieving them from the top of the sky, forming the optimal observation target set U. b .

[0029] "Without consuming the maneuvering resources of the space-based remote sensor" means that the space-based remote sensor has already maneuvered to its position before observing the next target.

[0030] Step S5: Add the remaining targets to set U one by one. b Select the target that consumes the least maneuvering resources from the space-based remote sensor and meets the maneuvering resource requirements, and add it to set U. b Repeat this process until the space-based remote sensor's maneuvering resources are exhausted.

[0031] Mobility resources can be adjusted based on the mobility of space-based remote sensors and their on-orbit operational status.

[0032] like Figure 2 As shown, assume the space-based remote sensor operates in a sun-synchronous orbit at an altitude of 500 km; based on hotspot information projected on the ground or target information acquired by other space-based remote sensors, it is known that there are 17 targets to be observed on the ground, forming a target set U, of which one is a high-priority target; the projection trajectory of the space-based remote sensor on the ground is denoted as trajectory L, and of the 17 ground targets, 14 targets are to the left of trajectory L, denoted as U. + Three targets are located to the right of trajectory L, denoted as U. - .

[0033] First, high-priority targets are extracted from the target set U. Using this target as the center and the observation field of view of the space-based remote sensor as the observation boundary, all other targets within this circle are clustered into one target. Targets not initially clustered are then clustered according to the principle of minimizing the number of targets in each subsequent cluster. Specifically, the remaining targets are sequentially clustered using themselves as the center and the observation field of view of the space-based remote sensor as the observation boundary. Targets with the most targets within this circle are then clustered. The remaining targets that did not initially participate in the clustering process continue until no targets can be clustered. The clustering results are shown below. Figure 3 Targets within a circle are clustered into one target. From 17 targets, they were clustered into 9 targets.

[0034] Then, during the observation process, the space-based remote sensor's maneuvering resources are consumed as little as possible while observing as many targets as possible. That is, throughout the entire observation process, the maneuvering amplitude of the space-based remote sensor is kept as small as possible. Based on the principle of a large number of ensemble targets and a small variance of the roll angle of the targets observed by the space-based remote sensor within the ensemble, U is determined. + For priority observation set, U - This is the candidate set. In set U... + In the optimal path observation node, the target clustered with high-priority targets (target number 4) is the one whose average roll angle is closest to that cluster. Based on the relationship between maneuver angles and overpass time intervals between targets, targets that do not consume maneuver energy are searched forward and backward from the optimal path observation node and added to the optimal observation path set U. b In the process, targets ①, ④, and ⑥ are added to set U through screening. b .

[0035] Finally, the set U that was not included b Other targets in U are added sequentially. b In the selection process, targets that consume the least maneuvering resources from space-based remote sensors and meet the maneuvering resource requirements are added to set U. b This process is repeated until the maneuvering resources of the space-based remote sensor are exhausted. Ultimately, the optimal observation path is planned as ①→④→⑥→⑧→⑨.

[0036] This invention also provides a multi-objective clustering and optimal observation path planning system. This system can be implemented by executing the steps of the multi-objective clustering and optimal observation path planning method. That is, those skilled in the art can understand the multi-objective clustering and optimal observation path planning method as a preferred embodiment of the multi-objective clustering and optimal observation path planning system. The system includes: Module M1: Collects targets to form an observation target set U.

[0037] Module M2: Extracts the highest priority target from the observation target set U in sequence according to priority. Other targets are clustered with the highest priority target as the center until no highest priority target can be extracted. The remaining targets are then clustered according to the principle of minimizing the number of targets after clustering, resulting in multiple clustered targets.

[0038] Module M3: Based on the distribution of clustered targets near the nadir point of the space-based remote sensor, select the best path observation nodes.

[0039] Module M4: Based on the optimal path observation nodes, targets that do not consume space-based remote sensor maneuvering resources are selected forward and backward according to the overpass time, forming the optimal observation target set U. b .

[0040] Module M5: Add the remaining targets to set U one by one. b Select the target that consumes the least maneuvering resources from the space-based remote sensor and meets the maneuvering resource requirements, and add it to set U. b Repeat this process until the space-based remote sensor's maneuvering resources are exhausted.

[0041] Those skilled in the art will understand that, besides implementing the system and its various devices, modules, and units provided by this invention in the form of purely computer-readable program code, the same functions can be achieved entirely through logical programming of the method steps, enabling the system and its various devices, modules, and units to function in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers. Therefore, the system and its various devices, modules, and units provided by this invention can be considered a hardware component, and the devices, modules, and units included therein for implementing various functions can also be considered structures within the hardware component; alternatively, the devices, modules, and units for implementing various functions can be considered both software modules implementing the method and structures within the hardware component.

