An unmanned ship path segmentation and weighting dynamic adjustment method and system

Abstract

This invention provides a method and system for weighted dynamic adjustment of unmanned surface vessel (USV) path segments, relating to the field of intelligent control technology. The method includes: Step 1, acquiring the USV's navigation environment information and mission requirements; determining a segmentation strategy based on obstacle distribution density and water complexity in the navigation environment information; and dynamically dividing the global navigation area into multiple navigation segments of varying lengths according to the segmentation strategy; Step 2, for each segment, real-time acquisition of water flow velocity data sensed by three fixed-position hydrological monitoring buoys within the segment; and constructing a water flow velocity field grid data for the segment based on the spatial coordinates of the three hydrological monitoring buoys and the water flow velocity data. This invention improves the navigation safety, efficiency, and autonomous navigation capabilities of USVs in complex waters.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology, and in particular to a method and system for dynamic adjustment of unmanned vessel path segmentation weighting. Background Technology

[0002] With the widespread application of unmanned surface vessels (USVs) in various water operations such as marine monitoring, port logistics, and waterway inspection, path planning, as a core technology for autonomous navigation, determines the safety, efficiency, and adaptability of their navigation. Path planning methods combining segmented and weighted approaches have become mainstream solutions due to their ability to effectively reduce global planning complexity and flexibly adapt to multi-objective optimization needs. However, existing technologies, such as a segmented weighted path planning method for port logistics USVs based on an improved RRT (Real-Time Response) approach, have shortcomings. Their fixed segmentation strategy cannot adapt to the complex environmental characteristics of different areas within port waters, such as obstacle distribution density and water flow velocity differences. Furthermore, the preset fixed weight coefficients cannot be dynamically adjusted according to real-time environmental changes during navigation. Additionally, they fail to conduct refined spatial quantitative analysis of the impact of water flow, simply incorporating water flow velocity as a single numerical factor without constructing a water flow velocity field and corresponding quantitative correction indicators. This results in the planned paths exhibiting high navigation energy consumption and insufficient obstacle avoidance accuracy in areas with complex port water flow and dense obstacles, making it difficult to meet the autonomous navigation needs of USVs in complex and dynamic water environments. Summary of the Invention

[0003] This invention provides a method and system for dynamic adjustment of unmanned surface vessel (USV) path segmentation with weighting, which improves the navigation safety, efficiency, and autonomous navigation capability of USVs in complex waters.

[0004] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0005] Firstly, a method for dynamically adjusting the path segmentation weighted by unmanned surface vessels, the method comprising:

[0006] Step 1: Obtain the navigation environment information and mission requirements of the unmanned vessel. Based on the obstacle distribution density and water complexity in the navigation environment information, determine the segmentation strategy and dynamically divide the global navigation area into multiple navigation sub-segments of different lengths according to the segmentation strategy.

[0007] Step 2: For each navigation segment, collect the water flow velocity data sensed by the hydrological monitoring buoys at three fixed positions within the navigation segment in real time. Based on the spatial coordinates of the three hydrological monitoring buoys and the water flow velocity data, construct the water flow velocity field grid data of the navigation segment.

[0008] Step 3: Perform smooth fitting on the water flow velocity field grid data to obtain the continuous spatial distribution function of water flow velocity within the navigation segment; calculate the gradient of the continuous spatial distribution function, extract the gradient field and integrate it along the segment to obtain the water flow influence correction coefficient of the navigation segment.

[0009] Step 4: The water flow influence correction coefficient is introduced as an additional term into the pre-constructed multi-objective weighted cost function. Through the preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental change information of the navigation sub-segment. The water flow influence correction coefficient is then combined to generate the local optimization path corresponding to each sub-segment.

[0010] Step 5: The locally optimized paths of the navigation sub-segments are spliced ​​and merged to obtain the complete global navigation path of the unmanned vessel. Based on the environmental change data perceived by the unmanned vessel in real time during navigation, the segmentation strategy and weight coefficients are iteratively corrected and the process is returned to step 1 to be re-executed, so as to realize the online dynamic adjustment of the path.

[0011] Furthermore, the navigation environment information and mission requirements of the unmanned surface vessel are acquired. Based on the obstacle distribution density and water complexity in the navigation environment information, a segmentation strategy is determined, and the global navigation area is dynamically divided into multiple navigation sub-segments of different lengths according to the segmentation strategy, including:

[0012] The navigation environment information and mission requirements of the unmanned vessel are obtained. The obstacle distribution density in the navigation environment information is rasterized to obtain the obstacle density distribution matrix. At the same time, the water complexity in the navigation environment information is quantitatively analyzed to obtain the quantitative value of water complexity.

[0013] Based on the obstacle density distribution matrix and the quantified value of water complexity, the sliding window analysis method is used to scan the global navigation area, calculate the comprehensive environmental characteristic value within each window, and determine the segment boundary points based on the abrupt change points of the comprehensive environmental characteristic value.

[0014] Based on the segment boundary points, and combined with the navigation mileage and time constraints in the task requirements, the length of the region between adjacent segment boundary points is optimized and adjusted to generate multiple navigation sub-segments with different lengths.

[0015] Furthermore, for each of the divided navigation segments, real-time water flow velocity data is collected from hydrological monitoring buoys at three fixed locations within the segment. Based on the spatial coordinates of the three hydrological monitoring buoys and the water flow velocity data, a water flow velocity field grid data for the navigation segment is constructed, including:

[0016] For each generated navigation segment, the water flow velocity sensing data output by the hydrological monitoring buoys at three fixed positions within the navigation segment is acquired in real time, and the water flow velocity sensing data is filtered to obtain the effective water flow velocity value corresponding to each hydrological monitoring buoy.

[0017] Using the spatial coordinates of three hydrological monitoring buoys and their corresponding effective water flow velocity values ​​as known sample points, the inverse distance weighted interpolation algorithm is used to perform spatial interpolation calculations on the unsampled points within the navigation segment, thereby obtaining the estimated water flow velocity values ​​at each grid node within the navigation segment.

[0018] The grid resolution is set according to the boundary range of the navigation segment. The navigation segment is then divided into regular grids based on the grid resolution, and the estimated water flow velocity at each grid node is filled into the corresponding grid node position to form the water flow velocity field grid data of the navigation segment.

[0019] Furthermore, a smooth fit is performed on the water flow velocity field grid data to obtain a continuous spatial distribution function of the water flow velocity within the navigation segment; the gradient of the continuous spatial distribution function is calculated, the gradient field is extracted and integrated along the segment to obtain the water flow influence correction coefficient for the navigation segment, including:

[0020] For the constructed water flow velocity field grid data of the navigation segment, the bicubic spline interpolation algorithm is used to perform smooth fitting on the grid data to generate a mathematical expression for the continuous variation of water flow velocity with spatial position within the navigation segment, which serves as the continuous spatial distribution function of water flow velocity within the navigation segment.

[0021] The first-order partial derivatives of the continuous spatial distribution function with respect to the abscissa and ordinate are obtained respectively. The gradient field of the water flow velocity in the navigation segment is constructed based on the first-order partial derivatives, and the water flow velocity change vector at each spatial location is obtained.

[0022] The gradient field is path integrated along the centerline of the navigation segment, and the result of the path integration is used as the correction coefficient for the water flow influence of the navigation segment.

[0023] Furthermore, the water flow impact correction coefficient is introduced as an additional term into the pre-constructed multi-objective weighted cost function. Through a preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental changes of the navigation sub-segments. Combined with the water flow impact correction coefficient, the local optimization paths corresponding to each sub-segment are generated, including:

[0024] The constructed water flow velocity field grid data is obtained as real-time environmental change information, and the obtained water flow influence correction coefficient is used as a penalty term to be weighted into the pre-constructed multi-objective weighted cost function. The multi-objective weighted cost function includes navigation energy consumption, path length, obstacle avoidance safety distance and path smoothness index.

[0025] Based on the water flow velocity vector and obstacle distribution characteristics in the real-time environmental change information, and combined with the preset initial weight coefficients, the dynamically adjusted weight coefficients corresponding to each indicator are calculated through a dynamic weight adjustment mechanism.

[0026] Based on a multi-objective weighted cost function with dynamically adjusted weighting coefficients and a correction coefficient for the influence of water flow, a path search is performed within the navigation segment with the objective of minimizing the weighted cost function value, generating a locally optimized path for that navigation segment.

[0027] Furthermore, the constructed water flow velocity field grid data is acquired as real-time environmental change information, and the obtained water flow influence correction coefficient is used as a penalty term and weighted into a pre-constructed multi-objective weighted cost function. The multi-objective weighted cost function includes navigation energy consumption, path length, obstacle avoidance safety distance, and path smoothness indicators, including:

[0028] The real-time water flow velocity vector field of the current navigation segment is extracted from the constructed water flow velocity field grid data, and the integral calculation result of the navigation segment is read from the calculated water flow influence correction coefficient. The spatiotemporal consistency of the real-time water flow velocity vector field and the water flow influence correction coefficient is verified.

[0029] Based on the water flow influence correction coefficient after the spatiotemporal consistency verification is passed, and combined with the preset penalty term weighting factor, an additional cost term containing the water flow influence correction coefficient is constructed, and the additional cost term is linearly weighted into the pre-constructed basic multi-objective weighted cost function. The basic multi-objective weighted cost function is composed of a linear weighted cost term of navigation energy consumption, a cost term of path length, a cost term of obstacle avoidance safety distance, and a cost term of path smoothness.

[0030] The navigation energy consumption cost term in the basic multi-objective weighted cost function is dynamically corrected based on the real-time water flow velocity vector field, thereby generating an updated multi-objective weighted cost function that takes into account the influence of real-time water flow.

[0031] Furthermore, the locally optimized paths of the navigation segments are spliced ​​and fused to obtain the complete global navigation path of the unmanned surface vessel (USV). Based on the environmental change data perceived in real time during the USV's navigation, the segmentation strategy and weight coefficients are iteratively corrected and the process is repeated in step 1 to achieve online dynamic adjustment of the path, including:

[0032] The local optimized path corresponding to each generated navigation segment is obtained. The local optimized paths of adjacent navigation segments are spliced ​​at the segment boundary points, and the path curvature at the splicing point is smoothed to generate a complete global navigation path for the unmanned vessel.

