A marine environment information fusion method based on a heterogeneous unmanned platform cluster

Abstract

The application discloses a kind of ocean environment information fusion methods based on heterogeneous unmanned platform cluster, comprising: based on heterogeneous unmanned platform cluster, set task information;According to task information, in combination with task constraint, through collaborative task planning algorithm, task allocation and path planning are carried out, and the observation path and the layout position of unmanned platform are generated;Unmanned platform cluster carries out collaborative observation, and the ocean environment information in the place is observed, and ocean environment data are perceived;Through ocean environment feature diagnosis algorithm, the data observed by heterogeneous unmanned platform cluster is fused and analyzed, and the characteristic parameter of jump layer is extracted;Through the isobath generation method based on thin plate spline function, the depth data observed by heterogeneous unmanned platform cluster is fused and analyzed, and isobath is fitted and constructed;Electronic chart and ocean environment information are fused, and visual display is carried out through rendering drawing technology.The application can realize the improvement of ocean environment feature perception recognition success rate and efficiency.

Description

Technical Field

[0001] This invention relates to the field of manned / unmanned intelligent collaborative command technology, and in particular to a method for fusion of marine environmental information based on heterogeneous unmanned platform clusters. Background Technology

[0002] With the increasing informatization of the modern marine environment, marine environmental perception capabilities have become a crucial factor affecting the full utilization of equipment's overall effectiveness. Marine environmental observation and acquisition methods are a key research direction in marine science and engineering. High-precision, high-resolution environmental parameters have significant application value for underwater acoustic propagation, target detection, and underwater acoustic communication. Marine environmental observation data can be obtained through sampling from fixed nodes such as underwater buoys / motor buoys or from mobile nodes carrying sensors. Due to the high cost of deploying fixed observation nodes over a large area in the ocean, while mobile nodes offer advantages such as flexible observation and wide coverage, adaptive sampling technology, represented by mobile observation nodes such as autonomous underwater vehicles (AUVs) and underwater gliders, has gradually become a research focus in recent years. Currently, using a single platform for marine environmental observation suffers from problems such as long detection times, low efficiency in acquiring results, and low resource utilization.

[0003] The technical problem to be solved by this invention is to integrate and process multi-source marine environmental data through collaborative observation of heterogeneous unmanned platform clusters, identify strata and construct geographic information to form a more refined marine environmental situation, which can be used to support important applications such as covert route planning, target detection, and underwater acoustic communication.

[0004] To address the aforementioned issues and fully leverage the advantages of collaborative observation, this paper proposes a marine environmental information fusion method based on a heterogeneous unmanned platform cluster. It focuses on observing marine environmental elements such as temperature, salinity, and depth within the collaborative context of a heterogeneous unmanned platform cluster. Through multi-source marine environmental data analysis and fusion technology, the efficiency of marine environmental information observation is improved. The types of unmanned platforms mainly include: surface unmanned surface vessels, wave boats, AUVs, manta ray simulators, gliders, observation buoys, and observation moorings. Each unmanned platform is equipped with at least one set of temperature, salinity, and depth sensors. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a marine environmental information fusion method based on a heterogeneous unmanned platform cluster, which addresses the shortcomings of the existing technology.

[0006] The technical solution adopted by this invention to solve its technical problem is: This invention provides a method for fusion of marine environmental information based on a heterogeneous unmanned platform cluster, the method comprising the following steps: Step 1: Based on the heterogeneous unmanned platform cluster, set the task information, including: the size of the sea area to be observed, the type and number of unmanned platforms, the starting position of the sea area to be observed, and the performance parameters of the unmanned platforms; Step 2: Based on the mission information and the marine environmental forecast field, and combined with mission constraints including communication distance and mission time, a collaborative mission planning algorithm is used to allocate tasks and plan paths, generating the observation path for each dynamic unmanned platform and the deployment position for each static unmanned platform. Step 3: The heterogeneous unmanned platform cluster conducts collaborative observation. Through the sensors it carries, including temperature, salinity, and depth sensors and multibeam sonar, it observes the marine environment information and senses marine environmental data, including temperature, salinity, and depth. Step 4: Using a marine environmental feature diagnosis algorithm, fuse and analyze the temperature and salinity data observed by the heterogeneous unmanned platform cluster, extract the feature parameters of the strata, and complete the identification of the strata; using a contour line generation method based on thin plate spline function, fuse and analyze the depth data observed by the heterogeneous unmanned platform cluster, and fit and construct contour lines. Step 5: Integrate the electronic nautical chart with marine environmental information, including strata and isobaths, and visualize it using rendering technology.

