A method and device for planning a search path for a maritime target

Abstract

The present application relates to a kind of offshore target search path planning method and device, method includes: at least based on a plurality of possible positions generation searcher to the initial search path of distress target;With the survival probability of distress target in each possible position as optimization goal is maximized, and with all searchers are covered when moving along search path, each possible position is covered by the search radius of at least one searcher, as constraint condition, initial search path is optimized, and the final search path after optimization is obtained, wherein, survival probability is calculated according to seawater temperature, whether distress target wears lifesaving equipment and distress time;Make searcher search distress target according to final search path, simultaneously according to seawater temperature, current searched possible position and distress time dynamically adjust the survival probability of distress target and final search path, search path is optimized by intelligent algorithm, more close to actual sea search situation, improve the success rate of sea search task.

Description

Technical Field

[0001] This invention belongs to the field of maritime search, specifically relating to a method and apparatus for planning a search path for maritime targets. Background Technology

[0002] With technological advancements, the maritime search environment has become increasingly complex and volatile, leading to a significant increase in the incidence of maritime accidents. Joint search for distressed targets at sea, as a crucial component of maritime operations, is finding increasingly widespread application. Joint search for distressed targets at sea, a vital part of maritime search and rescue operations, refers to the process of searching for targets encountering various dangerous situations (such as accidents, natural or man-made disasters) at sea using various sensors and platforms (such as aircraft and ships). This technological system, through a multi-source heterogeneous sensor network and cross-platform collaborative mechanisms, comprehensively utilizes high-resolution remote sensing satellites, fixed-wing patrol aircraft, shipborne UAVs, and surface vessels to dynamically search for distressed targets at sea encountering emergencies such as ship collisions, fires and explosions, extreme weather disasters, or human sabotage. By rationally planning target search paths, search routes can be optimized, reducing unnecessary area omissions and duplicate searches, thereby covering the most likely target areas in the shortest possible time and improving rescue efficiency and success rates.

[0003] Traditional maritime search relies too much on the subjective judgment and historical experience of decision-makers, lacks systematicity, and suffers from problems such as path redundancy, ambiguous priorities, and poor dynamic adaptability. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention provides a method and apparatus for planning a search path for maritime targets.

[0005] The present invention includes a method for planning a search path for a maritime target, comprising: acquiring distressed target and environmental data, and predicting multiple possible locations of the distressed target based on the environmental data; generating an initial search path for the distressed target by searchers based on at least multiple possible locations; optimizing the initial search path with the objective of maximizing the survival probability of the distressed target at each possible location, and with the constraint that each possible location is covered by the search radius of at least one searcher when all searchers move along the search path, to obtain an optimized final search path; enabling searchers to search for the distressed target according to the final search path, while dynamically adjusting the survival probability of the distressed target and the final search path according to seawater temperature, currently searched possible locations, and distress time; The formulas for calculating the probability of survival include:

[0006]

[0007] in, The probability of survival for the distressed target. For proportional parameters, For the time of the incident, For shape parameters, For seawater temperature, and Positive correlation The basic proportional parameters are based on the reference water temperature. The coefficient representing the influence of water temperature. The impact factor of lifesaving equipment, If the person in distress is wearing life-saving equipment, If the person in distress is not wearing life-saving equipment, .

[0008] Optionally, the constraints also include: ensuring that the total time for any searcher to complete the final search path does not exceed the maximum survival time of the distressed target, and ensuring that the final search path is a closed path, wherein the maximum survival time of the distressed target is... , This refers to the seawater temperature.

[0009] Optionally, when the number of searchers is 1, an initial search path for the distressed target is generated based on at least multiple possible locations, including: obtaining the initial location of the searchers and multiple candidate nodes; starting from the initial location, according to multiple possible locations and the survival probability of the distressed target, selecting one candidate node to add to the path queue each time, until each possible location is covered by the search radius of at least one searcher in the candidate node, and obtaining the initial search path according to the path queue.

[0010] Optionally, when the number of searchers is 1, an initial search path for the distressed target is generated based on at least multiple possible locations, including: obtaining the initial location of the searchers and multiple candidate nodes; starting from the initial location, selecting one candidate node to add to the path queue each time, based on the number of possible locations that the searcher can cover within the search radius of the candidate node, until each possible location is covered by the search radius of at least one searcher within the candidate node, and obtaining the initial search path based on the path queue.

[0011] Optionally, when the number of searchers is 1, an initial search path for the distressed target is generated based on at least multiple possible locations, including: obtaining the initial location of the searcher and multiple candidate nodes; randomly generating an initial path based on the initial location and performing random iterative adjustments on the initial path, while calculating the survival probability of the current initial path; accepting the current initial path if the survival probability of the current initial path is greater than the highest historical survival probability, otherwise discarding the current initial path; accepting the current initial path according to a preset probability if the survival probability of the current initial path is less than or equal to the highest historical survival probability, otherwise discarding the current initial path; until the iteration ends, the current initial path is used as the initial search path, wherein the preset probability decreases as the distress time increases.

[0012] Optionally, the initial search path includes multiple nodes. When the number of searchers is 1, the initial search path is optimized according to the optimization objective and constraints, including: reversing the node order of all path segments in the initial search path sequentially; retaining the current search path if the survival probability of the current search path after reversal is higher than that of the search path before reversal; otherwise, restoring the current search path to the search path before reversal; deleting each node in the initial search path sequentially; deleting the node if the current search path after deletion still satisfies the constraints and has a survival probability higher than that of the search path before deletion; otherwise, retaining the node; replacing each original node in the initial search path sequentially with its neighboring nodes; replacing the original node with its neighboring node if the current search path after replacement still satisfies the constraints and has a survival probability higher than that of the search path before replacement; otherwise, retaining the original node; and adding at least one node not belonging to the initial search path to the initial search path to increase the survival probability of the initial search path.

[0013] Optionally, when the number of searchers is greater than 1, an initial search path for the distressed target is generated based on at least multiple possible locations, including: obtaining the initial location of each searcher and multiple candidate nodes; generating an initial search path for each searcher in parallel based on the selection of all candidate nodes; generating an initial search path for one of the multiple searchers, including: starting from the initial location, selecting candidate nodes from those not currently selected by other searchers and adding them to the path queue based on the survival probability of the distressed target and its distance from the candidate nodes, until each possible location is covered by the search radius of at least one searcher in the candidate nodes, and obtaining the initial search path for that searcher based on the path queue.

[0014] Based on the same inventive concept, this invention also provides a maritime target search path planning device, comprising: a position prediction module for acquiring distressed target and environmental data, and predicting multiple possible locations of the distressed target based on the environmental data; an initial path generation module for generating an initial search path for the distressed target by searchers based on at least multiple possible locations; and a path optimization module for optimizing the initial search path with the objective of maximizing the survival probability of the distressed target at each possible location, and with the constraint that each possible location is covered by the search radius of at least one searcher when all searchers move along the search path, to obtain an optimized final search path; the formula for calculating the survival probability includes:

[0015]

[0016] in, The probability of survival for the distressed target. For proportional parameters, For the time of the incident, For shape parameters, For seawater temperature, and Positive correlation The basic proportional parameters are based on the reference water temperature. The coefficient representing the influence of water temperature. The impact factor of lifesaving equipment, If the person in distress is wearing life-saving equipment, If the person in distress is not wearing life-saving equipment, The dynamic adjustment module enables searchers to search for distressed targets based on the final search path, and dynamically adjusts the survival probability of the distressed target and the final search path based on seawater temperature, currently searched possible locations, and time of distress.

[0017] Based on the same inventive concept, the present invention also provides an electronic device, comprising: a processor; a memory for storing executable instructions; wherein the processor is configured to read the executable instructions from the memory and execute the executable instructions to implement any of the methods described above.

[0018] Based on the same inventive concept, the present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the processor performs any of the above-described methods.

[0019] The beneficial effects of this invention are as follows: By analyzing the drift distribution and dynamic decay characteristics of the survival probability of distressed targets at sea, this invention models the search task as a fixed-radius covering traveling salesman problem and introduces a dynamic survival probability model. It comprehensively considers the impact of factors such as water temperature, life-saving equipment, and time on the target's survival rate, and optimizes the search path through intelligent algorithms, making it more closely aligned with actual maritime search conditions and improving the success rate of maritime search missions. Furthermore, the formula provided by this invention causes the survival probability to decay exponentially over time, thereby shortening the total search time while maximizing the survival probability. The addition of a water temperature decay coefficient makes the survival probability decay non-linearly and faster as the temperature decreases. The inclusion of a life-saving equipment influence coefficient further aligns with the actual situation of distressed personnel. Attached Figure Description

[0020] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0021] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0022] Figure 1 A flowchart illustrating a method for planning a search path for a maritime target, provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the possible location of a distressed target, provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of an optimal search area provided by an embodiment of the present invention; Figure 4 This is a schematic diagram illustrating the relationship between maximum survival time and seawater temperature, provided as an embodiment of the present invention. Figure 5 A flowchart illustrating another method for planning a search path for a maritime target provided in an embodiment of the present invention; Figure 6 A flowchart illustrating another method for planning a search path for a maritime target provided in an embodiment of the present invention; Figure 7 This is a schematic diagram of a maritime target search path planning result provided in an embodiment of the present invention; Figure 8 This is a schematic diagram of another maritime target search path planning result provided in an embodiment of the present invention; Figure 9 This is a schematic diagram of another maritime target search path planning result provided in an embodiment of the present invention; Figure 10 A bar chart showing the experimental results of a maritime target search path planning method provided in an embodiment of the present invention; Figure 11 A box plot of experimental results for maritime target search path planning provided in an embodiment of the present invention; Figure 12 This is a schematic diagram of another maritime target search path planning result provided in an embodiment of the present invention; Figure 13 A line graph showing the experimental results of a maritime target search path planning method provided in an embodiment of the present invention; Figure 14 Box plot of another experimental result for maritime target search path planning provided in an embodiment of the present invention; Figure 15 This is a schematic diagram illustrating another possible location of a distressed target provided in an embodiment of the present invention; Figure 16 This is a schematic diagram of another maritime target search path planning result provided in an embodiment of the present invention; Figure 17 Box plot of another experimental result for maritime target search path planning provided in an embodiment of the present invention; Figure 18 This is a schematic diagram of the structure of a maritime target search path planning device provided in an embodiment of the present invention; Figure 19 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention.

