Search

CN-121702412-B - Dynamic path planning method, system and device based on ground wave radar

CN121702412BCN 121702412 BCN121702412 BCN 121702412BCN-121702412-B

Abstract

The application provides a dynamic path planning method, a system and a device based on ground wave radar, which comprise the steps of obtaining wide area dynamic data of an observation sea area through a shore-based high-frequency ground wave radar, obtaining short-range environment information around a mobile platform through a shipborne navigation radar, preprocessing and fusing the wide area dynamic data and the short-range environment information to construct a dynamic state vector containing multidimensional environment characteristics and uncertainty characterization, inputting the dynamic state vector into a path planning model based on reinforcement learning, optimizing through a multi-target rewarding function, and outputting an initial heading and a navigational speed of the mobile platform to form an initial observation path. According to the application, by constructing the shore-based and shipborne double-layer radar collaborative awareness network, the planning of the radar-mobile platform collaborative ocean dynamic observation path is realized, and technical support is provided for efficient observation in a dynamic ocean environment.

Inventors

  • YANG FAN
  • HU YUANDONG
  • ZHANG DONGLIANG
  • ZHAO WENXUAN

Assignees

  • 自然资源部珠海海洋中心(自然资源部珠海海洋预报台)

Dates

Publication Date
20260512
Application Date
20260224

Claims (8)

  1. 1. The dynamic path planning method based on the ground wave radar is characterized by comprising the following steps of: wide-area dynamic data of an observation sea area are obtained through a shore-based high-frequency ground wave radar, and short-range environment information around a mobile platform is obtained through a ship-borne navigation radar; preprocessing and fusing according to the wide-area dynamic data and the short-range environment information to construct a dynamic state vector containing multidimensional environment characteristics and uncertainty characterization; Inputting the dynamic state vector into a path planning model based on reinforcement learning, and outputting the initial heading and the navigational speed of the mobile platform through multi-objective rewarding function optimization to form an initial observation path; after the initial observation path is formed, the method further comprises: Controlling the mobile platform to execute the initial observation path, and continuously acquiring real-time radar observation data and platform state data; based on the real-time radar observation data and the platform state data, checking preset hierarchical re-planning triggering conditions in real time; if any re-planning triggering condition is triggered, invoking a lightweight reinforcement learning model, introducing an emergency priority weighting item into the multi-objective rewarding function, and generating and issuing an emergency re-planning path; After the emergency re-planning path is generated and issued, the method further comprises the following steps: Controlling the mobile platform to execute the emergency re-planning path, and feeding back path information of the mobile platform to the shore-based high-frequency ground wave radar to adjust the direction of a detection beam of the shore-based high-frequency ground wave radar; And simultaneously inputting the radar observation data into a marine numerical value assimilation module to modify a marine forecast model, and feeding back the modified forecast data to generate a new path plan.
  2. 2. The ground wave radar-based dynamic path planning method according to claim 1, wherein the dynamic state vector is: ; Wherein, the The method is characterized in that raster data is observed for a wide area of a shore radar, the dimension is 51 multiplied by 3, and the raster data corresponds to a flow field, wave height and a target density channel respectively; the method comprises the steps of short-range observation of raster data for the ship-borne radar, wherein the dimension is 51 multiplied by 2, and the raster data correspond to obstacle distribution and signal-to-noise ratio channels respectively; Is the electromagnetic interference intensity and frequency band vector; The method is a real-time rectangular coordinate of a mobile platform.
  3. 3. The method of claim 1, wherein the multi-objective rewards function comprises a composite rewards function of radar observation value rewards, obstacle avoidance rewards, observation effectiveness rewards and platform energy rewards, and the expression of the multi-objective rewards function is: ; Wherein alpha, beta, gamma and delta are self-adaptive weight coefficients and are dynamically adjusted according to the environment type; the method is characterized in that sea power element rewards are observed for radar, and the rewards are positively related to flow field gradients and wave height gradients; Dynamically assigning a barrier evasion reward according to the distance between the path and the barrier grid; A validity reward is observed for the radar, and judgment is carried out according to the SNR; And the energy consumption rewards of the platform and are inversely related to the actual power of the platform.
  4. 4. The ground wave radar-based dynamic path planning method of claim 1, wherein the re-planning triggering condition comprises at least one of a wide area environmental abrupt change, a short range risk proximity, and a radar observation failure; Triggering when the offset distance of the center of a flow field gradient area monitored in real time by a shore radar relative to a historical reference exceeds a first threshold value or the abrupt change value of wave height data exceeds a second threshold value; Triggering when the on-board radar detects that the distance between the moving obstacle and the platform is smaller than a third threshold value in real time or the short-range signal-to-noise ratio suddenly drops below a fourth threshold value in preset time; the triggering condition of radar observation failure is that when the electromagnetic interference intensity exceeds the anti-interference threshold value of the system and radar echo signals are continuously lost.
  5. 5. The dynamic path planning method based on ground wave radar according to claim 1, wherein the path information of the mobile platform is fed back to the shore-based high frequency ground wave radar to adjust its probe beam pointing direction, in particular by calculating the beam pointing angle by the following formula : ; Wherein, the For the location of the radar site, Is the center of the target area of the platform.
  6. 6. A dynamic path planning system based on ground wave radar, comprising: The collaborative perception data acquisition module is used for acquiring wide-area dynamic data of an observation sea area through a shore-based high-frequency ground wave radar and acquiring short-range environment information around the mobile platform through a ship-borne navigation radar; the dynamic state model construction module is used for preprocessing and fusing the wide-area dynamic data and the short-range environment information to construct a dynamic state vector containing multidimensional environment characteristics and uncertainty characterization; the initial observation path generation module is used for inputting the dynamic state vector into a path planning model based on reinforcement learning, outputting the initial heading and the navigational speed of the mobile platform through multi-target rewarding function optimization, and forming an initial observation path; after the initial observation path is formed, the method further comprises: Controlling the mobile platform to execute the initial observation path, and continuously acquiring real-time radar observation data and platform state data; based on the real-time radar observation data and the platform state data, checking preset hierarchical re-planning triggering conditions in real time; if any re-planning triggering condition is triggered, invoking a lightweight reinforcement learning model, introducing an emergency priority weighting item into the multi-objective rewarding function, and generating and issuing an emergency re-planning path; After the emergency re-planning path is generated and issued, the method further comprises the following steps: Controlling the mobile platform to execute the emergency re-planning path, and feeding back path information of the mobile platform to the shore-based high-frequency ground wave radar to adjust the direction of a detection beam of the shore-based high-frequency ground wave radar; And simultaneously inputting the radar observation data into a marine numerical value assimilation module to modify a marine forecast model, and feeding back the modified forecast data to generate a new path plan.
  7. 7. A dynamic path planning device based on ground wave radar, comprising: a memory unit for storing executable instructions, and A processing unit for interfacing with a memory to execute executable instructions to perform the method of any one of claims 1-5.
  8. 8. A computer readable storage medium, having stored thereon a computer program, the computer program being executable by a processor to implement the method of any of claims 1-5.