[0007] Furthermore, the method for task allocation in step 2 of the present invention specifically includes: Input: Latitude and longitude boundaries of the sea area to be observed, formation and node spacing, type, number and performance parameters of unmanned platforms, remaining range of unmanned platforms, nautical chart of the target sea area, current position of each unmanned platform, center coordinates and radius of obstacles, and ocean current information; Processing procedure: Step a1: Extract information from the pre-imported nautical chart, process obstacle and current information, generate restricted areas using the center coordinates and radius of obstacles, and use them as constraints in the clustering algorithm; mark the current areas and directions for path planning optimization. Step a2: Divide the mission area of ​​the unmanned platform into sub-regions and implement this using the K-means clustering algorithm; estimate the initial number of clusters based on the platform spacing and the area of ​​the sea to be observed. Initial cluster number = sea area / platform spacing 2; The constraints are that cluster centers avoid obstacles, and surface and underwater platforms are clustered hierarchically; the objective function is to minimize the sum of the distances from observation points to cluster centers; each cluster region is a sub-region, and its center coordinates and boundaries are labeled. Step a3: Assign tasks to the divided sub-regions and formulate a task assignment strategy using a genetic algorithm; Output: Preliminary path planning for each dynamic unmanned platform, including the location of waypoints and speed configuration between waypoints for each route.

[0008] Furthermore, the method for formulating a task allocation strategy using a genetic algorithm in step a3 of the present invention specifically includes: The optimization objective is to minimize the global task completion cost, and the objectives include: minimizing the total travel distance, balancing platform utilization, and maximizing task coverage. Set constraints: platform remaining range ≥ required travel distance, platform type matches sub-region requirements; The coding design is carried out such that each gene represents a sub-region, and the gene value is an assigned platform ID; The fitness function is designed with penalties for the platform exceeding its range or failing to meet the matching rules, with exceeding the range being the maximum penalty. Perform genetic operations and output the optimal solution.

[0009] Furthermore, the method for matching platform type with sub-regional requirements according to the present invention specifically includes: Platform types include: unmanned surface vessels (USVs), manta ray / gliders, buoys / submarine moorings, and AUVs; among which: Unmanned surface vessels (USVs) are suitable for long-distance sub-regional scenarios and have a high degree of matching requirements. The manta ray / glider design is suitable for complex terrain and areas with many obstacles, and is rated as medium in terms of matching requirements. Buoys / submersible moorings are suitable for long-term monitoring of key locations, and the matching degree of demand is fixed. AUVs are suitable for use in waters close to the mother ship to maintain communication quality and have a high degree of matching requirements.

[0010] Furthermore, the method for path planning in step 2 of the present invention specifically includes: Input: Each unmanned platform and its corresponding mission sub-region, performance parameters of each unmanned platform, nautical chart information of the target sea area, boundary of the mission sub-region, observation starting point assigned to each unmanned platform, corresponding unmanned platform formation, formation and node spacing; Processing procedure: Step b1: Data preprocessing; rasterize and initialize the nautical chart, convert the task sub-region and ocean current vector field, mark obstacle grids as impassable areas, add directional weights to ocean current grids, and define the speed configuration rules as follows: unmanned platforms should prioritize sailing at the preset recommended speed; if explicitly specified, the path segment will inherit the platform type's preset speed by default. Step b2: Based on the raster map generated after rasterization, use the improved A* path planning algorithm to output the preliminary path planning for each unmanned platform; Step b3, Path optimization; Perform curve fitting on the initially generated path plan, take the tangent direction of the preceding and following points, and generate a new path; Perform cubic spline interpolation on the speed configuration to ensure continuous acceleration; Output: Preliminary path planning for each dynamic unmanned platform, including the location of waypoints and speed configuration between waypoints for each route.

[0011] Furthermore, the improved A* path planning algorithm in step b2 of the present invention specifically includes: The multi-weighted cost function is:

[0012] Wherein, the base cost G(n) is the actual distance traveled, calculated based on the difference between latitude and longitude and depth, taking into account the curvature of the earth; the heuristic cost function H(n) is the Manhattan distance from the unmanned platform to the target point, including the depth difference penalty; the dynamic correction term C(n) is the ocean current influence coefficient, which corrects the travel energy consumption based on the downstream / upstream flow. The design includes platform motion constraints and remaining range constraints. Platform motion constraints include the maximum allowable turning angle for unmanned surface vessels and low-speed platforms, while glider-type platforms are only allowed to follow paths with smooth depth changes. The remaining range constraint is the total path distance calculated in real time. If the path exceeds the platform's remaining range, dynamic pruning is triggered, and the path is backtracked to the nearest valid node. Run the improved A* path planning algorithm to output the preliminary path plan for each unmanned platform, including the location of waypoints and the speed configuration between waypoints for each route.