[0033] During the unmanned vessel's navigation along the global navigation path, environmental change data sensed by the shipborne sensors are collected in real time. The environmental change data is compared with the constructed water flow velocity field grid data, and the deviation value between the current navigation environment and the constructed grid data is calculated.

[0034] The segmentation strategy and weight coefficients are iteratively corrected online based on the deviation value, and the corrected segmentation strategy and weight coefficients are used as input parameters to return to step 1 for re-execution, so as to realize the dynamic update and online adjustment of the global navigation path.

[0035] Secondly, the unmanned surface vessel path segmentation weighted dynamic adjustment system includes:

[0036] The acquisition module is used to acquire navigation environment information and mission requirements of the unmanned vessel. Based on the obstacle distribution density and water complexity in the navigation environment information, it determines the segmentation strategy and dynamically divides the global navigation area into multiple navigation sub-segments of different lengths according to the segmentation strategy.

[0037] The segmentation module is used to collect water flow velocity data sensed by hydrological monitoring buoys at three fixed positions within each segment of navigation in real time. Based on the spatial coordinates and water flow velocity data of the three hydrological monitoring buoys, a water flow velocity field grid data of the navigation segment is constructed.

[0038] The calculation module is used to smoothly fit the water flow velocity field grid data to obtain the continuous spatial distribution function of water flow velocity within the navigation segment; the gradient of the continuous spatial distribution function is calculated, the gradient field is extracted and integrated along the segment to obtain the water flow influence correction coefficient of the navigation segment;

[0039] The optimization module is used to introduce the water flow influence correction coefficient as an additional term into the pre-built multi-objective weighted cost function. Through the preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental change information of the navigation sub-segment, and the local optimization path corresponding to each sub-segment is generated together with the water flow influence correction coefficient.

[0040] The fusion module is used to splice and merge the locally optimized paths of the navigation sub-segments to obtain the complete global navigation path of the unmanned vessel. Based on the environmental change data perceived by the unmanned vessel in real time during navigation, the segmentation strategy and weight coefficients are iteratively corrected and the process is returned to step 1 for re-execution to achieve online dynamic adjustment of the path.

[0041] Thirdly, a computing device, comprising:

[0042] One or more processors;

[0043] A storage device for storing one or more programs that, when executed by one or more processors, cause the one or more processors to implement the method.

[0044] Fourthly, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0045] The above-described solution of the present invention has at least the following beneficial effects:

[0046] This approach utilizes a dynamic segmentation strategy that combines obstacle distribution density and water complexity with mission requirements. It employs a closed-loop technique that uses hydrological monitoring buoys to construct a water flow velocity field grid, obtains water flow impact correction coefficients through fitting and integration, and introduces these coefficients as an additional term into a multi-objective weighted cost function. It also dynamically adjusts the weight coefficients of the cost function based on real-time environmental changes and iteratively corrects the segmentation strategy and weight coefficients based on real-time environmental data during navigation. This overcomes the technical problems of existing unmanned surface vessel (USV) segmented weighted path planning technologies, such as fixed segmentation strategies failing to adapt to complex water environment characteristics, fixed weight coefficients failing to respond to real-time environmental changes, lack of refined spatial and quantitative analysis of water flow impact, and the absence of an online iterative correction mechanism. These problems lead to high navigation energy consumption and insufficient obstacle avoidance accuracy. Ultimately, this approach achieves environmental adaptation and precise quantification of hydrological impact in USV path planning, effectively improving navigation safety, efficiency, and autonomous navigation capabilities in complex dynamic waters, and meeting the operational needs of various scenarios such as marine monitoring, port logistics, and waterway inspection. Attached Figure Description

[0047] Figure 1 This is a flowchart illustrating a method for dynamic adjustment of unmanned vessel path segmentation weighting provided by an embodiment of the present invention.

[0048] Figure 2 This is a schematic diagram of an unmanned vessel path segmentation weighted dynamic adjustment system provided by an embodiment of the present invention.

[0049] Figure 3 This is a statistical diagram showing the distribution of obstacles in different aquatic environments.

[0050] Figure 4 This is a schematic diagram illustrating the dynamic adjustment trend of path segment weights.

[0051] Figure 5 This is a statistical diagram comparing the performance of different path planning algorithms. Detailed Implementation

[0052] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0053] like Figure 1 As shown, an embodiment of the present invention proposes a method for dynamic adjustment of unmanned surface vessel path segmentation with weights, the method comprising the following steps:

[0054] Step 1: Obtain the navigation environment information and mission requirements of the unmanned vessel. Based on the obstacle distribution density and water complexity in the navigation environment information, determine the segmentation strategy and dynamically divide the global navigation area into multiple navigation sub-segments of different lengths according to the segmentation strategy.

[0055] Step 2: For each navigation segment, collect the water flow velocity data sensed by the hydrological monitoring buoys at three fixed positions within the navigation segment in real time. Based on the spatial coordinates of the three hydrological monitoring buoys and the water flow velocity data, construct the water flow velocity field grid data of the navigation segment.

[0056] Step 3: Perform smooth fitting on the water flow velocity field grid data to obtain the continuous spatial distribution function of water flow velocity within the navigation segment; calculate the gradient of the continuous spatial distribution function, extract the gradient field and integrate it along the segment to obtain the water flow influence correction coefficient of the navigation segment.

[0057] Step 4: The water flow influence correction coefficient is introduced as an additional term into the pre-constructed multi-objective weighted cost function. Through the preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental change information of the navigation sub-segment. The water flow influence correction coefficient is then combined to generate the local optimization path corresponding to each sub-segment.

[0058] Step 5: The locally optimized paths of the navigation sub-segments are spliced ​​and merged to obtain the complete global navigation path of the unmanned vessel. Based on the environmental change data perceived by the unmanned vessel in real time during navigation, the segmentation strategy and weight coefficients are iteratively corrected and the process is returned to step 1 to be re-executed, so as to realize the online dynamic adjustment of the path.

[0059] In this embodiment of the invention, the following technical means are employed: dynamically dividing navigation segments of different lengths based on obstacle distribution density and water complexity; constructing a water flow velocity field grid using three hydrological monitoring buoys and obtaining a water flow influence correction coefficient through fitting and integration; introducing this coefficient as an additional term into a multi-objective weighted cost function and dynamically adjusting the weight coefficient in conjunction with the real-time environment; and iteratively correcting the segmentation strategy and weight coefficient based on real-time navigation environment data after splicing and fusion of locally optimized paths, and executing the process cyclically. This overcomes the technical problems of existing unmanned surface vessel (USV) segmented weighted path planning technologies, such as the inability of fixed segmentation strategies to adapt to complex water environments, the lack of refined spatial and quantitative analysis of water flow influence, the inability of weight coefficients to dynamically adjust with the environment, and the absence of an online dynamic path correction mechanism. This achieves environmental adaptability and accurate quantification of hydrological influence in USV path planning, enabling the planned path to be dynamically adjusted in real-time with the water environment, thereby improving the navigation safety, efficiency, and autonomous navigation capabilities of USVs in complex and dynamic waters.

[0060] In a preferred embodiment of the present invention, step 1 above may include:

[0061] Step 1.1: Acquire the navigation environment information and mission requirements of the unmanned surface vessel (USV). The obstacle distribution density in the navigation environment information is rasterized to obtain an obstacle density distribution matrix. Simultaneously, the water area complexity in the navigation environment information is quantitatively analyzed to obtain quantifiable values. Specifically, this includes: collaboratively acquiring the USV's global navigation environment information and operational mission requirements through the USV's onboard environmental perception sensors and the water area basic information database. The navigation environment information includes environmental data related to path planning, such as obstacle distribution locations, obstacle sizes, water area topography, and channel width variation characteristics within the global navigation area. The operational mission requirements include core mission indicators such as the USV's total navigation mileage constraints, mission completion time constraints, and navigation mission type. After information acquisition, the obstacle distribution density in the navigation environment information is rasterized. First, the continuous global navigation area is regularly divided according to a preset grid size, dividing the global area into multiple independent grid units of the same spatial size, and spatially numbering all grid units. The number of obstacles in each grid unit is counted, and the obstacle distribution density value corresponding to each grid unit is calculated. The obstacle distribution density values ​​of all grid units are arranged sequentially according to the spatial row and column order of the grid units to form a two-dimensional obstacle density distribution matrix, completing the digital representation of obstacle distribution density. A quantitative analysis of the water area complexity in the navigation environment information is then conducted. First, the core quantitative indicators affecting water area complexity are determined, including the average obstacle distribution density of each local area within the global navigation area, the degree of undulation of the water area topography, and the rate of change of channel width. Then, preset weighting coefficients are assigned to each core quantitative indicator, and the formula is used to... The water complexity quantification values ​​of each local area within the global navigation area are calculated, where For the quantification of complex water area, This represents the average obstacle distribution density in a local area. The degree of undulation in the water topography. The rate of change of the channel width, , , These are the preset weight coefficients corresponding to the three core quantitative indicators, and they satisfy... + + =1, completing the quantitative analysis and numerical expression of water area complexity.

[0062] Step 1.2: Based on the obstacle density distribution matrix and the quantified value of water complexity, a sliding window analysis method is used to scan the global navigation area. The comprehensive environmental characteristic value within each window is calculated, and segment boundary points are determined based on the abrupt change points of the comprehensive environmental characteristic value. Specifically, this includes: setting an appropriate sliding window size based on the overall spatial range of the global navigation area and the spatial variation scale of environmental characteristics; simultaneously setting the sliding window's movement step size, where the step size is smaller than the side length of the sliding window to ensure the continuity of the sliding window scan; using this sliding window to continuously slide and scan the rasterized global navigation area row by row and column by column; at each stop of the sliding window, extracting the obstacle distribution density value and the corresponding quantified value of water complexity for all raster cells within the window's coverage area; first, calculating the arithmetic mean of the obstacle distribution density values ​​for all raster cells within the window's coverage area using the arithmetic mean formula. Then, the arithmetic mean of the complex quantitative values ​​of all local water areas within the window coverage area is calculated using the arithmetic mean formula. Subsequently, preset environmental feature weighting coefficients are assigned to the two average values, using the formula... The comprehensive environmental feature value of the window location is calculated, where For comprehensive environmental characteristic values, This represents the average obstacle distribution density within the window. This represents the average of the quantified values ​​of water complexity within the window. , These are the preset environmental feature weight coefficients corresponding to the two average values, and they satisfy... + =1, after completing the full-range scan of the global navigation area within the sliding window, a continuous sequence of comprehensive environmental feature values ​​arranged along the scanning path of the sliding window is obtained; difference calculation is performed on this sequence, using the formula The change in the comprehensive environmental feature value at each scan location was calculated, where This represents the change in comprehensive environmental characteristic values. For the first The comprehensive environmental feature value of each scan location, For the first The comprehensive environmental characteristic value of each scan location; the change in the comprehensive environmental characteristic value. Scanning locations exceeding a preset threshold are identified as abrupt changes in the comprehensive environmental characteristic value. All abrupt changes are the segment boundary points of the global navigation area, thus completing the spatial location determination of the segment boundary points.