Description

Dynamic path planning method, system and device based on ground wave radar Technical Field The application belongs to the field of data application of ground wave radar detection signals, and particularly relates to a dynamic path planning method, system and device based on a ground wave radar. Background With the deep advancement of ocean national strategies, the real-time, global and accurate requirements of the fields of ocean power element monitoring, emergency maritime response, ocean resource exploration and the like on observed data are increasingly urgent, and an ocean dynamic observation technology becomes a core foundation for supporting ocean scientific research and ocean rights maintenance. The current ocean observation technology forms a composite system of shore-based fixed observation and mobile platform sampling, a shore-based radar can realize wide-area coverage, mobile platforms such as unmanned survey vessels, underwater gliders and the like can finish fine point sampling, and reinforcement learning algorithms are also gradually applied to mobile platform path planning, but the prior art still has a plurality of bottlenecks. The current ocean observation technology has the main problems of static observation path, inadaptability to dynamic ocean environment, limited single-point sampling coverage, insufficient cooperation of radar and a mobile platform, slow response of emergency re-planning, unbalanced value of observation data and the like. Therefore, a radar-mobile platform collaborative marine dynamic observation path planning scheme capable of providing technical support for efficient observation in a dynamic marine environment is lacking in the field. The statements made above merely provide background information related to the present disclosure and may not constitute prior art to the present disclosure except as may be expressly incorporated herein in any of the various aspects of the present disclosure. Disclosure of Invention The invention provides a dynamic path planning method, a system and a device based on ground wave radar, which constructs a shore-based and shipborne double-layer radar collaborative sensing network, and combining reinforcement learning and a lightweight emergency re-planning mechanism, introducing marine numerical assimilation to form closed-loop optimization, and finally realizing efficient, safe and self-adaptive planning of an observation path in a dynamic marine environment. According to a first aspect of the embodiment of the present application, there is provided a dynamic path planning method based on ground wave radar, including the steps of: wide-area dynamic data of an observation sea area are obtained through a shore-based high-frequency ground wave radar, and short-range environment information around a mobile platform is obtained through a ship-borne navigation radar; Preprocessing and fusing wide-area dynamic data and short-range environment information to construct a dynamic state vector containing multidimensional environment characteristics and uncertainty characterization; and inputting the dynamic state vector into a path planning model based on reinforcement learning, optimizing through a multi-target rewarding function, and outputting the initial heading and the navigational speed of the mobile platform to form an initial observation path. In some embodiments of the present application, after forming the initial observation path, further comprising: Controlling the mobile platform to execute an initial observation path, and continuously acquiring real-time radar observation data and platform state data; Based on real-time radar observation data and platform state data, checking preset hierarchical re-planning triggering conditions in real time; and if any re-planning triggering condition is triggered, invoking a lightweight reinforcement learning model, introducing an emergency priority weighting item into the multi-target rewarding function, and generating and issuing an emergency re-planning path. In some embodiments of the present application, after generating and issuing the emergency re-planning path, the method further includes: and S7, collaborative optimization and closed loop feedback, namely controlling the mobile platform to execute an emergency re-planning path, feeding back path information of the mobile platform to the shore-based high-frequency ground wave radar to adjust the direction of a detection beam of the shore-based high-frequency ground wave radar, inputting radar observation data into the ocean numerical assimilation module to correct an ocean forecast model, and feeding back the corrected forecast data to generate a new path plan. In some embodiments of the present application, the dynamic state vector is: ; Wherein, the The method is characterized in that raster data is observed for a wide area of a shore radar, the dimension is 51 multiplied by 3, and the raster data corresponds to a flow field, wave heigh