[0023] Among them, 10 is the location prediction module; 20 is the initial path generation module; 30 is the path optimization module; 1101 is the processor; 1102 is the memory; 1103 is the transceiver; and 1104 is the bus. Detailed Implementation

[0024] To better understand the above-mentioned objectives, features, and advantages of the embodiments of the present invention, the solutions of the embodiments of the present invention will be further described below. It should be noted that, unless otherwise specified, the embodiments of the present invention and the features thereof can be combined with each other.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of the embodiments of the present invention, but the embodiments of the present invention may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some embodiments of the embodiments of the present invention, and not all embodiments.

[0026] Current research in the field of maritime search has certain shortcomings: Most existing studies on maritime search strategies convert the sweep width into discrete spatial units using a rasterization method, essentially falling under the category of conventional path optimization problems. This fails to fully reflect the coupling relationship between the effective coverage area of ​​search equipment and the path trajectory. This simplification may lead to two typical problems in complex sea conditions: first, insufficient raster resolution creates actual coverage blind spots; second, the failure to accurately quantify the effective coverage area of ​​the sweep width results in "insufficient effective coverage." Particularly in multi-searcher collaborative operations, this discrete raster-based modeling method struggles to accurately characterize the spatial superposition effect of the search trajectory and the sweep width, failing to meet the high-precision search requirements in extreme environments.

[0027] In view of this, embodiments of the present invention provide a method for planning a search path for a maritime target, such as... Figure 1 As shown, it includes: S1. Obtain distress target and environmental data, and predict multiple possible locations of distress target based on environmental data.

[0028] Specifically, maritime distress targets include personnel, vessels, submarines, or other living targets operating at sea that are in danger in the marine environment due to accidents, natural disasters, or human factors and require location and rescue through search operations. Specifically, it covers the following two categories of targets: personnel targets, including ship crews, offshore workers, passengers, and other individuals engaged in maritime activities; and vehicle targets, such as merchant ships, fishing vessels, yachts, submarines, and maritime aircraft that have encountered accidents.

[0029] S2. Generate an initial search path for the distressed target based on at least multiple possible locations.

[0030] S3. The initial search path is optimized to maximize the survival probability of the distressed target at each possible location, with the constraint that each possible location is covered by the search radius of at least one searcher as all searchers move along the search path. The optimized final search path is obtained. The survival probability is calculated based on seawater temperature, whether the distressed target is wearing life-saving equipment, and the time of distress. Seawater temperature is directly proportional to the survival probability, and the time of distress is inversely proportional to the survival probability. Specifically, the formula for calculating the survival probability includes:

[0031]

[0032] in, The probability of survival of the distressed target. For proportional parameters, For the time when the target was in distress, For shape parameters, For seawater temperature, and Positive correlation, shape parameters If it is greater than 1, it means that the risk rate increases over time; These are the basic proportional parameters at the reference water temperature (0℃). The coefficient representing the influence of water temperature. The impact factor of lifesaving equipment, Indicate whether the person in distress is wearing life-saving equipment. If the person in distress is wearing life-saving equipment, If the person in distress is not wearing life-saving equipment, In one specific embodiment, the above parameters are taken as follows: Dimensions are ; Dimensions are ; ; It is understandable that the above parameter values ​​are based on statistical data from small-scale experiments, and in practical applications, they can be refitted using specific sea area datasets.

[0033] Specifically, this embodiment of the invention uses the effective scan width as the search radius. The effective scan width refers to the spatial range within which the detection equipment can effectively detect the search target. The detection equipment includes the visual and electronic detection equipment of the search personnel and is a key mathematical indicator used to quantify the target detection capability of the search team. In maritime search, the effective scan width is also called the sea-scanning width. Its core meaning is to define the effective range of the detection equipment, providing basic parameter basis for the problem of maritime target search under a fixed search radius.

[0034] Specifically, to improve search success rate and efficiency, the survival probability of distressed targets is a core criterion for measuring search efficiency in maritime search route planning, providing a quantitative indicator for route optimization. In maritime distress scenarios, the survival probability of a target is influenced by multiple factors. Environmentally, sea state and water temperature have a significant impact; rough sea conditions such as heavy rain or dense fog exacerbate the danger to distressed targets and reduce their survival probability. Low water temperatures can easily lead to hypothermia, endangering life, making survival even more difficult in cold waters. Regarding the distressed target itself, the availability of life-saving equipment and the target's physical condition are crucial; sufficient and high-performance life-saving equipment provides assurance and significantly improves the survival probability. Furthermore, those in better physical condition are better able to cope with harsh environments and sustain life. In terms of rescue forces, timeliness is extremely important; arriving at the scene as early as possible and avoiding prolonged waiting for rescue effectively improves the survival rate; that is, the shorter the search time, the higher the survival probability.

[0035] The Traveling Salesman Problem (TSP) is a classic combinatorial optimization problem. A classic TSP can be described as follows: A salesman needs to travel to several cities to sell his goods. Starting from one city, he needs to visit all the cities and return to his starting point. How should he choose his route to minimize the total distance traveled? The above embodiments of this invention can be considered a variant of the TSP, namely the Covering Salesman Problem (CSP) with a fixed radius. The target points to be searched can be considered as the cities to be covered by the TSP. Unlike the traditional TSP problem, in the maritime search scenario of this invention, the search device can simultaneously detect multiple target points through the coverage radius. This is equivalent to adding a constraint that vertices can be covered (i.e., the aforementioned constraint) to the TSP. Under certain conditions, a Hamiltonian cycle with the minimum length is constructed, where unvisited vertices must be within the coverage radius of at least one visited vertex.

[0036] S4. Enables the searcher to search for the distressed target according to the final search path, and dynamically adjusts the survival probability of the distressed target and the final search path based on seawater temperature, currently searched possible locations, and time of distress.

[0037] Based on the above constraints, this embodiment of the invention also proposes an optimization objective for dynamic priority survival probability. By quantifying the combined impact of time, environment, and individual disaster resilience, a real-time "search value weight" is assigned to each target point, thereby guiding the search path to prioritize areas with high survival probability. It should be understood that the aforementioned "weight" does not refer to a specific weight parameter, but rather that the search path determined by this embodiment based on survival probability will prioritize searching areas with higher survival probability, which is equivalent to giving that area a higher weight.

[0038] To address the issues of low search efficiency and insufficient dynamic prioritization in maritime search for distressed targets, this invention analyzes the drift distribution and dynamic decay characteristics of the survival probability of distressed targets. By applying the constraints in S3, the search task is modeled as a fixed-radius covering traveling salesman problem. Furthermore, a dynamic survival probability model is introduced through the optimization objective in S3, comprehensively considering the impact of factors such as water temperature, life-saving equipment, and time on the target's survival rate. Intelligent algorithms optimize the search path, better reflecting actual maritime search conditions and improving the success rate of maritime search missions. Simultaneously, the formula provided in this invention causes the survival probability to decay exponentially over time, thereby shortening the total search time while maximizing the survival probability. The inclusion of a water temperature decay coefficient allows the survival probability to decay non-linearly and more rapidly as temperature decreases. The introduction of a life-saving equipment influence coefficient further aligns with the actual situation of distressed personnel.

[0039] Predicting the distribution of distressed targets scientifically is a prerequisite for initiating a search mission. Accurate prediction of the distribution of distressed targets allows for the general delineation of the possible areas where they might appear, based on factors such as past accident data, marine environmental characteristics, and ship navigation patterns. This enables participating forces in the joint search to plan search routes more effectively, avoiding the waste of time and resources caused by blind searches. It also facilitates the efficient identification of key search areas, the rational allocation of search resources, the optimization of search routes, and increases the probability of finding distressed targets. During the period from receiving a distress signal to the arrival of rescue vessels at the accident site, the distressed target may drift due to marine meteorological and environmental factors. Therefore, scientifically predicting the target's drift trajectory becomes a core element in determining the accuracy of search area delineation. In specific implementation, S1 above predicts multiple possible locations of the distressed target based on environmental data, including using a Lagrange particle tracking algorithm to predict the drift trajectory of the distressed target.

[0040] Specifically, force analysis of a distressed target at sea yields the following results:

[0041] in, For the quality of the distressed target, For the speed of the distressed target, The Coriolis force is used to measure the effect of the Earth's rotation on ocean currents. The time elapsed since the target encountered danger. Express the first derivative of velocity with respect to time. The force exerted by the wind on the distressed target. This refers to the force exerted by ocean currents on a distressed target, including the force generated by wind on the movement of seawater and the force exerted by the movement of the ocean currents themselves.

[0042] Then, the displacement of the particle (i.e., the distressed target) is calculated using the Lagrange particle tracking algorithm:

[0043] in, This is the particle displacement. This represents the first derivative of the particle displacement with respect to time. The drift coefficient, Where is the diffusion coefficient. The values ​​are random numbers. The real-time position of the particles is obtained by calculating their displacement values ​​at each moment, enabling dynamic tracking of the particles. Since some coefficients in solving the drift prediction equation need to be determined based on empirical data and experimental analysis, the calculated trajectory may have some deviation. To assess the uncertainty level of the drift prediction results, the Monte Carlo algorithm is used to simulate the drift trajectories of five hundred particles, ultimately obtaining the particle distribution range.

[0044] In one specific embodiment of this invention, historical search case data from a certain sea area was selected for analysis and research. Based on marine environmental data such as ocean currents and wind direction in that sea area, the aforementioned drift prediction model was used to predict the possible location of the distressed target. An example prediction diagram is shown below. Figure 2 As shown. Specifically, Figure 2 The simulation experiment showed that 500 particles were randomly generated, with each update time step set to 1 hour, and a continuous simulation experiment was carried out for 72 hours. By recording the spatial position changes of each particle every hour, the probabilistic prediction of the drift distribution of the distressed target was achieved.