[0063] Step 1.3: Based on the segment boundary points and considering the navigation mileage and time constraints in the task requirements, optimize the length of the region between adjacent segment boundary points to generate multiple navigation segments of different lengths. Specifically, this includes: extracting the spatial geographic coordinates of all determined segment boundary points; calculating the straight-line distance between two adjacent segment boundary points using the spatial distance calculation formula based on the spatial geographic coordinates; determining the initial navigation segment length between adjacent boundary points using this straight-line distance; retrieving the navigation mileage and time constraint indicators from the task requirements; and calculating the planned navigation speed of the unmanned vessel in each local region based on the preset basic navigation speed of the unmanned vessel. First, calculate the average navigation speed of the unmanned vessel; then, correct the average navigation speed based on the quantified value of water complexity in each local region. The higher the quantified value of water complexity, the higher the planned navigation speed for that region. The lower the degree, the higher the planned navigation speed for each local area after correction. Based on the planned navigation speed of each local area, combined with the initial navigation segment length between adjacent segment boundary points, the planned navigation time for each initial segment is calculated. The planned navigation time of each initial segment is compared with a preset regional navigation time threshold, and the initial navigation segment length is compared with the mileage allocation requirements for each local area in the task requirements. If the planned navigation time of an initial segment exceeds the regional navigation time threshold, or the initial segment length does not match the local area mileage allocation requirements, the length of the area between adjacent segment boundary points is optimized and adjusted. During the adjustment, the final navigation segment length of the area is first calculated. If there is a deviation between the final navigation segment length and the initial navigation segment length, the spatial geographic coordinates of the segment boundary points of the area are fine-tuned to achieve the adaptation adjustment of the segment length. In the above manner, the length optimization adjustment of the area between all adjacent segment boundary points in the global navigation area is completed sequentially, and finally multiple navigation segments with different lengths are generated. The length of each navigation segment is adapted to the environmental characteristics of the corresponding area and the navigation mileage and time constraints of the task requirements.

[0064] In this embodiment of the invention, by rasterizing the obstacle distribution density, quantitatively analyzing the complexity of the water area, scanning the global navigation area using the sliding window analysis method, determining the segment boundary points based on the mutation points of comprehensive environmental feature values, and finally optimizing and adjusting the area length to generate navigation sub-segments of different lengths in combination with the navigation mileage and time constraints of the task, the technical means overcome the technical problems of existing unmanned vessel path segmentation strategies being unable to adapt to the differences in obstacle distribution and water complexity in different areas of the water, and the segmentation not being combined with the operational task requirements, resulting in poor adaptability of the segmentation to the actual environment and task. Thus, it achieves dual adaptive dynamic segmentation of the global navigation area based on environmental characteristics and task requirements, making the division of navigation sub-segments more in line with the actual environmental characteristics and operational task requirements of complex water areas.

[0065] In a preferred embodiment of the present invention, step 2 above may include:

[0066] Step 2.1: For each generated navigation segment, acquire the water flow velocity sensing data output by hydrological monitoring buoys at three fixed locations within the segment in real time, and filter the water flow velocity sensing data to obtain the effective water flow velocity value corresponding to each hydrological monitoring buoy. Specifically, for each generated navigation segment, establish a data transmission link through the wireless communication module between the hydrological monitoring buoy and the unmanned vessel, and acquire the raw water flow velocity sensing data collected and output by the hydrological monitoring buoys at three fixed spatial locations within the segment in real time. The raw data includes the buoy sampling timestamp, the magnitude and direction information of the water flow velocity, and the buoy's own equipment status data. To remove random noise, abnormal equipment fluctuations, etc., from the raw data... Effective data is obtained by filtering the acquired water flow velocity sensing data. A moving average filtering method is used to purify the data. The time length of the sliding window is set, and multiple sets of continuously collected raw water flow velocity data are included in the sliding window in chronological order. Then, the arithmetic mean of all raw water flow velocity data within the window is calculated to obtain the filtered water flow velocity value corresponding to the window. The sliding window is continuously slid and the above calculation process is repeated to finally obtain the effective water flow velocity value of each hydrological monitoring buoy in the continuous time dimension. This effective value retains the real-time change characteristics of water flow velocity and eliminates invalid interference data, thus completing the extraction of the effective water flow velocity value of a single buoy. All three buoys are processed in this way to obtain their respective effective water flow velocity values.

[0067] Step 2.2: Using the spatial coordinates and corresponding effective water velocity values ​​of the three hydrological monitoring buoys as known sample points, the inverse distance weighted interpolation algorithm is used to perform spatial interpolation calculations on the unsampled points within the navigation segment to obtain the estimated water velocity values ​​at each grid node within the navigation segment. Specifically, this includes: identifying all unsampled points within the navigation segment that require water velocity estimation; these unsampled points cover the entire spatial range of the navigation segment and are the basic nodes for subsequently constructing the water velocity field grid; using the spatial coordinates of the three hydrological monitoring buoys, including the horizontal and vertical coordinates and the corresponding effective water velocity values, as known sample points, and using the inverse distance weighted interpolation algorithm to estimate the water velocity at each unsampled point; calculating the spatial straight-line distance from the current unsampled point to the three known sample points, using the following formula: ,in For unsampled points to the th sample points ( The spatial straight-line distance between (=1, 2, 3), , ) represents the spatial coordinates of the unsampled points. , ) is the first The spatial coordinates of each hydrological monitoring buoy are used; an interpolation weight is assigned to each distance value, and the basic weight of each sample point is calculated first. That is, the reciprocal of the distance value, and then the normalized weight coefficient corresponding to each sample point is calculated using the formula: The normalized weight of a sample point is obtained by dividing its basic weight by the sum of the basic weights of the three sample points. For the first The basic weight of each hydrological monitoring buoy is calculated as the reciprocal of the straight-line distance from the unsampled point to the buoy. As the base weight for the first hydrological monitoring buoy, As the base weight for the second hydrological monitoring buoy, The base weight for the third hydrological monitoring buoy; Calculate the estimated flow velocity at unsampled points using the following formula: in These are estimated values ​​for the water flow velocity at unsampled points. , , These are the effective water flow velocity values ​​corresponding to the three sample points. , , The normalized weighting coefficients for the corresponding sample points are used. Following this calculation process, the water flow velocity estimation for all unsampled points within the navigation segment is completed sequentially, resulting in the water flow velocity estimation value for each unsampled point.

[0068] Step 2.3: Set the grid resolution according to the boundary range of the navigation sub-segment, perform regular grid subdivision on the navigation sub-segment based on the grid resolution, and fill the estimated water flow velocity at each grid node position to form the water flow velocity field grid data of the navigation sub-segment. Specifically, this includes: setting an appropriate grid resolution according to the actual spatial boundary range of the navigation sub-segment, such as the horizontal coordinate range of 0-1000 meters, the vertical coordinate range of 0-800 meters, and the required level of detail of environmental features. The grid resolution value is the side length of a single grid cell. The smaller the resolution value, the finer the grid cell, and the more accurately it can reflect the spatial distribution characteristics of water flow velocity. The principle is to balance computational efficiency and spatial resolution: a high resolution of 0.5 meters is used in sub-segments with dense obstacles and drastic water flow changes, such as port and channel intersections, while a moderate resolution of 2 meters is used in sub-segments with relatively calm environments, such as open water. Key geometric feature points within the navigation sub-segments are extracted, including the contour vertices of obstacles, the inflection points of sub-segment boundaries, and the deployment positions of hydrological monitoring buoys. These feature points are used as constraint points for Delaunay triangulation to ensure that the mesh structure fits the actual geometric boundaries of the water area during subsequent triangulation, avoiding boundary distortion problems caused by regular mesh triangulation.

[0069] Using the top-left corner of a navigation segment as the origin (e.g., 0 meters x, 800 meters y), the area is divided along the horizontal x and vertical x coordinates according to the set grid resolution side lengths. If the resolution of a segment is set to 1 meter, then starting from the origin, a node is marked for every 1 meter increase in the horizontal coordinate and every 1 meter decrease in the vertical coordinate. Following this rule, the continuous navigation segment space is divided into multiple rectangular grid units. The four vertices of each grid unit are the basic grid nodes. All basic grid nodes are assigned a unique spatial position number, completing the initial regular grid subdivision. For the initially subdivided regular grid, the Delaunay triangulation geometry algorithm is used for optimization, combining the previously extracted constraint points to integrate all basic grid nodes and constraint points into a subdivision point set, ensuring that there are no nodes with duplicate coordinates in the point set.

[0070] The core Delaunay triangulation rule is executed: the circumcircle of any triangle does not contain other nodes in the point set, and the nodes in the point set are connected sequentially to form a continuous triangular mesh covering the entire navigation segment space; the overlapping areas of the triangular mesh and the original regular mesh are merged: the regularity of the regular mesh in the smooth environment is preserved, while the triangular mesh is used to adapt to irregular geometric areas such as obstacle outlines and segment boundaries, ensuring that the mesh cells conform to the actual geometric features of the water area while taking into account the regularity of the calculation; all mesh nodes after merging are reassigned with unique spatial position numbers, completing the final mesh division of the navigation segment.

[0071] The calculated estimated water flow velocity values ​​for each grid node are then filled into the corresponding grid node positions one by one according to the spatial location number of the grid nodes. For example, for nodes numbered 10-20, with an x-coordinate of 10 meters and a y-coordinate of 20 meters, the estimated water flow velocity is 1.2 m / s and the flow direction is due east. This value and direction are then accurately filled into the node position. For a few grid nodes that are not covered due to boundary clipping, constraint point interleaving, etc., a simple arithmetic average is no longer used for supplementation. Instead, the Delaunay triangulation geometric algorithm is used to determine its neighboring nodes and calculate the weighted supplementation value. The specific process is as follows: Locate the Delaunay triangular unit to which the missing node belongs: find the unique triangular unit containing the missing node and determine the three vertices of the triangular unit; calculate the geometric weights from the missing node to the three neighboring nodes: calculate the Euclidean distance between the missing node and the three neighboring nodes respectively. The closer the distance, the greater the weight.