[0042] Specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the specific embodiments described above, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Unless otherwise specified, the embodiments and features described in this application can be arbitrarily combined with each other.

Claims

1. A multi-objective clustering and optimal observation path planning method, characterized in that, include: Step S1: Collect targets to form an observation target set U; Step S2: Extract the highest priority target from the observation target set U in sequence according to priority. Cluster the other targets with the highest priority target as the center until there is no highest priority target to extract. Then, cluster the remaining targets according to the principle of minimizing the number of targets after clustering to obtain multiple clustered targets. Step S3: Based on the distribution of clustered targets near the nadir point of the space-based remote sensor, select the best path observation nodes; Step S4: Based on the optimal path observation nodes, select targets that do not consume space-based remote sensor maneuvering resources by prioritizing and retrieving them from the top of the sky, forming the optimal observation target set U. b ; Step S5: Add the remaining targets to set U one by one. b Select the target that consumes the least maneuvering resources from the space-based remote sensor and meets the maneuvering resource requirements, and add it to set U. b Repeat this process until the space-based remote sensor's maneuvering resources are exhausted.

2. The multi-objective clustering and optimal observation path planning method according to claim 1, characterized in that, In step S1, the target includes hotspot area information on the ground or targets acquired by other space-based remote sensors; The targets in the observation target set U are arranged sequentially according to the observation time.

3. The multi-objective clustering and optimal observation path planning method according to claim 1, characterized in that, In step S2, targets within the circle with the highest observation priority as the center and half the width of the land surface observed by the space-based remote sensor as the radius are clustered into one observation target.

4. The multi-objective clustering and optimal observation path planning method according to claim 1, characterized in that, In step S3, the optimal path observation node is the target that consumes the least maneuvering resources in the observation path. In the vertical orbit direction, the target is divided into two sets, U+ and U-, with the nadir point trajectory as the dividing line. Based on the principle that the set has a large number of targets and the roll angle variance of the space-based remote sensor observation targets in the set is small, the priority observation set and the candidate set are determined. In the priority observation set, the target observed by the space-based remote sensor whose roll angle is closest to the average roll angle of the priority observation set is determined as the optimal path observation node.

5. The multi-objective clustering and optimal observation path planning method according to claim 1, characterized in that, In step S4, targets that do not consume space-based remote sensor maneuvering resources include: the space-based remote sensor has maneuvered into position before observing the next target of the target.

6. A multi-objective clustering and optimal observation path planning system, characterized in that, include: Module M1: Collects targets to form an observation target set U; Module M2: Extract the highest priority target from the observation target set U in sequence according to priority. Other targets are clustered with the highest priority target as the center until no highest priority target can be extracted. The remaining targets are then clustered according to the principle of minimizing the number of targets after clustering, resulting in multiple clustered targets. Module M3: Based on the distribution of clustered targets near the nadir point of the space-based remote sensor, select the best path observation nodes; Module M4: Based on the optimal path observation nodes, targets that do not consume space-based remote sensor maneuvering resources are selected forward and backward according to the overpass time, forming the optimal observation target set U. b ; Module M5: Add the remaining targets to set U one by one. b Select the target that consumes the least maneuvering resources from the space-based remote sensor and meets the maneuvering resource requirements, and add it to set U. b Repeat this process until the space-based remote sensor's maneuvering resources are exhausted.

7. The multi-objective clustering and optimal observation path planning system according to claim 6, characterized in that, In module M1, targets include hotspot information on the ground or targets acquired by other space-based remote sensors; The targets in the observation target set U are arranged sequentially according to the observation time.

8. The multi-objective clustering and optimal observation path planning system according to claim 6, characterized in that, In module M2, targets within the circle with the highest observation priority as the center and half the width of the land surface observed by the space-based remote sensor as the radius are clustered into one observation target.

9. The multi-objective clustering and optimal observation path planning system according to claim 6, characterized in that, In module M3, the optimal path observation node is the target that consumes the least maneuvering resources in the observation path. In the vertical track direction, the target is divided into two sets, U+ and U-, with the nadir point trajectory as the dividing line. Based on the principle that the set has a large number of targets and the roll angle variance of the space-based remote sensor observation targets within the set is small, the priority observation set and the candidate set are determined. In the priority observation set, the target observed by the space-based remote sensor whose roll angle is closest to the average roll angle of the priority observation set is determined as the optimal path observation node.

10. The multi-objective clustering and optimal observation path planning system according to claim 6, characterized in that, In module M4, targets that do not consume space-based remote sensor maneuvering resources include those where the space-based remote sensor has maneuvered into position before observing the next target of that target.

Images