[0045] After obtaining the target drift prediction results, the optimal execution period for the search task can be determined based on the target survival probability model. This involves selecting a time window with a high target survival probability to conduct the search operation, and then calculating the minimum coverage rectangle of all particles within that time period to delineate the search area and target points. To address the problem of determining the optimal search area, this invention provides two differentiated strategies: one is a complete coverage strategy focused on maximizing coverage of potential target locations, and the other is a high-value priority strategy focusing on high-probability drift areas. These two strategies are suitable for different emergency response needs; the former emphasizes the comprehensiveness of the search range, while the latter emphasizes the efficiency of resource allocation.

[0046] At the application level, this embodiment of the invention first constructs the spatial boundary of the baseline search region by calculating the convex hull geometry of the predicted scattered point set. This convex hull structure can effectively contain all predicted points and form a minimum closed convex polygon. Based on this, to improve the spatial regularity of the search region to adapt to the searcher's operational characteristics, this embodiment of the invention further introduces a minimum covering rectangle algorithm. Through geometric transformation, the convex hull region is converted into an axis-aligned rectangular region. This rectangle, while fully covering the convex hull range, significantly simplifies the spatial form, providing a standardized search region for subsequent scheduling and coordination of search forces. Specifically, S1 above also includes determining the optimal search region based on the predicted trajectory, and determining multiple candidate nodes of the search path in the following embodiments within the optimal search region, including using a scanning algorithm to generate convex polygons, and proposing an algorithm for calculating the minimum convex hull of a planar point set. For example, the algorithm is described as follows: First, traverse the point set. Find the point with the smallest ordinate. If there are points with the same ordinate, select the one with the smallest abscissa as the reference point. .by Using a reference point as the baseline, connect it to all other points in the point set, calculate the counterclockwise angle between the resulting line segment and the horizontal line, and sort them; when the angles are equal, determine the line segment based on the distance from the point to the horizontal line. The distances are then sorted a second time to obtain an ordered point sequence. Connecting adjacent points in the sequence forms an initial polygon, where... , ... These vertices must lie on the convex hull boundary. Finally, based on the geometric property that "vertices of a convex polygon must lie on the same side of any edge," the vertex sequence of the initial polygon is filtered, eliminating points not on the convex hull boundary, thus determining the final convex hull point set. After determining the final convex polygon region, this region is normalized, and the minimum area bounding rectangle of the convex polygon is constructed to determine the search area. The prediction results are transformed into the minimum bounding rectangle. By discretizing the continuous space into grid cells, a standardized geometric space is provided for the planning and design of maritime search force paths.

[0047] In the above Figure 2 In the illustrated embodiment, the optimal search area for particles was obtained using the minimum coverage rectangle algorithm during the period from 17:00 to 19:00 on the 29th. Figure 3 The red area is shown in the image.

[0048] Specifically, embodiments of the present invention provide a mathematical function for the change of survival probability over time after a maritime distress, which comprehensively considers seawater temperature. Time of flooding Key factors include (i.e., the time elapsed since the target encountered distress) and whether a personal flotation device (PFD) was worn. Among these, water temperature is the most significant influencing factor, especially in cold seawater, where survival time is significantly shortened. The maximum possible survival time at a given water temperature is also provided. In addition, the above constraints also include ensuring that the total time for any searcher to complete the search path does not exceed the maximum survival time of the distressed target. The relationship between maximum survival time in seawater and seawater temperature is as follows: Figure 4 As shown.

[0049] Furthermore, wearing a PFD significantly extends survival time. According to actual statistics, in a sea area with a constant water temperature, wearing a PFD can extend the median survival time from 3.6 hours to 29.3 hours.

[0050] Specifically, the process of constructing the above-mentioned survival probability model for distressed targets is as follows: First, we construct a basic model in which the survival probability decays exponentially over time:

[0051] Among them, the proportional parameter With water temperature (Unit: °C) Related:

[0052] Then, introduce the existence parameter of life-saving equipment into the model. Further refine the survival probability:

[0053] Based on the basic probability decay model, after adding the life-saving equipment correction factor and the morphology adjustment factor, a nonlinear combination of the final survival probability of the distressed target can be obtained:

[0054]

[0055]

[0056] in, The probability of survival of the distressed target. For proportional parameters, For the time when the target was in distress, The shape parameter reflects the characteristic that the survival probability decreases nonlinearly with decreasing temperature. For seawater temperature, The seawater temperature is The maximum possible survival time of a distressed target at that time. The basic proportional parameters are based on the reference water temperature. The coefficient representing the influence of water temperature. The impact factor of lifesaving equipment, This indicates whether the person in distress is wearing life-saving equipment. If the person in distress is wearing life-saving equipment, If the person in distress is not wearing life-saving equipment, .

[0057] In some embodiments, this invention proposes a single-searcher distress target search path planning algorithm, where the number of searchers is one, to enable a single searcher to plan a search path within a given ocean area based on the real-time updated survival probability of the distressed target. Specifically, the single-searcher search path planning model is as follows: First, define the variables in the model. Based on the predicted trajectory in the above embodiment, obtain the points of possible locations of the distressed target (i.e., multiple possible locations in the above embodiment). Let the set of these points be denoted as . Each distribution point has the same initial survival probability, set to 1. Search devices such as drones or ships travel at a fixed speed. The sensor coverage radius is [missing information]. The probability of target survival varies with the duration of the encounter. Dynamic decay, and affected by seawater temperature. Existence of life-saving equipment Influence.

[0058] The objective function of the single-searcher path planning model is to maximize the total survival probability score.

[0059] in, The survival probability of a distressed target under single-searcher conditions can be calculated based on the above survival probability. The calculation formula is obtained; This represents the total search time, which is equal to the time the target was in distress when the search was completed. This indicates that the search order in path planning affects the cumulative total search time. This represents the total number of steps taken by the searcher over the total search time (i.e., the maximum number of areas that can be searched). For the searcher in the first Position during the step, For the searcher in the first Position during the step; This indicates that the searcher is at the first The distance of the step to all other possible positions. For the searcher in the first The position at step (i.e., the first step) (Possible search locations) Indicates the searcher at the The position at the time of each step is also used to represent all other possible positions in the search path. Each time the searcher moves, they need to calculate the distance to all other possible positions to select the next possible target position and calculate the time for each step. Used to indicate the search path, , Used to indicate the first [item] on the path A possible location for the search.

[0060] When the searcher is a single searcher, the model's constraints include: Coverage constraint: Each possible location must be covered by the search radius of at least one node in the search path. Coverage, i.e. , Make Time constraint: The total search time shall not exceed the longest possible survival time of the target, i.e., the total time. Path continuity constraint: the final search path generated. It is a closed path.

[0061] In maritime distress target search scenarios, the survival probability of distressed targets decays exponentially over time due to limitations such as the timeliness of life-saving equipment and threats from the marine environment. Furthermore, different targets exhibit significant differences in their initial survival probability, decay rate, and rescue priority. A single coverage strategy is insufficient to meet the efficient search requirements in complex scenarios. Therefore, in some embodiments, this invention proposes a multi-searcher distress target search path planning algorithm, extending the search path planning from a single searcher to multiple searchers, thus better reflecting the actual situation of maritime distress target search scenarios. The multi-searcher target search model is as follows: First, define the variables in the model. Similarly, based on the distressed target drift prediction model, obtain the points where the distressed target might be located; let the set of these points be denoted as . Each distribution point has the same initial survival probability, set to 1. The number of searchers, such as drones or ships, is [number missing]. The searchers at a fixed speed The sensor coverage radius is [missing information]. The probability of target survival over time Dynamic decay, and affected by environmental parameters such as water temperature. Existence of life-saving equipment Influence.

[0062] The objective function of the multi-searcher path planning model is to maximize the total survival probability score for all searchers.

[0063] in, To determine the survival probability of a distressed target under multiple searchers, the above survival probability can be used as a basis. The calculation formula is obtained from, This represents the total search time, which is equal to the time the target was in distress when the search was completed. This indicates that the search order in path planning affects the cumulative search time. , Used to indicate the search path, , Used to indicate the first [item] on the path A possible location for the search.

[0064] When there are multiple searchers, the model's constraints include: Coverage constraint: Each possible location must be within the search radius of at least one node in the search path. Coverage, i.e. , Make Time constraint: The total search time for a single searcher shall not exceed the longest possible survival time of the target, i.e., the total time. Path continuity constraint: the final search path generated. It is a closed path.

[0065] Based on the aforementioned survival probability model and path planning model, this embodiment of the invention designs a suitable path planning algorithm to solve the model. The algorithm's objectives include achieving full coverage of drift prediction points and maximizing the cumulative spatiotemporal survival probability of the search path.

[0066] Achieving full coverage of drift prediction points is the fundamental solution objective of the TSP problem, requiring that the nodes in the final path cover all nodes in the dataset given a search radius. Applying the path planning model in the above embodiment, the search points of the final path can cover all possible distributions of target points on the target drift trajectory predicted by the ocean dynamics model. Maximizing the cumulative spatiotemporal survival probability of the search path, as a metric for the efficiency of maritime target search missions, requires maximizing the cumulative spatiotemporal survival probability score of the search path planned by a single searcher. This means prioritizing the search of denser distribution and higher survival rates of search points in the shortest possible time to improve the success rate of the search.

[0067] For a single-seeker path planning model, the single-seeker algorithm to achieve this goal can be divided into four steps: algorithm initialization (initial path construction), local optimization, global optimization, and result output. The algorithm initialization stage is equivalent to the process of generating the initial search path in the above embodiment, and the subsequent optimization stage is the process of optimizing the search path in the above embodiment. The algorithm flow is as follows: Figure 5 As shown.

[0068] In some embodiments, when the number of searchers is 1, the above S2 includes: S21. Obtain the searcher's initial position and multiple candidate nodes.