[0072] Example: If the distances from the missing node P to its neighbors A, B, and C are 2 meters, 1 meter, and 3 meters respectively, then:

[0073] The weight of node A = (1 / 2) ÷ [(1 / 2) + (1 / 1) + (1 / 3)] = 0.5 ÷ (0.5 + 1 + 0.333) ≈ 0.273;

[0074] The weight of node B = (1 / 1) ÷ 1.833 ≈ 0.545;

[0075] The weight of node C = (1 / 3) ÷ 1.833 ≈ 0.182;

[0076] Calculate the missing node's water flow velocity supplement value: Multiply the estimated water flow velocity values ​​of the three neighboring nodes by their corresponding geometric weights, and then add the products together to obtain the supplement value for the missing node;

[0077] Example: If the speed of node A is 1.0 m / s, the speed of node B is 1.5 m / s, and the speed of node C is 0.8 m / s, then the supplementary value = 1.0 × 0.273 + 1.5 × 0.545 + 0.8 × 0.182 ≈ 0.273 + 0.818 + 0.146 = 1.237 m / s;

[0078] The calculated supplementary values ​​are filled into the missing node positions to ensure that all grid nodes have corresponding water velocity values. After all grid nodes have been filled with water velocity values, they are integrated to form a two-dimensional data matrix with the grid nodes as the carriers. The rows of the matrix correspond to the horizontal coordinate numbers of the grid nodes, and the columns correspond to the vertical coordinate numbers. Each matrix element contains the magnitude and direction information of the water velocity of the corresponding grid node. This two-dimensional data matrix is ​​the water velocity field grid data for the navigation sub-segment.

[0079] In this embodiment of the invention, effective values ​​are obtained by filtering the water flow velocity sensing data of three fixed hydrological monitoring buoys. Then, using their spatial coordinates and effective velocity values ​​as sample points, the estimated water flow velocity values ​​of unsampled points are calculated through an inverse distance weighted interpolation algorithm. Finally, the grid resolution is set according to the boundary range of the navigation segment, and regular grids are formed by filling in the estimated values ​​to create a water flow velocity field grid data. Therefore, this invention overcomes the technical problems in the prior art, such as the lack of effective preprocessing of water flow velocity data, the use of water flow velocity as a single numerical value, and the failure to perform spatial grid modeling of water flow velocity within the navigation segment, which makes it impossible to accurately reflect the spatial distribution characteristics of water flow velocity. This invention achieves effective purification and spatialized, gridded, and quantitative expression of water flow velocity data within the navigation segment, accurately restoring the spatial distribution state of water flow velocity within the segment.

[0080] In a preferred embodiment of the present invention, step 3 above may include:

[0081] Step 3.1: For the constructed water flow velocity field grid data of the navigation segment, a bicubic spline interpolation algorithm is used to smooth the grid data, generating a mathematical expression for the continuous variation of water flow velocity with spatial location within the navigation segment. This expression serves as the continuous spatial distribution function of water flow velocity within the navigation segment. Specifically, this includes: preprocessing the constructed water flow velocity field grid data of the navigation segment to remove outliers caused by interpolation errors, ensuring that the water flow velocity values ​​of all grid nodes conform to the water flow characteristics of the navigation segment; and using a bicubic spline interpolation algorithm to smooth the preprocessed grid data. The core process is: determining the basis function of the bicubic spline interpolation. Two-dimensional bicubic spline basis functions are obtained by extending one-dimensional basis functions. Using the x-coordinate and y-coordinate of the navigation segment as independent variables and the water flow velocity of the grid nodes as the dependent variable, a system of bicubic spline interpolation equations is constructed. The constraint condition of the equation system is that the value of the fitting function at each grid node is completely consistent with the estimated value of the water flow velocity at that node; solving this interpolation equation system yields a mathematical expression with the x-axis and y-axis as inputs and the water flow velocity as the output. This expression is a continuous spatial distribution function that shows the continuous variation of water flow velocity with spatial position within a navigation segment. The function form satisfies that for any spatial position x and y within the navigation segment, a unique and continuous water flow velocity value can be obtained by substituting it into the function. This realizes the transformation from discrete grid data to continuous spatial distribution, accurately reflecting the spatial variation trend of water flow velocity within the navigation segment.

[0082] Step 3.2: Calculate the first-order partial derivatives of the continuous spatial distribution function with respect to the abscissa and ordinate respectively. Construct the gradient field of the water flow velocity within this navigation segment based on the first-order partial derivatives to obtain the water flow velocity change vector at each spatial location. Specifically, this includes: based on the generated continuous spatial distribution function of the water flow velocity... Find the first-order partial derivatives of the function with respect to the x-coordinate and the y-coordinate, respectively. The first-order partial derivative with respect to the x-coordinate is calculated by treating the y-coordinate as a constant and applying the differentiation rules for functions of one variable. China regarding Taking the derivative of the terms, we obtain the partial derivatives. This value represents the rate of change of water flow velocity along the horizontal axis; the first partial derivative with respect to the vertical axis y is calculated by treating the horizontal axis x as a constant and applying the differentiation rule for a single-variable function to the function. Taking the derivative of the term with respect to y, we obtain the partial derivative. This value characterizes the rate of change of water velocity along the vertical axis. A water velocity gradient field is constructed within the navigation segment using these two first-order partial derivatives as its core. The core component of the gradient field is the water velocity change vector corresponding to each spatial location within the navigation segment, which is determined by the partial derivative along the horizontal axis. Partial derivatives in the ordinate direction Composition, i.e., any spatial location The vector representing the change in water flow velocity is: By traversing all spatial locations within the navigation segment, calculating the first-order partial derivative of each location and generating the corresponding change vector, a water flow velocity gradient field covering the entire navigation segment is finally formed. This gradient field can accurately reflect the intensity and direction of the water flow velocity changes in the horizontal and vertical directions at each spatial location.

[0083] Step 3.3: Perform path integration on the gradient field along the centerline of the navigation sub-segment. Use the result of the path integration as the correction coefficient for the water flow influence of the navigation sub-segment. Specifically, this includes: determining the centerline of the navigation sub-segment, which is a pre-planned reference path traversing the navigation sub-segment. This path can completely cover the core navigation area of ​​the navigation sub-segment. Discretize the centerline into n consecutive micro-path units. The length of each path unit is small enough to ensure that the water flow velocity gradient field characteristics within the unit remain stable. Then, perform path integration on the water flow velocity gradient field along the discretized centerline path units. The trapezoidal integration method is used to complete the numerical calculation: For each micro-path unit, obtain the magnitude of the water flow velocity change vector at its starting and ending points. The magnitude calculation formula is: in The magnitude of the water flow velocity change vector at a single spatial location is obtained by substituting the partial derivative value at the starting point. Substituting the partial derivative value at the endpoint yields the modulus of the endpoint. Calculate the integral value of a single path cell, assuming the length of a single micropath cell is... Then the integral value of a single unit The calculation formula is: The integral values ​​of all the small path elements are summed sequentially to obtain the total path integral along the centerline. Let the water flow influence correction coefficient be... The formula for calculating the total integral is: in Let be the integral value of the i-th infinitesimal path unit. The total number of path units after discretization is K. The sum of the path integrals is the water flow influence correction coefficient of the navigation segment. This coefficient can quantitatively characterize the comprehensive influence of the spatial variation of water flow velocity in the entire navigation segment on the navigation of the unmanned vessel. The larger the coefficient value, the more drastic the water flow change in the segment and the more significant the impact on the navigation of the unmanned vessel.

[0084] In this embodiment of the invention, a continuous spatial distribution function of water flow velocity is obtained by smoothly fitting the grid data of the water flow velocity field using a bicubic spline interpolation algorithm. The first-order partial derivatives of the horizontal and vertical coordinates of this function are then used to construct the water flow velocity gradient field and obtain the velocity change vector at each spatial location. Finally, the water flow influence correction coefficient is obtained by path integration along the centerline of the navigation segment. This overcomes the technical problems of existing technologies that only consider water flow velocity as a discrete grid numerical value, which cannot reflect its continuous spatial variation law, and do not accurately quantify the actual impact of water flow on the navigation of unmanned vessels, lacking a unified quantitative index of water flow influence that can be directly integrated into path planning. Thus, it achieves the accurate transformation of water flow velocity from discrete grid data to continuous spatial distribution, accurately captures the spatial variation characteristics of water flow velocity within the segment, and obtains a correction coefficient that can quantitatively characterize the comprehensive impact of water flow on navigation through the path integration of the gradient field.

[0085] In a preferred embodiment of the present invention, step 4 above may include:

[0086] Step 4.1: Obtain the constructed water flow velocity field grid data as real-time environmental change information, and obtain the water flow influence correction coefficient. Introduce the water flow influence correction coefficient as a penalty term into the pre-constructed multi-objective weighted cost function. The multi-objective weighted cost function includes navigation energy consumption, path length, obstacle avoidance safety distance, and path smoothness indicators. Specifically, it includes: retrieving the constructed navigation segment water flow velocity field grid data as the real-time environmental change information for the current navigation segment; simultaneously retrieving the calculated water flow influence correction coefficient for the navigation segment, first verifying the data validity of the water flow velocity field grid data and the water flow influence correction coefficient, confirming the spatiotemporal consistency of the data, i.e., the sampling time of the water flow velocity field data matches the calculation time of the water flow influence correction coefficient, and the spatial range completely overlaps with the current navigation segment, discarding invalid or mismatched data; retrieving the pre-constructed basic multi-objective weighted cost function, which includes cost terms corresponding to four core optimization indicators: navigation energy consumption cost term, path length cost term, obstacle avoidance safety distance cost term, and path smoothness cost term, clarifying the calculation method of each cost term:

[0087] Navigation energy consumption cost item Path length cost term The candidate path length is Let the basic energy consumption per unit mileage of the unmanned vessel be valued as follows: Let the straight-line distance between the start and end points of the navigation segment be... ;

[0088] Obstacle avoidance safety distance cost term: Let the shortest distance from a node on the candidate path to surrounding obstacles be... The preset safety threshold is Then the obstacle avoidance cost of this node Iterate through all nodes of the candidate path and take the average obstacle avoidance cost of all nodes as the final obstacle avoidance safe distance cost. where n is the total number of candidate path nodes;

[0089] Path smoothness cost term: Let the turning angle of adjacent path segments on the candidate path be... Then the path smoothness cost term .