[0013] Furthermore, step 4 of the present invention specifically includes: Step 4.1, Data Preprocessing: The raw marine environmental data is preprocessed using methods including outlier detection, filtering, and missing value handling. Step 4.2, Layer Crossing Identification: The temperature and salinity data obtained from the unmanned platform are fused with the background field data using a weighted average method. The fused value = measured data + background field data * weight. The weight is determined by the closeness between the background field data and the measured data, and the degree of closeness is proportional to the weight. The background field data is then corrected using the measured data. Based on the fused background field, temperature profile data and salinity profile data are extracted for layer crossing diagnosis. Step 4.3, Contour Line Fitting and Construction: Based on the depth data measured by the unmanned platform, the contour line is constructed by using a thin plate spline function-based contour line generation method.

[0014] Furthermore, the method for handling missing values ​​in step 4.1 of the present invention specifically includes: Handling small-scale continuous missing data: For cases where the continuous missing range in the original marine environmental data is less than a certain data segment, piecewise linear interpolation is used to fill in the missing data; by taking values ​​at both ends of the missing interval and calculating the intermediate missing value according to a linear relationship, it is suitable for cases of short-term data loss. Handling large-scale missing data: For cases where the continuous missing range is greater than a certain data segment, the missing range is marked. Generate missing data: In the missing interval, input context information to allow the generator to generate reasonable numerical placeholders to fill the missing parts and maintain the temporal continuity of the data.

[0015] Furthermore, the specific process of the layer skipping diagnosis of the present invention is as follows: The formula for diagnosing hierarchical changes is:

[0016] Where Q is the temperature gradient value, representing the amount of temperature change per unit depth; ΔT is the temperature difference between two layers of seawater; ΔZ is the depth difference between two layers of seawater; when the temperature gradient Q of a certain water layer exceeds a preset critical value, the water layer is defined as a thermocline. The contour line generation method based on the thin plate spline function TPS is used to achieve contour line fitting and construction. The specific process is as follows: Data preparation: The input data consists of an irregularly distributed set of sounding points, which undergoes preprocessing including outlier removal, coordinate system unification, and data normalization. Constructing a TPS interpolation model: By constructing a smooth surface to fit all sounding points, isobaths are extracted from this surface. The objective of the thin-plate spline function is to minimize the bending energy of the surface while ensuring that the interpolated surface strictly passes through all sounding points. The mathematical expression is:

[0017] Where X=(x,y) are planar coordinates. Let be the plane coordinates of the i-th sounding point. It is the radial basis function of the two-dimensional TPS, where r is the distance; The weighting coefficients are determined by solving a system of linear equations. , , These are the coefficients for the linear term, used to eliminate the effects of overall translation and rotation; Solution weight and linear term coefficients , , This makes the interpolated surface satisfy:

[0018] Generate a regular mesh: A regular grid is generated within the study area to calculate the interpolated depth values; the grid extent is defined, the grid resolution is set, and the grid point coordinates are generated. ; Calculate the depth value of the grid points: For each grid point Substitute into the TPS function to calculate the depth:

[0019] Obtain the depth dataset of the regular grid ; Extracting contour lines: The depth values ​​of the regular grid are binarized, the Marching Squares algorithm is applied to extract contour line segments, the segments are connected to form closed contour lines, and the contour lines are smoothed.

[0020] This invention provides a marine environmental information fusion system based on a heterogeneous unmanned platform cluster, comprising: Memory, used to store executable computer programs; The processor, when executing executable computer programs stored in memory, implements a method for fusion of marine environmental information based on a heterogeneous unmanned platform cluster.

[0021] The beneficial effects of this invention are: (i) Heterogeneous unmanned platform clusters have high spatial and temporal parallelism, which can effectively expand the spatiotemporal coverage of underwater missions. Through the overlapping coverage of multiple machines, the uncertainty of data acquisition by a single platform and the limitation of measurement range can be compensated.

[0022] (ii) The present invention can improve the success rate and efficiency of marine environmental feature perception and recognition, and the success rate can be increased by more than 10% compared with existing methods.