[0069] In practical implementation, after determining the optimal search area, to effectively support the joint search path planning for distressed targets at sea, it is necessary to obtain discrete latitude and longitude coordinate data within the optimal search area. To facilitate subsequent algorithm solving, this embodiment of the invention converts the latitude and longitude coordinates into a grid coordinate system, and maps the particle coordinates to a Cartesian coordinate system with one corner of the smallest bounding rectangle as the origin and the two sides of that corner as the X and Y axes. Simultaneously, candidate nodes for subsequent search paths are determined based on the intersection of the X and Y axes in the Cartesian coordinate system. Specifically, in the algorithm initialization module, considering that maritime target search missions typically need to depart from a specific base, to enhance the algorithm's fit with actual engineering scenarios, this study sets the path starting point as a flexibly input fixed coordinate parameter, allowing users to adjust it according to different mission requirements, ensuring that the search path generated by the algorithm better conforms to real operational procedures.

[0070] S221. Starting from the initial position, based on multiple possible positions and the survival probability of the distressed target, select a candidate node each time to add it to the path queue, until each possible position is covered by the search radius of at least one searcher in the candidate node, and obtain the initial search path based on the path queue.

[0071] Alternatively, in other embodiments, when the number of searchers is 1, the above S2 includes: S21. Obtain the searcher's initial position and multiple candidate nodes.

[0072] S222. Starting from the initial position, based on the number of possible positions that the searcher can cover within the search radius of the candidate node, select one candidate node at a time and add it to the path queue until each possible position is covered by the search radius of at least one searcher within the candidate node. The initial search path is obtained based on the path queue.

[0073] Alternatively, in other embodiments, when the number of searchers is 1, the above S2 includes: S21. Obtain the searcher's initial position and multiple candidate nodes.

[0074] S223. Based on the initial position, randomly generate an initial path and perform random iterative adjustments to the initial path. At the same time, calculate the survival probability of the current initial path. If the survival probability of the current initial path is greater than the highest historical survival probability, accept the current initial path; otherwise, discard the current initial path. If the survival probability of the current initial path is less than or equal to the highest historical survival probability, accept the current initial path according to a preset probability; otherwise, discard the current initial path. Until the iteration ends, use the current initial path as the initial search path. The preset probability decreases as the time of encounter increases.

[0075] Regarding the construction of the initial solution (i.e., the initial path), to improve the algorithm's adaptability and stability to complex search environments, this embodiment of the invention designs three differentiated initial solution generation algorithms: greedy search based on the survival probability of the distressed target (S221 above), heuristic search based on coverage gain and distance (S222 above), and heuristic search based on simulated annealing (S223 above). The following is a detailed explanation of the design ideas for these three different initialization algorithms.

[0076] First, the classic greedy algorithm, as an algorithm with significant application value in fields such as information processing, demonstrates unique research value due to its practicality and efficiency. The greedy algorithm expands only one candidate atom in each iteration, adding it incrementally to continuously build an approximation model of the target signal. It does not rely on complex external auxiliary mechanisms to approximate the original signal, but instead focuses on the iterative process itself, relying on continuous and in-depth iteration to achieve forward search. This iteration-driven optimization approach allows the algorithm to achieve relatively ideal signal reconstruction accuracy in scenarios such as sparse representation and compressed sensing while reducing computational complexity.

[0077] The greedy search based on the survival probability of distressed targets employs a greedy coverage gain maximization strategy in the initialization phase. First, it sets the node coordinates of possible locations of distressed targets based on the coordinates of the dataset. And construct the initial node value set accordingly. Define the set of possible uncovered locations as follows: .like If not empty, then the maximum survival time will be guaranteed. Under the premise of calculating the coverage gain of each vertex (i.e., candidate node) in the set into the path, one by one. and the arrival time to reach that vertex The coverage gain score for that point is calculated based on the survival probability model. Candidate nodes that can cover the most high-survival-probability undetected nodes are placed in the initial path to maximize the current coverage gain score. This process is iterated repeatedly until all possible locations are covered. If empty, the initial search path is obtained. The pseudocode for its initialization phase is shown in Table 1:

[0078] Heuristic algorithms are a class of algorithms that efficiently solve complex problems by utilizing problem-specific heuristics such as empirical rules, intuition, and local optima strategies. They do not pursue a strictly mathematically optimal solution, but rather quickly find approximate or feasible solutions that meet practical requirements with reasonable computational cost. Heuristic algorithms can be broadly classified into two categories: single-solution heuristics based on iterative optimization of a single solution and swarm heuristics based on the cooperative evolution of multiple solutions. Single-solution heuristic algorithms mainly include greedy algorithms, local search, and simulated annealing algorithms, which construct solutions by designing neighborhood operations such as swapping and insertion, and local search strategies. Swarm heuristic algorithms mainly include genetic algorithms, ant colony algorithms, and particle swarm optimization algorithms, which generate solutions by designing population initialization, evolutionary operators, and cooperative mechanisms.

[0079] The initialization phase of the heuristic search based on coverage gain and distance differs from the greedy search of the previous approach. Instead, it employs a hybrid heuristic search strategy to generate an initial solution. First, a set of node coordinates is set according to the dataset coordinates. And construct the initial node value set accordingly. To ensure coverage integrity takes priority, a set of possible uncovered locations is defined. If it is not empty, then calculate the coverage gain of the path to include the vertices (i.e., candidate nodes). While ensuring that the survival time does not exceed the maximum survival time Under the premise of this, in each iteration, the candidate point that can cover the most undetected nodes is selected; secondly, a path efficiency trade-off is added, and a distance penalty term is introduced. To avoid path detours caused by simply pursuing coverage gain, a strategy is developed that combines both distance cost and coverage gain, assigning appropriate weights to select the next node in the path. This strategy balances the dual objectives of distance cost and coverage gain, ensuring coverage of all predicted points while providing a high-quality starting point for subsequent optimization. It iterates repeatedly until all possible locations are covered. If empty, the initial search path is obtained. The pseudocode for its initialization phase is shown in Table 2:

[0080] Simulated Annealing (SA) is inspired by the cooling process of solid-state crystal materials. Its core lies in transforming the annealing mechanism of a physical system into an algorithm design criterion, finding a practical balance between computational efficiency and solution quality through a dynamic "exploration-exploitation" approach. This algorithm is a metaheuristic that achieves global optimization by simulating the annealing process of a physical system, making it particularly suitable for solving complex non-convex optimization problems with multiple local optima. Its key advantage lies in balancing the depth of local search with the breadth of global search by probabilistically accepting inferior solutions, thus effectively avoiding the pitfalls of traditional greedy algorithms that easily get trapped in local optima.

[0081] The heuristic search algorithm based on simulated annealing employs the simulated annealing framework. While ensuring a fixed starting point for the search path, it utilizes a temperature control mechanism to conduct neighborhood searches, thereby generating high-quality initial solutions. Compared to the previous two algorithms, this algorithm introduces a probabilistic acceptance mechanism unique to simulated annealing, allowing for the temporary acceptance of inferior solutions during the search process. This avoids getting trapped in local optima, improves global search capabilities, and makes it more adaptable to complex and ever-changing maritime target search scenarios. Furthermore, to further enhance the diversity of initial solutions, the algorithm incorporates three neighborhood operations—node swapping, path reversal, and random insertion—into the initial solution generation process.

[0082] The specific execution flow of the algorithm is as follows: First, clarify the core SA parameters required for initializing the simulated annealing algorithm: initial temperature Termination temperature Cooling coefficient and current temperature Subsequently, multiple initial solutions (i.e., initial search paths) are generated through random selection, and the current score corresponding to each initial solution is recorded in real time. and the all-time best score This provides a baseline for subsequent optimization. After entering the simulated annealing main loop, the algorithm sequentially applies three neighborhood operations—node swapping, path reversal, and random perturbation insertion—to perturb and update the current solution, continuously generating neighborhood solutions and accurately calculating the score corresponding to the new path. The algorithm determines whether to accept a new solution with a specific probability: if the new solution is better than the current solution, it is accepted unconditionally; if the new solution is inferior, it is accepted with a certain probability based on the difference in scores between the current temperature and the old and new solutions. As the iteration process progresses, the temperature decreases according to a preset cooling coefficient, and the probability of the algorithm accepting inferior solutions also decreases, causing the search to gradually focus on the region of high-quality solutions. The algorithm terminates when the maximum number of iterations is reached, or when the quality of the solution no longer improves after multiple updates, and outputs the final initial search path. The pseudocode for its initialization phase is shown in Table 3.

[0083]

[0084] For initial path construction, the above embodiments of the present invention selected three methods to generate initial solutions: greedy search based on survival probability, hybrid heuristic, and simulated annealing search heuristic. Each algorithm has its own advantages and disadvantages. Greedy search based on the survival probability of the distressed target is fast and simple to implement, but it is prone to getting trapped in local optima and the initial solution is not good enough, making it suitable for small-scale, simple scenarios. Heuristic search based on coverage gain and distance can improve path compactness, and its search logic is more reasonable, but it is also prone to getting trapped in local optima and the weight parameters need to be adjusted, making it suitable for medium-scale search tasks. Heuristic search based on the simulated annealing algorithm has strong global search capabilities, generates higher-quality initial solutions, and can adapt to environmental parameters, but its computational cost and implementation complexity are high, making it suitable for large-scale, complex scenarios. Those skilled in the art can choose one of the above three initial path generation methods for application according to actual needs.

[0085] After the initial search path is generated, such as Figure 5As shown, this embodiment of the invention optimizes the local path of the initial solution through three steps: 2-opt path smoothing, redundant node removal, and neighbor node replacement optimization. In some embodiments, the initial search path includes multiple nodes. When the number of searchers is 1, the optimization of the initial search path in step S3 above, based on the optimization objective and constraints, includes: S31. Reverse the node order of all path segments in the initial search path in turn. If the survival probability of the current search path after reversal is higher than the survival probability of the search path before reversal, retain the current search path; otherwise, restore the current search path to the search path before reversal.