[0090] The basic multi-objective weighted cost function is calculated as follows: Let the preset initial weight coefficients for navigation energy consumption, path length, obstacle avoidance safety distance, and path smoothness be respectively... , , , satisfy + + + =1, then the basic multi-objective weighted cost function is: .

[0091] The water flow impact correction coefficient is introduced as a penalty term into the basic multi-objective weighted cost function: Let the weight factor of the penalty term for the water flow impact correction coefficient be... Based on the preset anti-interference capability of unmanned vessels, the weaker the anti-interference capability, The larger the value, the higher the penalty item value. The multi-objective weighted cost function after introducing the penalty term is: .

[0092] Step 4.2: Based on the water flow velocity vector and obstacle distribution characteristics in the real-time environmental change information, and combined with the preset initial weight coefficients, calculate the dynamically adjusted weight coefficients corresponding to each index through a dynamic weight adjustment mechanism. Specifically, this includes: extracting the water flow velocity vector of each grid node from the water flow velocity field grid data, including the velocity magnitude and direction; calculating the mean and variance of the water flow velocity vector within the navigation sub-segment; the mean reflects the overall intensity of the water flow within the sub-segment, and the variance reflects the degree of spatial variation of the water flow velocity; extracting the obstacle distribution characteristics of the current navigation sub-segment from the generated obstacle density distribution matrix; calculating the mean of the obstacle distribution density within the sub-segment; the mean reflects the density of obstacles within the sub-segment.

[0093] The preset initial weight coefficients for each optimization index are retrieved: initial weights for navigation energy consumption, path length, obstacle avoidance safety distance, and path smoothness. The sum of these initial weight coefficients is 1. A dynamic weight adjustment mechanism is constructed based on the extracted water flow and obstacle features to adjust the weights of each index across different dimensions.

[0094] Dynamic weighting of navigation energy consumption: The preset energy consumption adjustment coefficient is set as follows. Adjustment factor , If a base speed is preset for the unmanned surface vessel, then the dynamic weight of its navigation energy consumption will be determined. ;

[0095] Obstacle avoidance safety distance dynamic weight: Let the preset obstacle avoidance adjustment coefficient be... Adjustment factor , The obstacle avoidance safety distance is dynamically weighted based on a preset obstacle density benchmark value. ;

[0096] Path length dynamic weight: adjustment factor The path length dynamic weight ;

[0097] Path smoothness dynamic weights: To ensure the sum of the weights is 1, the path smoothness dynamic weights... .

[0098] Finally, the weight coefficients of each indicator after dynamic adjustment are obtained. , , , ,satisfy + + + =1, and adapts to changes in response to water flow and obstacle characteristics.

[0099] After completing the above calculations, the dynamically adjusted weight coefficients corresponding to each optimization index are obtained. These coefficients can adapt to the real-time characteristics of the water flow and obstacles in the current sub-segment, rather than remaining fixed.

[0100] Step 4.3: Based on the dynamically adjusted weighting coefficients and the multi-objective weighted cost function incorporating the water flow influence correction coefficient, a path search is performed within the navigation sub-segment with the objective of minimizing the weighted cost function value, generating the corresponding locally optimized path for that navigation sub-segment. Specifically, this includes: determining the path search range for the current navigation sub-segment; using the defined navigation sub-segment boundaries as constraints; using the water flow velocity field grid nodes within the sub-segment as feasible nodes for the path search; simultaneously removing infeasible nodes within obstacle coverage areas; and constructing a set of feasible nodes within the sub-segment. The dynamically adjusted weighting coefficients are then substituted into the multi-objective weighted cost function incorporating the water flow influence correction coefficient. .

[0101] With minimizing the comprehensive weighted cost function as the core objective, the A* path search algorithm is used to perform path search within a sub-segment. The starting point of the path search is set as the initial boundary point of the current navigation sub-segment, and the ending point is set as the termination boundary point of the sub-segment. Starting from the starting point, the algorithm traverses its adjacent feasible nodes, calculates the actual cost from each adjacent node to the starting point, and the estimated cost from the node to the ending point, and adds the two to obtain the total cost of the node. The node with the minimum total cost is selected as the next path node, and the above traversal and calculation process is repeated until the ending node is found. If a node with a sudden increase in total cost is encountered during the search, the algorithm backtracks to the previous node and retraces other adjacent feasible nodes to ensure that the path always extends in the direction with the minimum total cost. After completing the path search from the starting point to the ending point, the searched nodes are connected in sequence to form a continuous path trajectory. This trajectory is the locally optimized path corresponding to the current navigation sub-segment. This path satisfies the real-time environmental characteristics of water flow and obstacles, and also achieves a comprehensive final result in terms of navigation energy consumption, path length, obstacle avoidance safety, and path smoothness.

[0102] In this embodiment of the invention, water flow velocity field grid data is used as real-time environmental change information, and water flow influence correction coefficient is introduced as a penalty term into a multi-objective weighted cost function that includes navigation energy consumption, path length, obstacle avoidance safety distance, and path smoothness. Then, by combining water flow velocity vectors and obstacle distribution characteristics through a dynamic weight adjustment mechanism, dynamic weight coefficients for each indicator are obtained. Finally, the goal is to minimize the weighted cost function value to perform path search within the navigation segment and generate a locally optimized path. This overcomes the technical problems of existing technologies where the multi-objective weighted cost function does not incorporate refined water flow influence quantification indicators, and the weight coefficients are preset fixed values ​​that cannot be dynamically adjusted according to real-time environmental characteristics such as water flow and obstacles, resulting in local path optimization failing to conform to the actual environmental characteristics of the water area. This achieves a deep integration of water flow influence and multi-objective optimization indicators, allowing the weight coefficients to adaptively and dynamically adjust with real-time environmental characteristics. Ultimately, the generated locally optimized path for the sub-segment can comprehensively consider navigation energy consumption, path length, obstacle avoidance safety, and smoothness, improving the environmental adaptability and optimization effect of local path planning.

[0103] In a preferred embodiment of the present invention, step 4.1 above may include:

[0104] Step 4.11: Extract the real-time water flow velocity vector field of the current navigation segment from the constructed water flow velocity field grid data, and read the integral calculation result of the navigation segment from the calculated water flow influence correction coefficient. Perform spatiotemporal consistency verification on the real-time water flow velocity vector field and the water flow influence correction coefficient. Specifically, this includes: retrieving the constructed water flow velocity field grid data of the navigation segment, extracting the water flow velocity vector information of all grid nodes within the spatial boundary range of the current navigation segment, including the magnitude and direction of the water flow velocity of each node, based on the minimum and maximum values ​​of the horizontal coordinate and the minimum and maximum values ​​of the vertical coordinate, and integrating them to form the real-time water flow velocity vector field of the current navigation segment, ensuring that the extracted vector field completely covers the entire spatial region of the current navigation segment without any data omissions.

[0105] The calculated water flow influence correction coefficient for the navigation segment is read, which is the sum of the path integrals along the centerline. Key attribute information such as the calculation timestamp and spatial identifier of the navigation segment corresponding to this coefficient is identified. A spatiotemporal consistency check is performed on the real-time water flow velocity vector field and the water flow influence correction coefficient to ensure that the two types of data match in both time and space. The data sampling timestamps of the hydrological monitoring buoys in the real-time water flow velocity vector field and the integral calculation timestamp of the water flow influence correction coefficient are extracted. The difference between the two timestamps is calculated. If the difference is less than a preset time threshold, such as 10 seconds, the time dimension is considered consistent. If the difference exceeds the threshold, the water flow velocity field grid data matching the timestamp of the water flow influence correction coefficient is retrieved again until the time check passes. The real-time water flow velocity is then verified. The spatial coverage of the vector field and the spatial boundary of the navigation sub-segment to which the water flow influence correction coefficient belongs are verified. It is confirmed that the abscissa and ordinate range of the vector field completely include the boundary range of the current navigation sub-segment, and that the grid resolution of the vector field is consistent with the grid resolution used when calculating the correction coefficient. If the spatial range or resolution does not match, the vector field is spatially clipped or resampled until the spatial verification passes. All node data of the real-time water flow velocity vector field are traversed, and outliers such as zero values ​​and values ​​exceeding the reasonable water flow velocity range (e.g., greater than the historical maximum water flow velocity of the water area) are removed. At the same time, it is checked whether the value of the water flow influence correction coefficient is a non-negative integral result without negative meaning. If there are outliers, the correction coefficient is recalculated. After ensuring that both types of data are valid, the spatiotemporal consistency verification is completed.

[0106] Step 4.12: Based on the water flow influence correction coefficient after the spatiotemporal consistency verification is passed, and combined with the preset penalty weight factor, an additional cost term including the water flow influence correction coefficient is constructed. The additional cost term is then linearly weighted into the pre-constructed basic multi-objective weighted cost function. The basic multi-objective weighted cost function is composed of a navigation energy consumption cost term, a path length cost term, an obstacle avoidance safety distance cost term, and a path smoothness cost term, which are linearly weighted. Specifically, the basic multi-objective weighted cost function is composed of a navigation energy consumption cost term, a path length cost term, an obstacle avoidance safety distance cost term, and a path smoothness cost term, which are linearly weighted. Navigation energy consumption cost term: The basic energy consumption per unit distance of the unmanned vessel is set to 5 kWh / km. The actual length of the candidate path is 2 km. The calculation method of the basic navigation energy consumption cost is 5 kWh / km × 2 km, and the result is 10 kWh.

[0107] Path length cost: The straight-line distance between the start and end points of the current navigation segment is 1.8 kilometers, and the actual length of the candidate path is 2 kilometers. Therefore, the path length cost is calculated as 2 kilometers ÷ 1.8 kilometers, which yields a result of approximately 1.11.

