A collision avoidance method for large commercial ships based on collision avoidance rules

Abstract

The present invention discloses a method for autonomous navigation and collision avoidance of large merchant ships based on collision avoidance rule constraints, including obtaining navigation data of a target large merchant ship and hydrological and meteorological environmental data of the waters where the ship is located, generating a static planning route without static collision; selecting collision test points in path nodes, and calculating the abnormal navigation probability of the target large merchant ship according to waters change data; obtaining a dynamic ship domain to assess the collision risk and judge the encounter situation; constructing a ship collision avoidance model, establishing model constraints according to collision avoidance rules, obtaining geometric information in the encounter situation, and importing the geometric information into the ship collision avoidance model to obtain a collision avoidance strategy. The present invention realizes effective obstacle avoidance under the constraints of international collision avoidance rules, introduces a dynamic ship domain to adjust the ship domain according to the encounter situation, and effectively improves the efficiency of autonomous navigation and obstacle avoidance of large merchant ships.

Description

Technical Field

[0001] The present invention relates to the field of ship navigation technology, and more specifically, to an autonomous navigation and collision avoidance method for large commercial ships based on collision avoidance rule constraints. Background Art

[0002] In recent years, with the rapid increase in the size and speed of ships and their number, the problem of navigation safety has become increasingly serious. Among them, maritime ship collision is the main cause of endangering the safety of ship navigation, accounting for a large proportion of maritime ship traffic accidents. Once a ship collides during navigation, the high-value ship and the cargo it transports will suffer losses, and the lives of crew members and passengers will also be threatened. In addition, once a collision accident causes pollution to the marine environment, the cost is often huge. Ship collision avoidance requires comprehensive analysis of various information in a short period of time and consideration of the influence of multiple factors.

[0003] Data from various ship sensors can be directly displayed to the driver through relevant display devices, but it is impossible to provide comprehensive information fusion display and collision avoidance suggestions. Drivers need to evaluate the risk of collision based on subjective experience, especially when entering and leaving the port, they are prone to misjudgment and cause collision accidents. With the breakthrough of new technologies such as data mining, artificial intelligence, sensors, and the Internet of Things, smart ships and smart shipping have become an inevitable trend in the development of the future shipping industry, and the spatiotemporal dynamic prediction of ship collision avoidance is a key issue facing the autonomous navigation of smart ships and collision avoidance decisions of smart shipping in the future. Therefore, how to accurately predict ship collisions, make the prediction results more in line with the actual situation, and avoid decision-making misunderstandings when making collision avoidance decisions is an urgent problem to be solved. Summary of the invention

[0004] In order to solve the above technical problems, the present invention proposes a large merchant ship autonomous navigation collision avoidance method based on collision avoidance rule constraints, which provides a decision-making basis for dynamic collision avoidance of large merchant ships autonomous navigation.

[0005] The first aspect of the present invention provides a large commercial ship autonomous navigation and collision avoidance method based on collision avoidance rule constraints, comprising:

[0006] Obtain the navigation data of the target large commercial ship and the hydrological and meteorological environment data of the waters where it is located, plan the route in the current waters based on the navigation data, and generate a static and collision-free static planning route;

[0007] Divide the path nodes according to the static planned route, select collision test points in the path nodes, obtain water area change data according to the hydrological and meteorological environment data of the collision test points, and calculate the abnormal navigation probability of the target large commercial ship;

[0008] Acquire a dynamic ship area according to the abnormal navigation probability, evaluate the collision risk according to the dynamic ship area, and judge the encounter situation;

[0009] A ship collision avoidance model is constructed, model constraints are established according to collision avoidance rules, geometric information in the encounter situation is obtained, and the geometric information is imported into the ship collision avoidance model to obtain a collision avoidance strategy.

[0010] In this scheme, the navigation data of the target large commercial ship and the hydrological and meteorological environmental data of the waters where it is located are obtained, specifically:

[0011] Obtain the initial heading, speed and basic ship maneuvering parameters of the target large merchant ship according to the preset sensors, obtain the AIS data of the waters where the target large merchant ship is currently located, and perform data standardization and data cleaning on the AIS data;

[0012] Select ship dynamic information and ship static information according to the pre-processed AIS data, divide the current waters into several grid areas, obtain a two-dimensional grid map, and extract the geographical elements of ship dynamic information and ship static information;

[0013] Matching the geographic elements with the two-dimensional grid map to obtain a ship density heat map, presetting a ship density threshold to filter high-flow areas in the ship density heat map, and setting a preset area label in the two-dimensional grid map;

[0014] The hydrological and meteorological environmental data of the current waters are obtained, the two-dimensional grid map is filled with data, and the safe encounter distance in the corresponding area is initialized based on the hydrological and meteorological environmental data and ship flow in different areas.