[0086] Specifically, the 2-opt algorithm is based on a local adjustment of "edge swapping." It arbitrarily selects two non-adjacent edges, deletes them, and splits the original path into two segments. By reconnecting the endpoints of these two segments, a new feasible path is generated; essentially, it reverses a subsequence of the path. This optimization algorithm can improve the quality of the solution to a certain extent through simple local adjustments, thereby improving the overall solution performance. This embodiment of the invention uses the 2-opt algorithm to eliminate intersections by reversing path segments, improving the path geometry and indirectly increasing the score within the effective rescue time window. The main idea is: first, traverse all possible path segments, then reverse the path segments to generate a new path. If the new path is better, i.e., has a higher cumulative spatiotemporal survival probability score, then the modification is retained.

[0087] S32. Delete each node in the initial search path in turn. If the current search path after deleting the node still satisfies the constraints and the survival probability is higher than the survival probability of the search path before deleting the node, delete the node; otherwise, keep the node.

[0088] Specifically, redundant node removal aims to delete nodes in the path that do not contribute to coverage integrity. The main idea is to scan the path from left to right and dynamically adjust the path structure after deletion. First, a necessity check is performed, traversing all path nodes except the starting point and checking whether the problem coverage constraint is still satisfied after deletion. Second, the cumulative spatiotemporal survival probability score is compared; if the cumulative score after deletion is higher than the original score, then deletion is performed. This operation removes intermediate nodes that do not affect coverage integrity, shortening the path length and reducing time consumption.

[0089] S33. Replace each original node in the initial search path with its neighbor node in turn. If the current search path after replacing the original node still satisfies the constraints and the survival probability is higher than the survival probability of the search path before replacing the original node, replace the original node with its neighbor node; otherwise, retain the original node.

[0090] Specifically, neighboring nodes replace locally adjusted path nodes to explore better solutions within the coverage area. The main idea is as follows: First, a candidate set is generated by searching for neighboring nodes not present in the path within the coverage radius of each node except the starting point. Second, replacement verification is performed by replacing the original node with a neighboring node. The new path undergoes a coverage integrity check, and the cumulative spatiotemporal survival probability score is calculated. Finally, the best replacement scheme with the highest cumulative score is selected.

[0091] In some embodiments, such as Figure 5 As shown, after completing the inner loop of local search optimization of the search path, this invention adds a random perturbation factor to optimize the global search path. Specifically, when the number of searchers is 1, the optimization of the initial search path in S3 above based on the optimization objective and constraints also includes: S34. Add at least one node that does not belong to the initial search path to the initial search path to increase the survival probability of the initial search path.

[0092] This invention, through the random insertion of new nodes, breaks the current solution structure, effectively balancing the problem of the 2-opt algorithm getting trapped in local optima by greedily selecting the optimal edge for exchange, and enhancing the algorithm's ability to escape local optima. The main idea is as follows: First, traverse all nodes and select nodes not currently in the path. Second, randomly select K nodes and estimate their optimal insertion positions, i.e., the path that maximizes the cumulative spatiotemporal survival probability score after insertion. Finally, perform the insertion and update the path.

[0093] The perturbation injection employs a multi-stage iterative optimization mechanism. The method provided in this embodiment alternately executes local optimization and global perturbation, gradually approaching the global optimum by controlling the total number of iterations. This operation introduces new coverage possibilities, which may temporarily reduce path efficiency. However, through subsequent 2-opt optimization, removal of redundant nodes, and coordinated adjustments of the neighborhood replacement local algorithm, a better solution may eventually be found, effectively preventing the algorithm from getting trapped in local optima.

[0094] For multi-searcher path planning models, where the number of searchers is greater than one, the algorithm is similar to that of the single-searcher model. However, it requires multi-searcher collaboration to cover the drift prediction points and to maximize the cumulative spatiotemporal survival probability score of the search path using parallel time compression. Since simulated annealing is a single-solution heuristic algorithm that always searches and optimizes around a single solution, it is not suitable for multi-searcher joint search planning. Therefore, this embodiment of the invention designs a comprehensive heuristic algorithm based on coverage gain and distance to solve the problem. The algorithm can be divided into four sections: algorithm initialization, parallel search, optimization processing, and result output and visualization. Algorithm initialization and parallel search are equivalent to the initial search path generation stage, and optimization processing is equivalent to the path optimization stage. The algorithm flow is as follows: Figure 6 As shown.

[0095] In some embodiments, when the number of searchers is greater than 1, the above S2 includes: S201: Obtain the initial position of each searcher and multiple candidate nodes. In practice, S201 is similar to S21 above, and will not be described again.

[0096] S202. Generate an initial search path in parallel for each searcher based on the selection of all candidate nodes.

[0097] The process of generating an initial search path for one of the multiple searchers in S202 includes: starting from the initial position, selecting candidate nodes from those not currently selected by other searchers and adding them to the path queue based on the survival probability of the distressed target and its distance from the candidate nodes, until each possible position is covered by the search radius of at least one searcher in the candidate nodes, and obtaining the initial search path of that searcher based on the path queue.

[0098] like Figure 6 As shown, the algorithm initialization phase of the multi-searcher path planning model is mainly used to complete the basic configuration of multi-searcher collaborative search, verify the legality of the input initial coordinates and map them to the nearest discrete node, then update the global coverage state, that is, mark the initial coverage node and the first coverage time, and create an object containing the path sequence, actual coordinate trajectory and local coverage state for each searcher. At the same time, the state synchronization in a multi-threaded environment is ensured through thread locks.

[0099] In specific implementation, the parallel search phase (i.e., S202 above) achieves asynchronous collaborative coverage by multiple searchers through a multi-threaded architecture. Each searcher makes autonomous decisions in an independent thread based on the dynamic global coverage state: it quickly locates nearby uncovered nodes using KDTree, selects the top 50 candidate nodes using a hybrid score of coverage gain (70%) and distance cost (30%), evaluates the expected score after movement using a survival probability time decay model, selects the optimal expansion node, and updates the path. Simultaneously, the 2-opt optimization in the above embodiment can be used to refine the path in real time, and the global coverage state and first coverage time are updated atomically using thread locks. Ultimately, the spatiotemporal collaborative optimization of multiple paths is completed within a limited number of iterations, achieving priority coverage of high-risk areas and dynamic balanced allocation of search resources, significantly compressing the overall response time and improving the global survival probability.

[0100] In some embodiments, when the number of searchers is greater than 1, the optimization of the initial search path for each searcher in step S3 above, based on the optimization objective and constraints, includes: Redundant node removal and deep path optimization simplify the final path and improve efficiency. First, the path of each searcher is traversed, and the necessity of each node is verified in reverse—if deleting a node still allows the remaining path to cover the entire area, it is permanently removed to avoid redundant movement. Then, a high-iteration 2-opt optimization is applied to the simplified path to minimize the total movement distance. Finally, while ensuring global coverage integrity, the overall task time is shortened by eliminating invalid path segments, indirectly increasing the survival probability of search nodes. The specific optimization process can be found in the implementation example of the single-searcher path planning model described above, and will not be repeated here.

[0101] The embodiments of this invention selected the VLSI dataset provided by Andre Rohe, and experiments were conducted based on the research of the Institute for Disk Mathematics at the University of Bonn. The instance size in the VLSI dataset ranges from 131 cities to 744,710 cities. In these examples, the travel cost between cities is specified by Euclidean distance rounded to the nearest integer.

[0102] In the single-searcher target search planning algorithm, the parameters mainly include three categories: environmental parameters, algorithm control parameters, and survival probability model parameters. Environmental parameters are primarily determined based on the actual situation of searching for distressed targets at sea. Experimental settings include: seawater temperature... Search radius of the searcher Searcher movement speed The distressed target is confirmed to be carrying personal survival equipment. Algorithm control parameters are parameters involved in the algorithm's update, iteration, and execution in a single-seeker objective search planning algorithm. Key settings include: maximum number of global optimization iterations. Number of randomly inserted nodes during the perturbation phase Simulated annealing initialization iteration count The survival probability model parameters can be found in the above embodiments.

[0103] The specific experimental results are as follows: To verify the effectiveness of the algorithm, this embodiment of the invention uses three datasets of different sizes (xqf131, xqg237, pka379) for testing. The results of the embodiments using the three different initial path construction methods are compared with those of the conventional CSP algorithm in related technologies. The results of each experimental group are obtained by averaging 10 independent repeated tests. On the xqf131 dataset, the conventional CSP algorithm scored 120.10, the greedy search algorithm of this embodiment scored 124.38, the heuristic search algorithm of this embodiment scored 124.78, and the simulated annealing search algorithm of this embodiment scored 130.54. On the xqg237 dataset, the conventional CSP algorithm scored 196.48, the greedy search algorithm scored 196.62, the heuristic search algorithm scored 213.58, and the simulated annealing search algorithm scored 230.54. On the pka379 dataset, the conventional CSP algorithm scored 280.48, the greedy search algorithm scored 290.52, the heuristic search algorithm scored 301.07, and the simulated annealing search algorithm scored 364.59. Specifically, all of the above scores are positively correlated with the survival probability; that is, the higher the score, the higher the survival probability of the distressed target.

[0104] The above data analysis shows that, when the dataset size is relatively small, the method provided in this embodiment of the invention has only a slight advantage over the conventional CSP search algorithm. As the dataset size gradually increases, the performance disadvantage of the conventional CSP algorithm gradually becomes apparent due to the randomness of the search and the limitation of the searcher's speed. The method provided in this embodiment of the invention, based on a real-time survival probability update model, can more effectively handle larger-scale datasets and provide an effective search path planning scheme through different initial solution construction schemes.