[0108] Obstacle avoidance safety distance cost: The preset obstacle avoidance safety threshold is 5 meters. The candidate path contains 10 nodes, of which 3 nodes have shortest distances to surrounding obstacles of 3 meters, 4 meters, and 2 meters respectively, and the shortest distances of the remaining 7 nodes are all ≥ 5 meters. First, calculate the obstacle avoidance cost of a single node: the first node is 5 meters - 3 meters = 2 meters, the second node is 5 meters - 4 meters = 1 meter, the third node is 5 meters - 2 meters = 3 meters, and the cost of the remaining nodes is 0. Then, calculate the arithmetic mean of the costs of all nodes, i.e., (2 + 1 + 3 + 0 + 0 + 0 + 0 + 0 + 0) ÷ 10, and the result is 0.6 meters.

[0109] Path smoothness cost: The candidate path contains 9 adjacent path segments, and the absolute values ​​of the turning angles of each segment are 5°, 3°, 4°, 2°, 6°, 1°, 3°, 2°, and 4° respectively. First, add up all the absolute values ​​of the turning angles, i.e., 5+3+4+2+6+1+3+2+4=30°, and then divide by the total number of path segments 9 to get a result of approximately 3.33°.

[0110] The basic multi-objective weighted cost function is as follows: the initial weight coefficients for navigation energy consumption, path length, obstacle avoidance safety distance, and path smoothness are preset to be 0.4, 0.2, 0.3, and 0.1, respectively. The sum of the four coefficients is 1. The basic function is calculated as 10×0.4+1.11×0.2+0.6×0.3+3.33×0.1, which is calculated step by step to obtain 4+0.222+0.18+0.333=4.735.

[0111] Based on the water flow impact correction coefficient after passing the spatiotemporal consistency verification, an additional cost term including this coefficient is constructed. The specific calculation is as follows: the calculated water flow impact correction coefficient is set to 8, and the preset penalty term weight factor for this small and medium-sized unmanned vessel is 0.5; then the calculation method of the additional cost term is 8×0.5, and the result is 4. This value only reflects the additional impact of water flow spatial changes on path cost; the additional cost term is linearly weighted into the basic multi-objective weighted cost function: the initial update of the cost function is to add the additional cost term 4 to the basic function result 4.735, and the result is 8.735. This process is a linear weighted integration, which does not change the weight ratio of each cost term in the basic function, and only quantifies the penalty effect of water flow spatial changes through the additional term.

[0112] Step 4.13: Dynamically correct the navigation energy consumption cost term in the basic multi-objective weighted cost function based on the real-time water flow velocity vector field, generating an updated multi-objective weighted cost function that considers the influence of real-time water flow. Specifically, this includes: First, analyzing the influence mechanism of the real-time water flow velocity vector field on the navigation energy consumption of the unmanned vessel: When the unmanned vessel navigates in a water flow environment, downstream flow reduces navigation energy consumption, while upstream or cross-current flow increases it. The degree of energy consumption change is directly related to the magnitude of the water flow velocity and the angle between the water flow direction and the navigation direction. The navigation energy consumption cost term in the basic multi-objective weighted cost function is then dynamically corrected. The specific correction process is as follows:

[0113] Calculate the water flow energy consumption correction coefficient: For each node on the candidate path, extract the magnitude and direction of the water flow velocity in the real-time water flow velocity vector field, calculate the angle between the water flow direction and the preset navigation direction of the unmanned vessel at the node; and then calculate the energy consumption correction coefficient based on the angle.

[0114] Corrected energy consumption per unit mile: Multiply the base energy consumption per unit mile of the unmanned vessel by the arithmetic mean of the energy consumption correction coefficients of all nodes on the candidate path to obtain the corrected energy consumption per unit mile that takes into account the influence of real-time water flow.

[0115] Recalculate the navigation energy cost item: Replace the base energy cost item with the corrected energy consumption value per unit distance and recalculate the navigation energy cost item.

[0116] Generate an updated multi-objective weighted cost function: Replace the original navigation energy consumption cost term in the basic multi-objective weighted cost function with the modified navigation energy consumption cost term, while retaining the added additional cost term, and finally generate an updated multi-objective weighted cost function that takes into account the influence of real-time water flow.

[0117] In this embodiment of the invention, by performing spatiotemporal consistency verification on the real-time water flow velocity vector field and the water flow influence correction coefficient, and combining the penalty term weighting factor to construct the verified water flow influence correction coefficient as an additional cost term and linearly weighting it into the basic multi-objective weighted cost function, and simultaneously dynamically correcting the navigation energy consumption cost term in the basic function according to the real-time water flow velocity vector field to generate an updated multi-objective weighted cost function, this invention overcomes the technical problems in the prior art where the lack of data consistency verification when integrating water flow influence indicators into the cost function easily introduces errors, and the failure to construct a dedicated additional cost term for water flow influence and to dynamically correct the navigation energy consumption term in conjunction with real-time water flow, resulting in the multi-objective weighted cost function failing to accurately and rigorously reflect the actual impact of real-time water flow on various navigation optimization indicators. This invention thus ensures the effectiveness and accuracy of water flow-related quantitative data, allowing water flow influence to be deeply integrated into the multi-objective weighted optimization system in the form of a standardized and rigorous additional cost term, while simultaneously achieving precise adaptation of the navigation energy consumption term to the real-time water flow environment, improving the fit of the multi-objective weighted cost function to complex water flow environments and the accuracy of quantitative assessment.

[0118] In a preferred embodiment of the present invention, step 5 above may include:

[0119] Step 5.1: Obtain the local optimized path corresponding to each generated navigation segment. Perform path splicing on the local optimized paths of adjacent navigation segments at the segment boundary points, and smooth the path curvature at the splicing points to generate a complete global navigation path for the unmanned vessel. Specifically, this includes: retrieving the local optimized path corresponding to each navigation segment generated in the previous step; clarifying the starting and ending boundary points of each local optimized path, as well as the spatial coordinates and path curvature information of all nodes on the path; ensuring that the coordinate system of all local optimized paths is consistent with the coordinate system of the global navigation area to avoid splicing failure due to coordinate deviations; and performing the splicing of local optimized paths of adjacent navigation segments using the common segment boundary point of two adjacent navigation segments as the splicing reference. Accurately align the ending boundary point of the local optimized path of the previous navigation segment with the starting boundary point of the local optimized path of the next navigation segment, ensuring that the spatial coordinates of the two boundary points completely coincide, thus achieving the initial splicing of adjacent local paths. After the initial splicing is completed, focus on smoothing the path curvature at the splicing points.

[0120] Calculate the path curvature at the splicing point. Extract the spatial coordinates of the last three nodes of the local path of the previous sub-segment, and calculate the tangent direction of the last path using the slope of the line connecting adjacent nodes. Then extract the spatial coordinates of the first three nodes of the local path of the next sub-segment, and calculate the tangent direction of the starting path. Subsequently, calculate the angle between the two tangent directions, and combine this with the lengths of the path segments before and after the splicing point to calculate the initial curvature at the splicing point. Wherein, the angle between the tangent direction of the path at the end of the segment preceding the splicing point and the tangent direction of the path at the beginning of the segment following the splicing point is... The length of the path segment before the splicing point is The length of the path segment after the splicing point is The initial curvature at the splicing point is ;

[0121] The system determines whether the initial curvature exceeds a preset curvature threshold. If the initial curvature is less than or equal to the preset threshold, the path smoothness at the splicing point meets the requirements and no further processing is needed. If the initial curvature is greater than the preset threshold, a circular arc transition method is used for smoothing. First, the radius of the transition arc is determined based on the preset maximum allowable curvature, calculated as the reciprocal of the preset maximum allowable curvature; the larger the curvature, the smaller the radius of the transition arc. Then, the center position of the transition arc is determined based on the tangent direction of the paths before and after the splicing point, ensuring that the transition arc is tangent to the end path of the previous sub-segment and the starting path of the next sub-segment. Finally, the transition arc segment is inserted at the splicing point to seamlessly connect the local optimized paths before and after with the transition arc segment. The curvature of the connected path is recalculated until the curvature at the splicing point is less than or equal to the preset threshold, completing the smooth transition. After all the local optimized paths of all adjacent navigation sub-segments have been spliced ​​and smoothed according to the above process, all local paths and transition segments are connected sequentially to form a continuous, smooth, and curvature-free complete path trajectory, which is the global navigation path of the unmanned vessel.

[0122] Step 5.2: During the unmanned surface vessel's (USV) navigation along the global navigation path, real-time environmental change data perceived by onboard sensors is collected. This environmental change data is compared with the constructed water flow velocity field grid data to calculate the deviation between the current navigation environment and the constructed grid data. Specifically, this includes: real-time collection of current navigation environment change data during the USV's navigation along the generated global navigation path, with the collection frequency consistent with the sampling frequency of the hydrological monitoring buoy to ensure the spatiotemporal synchronization of the data. The collected environmental change data mainly includes: the magnitude and direction of water flow velocity at the current navigation position, the real-time distribution and size of surrounding obstacles, and changes in the water topography at the current location, among other core environmental parameters. The real-time collected environmental change data is then accurately compared with the constructed water flow velocity field grid data and the generated obstacle density distribution matrix, with a focus on calculating the deviations in water flow velocity and obstacle distribution.

[0123] Water flow velocity deviation calculation: For each water flow velocity sampling point collected in real time, map it to the grid node in the water flow velocity field grid that is closest to its spatial coordinates, and extract the estimated water flow velocity value of that grid node; calculate the difference between the real-time sampled water flow velocity value and the estimated value of the corresponding grid node, and take the absolute value of the difference as the water flow velocity deviation of a single sampling point; traverse all real-time water flow velocity sampling points, calculate the arithmetic mean of the deviations of all sampling points, and obtain the average water flow velocity deviation value.

[0124] Obstacle distribution deviation calculation: The real-time acquired obstacle distribution locations are mapped to the corresponding grid cells in the generated obstacle density distribution matrix, and the obstacle distribution density value of the grid cell is extracted; the difference between the actual density value obtained by calculating the number of obstacles in the real-time acquired grid cell and the corresponding grid density value in the matrix is ​​calculated, and the absolute value of the difference is taken as the obstacle distribution deviation of a single grid cell; all grid cells with obstacle distributions are traversed, and the arithmetic mean of the deviations of all cells is calculated to obtain the average obstacle distribution deviation value.