[0015] In this solution, the route planning in the current waters is performed based on the navigation data to generate a static collision-free static planning route, specifically:

[0016] Using big data to obtain historical navigation data of the current waters, statistically analyzing the historical navigation data to obtain underwater and surface landmark information of the current waters, and marking and displaying the underwater and surface landmark information as static obstacles on a two-dimensional grid map of the current waters;

[0017] The genetic algorithm is used to optimize the whale algorithm to build a route planning model. The navigation data of the target large commercial ship and the annotated two-dimensional grid map are encoded as the model input, and the length and smoothness of the planned path are used as the fitness function.

[0018] Mark the optimal whale position in the route planning model, update the whale position through genetic algorithm selection, mutation and crossover combined with adaptive weights, and preset safe navigation constraints in the neighborhood of static obstacles during the whale position update process;

[0019] The heading is guided according to the safety navigation constraints, and when the iteration is satisfied, an updated optimal whale position is obtained, and a statically planned route is generated through the path points corresponding to the optimal whale position.

[0020] In this scheme, in the annotated two-dimensional grid map corresponding to the current waters, the navigation direction of the ship in the vicinity of the static obstacle is preset, and the weight information is set according to the hydrological and meteorological environmental data of different areas, the ship density thermal value and the captain parameter;

[0021] The weight information is used to weight the initialized safe encounter distance to obtain a safe distance, and a safe navigation constraint of a static obstacle neighborhood is generated according to the navigation direction of the ship and the safe distance.

[0022] In this solution, a collision test point is selected in the path node, and water area change data is obtained according to the hydrological and meteorological environment data of the collision test point, specifically:

[0023] The statically planned route of the target large commercial ship in the current waters is divided to generate path nodes, and the selection weight is set according to the ship density in the area where the path node is located. The larger the selection weight, the more collision test points are selected;

[0024] Acquire wave dispersion data according to the hydrological and meteorological environmental data of the collision test points of different path nodes, construct a wave dispersion data sequence and a meteorological data sequence of the collision test points, and import the wave dispersion data sequence and the meteorological data sequence into the Bi-LSTM network;

[0025] The attention mechanism is introduced into the Bi-LSTM network to extract the time series features of wave dispersion and meteorology. The attention weight is used to highlight the time series features that are too different from other features. The water area change data is obtained based on the weighted wave dispersion time series features and meteorological time series features.

[0026] Constructing a search tag according to the water area change data of the current water area, using the search tag to perform data search in a preset data space to obtain a ship navigation data set, and marking abnormal navigation data in the ship navigation data set;

[0027] The abnormal navigation probability of the target large commercial ship is obtained according to the data volume ratio of the abnormal navigation data to the ship navigation data set.

[0028] In this solution, the dynamic ship area is obtained according to the abnormal navigation probability, the collision risk is evaluated according to the dynamic ship area, and the encounter situation is judged, specifically:

[0029] The abnormal navigation probability is compared with a preset probability threshold, and if the abnormal navigation probability is less than or equal to the preset probability threshold, a dynamic ship area is generated according to the collision avoidance size changes of the bow and the stern;

[0030] If the abnormal navigation probability is greater than a preset probability threshold, a dynamic ship domain gain is generated, and the dynamic ship domain of the target large commercial ship is adjusted using the dynamic ship domain gain;

[0031] According to the dynamic ship domain of the target large merchant ship and other ships, the minimum encounter distance, minimum encounter time and relative distance are obtained, and the index weights of the minimum encounter distance, minimum encounter time and relative distance are obtained to calculate the collision risk between the target large merchant ship and other ships, and the targets with collision danger are marked;

[0032] The encounter characteristics of the marked ships are obtained, the encounter situation stipulated by the collision avoidance rules is identified according to the encounter characteristics, and the size of the dynamic ship area is adjusted according to the encounter situation.