[0023] (iii) This invention helps to form a more refined marine environmental situation, and improves stealth and combat effectiveness. Attached Figure Description

[0024] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a flowchart of the marine environmental information fusion method based on a heterogeneous unmanned platform cluster according to an embodiment of the present invention; Figure 2 This is a diagram showing the platform types, quantities, and performance parameters of an embodiment of the present invention; Figure 3 This is a platform type and regional demand matching diagram according to an embodiment of the present invention; Figure 4This is a flowchart of the layer-by-layer recognition process according to an embodiment of the present invention; Figure 5 This is a flowchart illustrating the construction process of contour lines according to an embodiment of the present invention; Figure 6 This is a visual flowchart illustrating an embodiment of the present invention; Figure 7 This is a flowchart of the hierarchical diagnosis process according to an embodiment of the present invention. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0026] Example 1 The technical solution of the marine environmental information fusion method based on heterogeneous unmanned platform clusters in this invention is shown below, and the specific steps are detailed in the appendix. Figure 1 .

[0027] The first step is to set up the task.

[0028] Set the size of the sea area to be observed, the type and number of unmanned platforms, the starting position of the sea area to be observed, and the performance parameters of the unmanned platforms.

[0029] The second step is collaborative task planning.

[0030] Based on the information input in the first step, and in conjunction with constraints such as communication distance and mission time, a collaborative mission planning algorithm is used to allocate tasks and plan paths, generating the observation path for each dynamic unmanned platform and the deployment location for each static unmanned platform.

[0031] The third step is collaborative observation.

[0032] The unmanned platform cluster observes the marine environment by carrying temperature, salinity, and depth sensors and multibeam sonar, sensing important marine environmental data such as temperature, salinity, and depth.

[0033] The fourth step is the fusion and processing of multi-source marine environmental observation data.

[0034] By using marine environmental feature diagnostic algorithms, temperature and salinity data observed by heterogeneous unmanned platform clusters are fused and analyzed to extract the characteristic parameters of the strata and complete the identification of the strata.

[0035] By using a thin plate spline function (TPS)-based method for generating contour lines, we can fuse and analyze the depth data observed by a cluster of heterogeneous unmanned platforms and fit and construct contour lines.

[0036] Step 5: Visual presentation.

[0037] Electronic nautical charts are integrated with marine environmental information such as strata and isobaths, and then visualized using rendering techniques.

[0038] Example 2 Based on Example 1, this invention provides a more detailed description of the invention by combining specific experimental data and scenarios with accompanying drawings and examples.

[0039] The first step is to set up the task.

[0040] The area to be observed is defined as a square sea area of ​​100km × 100km. The type, quantity, and performance parameters of the unmanned platforms are as follows (see attached). Figure 2 (As shown).

[0041] The second step is collaborative task planning.

[0042] (1) Task allocation Input: Latitude and longitude boundaries of the sea area to be observed, formation and node spacing, type, number and performance parameters of unmanned platforms, remaining range of unmanned platforms, nautical chart of the target sea area, current position of each unmanned platform, center coordinates and radius of obstacles, and ocean current information.

[0043] Processing procedure: 1) First, extract information from the pre-imported nautical charts, process the information on obstacles and ocean currents, generate restricted areas using the center coordinates and radius of the obstacles (such as expanding the radius to a safe distance), and use them as constraints in the clustering algorithm (such as removing solutions whose cluster centers fall within the restricted areas); mark the ocean current areas and directions for optimizing path planning (such as saving energy by sailing with the current).

[0044] 2) The mission area of ​​the unmanned platform is divided into sub-regions, and the K-means clustering algorithm is used. The initial number of clusters is estimated based on the platform spacing and the area of ​​the sea to be observed: initial number of clusters = sea area / platform spacing 2; the constraint is that the cluster centers avoid obstacles, and the surface / underwater platforms are clustered hierarchically; the objective function is to minimize the sum of the distances from the observation points to the cluster centers; each cluster region is a sub-region, and its center coordinates and boundaries are marked.

[0045] 3) Task allocation is performed for the divided sub-regions, using a genetic algorithm to formulate a task allocation strategy. First, the optimization objective is to minimize the global task completion cost, including: minimizing total travel distance, balancing platform utilization (avoiding overload on some platforms), and maximizing task coverage. Second, constraints are set: remaining platform range ≥ required travel distance, and platform type matches sub-region requirements (matching rules are attached). Figure 3Then, the encoding design is performed, where each gene represents a sub-region and the gene value is the assigned platform ID; a fitness function is designed, with penalties for platform overrun or failure to meet matching rules (where overrun is the maximum penalty); finally, genetic operations are performed, and the optimal solution is output.

[0046] Output: Each unmanned platform and its corresponding task sub-region, the center point of the task sub-region, the boundary of the task sub-region, the observation starting point assigned to each unmanned platform, and the corresponding unmanned platform formation.

[0047] (2) Path planning Inputs: each unmanned platform and its corresponding mission sub-region, performance parameters of each unmanned platform, nautical chart information of the target sea area, boundary of the mission sub-region, observation starting point assigned to each unmanned platform, corresponding unmanned platform formation, formation and node spacing.