[0105] This invention also compares and analyzes the performance of the three different initial solution construction algorithms mentioned above. Three TSP standard datasets of different sizes were selected for testing, with the starting point coordinates all set to (0, 0). The parameter settings can be referenced in the above embodiments. The solution path planning results of the three initial solution construction methods are as follows: Figures 7 to 9 As shown, where, Figure 7 A comparison of path planning results based on the xqf131 dataset is presented for three initial solution construction methods. Figure 8 A comparison of path planning results based on the xqg237 dataset is presented for three initial solution construction methods. Figure 9 A comparison of path planning results based on the pka379 dataset is presented for three initial solution construction methods. Specifically, Figure 7 (a) shows the path planning results of the greedy search method based on the cumulative survival probability score on the xqf131 dataset. Figure 7(b) shows the path planning results based on the xqf131 dataset for the distance- and coverage gain-based heuristic search method. Figure 7 (c) The path planning results of the simulated annealing search method based on the xqf131 dataset; Figure 8 (a) shows the path planning results of the greedy search method based on the cumulative survival probability score on the xqg237 dataset. Figure 8 (b) shows the path planning results based on the xqg237 dataset for the distance- and coverage gain-based heuristic search method. Figure 8 (c) shows the path planning results of the simulated annealing search method based on the xqg237 dataset; Figure 9 (a) shows the path planning results of the greedy search method based on the cumulative survival probability score using the pka379 dataset. Figure 9 (b) shows the path planning results based on the pka379 dataset for the distance- and coverage gain-based heuristic search method. Figure 9 (c) shows the path planning results of the simulated annealing search method based on the pka379 dataset.

[0106] This invention compares the initial path generation methods of three algorithms: greedy search based on the survival probability of distressed targets, heuristic search based on coverage gain and distance, and heuristic search based on simulated annealing. The cumulative spatiotemporal survival probability of the algorithms is used as the core indicator. Figure 10 As shown in the bar chart, the heuristic search algorithm based on simulated annealing performs best, and its advantage becomes more pronounced as the data point size increases. The next best heuristic search algorithm is based on coverage gain and distance, which performs slightly better than the greedy search algorithm based on the survival probability of distressed targets.

[0107] Furthermore, this invention embodiment tests the stability of three initial solution construction methods. Based on the standard TSP dataset xqf131, this invention embodiment conducts analysis experiments, setting up three sets of experiments: greedy search, heuristic search, and simulated annealing search. Each set is run independently 30 times, and statistical analysis is performed using the cumulative spatiotemporal survival probability as the core indicator. The corresponding algorithm result box plots are shown below. Figure 11 As shown. Figure 11The specific data for the box plots shown are as follows: the average score for greedy search is 119.88, the range is [119.08, 120.41], the median is 120.00, and the IQR (Interquartile Range) is 0.63; the average score for heuristic search is 119.83, the range is [119.05, 120.41], the median is 119.87, and the IQR is 0.32; the average score for simulated annealing search is 130.47, the range is [130.32, 130.58], the median is 130.49, and the IQR is 0.08.

[0108] Experimental results show that, in terms of data dispersion, simulated annealing search exhibits superior stability, with the lowest IOR and range among the three experimental groups. The heuristic search algorithm shows slightly better stability than greedy search. Although the three methods differ somewhat in average performance, their stability is generally good: the range range for all groups is less than 1.5% of the mean, and the IQR is less than 0.7, indicating that regardless of the initialization strategy used, the algorithm provided in this embodiment can maintain consistent output in repeated experiments.

[0109] In the multi-searcher-based target search planning algorithm, the parameters still include three categories: environmental parameters, algorithm control parameters, and survival probability model parameters. Unlike the single-searcher model described above, the multi-searcher model adds a searcher number parameter m=3 to the environmental parameters, while the remaining algorithm control parameters and survival probability model parameters remain unchanged.

[0110] For the joint search path planning problem of distressed targets at sea by multiple searchers, this embodiment of the invention uses the same dataset as the single-searcher target search for result analysis. First, a dataset "xqf131.tsp" with 131 city nodes is used, with the number of searchers set sequentially to 2, 3, and 4. Based on the distribution of the dataset, the starting point coordinates are set to (0, 0), (0, 60), (100, 0), and (100, 60), while other parameters remain unchanged. The algorithm result is shown in the figure below. Figure 12 As shown, where, Figure 12 The comparison of path planning results is shown for 2, 3, and 4 searchers (specifically, Figure 12 (a) shows the path planning results for a searcher count of 2. Figure 12 (b) shows the path planning results for a searcher count of 3. Figure 12 (c) shows the path planning results for a searcher with 4 searchers. The line segments and circles of different colors represent the final search paths and search radii of different searchers. The horizontal axis is the X-axis of the sea surface and the vertical axis is the Y-axis of the sea surface.

[0111] This invention addresses the impact of the number of searchers on the cumulative spatiotemporal survival probability score of a target by conducting controlled variable experiments on four standard datasets of different sizes to explore the optimal scale of search resource allocation. Figure 13 These are experimental results from 2 to 10 searchers on four standard datasets of different sizes. The results show that the cumulative spatiotemporal survival probability score of the distressed target increases with the number of searchers. Figure 13 It can be seen that when the number of searchers is small, the cumulative spatiotemporal survival probability score increases significantly with the increase of search resource investment. When the number of searchers exceeds 4, the score curves on each dataset show a characteristic of diminishing marginal returns, indicating that the improvement in search performance is very small at this point. Therefore, setting 3 or 4 searchers is more appropriate in joint search path planning for distressed targets by multiple searchers.

[0112] To verify the stability of the multi-searcher algorithm under different resource configurations, this embodiment of the invention conducts a comparative experiment based on the standard TSP dataset "xqf131.tsp". The number of searchers is set to 2, 3, and 4 sequentially, and each group is run independently 30 times. Statistical analysis is performed using the cumulative spatiotemporal survival probability as the core indicator, and box plots of the results are generated. Other parameter settings remain unchanged. The algorithm results are shown in the figure below. Figure 14 As shown, Figure 14 The specific data from the box plots are as follows: For 2 searchers, the average score was 122.92, with a range of [121.58, 124.38], a median of 123.04, and an IQR of 0.67; for 3 searchers, the average score was 125.56, with a range of [123.90, 126.67], a median of 125.63, and an IQR of 0.84; for 4 searchers, the average score was 126.85, with a range of [125.70, 127.61], a median of 126.93, and an IQR of 0.77. The experimental results show that as the number of searchers increases, the cumulative spatiotemporal survival probability of the average value steadily increases. The box plot characteristics show that the median and mean of each group are highly similar, and the data distribution shows no significant skewness. The IQR values ​​were 0.67, 0.84, and 0.77, respectively, all at low levels, indicating good consistency of the algorithm output within the group. In terms of dispersion, the standard deviation of each group was controlled within 0.7, and the range remained within a narrow fluctuation range, reflecting that the multi-searcher algorithm provided in this embodiment of the invention can maintain stable optimization capabilities under different resource scales.

[0113] To verify the effectiveness of the multi-searcher algorithm provided in this invention for path planning of distressed targets at sea, comparative experiments were conducted using standard datasets of different sizes. The multi-searcher algorithm provided in this invention was compared with the conventional multi-searcher CSP algorithm in related technologies that do not consider survival probability priority. The experiments employed 10 independent repeated tests and averaged results to ensure the reliability and stability of the results. The experimental results are as follows: With two searchers, the multi-searcher algorithm provided in this invention scored 124.35, 210.32, and 257.40 on the xqf131, xqg237, and pka379 datasets, respectively, while the algorithms in related technologies scored 114.17, 180.09, and 211.41, respectively. With three searchers, the multi-searcher algorithm provided in this embodiment of the invention scores 125.46, 216.23, and 291.68 on the xqf131, xqg237, and pka379 datasets, respectively, while the scores of algorithms in related technologies are 114.55, 196.11, and 231.88, respectively. The above data clearly shows that under different searcher and test set conditions, the final score of the algorithm provided in this embodiment of the invention is higher than that of conventional algorithms. As the dataset size continues to increase, the superiority of the algorithm provided in this embodiment of the invention in terms of overall performance becomes increasingly prominent, and its score improvement shows a significant increasing trend. The above experimental results confirm the effectiveness of the algorithm provided in this embodiment of the invention. Conventional algorithms, because they do not incorporate the target survival probability into the dynamic decision-making system, struggle to accurately match actual needs in resource allocation, resulting in a lack of flexibility and efficiency in search path planning. The target survival probability priority strategy in the algorithm provided in this embodiment of the invention can achieve precise allocation of search resources. In complex maritime operation environments, it can dynamically plan search paths based on the target survival probability and prioritize coverage search of areas with high survival probability, thereby significantly improving the success rate of rescue.

[0114] To explore the performance differences between the multi-searcher algorithm and the single-searcher algorithm provided in this invention for path planning of targets in distress at sea, and to understand the advantages of joint search over unit search, this invention conducts comparative experiments on three standard datasets of different sizes: xqf131, xqg237, and pka379, and analyzes the path planning results in single and multi-searcher scenarios. The experiments employ a method of averaging 10 independent repeated tests to ensure the reliability and stability of the results. The specific experimental results are as follows: The single-searcher algorithm scored 124.78, 213.58, and 301.07 on the xqf131, xqg237, and pka379 datasets, respectively; the multi-searcher algorithm with two searchers scored 124.35, 210.32, and 257.40 on the xqf131, xqg237, and pka379 datasets, respectively; the multi-searcher algorithm with three searchers scored 125.46, 216.23, and 291.68 on the xqf131, xqg237, and pka379 datasets, respectively; and the multi-searcher algorithm with 24 searchers scored 126.75, 219.96, and 313.32 on the xqf131, xqg237, and pka379 datasets, respectively.

[0115] Experimental results show that the multi-searcher algorithm outperforms the single-searcher algorithm when the number of searchers increases to 3 or more. Furthermore, as the dataset size increases, the advantages of the multi-searcher joint search mode in terms of path planning accuracy and resource utilization efficiency become increasingly prominent.

[0116] In terms of algorithm implementation, the single-searcher algorithm employs a centralized iterative search, combining a heuristic path initialization (70% coverage gain + 30% distance), an improvement process (redundancy removal + node replacement), and a perturbation mechanism (random insertion) with offline 2-opt optimization to achieve single-path optimization. Since the improvement and perturbation mechanisms are implemented in each iteration, it slightly outperforms the multi-searcher algorithm when the number of searchers is small, but its algorithmic complexity is higher. The multi-searcher algorithm, on the other hand, uses distributed multi-threaded optimization. Each searcher makes independent decisions using KDTree spatial indexing and a hybrid scoring function, synchronizes coverage states with a global lock, and incorporates online 2-opt optimization and post-processing redundancy removal, forming a hierarchical optimization system that yields the final path planning result more quickly.