[0033] In this scheme, the minimum encounter distance, minimum encounter time and relative distance are obtained based on the dynamic ship domain of the target large merchant ship and other ships for analysis, and the encounter characteristics are obtained by comparing with the safe encounter distance. The navigation situation of other ships' positions and motion parameters is predicted to judge the encounter situation, and the encounter situation agreed upon by the collision avoidance rules is identified based on the encounter characteristics and encounter situation.

[0034] In this scheme, a ship collision avoidance model is constructed, specifically:

[0035] Acquire the international collision avoidance rules for knowledge extraction, perform structured processing on the extracted collision avoidance rules knowledge, acquire historical collision avoidance data of ships, divide the historical collision avoidance data by using the collision avoidance rules knowledge, and generate collision avoidance experience data subsets for different encounter situations;

[0036] Building a ship collision avoidance model based on deep learning, using a subset of collision avoidance experience data corresponding to collision avoidance rule knowledge to train the model, generating model classification constraints, and obtaining the encounter attributes of the marked ships according to the model classification constraints;

[0037] Extracting change features according to the time series of the dynamic ship domain changes of the marked ship, using the change features to predict the movement trend of the marked ship, and judging the avoidance attribute of the target large merchant ship to the marked ship through the movement trend and the encounter attribute;

[0038] Determine the avoidance direction based on the avoidance attribute and the relative angle between the target large merchant ship and the marked ship, and judge whether the dynamic ship domain of the marked ship is greater than the safe encounter distance. If less than, adjust according to the safe encounter distance;

[0039] The target path point is selected in the static planning route, and the obstacle avoidance path planning is performed outside the dynamic ship domain of the marked ship according to the current position of the target large merchant ship and the target path point.

[0040] The present invention discloses a method for autonomous navigation and collision avoidance of large merchant ships based on collision avoidance rule constraints, including obtaining navigation data of a target large merchant ship and hydrological and meteorological environmental data of the waters where the ship is located, generating a static planning route without collision; selecting collision test points in path nodes, and calculating the abnormal navigation probability of the target large merchant ship according to waters change data; obtaining a dynamic ship domain to assess the collision risk and judge the encounter situation; constructing a ship collision avoidance model, establishing model constraints according to collision avoidance rules, obtaining geometric information in the encounter situation, and importing the geometric information into the ship collision avoidance model to obtain a collision avoidance strategy. The present invention realizes effective obstacle avoidance under the constraints of international collision avoidance rules, introduces a dynamic ship domain to adjust the ship domain according to the encounter situation, accurately analyzes the risk of ship collision, effectively improves the efficiency of autonomous navigation and obstacle avoidance of large merchant ships, and the collision avoidance process complies with the International Regulations for Preventing Collisions at Sea, ensuring the standardization and safety of collision avoidance actions. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 A flow chart of a method for autonomous navigation and collision avoidance of a large commercial ship based on collision avoidance rule constraints of the present invention is shown;

[0042] Figure 2 The flowchart of the present invention for obtaining a statically planned route without collision in the current waters is shown;

[0043] Figure 3 The flow chart of constructing a ship collision avoidance model according to the present invention is shown;

[0044] Figure 4 A block diagram of a large commercial ship autonomous navigation and collision avoidance system based on collision avoidance rule constraints of the present invention is shown. DETAILED DESCRIPTION

[0045] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0046] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the protection scope of the present invention is not limited to the specific embodiments disclosed below.

[0047] Figure 1 A flow chart of a method for autonomous navigation and collision avoidance of a large commercial ship based on collision avoidance rule constraints of the present invention is shown.

[0048] like Figure 1 As shown, the first aspect of the present invention provides a large commercial ship autonomous navigation and collision avoidance method based on collision avoidance rule constraints, comprising:

[0049] S102, obtaining navigation data of the target large commercial ship and hydrological and meteorological environmental data of the waters where the ship is located, planning a route in the current waters based on the navigation data, and generating a statically planned route without collision;

[0050] S104, dividing the path nodes according to the static planned route, selecting collision test points in the path nodes, obtaining water area change data according to the hydrological and meteorological environment data of the collision test points, and calculating the abnormal navigation probability of the target large commercial ship;

[0051] S106, acquiring a dynamic ship area according to the abnormal navigation probability, evaluating a collision risk according to the dynamic ship area, and determining an encounter situation;

[0052] S108, constructing a ship collision avoidance model, establishing model constraints according to collision avoidance rules, obtaining geometric information in the encounter situation, and importing the geometric information into the ship collision avoidance model to obtain a collision avoidance strategy.