[0125] Comprehensive deviation value calculation: Deviation weight coefficients are set for the average deviation value of water flow velocity and the average deviation value of obstacle distribution, and the sum of the two weight coefficients is 1. In water areas where water flow has a greater impact, the weight of water flow velocity deviation can be appropriately increased, and in areas with dense obstacles, the weight of obstacle distribution deviation can be appropriately increased. The comprehensive deviation value is calculated to obtain the comprehensive deviation value between the current navigation environment and the pre-constructed grid data. This value can quantitatively characterize the degree of difference between the current environment and the environment at the time of planning. The larger the deviation value, the more drastic the environmental change, and the higher the necessity for path adjustment.

[0126] Step 5.3 involves iteratively correcting the segmentation strategy and weighting coefficients online based on the deviation value, and then returning to Step 1 as input parameters to re-execute the process. This achieves dynamic updating and online adjustment of the global navigation path. Specifically, this includes: first, setting a comprehensive deviation threshold, which is preset based on the unmanned vessel's navigation stability requirements and environmental adaptability. The smaller the comprehensive deviation threshold, the higher the sensitivity to environmental changes and the more frequent the path adjustment. The calculated comprehensive deviation value is compared with the preset comprehensive deviation threshold. If the comprehensive deviation value is less than or equal to the preset threshold, it indicates that the current navigation environment is relatively similar to the planned environment, and there is no need to correct the segmentation strategy and weighting coefficients. The unmanned vessel can continue to navigate along the current global navigation path.

[0127] If the overall deviation value exceeds the preset threshold, an online iterative correction mechanism is activated to correct the segmentation strategy and weight coefficients respectively. The specific correction process is as follows:

[0128] Segmentation Strategy Correction: For navigation areas with large overall deviation values, the core processes of steps 1.1 to 1.3 are re-executed to adjust segment boundary points and sub-segment lengths. Obstacle distribution density and water complexity data for this area are re-collected, and the obstacle density distribution matrix and water complexity quantification values ​​are recalculated. Then, the sliding window analysis method is used to re-scan the area, redetermine the abrupt change points of the overall environmental characteristic values, adjust the spatial position of the segment boundary points, and further divide the area with large deviations into shorter navigation sub-segments to improve the sub-segment adaptability to environmental changes. At the same time, combined with the navigation mileage and time constraints required by the mission, the length of each sub-segment is re-optimized to ensure that the corrected segmentation strategy is adapted to the current real-time environmental characteristics and mission requirements.

[0129] Weighting coefficient correction: Based on the type of comprehensive deviation, targeted dynamic weighting coefficients and penalty term weighting factors are applied. If the proportion of water flow velocity deviation is too large, it indicates that the current water flow environment is changing drastically, and the dynamic weighting coefficient for increasing navigation energy consumption needs to be calculated as follows: , This is the dynamic weighting coefficient for navigation energy consumption before correction. It is the percentage of water flow velocity deviation. This is the corrected dynamic weighting coefficient for navigation energy consumption. The calculation method for the penalty term weighting factor of the water flow influence correction coefficient is as follows: , This is the weighting factor of the penalty term for the water flow influence correction coefficient before correction. It is the percentage of water flow velocity deviation. This is the corrected penalty weighting factor; if the obstacle distribution deviation accounts for too large a proportion, it indicates that the current obstacle distribution is changing drastically, and the dynamic weighting coefficient of the obstacle avoidance safety distance needs to be increased. The calculation method is as follows: , This is the dynamic weighting coefficient of the obstacle avoidance safety distance before correction. It is the percentage of obstacle distribution deviation. It is the revised dynamic weight coefficient of obstacle avoidance safety distance; at the same time, the weight coefficients of path length and path smoothness are adjusted to ensure that the sum of the dynamic weight coefficients of the four optimization indicators is still 1. The revised weight coefficients are more in line with the core influencing factors of the current environment.

[0130] After the segmentation strategy and weight coefficients are corrected, the corrected segmentation strategy and weight coefficients are used as new input parameters. The entire path planning process is then re-executed in step 1. This involves re-performing the dynamic segmentation of the global navigation area, the construction of the water flow velocity field, the calculation of the water flow influence correction coefficient, the generation of local optimized paths, and the splicing and smoothing of global paths. This generates a new global navigation path that adapts to the current real-time environment, enabling dynamic updates and online adjustments to the global navigation path. Steps 5.2 to 5.3 are repeated. Throughout the entire navigation process of the unmanned vessel, environmental changes, calculation deviations, and iterative corrections are continuously monitored to ensure that the navigation path of the unmanned vessel always remains adapted to the real-time water environment.

[0131] In this embodiment of the invention, a complete global navigation path is generated by splicing the locally optimized paths of each navigation segment at the segment boundary points and smoothing the curvature of the splicing points. Real-time environmental data from onboard sensors is collected during the unmanned surface vessel's (USV) navigation and compared with the water flow velocity field grid data to calculate the deviation value. Then, based on the deviation value, the segmentation strategy and weight coefficients are iteratively corrected online, and step 1 is returned to be re-executed to achieve dynamic updates of the global path. This overcomes the technical problems of existing technologies where the lack of curvature smoothing during local path splicing easily leads to sudden changes in global path curvature, affecting navigation stability. Furthermore, the absence of an environmental data comparison and deviation analysis mechanism during navigation, as well as the lack of online iterative correction of segmentation strategies and weight coefficients and a path cyclic update mechanism, results in a static global path that cannot adapt to the dynamically changing aquatic environment during navigation. This invention achieves smooth splicing of the global navigation path, ensuring the stability and smoothness of USV navigation, while accurately sensing dynamic changes in the aquatic environment during navigation. Through online iterative correction and path cyclic updates, the global navigation path is dynamically adjusted in real time, ensuring that the USV's navigation path always conforms to the real-time changing aquatic environment.

[0132] like Figure 2 As shown, embodiments of the present invention also provide an unmanned surface vessel path segmentation weighted dynamic adjustment system, comprising:

[0133] The acquisition module is used to acquire navigation environment information and mission requirements of the unmanned vessel. Based on the obstacle distribution density and water complexity in the navigation environment information, it determines the segmentation strategy and dynamically divides the global navigation area into multiple navigation sub-segments of different lengths according to the segmentation strategy.

[0134] The segmentation module is used to collect water flow velocity data sensed by hydrological monitoring buoys at three fixed positions within each segment of navigation in real time. Based on the spatial coordinates and water flow velocity data of the three hydrological monitoring buoys, a water flow velocity field grid data of the navigation segment is constructed.

[0135] The calculation module is used to smoothly fit the water flow velocity field grid data to obtain the continuous spatial distribution function of water flow velocity within the navigation segment; the gradient of the continuous spatial distribution function is calculated, the gradient field is extracted and integrated along the segment to obtain the water flow influence correction coefficient of the navigation segment;

[0136] The optimization module is used to introduce the water flow influence correction coefficient as an additional term into the pre-built multi-objective weighted cost function. Through the preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental change information of the navigation sub-segment, and the local optimization path corresponding to each sub-segment is generated together with the water flow influence correction coefficient.

[0137] The fusion module is used to splice and merge the locally optimized paths of the navigation sub-segments to obtain the complete global navigation path of the unmanned vessel. Based on the environmental change data perceived by the unmanned vessel in real time during navigation, the segmentation strategy and weight coefficients are iteratively corrected and the process is returned to step 1 for re-execution to achieve online dynamic adjustment of the path.

[0138] It should be noted that this system is a system corresponding to the above method. All implementation methods in the above method embodiments are applicable to this embodiment and can achieve the same technical effect.

[0139] Embodiments of the present invention also provide a computing device, including: a processor and a memory storing a computer program, wherein the computer program, when executed by the processor, performs the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0140] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0141] Experimental example:

[0142] I. Experimental Background

[0143] This experiment validates the unmanned surface vessel (USV) path segmentation weighted dynamic adjustment method and system. The experimental scenario is a port area and surrounding waterways (total area of ​​approximately five square kilometers). The experiment aims to verify the system's multi-sensor fusion perception capability, the effect of path segmentation weighted dynamic adjustment, and its real-time path replanning performance in dynamic obstacle environments.

[0144] II. Experimental Equipment and Parameters

[0145] The experimental equipment includes: one unmanned surface vessel (4.5 meters long, maximum speed of 8 knots), one multibeam sonar system (detection range of 200 meters), one lidar system (detection range of 100 meters), two millimeter-wave radars (operating frequency of 77 GHz), four industrial cameras (2 million pixels each), and one GPS / INS integrated navigation system.

[0146] Experimental parameter settings: The path segment length is adjustable from 50 to 200 meters, and the weighting coefficients include safety (0.3 to 0.45), efficiency (0.25 to 0.4), smoothness (0.15 to 0.2), and energy consumption (0.1 to 0.15); the path replanning trigger condition is that a new obstacle is detected at a distance of less than 50 meters or the current path deviation exceeds 5 meters.

[0147] III. Experimental Procedure

[0148] Figure 3 This presentation displays statistical data on obstacle distribution in different aquatic environments, conducted across five typical aquatic environments. Open waters had the fewest obstacles (12); port channels had 45; inland waterways had the most (68); nearshore shoals had 35; and complex waters (including bridges, docks, and other vessels) had the most (92). Obstacle types included static obstacles (reefs, shipwrecks, buoys) and dynamic obstacles (other vessels, floating objects).

[0149] Figure 4 The diagram demonstrates the dynamic adjustment of path segment weights based on the characteristics of the aquatic environment. In narrow channels (segments one to four), the safety weight increases to 0.42 to 0.45, while the efficiency weight decreases to 0.25 to 0.28. In open waters (segments eight to ten), the efficiency weight increases to 0.4 to 0.45, while the safety weight decreases to 0.28 to 0.3. The smoothness weight remains at 0.17 to 0.2, and the energy consumption weight remains at 0.1 to 0.12.

[0150] Figure 5 This paper presents a performance comparison of different path planning algorithms, specifically comparing the performance of four algorithms. The A* algorithm plans a path length of 1250 meters with an average deviation of 3.2 meters; the RRT algorithm plans a path length of 1380 meters with an average deviation of 4.5 meters; the artificial potential field method plans a path length of 1320 meters with an average deviation of 2.8 meters; and the method of this invention plans a path length of 1180 meters (shortest) with an average deviation of 1.5 meters (smallest). The method of this invention outperforms traditional algorithms in both path length and tracking accuracy.