[0117] In terms of algorithmic objectives, the single-agent algorithm focuses on the spatiotemporal optimization of a single path, balancing path length and time decay effects in continuous coverage. Its objective function is the cumulative survival probability on a single timeline. The multi-agent algorithm, on the other hand, maximizes the global survival score through multi-agent collaborative coverage and parallel time compression. Its mathematical model focuses on the timestamp of the earliest covered agent for each node, reducing response latency in high-risk areas through spatial division of labor, thus better aligning with the resource allocation and strategy formulation requirements in maritime distress target search scenarios.

[0118] In the historical search case analysis of the Zhuhai-Hong Kong-Macau waters, this embodiment of the invention uses people who have fallen overboard as the search targets, and conducts practical application research based on specific drift prediction model parameters and environmental parameters. According to the drift prediction model mentioned above, the distribution prediction data of the distressed targets in the sea area is obtained, with each point representing a possible location of the distressed target. Example application data is obtained after determining the time period, search area, and coordinate transformation.

[0119] Based on the constructed drift prediction model, the following key parameters were input: the prediction start time was set to 16:00 on April 29, 2023; the prediction trajectory duration was 3 hours; and the prediction starting point coordinates were 116°02′E, 21°17′N. Simultaneously, environmental parameters were incorporated into the model calculation, including a wind speed of 1.116 km / h and an ocean current speed of 4.08 km / h. The model simulates the drift process of a distressed target in a complex marine environment by coupling dynamic factors such as ocean currents and wind fields, generating a series of discrete point sets that characterize the target's potential location, such as... Figure 2 As shown.

[0120] For discrete point sets generated based on a drift prediction model, a minimum convex hull algorithm for planar point sets is used to determine the core distribution area of ​​distressed targets. First, geometric calculations are performed to obtain the smallest closed convex polygon containing all predicted points, effectively eliminating redundant internal points and forming the smallest convex region covering the potential locations of the targets. Based on this, the minimum area bounding rectangle of the convex polygon is further solved. Rotation optimization calculations minimize the area of ​​the rectangle while completely enclosing the convex hull, thus constructing a compact search boundary. The effect is as follows: Figure 3 As shown.

[0121] To adapt to the computational requirements of the joint search path planning algorithm for maritime distress targets in this embodiment of the invention, coordinate transformation of the predicted distribution points is necessary. Using Geographic Information System (GIS) technology, the original latitude and longitude coordinates of the distribution points in the example data are converted into coordinate data in a Cartesian coordinate system via Mercator projection, such as... Figure 15 As shown.

[0122] To verify the effectiveness and reliability of the multi-searcher algorithm in real-world maritime search scenarios, this study conducted example tests based on specific parameter settings. In the example data experiment, the algorithm set the marine ambient temperature to 10℃, the distressed target to wear a PFD (Protective Device), and standardized the constraints on the search resource configuration: the searchers were fixed-wing aircraft, the number of searchers was 3, the search radius was fixed at 0.1 nautical miles, the moving speed was set to 0.5 nautical miles per hour, and the starting positions of the searchers were (0,0), (1,0), and (0,1.4).

[0123] The algorithm's execution result is as follows Figure 16 As shown, the different colored line segments and circles represent the final search paths and search radii of different searchers, with the horizontal axis representing the X-axis of the sea surface and the vertical axis representing the Y-axis of the sea surface. It can be seen that the multi-searcher algorithm provided in this embodiment of the invention achieves the basic goal of full coverage of all search points, verifying its effectiveness and feasibility in the search path planning task.

[0124] To evaluate the robustness of the algorithm in dynamic environments, this embodiment of the invention uses the total cumulative spatiotemporal survival probability score of multiple searchers in each run as the core experimental metric, and conducts 30 independent and repeated experiments. Statistical analysis of the experimental data is performed, and box plots are generated as follows. Figure 17 As shown, the average cumulative spatiotemporal survival probability score of the algorithm is 768.38, with a standard deviation of 13.77 and a median of 768.78. The standard deviation / mean ratio is 1.79%, far below the common threshold of 5%, indicating that in this maritime target search example, the algorithm's environmental parameters are reasonably set, the multi-agent cooperation mechanism is robust, and the algorithm has strong stability. Furthermore, the worst score (727.81) still reaches 94.72% of the average, and the low fluctuation characteristic indicates that the algorithm has stable collaborative optimization capabilities under the parameter configuration, maintaining a high level of search efficiency even when faced with initial condition perturbations.

[0125] From the practical perspective of joint search for distressed targets at sea, the multi-search algorithm provided in this invention overcomes the limitations of traditional single-point search, demonstrating stability and efficiency in joint search tasks. The algorithm quantifies the search priority of uncovered points in real time through a target survival probability model, incorporating environmental parameters and target status, enabling searchers to dynamically focus on high-value areas and effectively addressing the shortcomings of fixed target priorities in traditional methods. Example applications show that this algorithm significantly improves the balance of task allocation and the completeness of path coverage during multi-searcher collaborative operations. It can also achieve intelligent scheduling of search resources under complex sea conditions.

[0126] It should be noted that the method of this embodiment can be executed by a single device, such as a computer or server. The method of this embodiment can also be applied to a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method of this embodiment, and the multiple devices will interact with each other to complete the above method.

[0127] It should be noted that some embodiments of the present invention have been described above. Furthermore, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0128] Based on the same inventive concept, corresponding to any of the methods in the above embodiments, the present invention also provides a maritime target search path planning device, such as... Figure 18 As shown, it includes: The location prediction module 10 is used to acquire distressed target and environmental data, and predict multiple possible locations of the distressed target based on the environmental data.

[0129] The initial path generation module 20 is used to generate an initial search path for the distressed target by the searcher based on at least multiple possible locations.

[0130] The path optimization module 30 is used to optimize the initial search path with the objective of maximizing the survival probability of the distressed target at each possible location, and with the constraint that each possible location is covered by the search radius of at least one searcher as all searchers move along the search path. The optimized final search path is obtained by optimizing the survival probability, where the survival probability is calculated based on seawater temperature, whether the distressed target is wearing life-saving equipment, and the time of distress. Seawater temperature is directly proportional to the survival probability, and the time of distress is inversely proportional to the survival probability. The formula for calculating the survival probability includes:

[0131]

[0132] in, The probability of survival for the distressed target. For proportional parameters, For the time of the incident, For shape parameters, For seawater temperature, and Positive correlation The basic proportional parameters are based on the reference water temperature. The coefficient representing the influence of water temperature. The impact factor of lifesaving equipment, If the person in distress is wearing life-saving equipment, If the person in distress is not wearing life-saving equipment, .

[0133] The dynamic adjustment module 40 is used to enable the searcher to search for the distressed target according to the final search path, and to dynamically adjust the survival probability of the distressed target and the final search path according to the seawater temperature, the currently searched possible locations, and the time of distress.

[0134] To address the issues of low search efficiency and insufficient consideration of dynamic priority in maritime search, this invention analyzes the drift distribution and dynamic decay characteristics of the survival probability of maritime targets, models the search task as a fixed-radius covering traveling salesman problem, and introduces a dynamic survival probability model. This model comprehensively considers the impact of factors such as water temperature, life-saving equipment, and time on the survival rate of targets, and optimizes the search path through intelligent algorithms to better reflect the actual situation of maritime search and improve the success rate of maritime search missions.

[0135] For ease of description, the above apparatus is described in terms of its functions, divided into various modules. Of course, in implementing this invention, the functions of each module can be implemented in one or more software and / or hardware components.

[0136] The apparatus described above is used to implement the corresponding maritime target search path planning method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0137] Figure 19 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention is shown.

[0138] like Figure 19 As shown, the electronic device may include a processor 1101 and a memory 1102 storing computer program instructions.

[0139] Specifically, the processor 1101 may include a central processing unit (CPU), an application specific integrated circuit (ASIC), or one or more integrated circuits that can be configured to implement the embodiments of the present invention.

[0140] Memory 1102 may include a mass storage device for information or instructions. For example, and not limitingly, memory 1102 may include a hard disk drive (HDD), a floppy disk drive, flash memory, optical disk, magneto-optical disk, magnetic tape, or a Universal Serial Bus (USB) drive, or a combination of two or more of these. Where appropriate, memory 1102 may include removable or non-removable (or fixed) media. Where appropriate, memory 1102 may be internal or external to the integrated gateway device. In a particular embodiment, memory 1102 is a non-volatile solid-state memory. In a particular embodiment, memory 1102 includes read-only memory (ROM). Where appropriate, the ROM may be a mask-programmed ROM, a programmable ROM (PROM), an erasable PROM (EPROM), an electrically erasable programmable PROM (EEPROM), an electrically alterable ROM (EAROM), or flash memory, or a combination of two or more of these.

[0141] The processor 1101 reads and executes computer program instructions stored in the memory 1102 to perform the steps of the maritime target search path planning method provided in the embodiments of the present invention.

[0142] In one example, the electronic device may also include a transceiver 1103 and a bus 1104. Wherein, as... Figure 19 As shown, the processor 1101, memory 1102 and transceiver 1103 are connected via bus 1104 and communicate with each other.

[0143] Bus 1104 may include hardware, software, or both. For example, and not limitingly, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Extended Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a Hyper Transport (HT) interconnect, an Industrial Standard Architecture (ISA) bus, an Infinite Bandwidth Interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local Bus (VLB) bus, or other suitable buses, or a combination of two or more of these. Where appropriate, bus 1104 may include one or more buses. Although specific buses are described and illustrated in the embodiments of the present invention, the present invention is contemplated by any suitable bus or interconnect.

[0144] The following are embodiments of a computer-readable storage medium provided in this invention. This computer-readable storage medium and the maritime target search path planning method of the above embodiments belong to the same inventive concept. For details not described in detail in the embodiments of the computer-readable storage medium, please refer to the embodiments of the maritime target search path planning method described above.