[0048] Processing procedure: 1) Data preprocessing: First, the nautical chart is rasterized and initialized, and the task sub-region and ocean current vector field are converted. Obstacle grids are marked as impassable areas, and ocean current grids are given directional weights (0.8 for downstream cost coefficient and 1.5 for upstream cost coefficient). The speed configuration rules are defined as follows: the unmanned platform should prioritize sailing at the preset recommended speed; if explicitly specified, the path segment inherits the preset speed of the platform type by default; replanning can perform a certain degree of route planning.

[0049] 2) Based on the generated raster map, an improved A* path planning algorithm is used; the multi-weight cost function is: The system is structured as follows: the base cost G(n) is the actual travel distance (calculated using latitude and longitude depth differences, taking into account the Earth's curvature); the heuristic cost function H(n) is the Manhattan distance from the unmanned platform to the target point (including depth difference penalties); and the dynamic correction term C(n) is the ocean current influence coefficient (correcting travel energy consumption with / against the current). Platform motion constraints and remaining range constraints are designed. Platform motion constraints include a maximum allowable turning angle of 30° / s for unmanned surface vessels (15 knots), a maximum allowable turning angle of 15° / s for low-speed platforms (4 knots), and glider-type platforms are only allowed smooth depth-change paths (vertical slope ≤ 1:10). The remaining range constraint is the total path distance calculated in real-time; if the remaining range is exceeded, dynamic pruning (backtracking to the nearest valid node) is triggered. The improved A* path planning algorithm is run to output the preliminary path plan for each unmanned platform, including the location of each path point and the speed configuration between points.

[0050] 3) Path optimization: Curve fitting is performed on the initially generated path plan, and the tangent direction of the preceding and following points is taken to generate a new path; cubic spline interpolation is performed on the speed configuration to make the acceleration continuous.

[0051] Output: Preliminary path planning for each dynamic unmanned platform, including the location of waypoints and speed configuration between waypoints for each route.

[0052] The third step is collaborative observation.

[0053] The unmanned platform conducts collaborative observations along the planned route or location. The information to be observed includes: the position of the unmanned platform in the three-dimensional coordinate system, marine environmental data such as temperature, salinity, and depth acquired at this location, whether the current location is within the observation background field, and how many times the sample has been taken.

[0054] The fourth step is the fusion and processing of multi-source marine environmental observation data.

[0055] (1) Data preprocessing First, the raw marine environmental data is preprocessed using outlier detection and filtering techniques. The handling of missing values ​​is shown below.

[0056] Handling small-range consecutive missing data: For cases where the range of consecutive missing data is small, piecewise linear interpolation is used for completion. This method takes values ​​at both ends of the missing interval and calculates the intermediate missing value according to a linear relationship, making it suitable for situations where data is lost over a short period of time.

[0057] Handling large-scale missing data: For data segments with a large range of consecutive missing data, the missing range is marked.

[0058] Generate missing data: In the missing interval, input context information to allow the generator to generate reasonable numerical placeholders to fill in the missing parts and maintain the temporal continuity of the data as much as possible.

[0059] (2) Layer-by-layer identification Temperature and salinity data obtained from unmanned platforms are fused with background field data using a weighted average method. The fused value = measured data + background field data * weight. The weight is determined by the similarity between the background field data and the measured data; the closer they are, the larger the weight, and vice versa. Using the measured data to correct the background field data can improve the prediction accuracy of the background field.

[0060] Based on the fused background field, temperature and salinity profile data are extracted for layer transition diagnosis. The process is shown in the attached figure. Figure 4 As shown.

[0061] (3) Construction of contour lines by fitting This invention, based on depth data measured by an unmanned platform, employs a contour line generation method based on Thin Plate Spline Function (TPS) to achieve contour line fitting and construction. The process is shown in the attached figure. Figure 5 As shown.

[0062] In a preferred embodiment of the present invention, the multi-agent adaptive sampling path planning problem is essentially a constrained optimization global path planning problem. It requires constructing an adaptive sampling path planning model based on the marine environmental forecast field and the characteristics of the multi-agent measurement equipment. Through optimization of the modeling problem, the optimal sampling path for the multi-agents is obtained, thereby acquiring observational data of important environmental field samples. Then, the high-resolution data is assimilated with the low-resolution field predicted by the model to construct a high-resolution and high-precision estimated environmental field, providing data support for thermocline diagnosis. The flowchart is as follows: Figure 7 As shown.

[0063] The core formula for thermocline diagnosis is:

[0064] in: Q: Temperature gradient value (unit: °C / m), representing the amount of temperature change per unit depth.