[0151] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for dynamically adjusting the path of an unmanned surface vessel using segmented weighting, characterized in that: The method includes: Step 1: Obtain the navigation environment information and mission requirements of the unmanned vessel. Based on the obstacle distribution density and water complexity in the navigation environment information, determine the segmentation strategy and dynamically divide the global navigation area into multiple navigation sub-segments of different lengths according to the segmentation strategy. Step 2: For each navigation segment, collect the water flow velocity data sensed by the hydrological monitoring buoys at three fixed positions within the navigation segment in real time. Based on the spatial coordinates of the three hydrological monitoring buoys and the water flow velocity data, construct the water flow velocity field grid data of the navigation segment. Step 3: Perform smooth fitting on the water flow velocity field grid data to obtain the continuous spatial distribution function of water flow velocity within the navigation segment; calculate the gradient of the continuous spatial distribution function, extract the gradient field and integrate it along the segment to obtain the water flow influence correction coefficient of the navigation segment. Step 4: The water flow influence correction coefficient is introduced as an additional term into the pre-constructed multi-objective weighted cost function. Through the preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental change information of the navigation sub-segment. The water flow influence correction coefficient is then combined to generate the local optimization path corresponding to each sub-segment. Step 5: The locally optimized paths of the navigation sub-segments are spliced ​​and merged to obtain the complete global navigation path of the unmanned vessel. Based on the environmental change data perceived by the unmanned vessel in real time during navigation, the segmentation strategy and weight coefficients are iteratively corrected and the process is returned to step 1 to be re-executed, so as to realize the online dynamic adjustment of the path.

2. The unmanned surface vessel path segmentation weighted dynamic adjustment method according to claim 1, characterized in that, Step 1: Obtain the navigation environment information and mission requirements of the unmanned surface vessel. Based on the obstacle distribution density and water complexity in the navigation environment information, determine the segmentation strategy and dynamically divide the global navigation area into multiple navigation sub-segments of different lengths according to the segmentation strategy, including: The navigation environment information and mission requirements of the unmanned vessel are obtained. The obstacle distribution density in the navigation environment information is rasterized to obtain the obstacle density distribution matrix. At the same time, the water complexity in the navigation environment information is quantitatively analyzed to obtain the quantitative value of water complexity. Based on the obstacle density distribution matrix and the quantified value of water complexity, the sliding window analysis method is used to scan the global navigation area, calculate the comprehensive environmental characteristic value within each window, and determine the segment boundary points based on the abrupt change points of the comprehensive environmental characteristic value. Based on the segment boundary points, and combined with the navigation mileage and time constraints in the task requirements, the length of the region between adjacent segment boundary points is optimized and adjusted to generate multiple navigation sub-segments with different lengths.

3. The unmanned surface vessel path segmentation weighted dynamic adjustment method according to claim 2, characterized in that, Step 2: For each divided navigation segment, real-time water flow velocity data is collected from hydrological monitoring buoys at three fixed locations within the segment. Based on the spatial coordinates of the three hydrological monitoring buoys and the water flow velocity data, a water flow velocity field grid data for the navigation segment is constructed, including: For each generated navigation segment, the water flow velocity sensing data output by the hydrological monitoring buoys at three fixed positions within the navigation segment is acquired in real time, and the water flow velocity sensing data is filtered to obtain the effective water flow velocity value corresponding to each hydrological monitoring buoy. Using the spatial coordinates of three hydrological monitoring buoys and their corresponding effective water flow velocity values ​​as known sample points, the inverse distance weighted interpolation algorithm is used to perform spatial interpolation calculations on the unsampled points within the navigation segment, thereby obtaining the estimated water flow velocity values ​​at each grid node within the navigation segment. The grid resolution is set according to the boundary range of the navigation segment. The navigation segment is then divided into regular grids based on the grid resolution, and the estimated water flow velocity at each grid node is filled into the corresponding grid node position to form the water flow velocity field grid data of the navigation segment.

4. The unmanned surface vessel path segmentation weighted dynamic adjustment method according to claim 3, characterized in that, Step 3: Perform smooth fitting on the water flow velocity field grid data to obtain the continuous spatial distribution function of water flow velocity within the navigation segment; The gradient of the continuous spatial distribution function is calculated, the gradient field is extracted and integrated along the sub-segment to obtain the water flow influence correction coefficient for the navigation sub-segment, including: For the constructed water flow velocity field grid data of the navigation segment, the bicubic spline interpolation algorithm is used to perform smooth fitting on the grid data to generate a mathematical expression for the continuous variation of water flow velocity with spatial position within the navigation segment, which serves as the continuous spatial distribution function of water flow velocity within the navigation segment. The first-order partial derivatives of the continuous spatial distribution function with respect to the abscissa and ordinate are obtained respectively. The gradient field of the water flow velocity in the navigation segment is constructed based on the first-order partial derivatives, and the water flow velocity change vector at each spatial location is obtained. The gradient field is path integrated along the centerline of the navigation segment, and the result of the path integration is used as the correction coefficient for the water flow influence of the navigation segment.

5. The unmanned surface vessel path segmentation weighted dynamic adjustment method according to claim 4, characterized in that, Step 4: The water flow impact correction coefficient is introduced as an additional term into the pre-constructed multi-objective weighted cost function. Through a preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental changes of the navigation sub-segments. Combined with the water flow impact correction coefficient, the local optimization paths corresponding to each sub-segment are generated, including: The constructed water flow velocity field grid data is obtained as real-time environmental change information, and the obtained water flow influence correction coefficient is used as a penalty term to be weighted into the pre-constructed multi-objective weighted cost function. The multi-objective weighted cost function includes navigation energy consumption, path length, obstacle avoidance safety distance and path smoothness index. Based on the water flow velocity vector and obstacle distribution characteristics in the real-time environmental change information, and combined with the preset initial weight coefficients, the dynamically adjusted weight coefficients corresponding to each indicator are calculated through a dynamic weight adjustment mechanism. Based on a multi-objective weighted cost function with dynamically adjusted weighting coefficients and a correction coefficient for the influence of water flow, a path search is performed within the navigation segment with the objective of minimizing the weighted cost function value, generating a locally optimized path for that navigation segment.

6. The unmanned surface vessel path segmentation weighted dynamic adjustment method according to claim 5, characterized in that, Step 4.1: Obtain the constructed water flow velocity field grid data as real-time environmental change information, and obtain the water flow influence correction coefficient. Introduce this correction coefficient as a penalty term into the pre-constructed multi-objective weighted cost function. The multi-objective weighted cost function includes navigation energy consumption, path length, obstacle avoidance safety distance, and path smoothness indices, including: The real-time water flow velocity vector field of the current navigation segment is extracted from the constructed water flow velocity field grid data, and the integral calculation result of the navigation segment is read from the calculated water flow influence correction coefficient. The spatiotemporal consistency of the real-time water flow velocity vector field and the water flow influence correction coefficient is verified. Based on the water flow influence correction coefficient after the spatiotemporal consistency verification is passed, and combined with the preset penalty term weighting factor, an additional cost term containing the water flow influence correction coefficient is constructed, and the additional cost term is linearly weighted into the pre-constructed basic multi-objective weighted cost function. The basic multi-objective weighted cost function is composed of a linear weighted cost term of navigation energy consumption, a cost term of path length, a cost term of obstacle avoidance safety distance, and a cost term of path smoothness. The navigation energy consumption cost term in the basic multi-objective weighted cost function is dynamically corrected based on the real-time water flow velocity vector field, thereby generating an updated multi-objective weighted cost function that takes into account the influence of real-time water flow.

7. The unmanned surface vessel path segmentation weighted dynamic adjustment method according to claim 6, characterized in that, Step 5 involves splicing and merging the locally optimized paths of the navigation segments to obtain the complete global navigation path of the unmanned surface vessel (USV). Based on real-time environmental change data perceived by the USV during navigation, the segmentation strategy and weight coefficients are iteratively corrected, and the process is repeated in step 1 to achieve online dynamic adjustment of the path. This includes: The local optimized path corresponding to each generated navigation segment is obtained. The local optimized paths of adjacent navigation segments are spliced ​​at the segment boundary points, and the path curvature at the splicing point is smoothed to generate a complete global navigation path for the unmanned vessel. During the unmanned vessel's navigation along the global navigation path, environmental change data sensed by the shipborne sensors are collected in real time. The environmental change data is compared with the constructed water flow velocity field grid data, and the deviation value between the current navigation environment and the constructed grid data is calculated. The segmentation strategy and weight coefficients are iteratively corrected online based on the deviation value, and the corrected segmentation strategy and weight coefficients are used as input parameters to return to step 1 for re-execution, so as to realize the dynamic update and online adjustment of the global navigation path.

8. An unmanned surface vessel path segmentation weighted dynamic adjustment system, wherein the system implements the method as described in any one of claims 1 to 7, characterized in that, include: The acquisition module is used to acquire navigation environment information and mission requirements of the unmanned vessel. Based on the obstacle distribution density and water complexity in the navigation environment information, it determines the segmentation strategy and dynamically divides the global navigation area into multiple navigation sub-segments of different lengths according to the segmentation strategy. The segmentation module is used to collect water flow velocity data sensed by hydrological monitoring buoys at three fixed positions within each segment of navigation in real time. Based on the spatial coordinates and water flow velocity data of the three hydrological monitoring buoys, a water flow velocity field grid data of the navigation segment is constructed. The calculation module is used to smoothly fit the water flow velocity field grid data to obtain the continuous spatial distribution function of water flow velocity within the navigation segment; Gradient calculation is performed on the continuous spatial distribution function, the gradient field is extracted and integrated along the sub-segment to obtain the water flow influence correction coefficient of the navigation sub-segment; The optimization module is used to introduce the water flow influence correction coefficient as an additional term into the pre-built multi-objective weighted cost function. Through the preset dynamic weight adjustment mechanism, the weight coefficients of each indicator in the multi-objective weighted cost function are adjusted according to the real-time environmental change information of the navigation sub-segment, and the local optimization path corresponding to each sub-segment is generated together with the water flow influence correction coefficient. The fusion module is used to stitch together and merge the locally optimized paths of the navigation segments to obtain the complete global navigation path of the unmanned vessel. Based on the environmental change data perceived in real time during the unmanned vessel's navigation, the segmentation strategy and weight coefficients are iteratively corrected and the process is repeated in step 1 to achieve online dynamic adjustment of the path.

9. A computing device, characterized in that, include: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the one or more processors to implement the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the method as described in any one of claims 1 to 7.

Images