[0145] This embodiment provides a storage medium containing computer-executable instructions, which, when executed by a computer processor, are used to perform a method for planning a search path for a maritime target.

[0146] Of course, the computer-executable instructions provided in the embodiments of the present invention are not limited to the above-described method operations, but can also execute related operations in the maritime target search path planning method provided in any embodiment of the present invention.

[0147] Based on the above description of the implementation methods, those skilled in the art can clearly understand that the present invention can be implemented using software and necessary general-purpose hardware, and of course, it can also be implemented using hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as a computer floppy disk, read-only memory (ROM), random access memory (RAM), flash memory, hard disk, or optical disk, etc., including several instructions to cause a computer cloud platform (which can be a personal computer, server, or network cloud platform, etc.) to execute the maritime target search path planning method provided in the various embodiments of the present invention.

[0148] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes the aforementioned element.

[0149] The above description is merely a specific embodiment of the present invention, enabling those skilled in the art to understand or implement the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments described above, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for planning a search path for maritime targets, characterized in that, include: Acquire distress target and environmental data, and predict multiple possible locations of distress target based on the environmental data; An initial search path for the distressed target is generated by the searcher based on at least the multiple possible locations; The initial search path is optimized with the goal of maximizing the survival probability of the distressed target at each of the possible locations, and with the constraint that each of the possible locations is covered by the search radius of at least one of the searchers as all the searchers move along the search path, so as to obtain the optimized final search path. The searcher searches for the distressed target according to the final search path, and dynamically adjusts the survival probability of the distressed target and the final search path based on seawater temperature, currently searched possible locations, and time of distress. The formula for calculating the survival probability includes: in, The probability of survival of the distressed target. For proportional parameters, The time of the incident, For shape parameters, The seawater temperature is... and Positive correlation The basic proportional parameters are based on the reference water temperature. The coefficient representing the influence of water temperature. The impact factor of lifesaving equipment, If the distressed target is wearing life-saving equipment, If the distressed target is not wearing life-saving equipment, ; The constraints also include: The total time for any of the searchers to complete the final search path shall not exceed the maximum survival time of the distressed target, and the final search path shall be a closed path, wherein the maximum survival time of the distressed target is... , The seawater temperature; The initial search path includes multiple nodes; When the number of searchers is 1, a single searcher plans a search path within a given ocean area based on the real-time updated survival probability of the distressed target. The objective function of the single-searcher path planning model is to maximize the total survival probability score. in, The survival probability of a distressed target under single-searcher conditions, based on the above survival probability. The calculation formula is obtained; This represents the total search time, which is equal to the time the target was in distress when the search was completed. This indicates that the search order in path planning affects the cumulative total search time. This represents the total number of steps taken by the searcher over the total search time. For the searcher in the first Position during the step, For the searcher in the first Position during the step; This indicates that the searcher is at the first The distance of the step to all other possible positions. For the searcher in the first Position during the step, Indicates the searcher at the The position at the time of each step is also used to represent all other possible positions in the search path. Each time the searcher moves, they need to calculate the distance to all other possible positions to select the next possible target position and calculate the time for each step. Used to indicate the search path, , Used to indicate the first [item] on the path One possible location for the search; The initial search path is optimized according to the optimization objective and the constraints, including: The nodes of all path segments in the initial search path are reversed sequentially. If the survival probability of the current search path after reversal is higher than that of the search path before reversal, the current search path is retained; otherwise, the current search path is restored to the search path before reversal. Delete each node in the initial search path in sequence. If the current search path after deleting the node still satisfies the constraint and the survival probability is higher than the survival probability of the search path before deleting the node, delete the node; otherwise, keep the node. Replace each original node in the initial search path with its neighbor node in turn. If the current search path after replacing the original node still satisfies the constraint condition and the survival probability is higher than the survival probability of the search path before replacing the original node, replace the original node with its neighbor node; otherwise, retain the original node. Add at least one node that does not belong to the initial search path to the initial search path to increase the survival probability of the initial search path; First, traverse all nodes and select nodes that are not in the current path. Second, randomly select K nodes and estimate their optimal insertion positions, which are the paths that will result in the highest cumulative spatiotemporal survival probability scores after insertion. Finally, insert the nodes and update the path.

2. The maritime target search path planning method as described in claim 1, characterized in that, When the number of searchers is 1, generating an initial search path for the distressed target based at least on the plurality of possible locations includes: Obtain the initial position of the searcher and multiple candidate nodes; Starting from the initial position, based on the multiple possible positions and the survival probability of the distressed target, a candidate node is selected each time and added to the path queue until each possible position is covered by the search radius of the candidate node by at least one searcher, and the initial search path is obtained based on the path queue.

3. The maritime target search path planning method as described in claim 1, characterized in that, When the number of searchers is 1, generating an initial search path for the distressed target based at least on the plurality of possible locations includes: Obtain the initial position of the searcher and multiple candidate nodes; Starting from the initial position, based on the number of possible positions that the searcher can cover with the search radius of the candidate node, one candidate node is selected and added to the path queue each time, until each possible position is covered by the search radius of at least one searcher with the candidate node, and the initial search path is obtained based on the path queue.

4. The maritime target search path planning method as described in claim 1, characterized in that, When the number of searchers is 1, generating an initial search path for the distressed target based at least on the plurality of possible locations includes: Obtain the initial position of the searcher and multiple candidate nodes; Based on the initial position, an initial path is randomly generated and iteratively adjusted. Simultaneously, the survival probability of the current initial path is calculated. If the survival probability of the current initial path is greater than the highest historical survival probability, the current initial path is accepted; otherwise, it is discarded. If the survival probability of the current initial path is less than or equal to the highest historical survival probability, the current initial path is accepted according to a preset probability; otherwise, it is discarded. This process continues until the iteration ends, at which point the current initial path is used as the initial search path. The preset probability decreases as the time of encounter increases.

5. The maritime target search path planning method as described in claim 1, characterized in that, When the number of searchers is greater than 1, generating an initial search path for the distressed target based at least on the plurality of possible locations includes: Obtain the initial position of each searcher and multiple candidate nodes; The initial search path is generated in parallel for each searcher based on the selection of all the candidate nodes; Generating the initial search path for one of multiple searchers includes: Starting from the initial position, based on the survival probability of the distressed target and its distance from the candidate node, a candidate node is selected from the candidate nodes that have not been selected by other searchers and added to the path queue, until each possible position is covered by the search radius of the candidate node by at least one searcher, and the initial search path of the searcher is obtained based on the path queue.

6. A maritime target search path planning device, characterized in that, include: The location prediction module is used to acquire distressed target and environmental data, and predict multiple possible locations of the distressed target based on the environmental data. An initial path generation module is used to generate an initial search path for the distressed target by the searcher based on at least the plurality of possible locations; The path optimization module is used to optimize the initial search path with the optimization objective of maximizing the survival probability of the distressed target at each of the possible locations, and with the constraint that each of the possible locations is covered by the search radius of at least one of the searchers when all the searchers move along the search path, so as to obtain the optimized final search path. The formula for calculating the survival probability include: in, The probability of survival of the distressed target. For proportional parameters, For the time of the incident, For shape parameters, For seawater temperature, and Positive correlation The basic proportional parameters are based on the reference water temperature. The coefficient representing the influence of water temperature. The impact factor of lifesaving equipment, If the distressed target is wearing life-saving equipment, If the distressed target is not wearing life-saving equipment, ; The dynamic adjustment module is used to enable the searcher to search for the distressed target according to the final search path, and to dynamically adjust the survival probability of the distressed target and the final search path according to the seawater temperature, the currently searched possible locations and the time of distress. The constraints also include: The total time for any of the searchers to complete the final search path shall not exceed the maximum survival time of the distressed target, and the final search path shall be a closed path, wherein the maximum survival time of the distressed target is... , The seawater temperature; The initial search path includes multiple nodes; When the number of searchers is 1, a single searcher plans a search path within a given ocean area based on the real-time updated survival probability of the distressed target. The objective function of the single-searcher path planning model is to maximize the total survival probability score. in, The survival probability of a distressed target under single-searcher conditions, based on the above survival probability. The calculation formula is obtained; This represents the total search time, which is equal to the time the target was in distress when the search was completed. This indicates that the search order in path planning affects the cumulative total search time. This represents the total number of steps taken by the searcher over the total search time. For the searcher in the first Position during the step, For the searcher in the first Position during the step; This indicates that the searcher is at the first The distance of the step to all other possible positions. For the searcher in the first Position during the step, Indicates the searcher at the The position at the time of each step is also used to represent all other possible positions in the search path. Each time the searcher moves, they need to calculate the distance to all other possible positions to select the next possible target position and calculate the time for each step. Used to indicate the search path, , Used to indicate the first [item] on the path One possible location for the search; The initial search path is optimized according to the optimization objective and the constraints, including: The nodes of all path segments in the initial search path are reversed sequentially. If the survival probability of the current search path after reversal is higher than that of the search path before reversal, the current search path is retained; otherwise, the current search path is restored to the search path before reversal. Delete each node in the initial search path in sequence. If the current search path after deleting the node still satisfies the constraint and the survival probability is higher than the survival probability of the search path before deleting the node, delete the node; otherwise, keep the node. Replace each original node in the initial search path with its neighbor node in turn. If the current search path after replacing the original node still satisfies the constraint condition and the survival probability is higher than the survival probability of the search path before replacing the original node, replace the original node with its neighbor node; otherwise, retain the original node. Add at least one node that does not belong to the initial search path to the initial search path to increase the survival probability of the initial search path; First, traverse all nodes and select nodes that are not in the current path. Second, randomly select K nodes and estimate their optimal insertion positions, which are the paths that will result in the highest cumulative spatiotemporal survival probability scores after insertion. Finally, insert the nodes and update the path.

7. An electronic device, characterized in that, include: processor; A memory for storing executable instructions; wherein the processor is configured to read the executable instructions from the memory and execute the executable instructions to implement the maritime target search path planning method according to any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, The storage medium stores a computer program that, when executed by a processor, causes the processor to implement the maritime target search path planning method according to any one of claims 1 to 5.

Images