[0065] ΔT: Temperature difference between two layers of seawater (unit: °C).

[0066] ΔZ: The depth difference between two layers of seawater (unit: m).

[0067] Judgment rules: A water layer is defined as a thermocline when its temperature gradient Q exceeds a preset critical value. The critical value is categorized into two types based on the water depth: Shallow sea (water depth < 200m): The critical value is 0.2℃ / m. If Q ≥ 0.2℃ / m of a certain water layer, it is determined to be a thermocline.

[0068] Deep sea (depth > 200m): The critical value is 0.05℃ / m. If Q ≥ 0.05℃ / m of a certain water layer, it is determined to be a thermocline.

[0069] In a preferred embodiment of the present invention, the contour line generation method based on Thin Plate Spline (TPS) is a nonparametric interpolation technique suitable for processing irregularly distributed discrete sounding data (such as point cloud data acquired by multibeam sonar and single-beam echo sounders). Its core idea is to construct a smooth surface to fit all sounding points, and then extract contour lines from this surface.

[0070] Thin-plate spline functions are a radial basis function interpolation method that aims to minimize the bending energy of a surface while ensuring that the interpolated surface passes strictly through all sounding points. The mathematical expression is:

[0071] in: X=(x,y) represents planar coordinates (such as latitude and longitude or projected coordinates).

[0072] Let be the plane coordinates of the i-th sounding point.

[0073] (Radial basis function of 2D TPS, where r is the distance).

[0074] The weighting coefficients are determined by solving a system of linear equations.

[0075] , , These are the coefficients of the linear term, used to eliminate the effects of overall translation and rotation.

[0076] (1) Data preparation Input data: an irregularly distributed set of sounding points Data preprocessing: Remove outliers (such as those with negative depth or depths much greater than those of surrounding points).

[0077] Unify the coordinate system (e.g., convert to UTM projected coordinates).

[0078] Normalize the data (optional, to speed up numerical computation).

[0079] (2) Constructing the TPS interpolation model Solving for weights and linear term coefficients , , This makes the interpolated surface satisfy:

[0080] (3) Generate regular mesh Generate a regular grid (such as a latitude and longitude grid or a UTM grid) within the study area to calculate the interpolated depth value. Define the grid range (such as latitude and longitude range or UTM range).

[0081] Set the grid resolution (e.g., spacing of 10 meters or 1 arcsecond).

[0082] Generate grid point coordinates .

[0083] (4) Calculate the depth value of the grid point For each grid point Substitute into the TPS function to calculate the depth:

[0084] Obtain the depth dataset of the regular grid .

[0085] (5) Extracting contour lines Binarize the depth values ​​of the regular grid (e.g., to determine if they are equal to the target depth). The Marching Squares algorithm is used to extract contour lines. Connect the line segments to form a closed contour line (if necessary). Smooth contour lines (e.g., B-spline curves).

[0086] Step 5: Visual presentation.

[0087] Electronic nautical charts are integrated with marine environmental information such as strata and isobaths, and visualized using rendering techniques, as shown in the attached figure. Figure 6 As shown, you can view the contour lines on the electronic nautical chart and view the cross-sectional profile at a specified location.

[0088] Example 3 This invention provides a marine environmental information fusion system based on a heterogeneous unmanned platform cluster, comprising: Memory, used to store executable computer programs; The processor, when executing executable computer programs stored in memory, implements a method for fusion of marine environmental information based on a heterogeneous unmanned platform cluster.

[0089] In this embodiment, the memory (i.e., the readable storage medium) includes flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and programmable read-only memory (PROM). It can also be an external storage device of the computer device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device. Of course, the memory can also include both internal storage units and external storage devices of the computer device. In this embodiment, the memory is typically used to store the operating system and various application software installed on the computer device, such as the program code of the marine environmental information fusion method based on a heterogeneous unmanned platform cluster in the method embodiment. Furthermore, the memory can also be used to temporarily store various types of data that have been output or will be output.

[0090] In some embodiments, the processor may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor is typically used to control the overall operation of a computer device. In this embodiment, the processor is used to run program code stored in memory or process data, such as running a program for a marine environmental information fusion method based on a heterogeneous unmanned platform cluster.

[0091] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0092] It should be understood that those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A method for fusion of marine environmental information based on heterogeneous unmanned platform clusters, characterized in that, The method includes the following steps: Step 1: Based on the heterogeneous unmanned platform cluster, set the task information, including: the size of the sea area to be observed, the type and number of unmanned platforms, the starting position of the sea area to be observed, and the performance parameters of the unmanned platforms; Step 2: Based on the mission information and the marine environmental forecast field, and combined with mission constraints including communication distance and mission time, a collaborative mission planning algorithm is used to allocate tasks and plan paths, generating the observation path for each dynamic unmanned platform and the deployment position for each static unmanned platform. Step 3: The heterogeneous unmanned platform cluster conducts collaborative observation. Through the sensors it carries, including temperature, salinity, and depth sensors and multibeam sonar, it observes the marine environment information and senses marine environmental data, including temperature, salinity, and depth. Step 4: Using a marine environmental feature diagnosis algorithm, fuse and analyze the temperature and salinity data observed by the heterogeneous unmanned platform cluster, extract the feature parameters of the strata, and complete the identification of the strata; using a contour line generation method based on thin plate spline function, fuse and analyze the depth data observed by the heterogeneous unmanned platform cluster, and fit and construct contour lines. Step 5: Integrate the electronic nautical chart with marine environmental information, including strata and isobaths, and visualize it using rendering technology.

2. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 1, characterized in that, The method for task allocation in step 2 specifically includes: Input: Latitude and longitude boundaries of the sea area to be observed, formation and node spacing, type, number and performance parameters of unmanned platforms, remaining range of unmanned platforms, nautical chart of the target sea area, current position of each unmanned platform, center coordinates and radius of obstacles, and ocean current information; Processing procedure: Step a1: Extract information from the pre-imported nautical chart, process obstacle and current information, generate restricted areas using the center coordinates and radius of obstacles, and use them as constraints in the clustering algorithm; mark the current areas and directions for path planning optimization. Step a2: Divide the mission area of ​​the unmanned platform into sub-regions and implement this using the K-means clustering algorithm; estimate the initial number of clusters based on the platform spacing and the area of ​​the sea to be observed. Initial number of clusters = sea area / platform spacing 2 ; The constraints are that cluster centers avoid obstacles, and surface and underwater platforms are clustered hierarchically; the objective function is to minimize the sum of the distances from observation points to cluster centers; each cluster region is a sub-region, and its center coordinates and boundaries are labeled. Step a3: Assign tasks to the divided sub-regions and formulate a task assignment strategy using a genetic algorithm; Output: Preliminary path planning for each dynamic unmanned platform, including the location of waypoints and speed configuration between waypoints for each route.

3. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 2, characterized in that, The method for formulating a task allocation strategy using a genetic algorithm in step a3 specifically includes: The optimization objective is to minimize the global task completion cost, and the objectives include: minimizing the total travel distance, balancing platform utilization, and maximizing task coverage. Set constraints: platform remaining range ≥ required travel distance, platform type matches sub-region requirements; The coding design is carried out such that each gene represents a sub-region, and the gene value is an assigned platform ID; The fitness function is designed with penalties for the platform exceeding its range or failing to meet the matching rules, with exceeding the range being the maximum penalty. Perform genetic operations and output the optimal solution.

4. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 3, characterized in that, The method for matching platform type with sub-regional needs specifically includes: Platform types include: unmanned surface vessels (USVs), manta ray / gliders, buoys / submarine moorings, and AUVs; among which: Unmanned surface vessels (USVs) are suitable for long-distance sub-regional scenarios and have a high degree of matching requirements. The manta ray / glider design is suitable for complex terrain and areas with many obstacles, and is rated as medium in terms of matching requirements. Buoys / submersible moorings are suitable for long-term monitoring of key locations, and the matching degree of demand is fixed. AUVs are suitable for use in waters close to the mother ship to maintain communication quality and have a high degree of matching requirements.

5. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 2, characterized in that, The method for path planning in step 2 specifically includes: Input: Each unmanned platform and its corresponding mission sub-region, performance parameters of each unmanned platform, nautical chart information of the target sea area, boundary of the mission sub-region, observation starting point assigned to each unmanned platform, corresponding unmanned platform formation, formation and node spacing; Processing procedure: Step b1: Data preprocessing; rasterize and initialize the nautical chart, convert the task sub-region and ocean current vector field, mark obstacle grids as impassable areas, add directional weights to ocean current grids, and define the speed configuration rules as follows: unmanned platforms should prioritize sailing at the preset recommended speed; if explicitly specified, the path segment will inherit the platform type's preset speed by default. Step b2: Based on the raster map generated after rasterization, use the improved A* path planning algorithm to output the preliminary path planning for each unmanned platform; Step b3, Path optimization; Perform curve fitting on the initially generated path plan, take the tangent direction of the preceding and following points, and generate a new path; Perform cubic spline interpolation on the speed configuration to ensure continuous acceleration; Output: Preliminary path planning for each dynamic unmanned platform, including the location of waypoints and speed configuration between waypoints for each route.

6. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 5, characterized in that, The improved A* path planning algorithm in step b2 specifically includes: The multi-weighted cost function is: Wherein, the base cost G(n) is the actual distance traveled, calculated based on the difference between latitude and longitude and depth, taking into account the curvature of the earth; the heuristic cost function H(n) is the Manhattan distance from the unmanned platform to the target point, including the depth difference penalty; the dynamic correction term C(n) is the ocean current influence coefficient, which corrects the travel energy consumption based on the downstream / upstream flow. The design includes platform motion constraints and remaining range constraints. Platform motion constraints include the maximum allowable turning angle for unmanned surface vessels and low-speed platforms, while glider-type platforms are only allowed to follow paths with smooth depth changes. The remaining range constraint is the total path distance calculated in real time. If the path exceeds the platform's remaining range, dynamic pruning is triggered, and the path is backtracked to the nearest valid node. Run the improved A* path planning algorithm to output the preliminary path plan for each unmanned platform, including the location of waypoints and the speed configuration between waypoints for each route.

7. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 1, characterized in that, The method in step 4 specifically includes: Step 4.1, Data Preprocessing: The raw marine environmental data is preprocessed using methods including outlier detection, filtering, and missing value handling. Step 4.2, Layer Crossing Identification: The temperature and salinity data obtained from the unmanned platform are fused with the background field data using a weighted average method. The fused value = measured data + background field data * weight. The weight is determined by the closeness between the background field data and the measured data, and the degree of closeness is proportional to the weight. The background field data is then corrected using the measured data. Based on the fused background field, temperature profile data and salinity profile data are extracted for layer crossing diagnosis. Step 4.3, Contour Line Fitting and Construction: Based on the depth data measured by the unmanned platform, the contour line is constructed by using a thin plate spline function-based contour line generation method.

8. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 7, characterized in that, The method for handling missing values ​​in step 4.1 specifically includes: Handling small-scale continuous missing data: For cases where the continuous missing range in the original marine environmental data is less than a certain data segment, piecewise linear interpolation is used to fill in the missing data; by taking values ​​at both ends of the missing interval and calculating the intermediate missing value according to a linear relationship, it is suitable for cases of short-term data loss. Handling large-scale missing data: For cases where the continuous missing range is greater than a certain data segment, the missing range is marked. Generate missing data: In the missing interval, input context information to allow the generator to generate reasonable numerical placeholders to fill the missing parts and maintain the temporal continuity of the data.

9. The marine environmental information fusion method based on heterogeneous unmanned platform clusters according to claim 7, characterized in that, The specific process for the hierarchical diagnosis is as follows: The formula for diagnosing hierarchical changes is: Where Q is the temperature gradient value, representing the amount of temperature change per unit depth; ΔT is the temperature difference between two layers of seawater; ΔZ is the depth difference between two layers of seawater; when the temperature gradient Q of a certain water layer exceeds a preset critical value, the water layer is defined as a thermocline. The contour line generation method based on the thin plate spline function TPS is used to achieve contour line fitting and construction. The specific process is as follows: Data preparation: The input data consists of an irregularly distributed set of sounding points, which undergoes preprocessing including outlier removal, coordinate system unification, and data normalization. Constructing a TPS interpolation model: By constructing a smooth surface to fit all sounding points, isobaths are extracted from this surface. The objective of the thin-plate spline function is to minimize the bending energy of the surface while ensuring that the interpolated surface strictly passes through all sounding points. The mathematical expression is: Where X=(x,y) are planar coordinates. Let be the plane coordinates of the i-th sounding point. It is the radial basis function of the two-dimensional TPS, where r is the distance; The weighting coefficients are determined by solving a system of linear equations. , , These are the coefficients for the linear term, used to eliminate the effects of overall translation and rotation; Solution weight and linear term coefficients , , This makes the interpolated surface satisfy: Generate a regular mesh: A regular grid is generated within the study area to calculate the interpolated depth values; the grid extent is defined, the grid resolution is set, and the grid point coordinates are generated. ; Calculate the depth value of the grid points: For each grid point Substitute into the TPS function to calculate the depth: Obtain the depth dataset of the regular grid ; Extracting contour lines: The depth values ​​of the regular grid are binarized, the Marching Squares algorithm is applied to extract contour line segments, the segments are connected to form closed contour lines, and the contour lines are smoothed.

10. A marine environmental information fusion system based on a heterogeneous unmanned platform cluster, characterized in that, include: Memory, used to store executable computer programs; The processor, when executing an executable computer program stored in memory, implements the marine environmental information fusion method based on a heterogeneous unmanned platform cluster as described in any one of claims 1 to 9.

Images