[0053] It should be noted that the initial heading, speed and basic ship maneuvering parameters of the target large merchant ship are obtained from navigation equipment such as ship-borne radar, gyrocompass, speedometer, GPS, ECDIS, visibility meter, etc., and the AIS data of the current waters where the target large merchant ship is located is obtained, and the AIS data is standardized to achieve the standardization of the AIS information sub-segment name, and the unreasonable data is removed or modified by data cleaning, and the missing data is interpolated; the ship dynamic information and ship static information are selected according to the pre-processed AIS data, including the number of ships, ship spacing, speed distribution, track distribution, etc.; the current waters are divided into several grid areas based on the nautical chart data of the current waters, and a two-dimensional grid map is obtained to extract the ship dynamic information and ship static information. The geographical elements of the static information of the ship, including the ship's deviation angle, the ship's position and the distance between the ship and the channel boundary, etc.; the geographical elements are matched with the two-dimensional grid map to obtain a ship density heat map, and a preset ship density threshold is used to filter high-flow areas in the ship density heat map, and preset area labels are set in the two-dimensional grid map; and the hydrological and meteorological environmental data of the current waters are obtained, and the two-dimensional grid map is filled with data, and the safe encounter distance in the corresponding area is initialized based on the hydrological and meteorological environmental data and ship flow in different areas. When there is an abnormal weather, abnormal wave dispersion or channel congestion, the corresponding safe encounter distance is used. In addition, the safe encounter distance can also be adaptively adjusted using the ship type, speed and situation.

[0054] Figure 2 The flowchart of the present invention for obtaining a statically planned route without collision in the current water area is shown.

[0055] According to an embodiment of the present invention, a route planning is performed in the current waters based on the navigation data to generate a static collision-free static planning route, specifically:

[0056] S202, using big data means to obtain historical navigation data of the current waters, performing statistical analysis based on the historical navigation data to obtain underwater and surface landmark information of the current waters, and marking and displaying the underwater and surface landmark information as static obstacles on a two-dimensional grid map of the current waters;

[0057] S204, using a genetic algorithm to optimize the whale algorithm to construct a route planning model, encoding the navigation data of the target large commercial ship and the annotated two-dimensional grid map as model input, and using the length and smoothness of the planned path as a fitness function;

[0058] S206, marking the optimal whale position in the route planning model, updating the whale position through selection, mutation and crossover of the genetic algorithm combined with adaptive weights, and in the process of updating the whale position, presetting the safe navigation constraints of the static obstacle neighborhood;

[0059] S208, guiding the heading according to the safety navigation constraint, obtaining an updated optimal whale position when the iteration is satisfied, and generating a statically planned route through the path points corresponding to the optimal whale position.

[0060] It should be noted that the whale algorithm achieves optimization by simulating the foraging strategy of humpback whales, introduces the selection, crossover and mutation steps in the genetic algorithm into the whale algorithm, mutates the optimal whale position in a good direction, and uses the global optimal whale position to perform crossover operations with individual whales to generate new individuals, improves the global search capability, and introduces adaptive weight update to optimize the local search capability, and adaptively mutates and updates the output value to solve the problem that the whale algorithm is prone to fall into the local optimum; according to the historical routes of ships of different modes, several straight path segments from the starting point to the end point that do not cross obstacles are generated in the current waters, and the initial whale population of the collision-free path is obtained, which improves the quality of the initial population while ensuring diversity, and ensures the search efficiency of the subsequent whale algorithm. The length and smoothness of the planned path are used as the fitness function. A shorter path can improve the traffic efficiency of large merchant ships and ensure economy, while the large-scale turning of large merchant ships is likely to cause path point deviation. The better the smoothness, the smaller the amplitude of large merchant ships when performing the turning.

[0061] Due to factors such as the length, width and maneuvering characteristics of large merchant ships, if the selected path is too close, it is easy to collide with existing obstacles. In the annotated two-dimensional grid map corresponding to the current waters, the ship's navigation direction in the vicinity of static obstacles is preset, and weight information is set according to the hydrological and meteorological environmental data, ship density thermal values ​​and captain parameters of different regions; the weight information is used to weight the initialized safe encounter distance to obtain the safe distance, and the safe navigation constraints of the static obstacle neighborhood are generated according to the ship's navigation direction and the safe distance. In addition, the path points corresponding to the obtained whale position are simplified, and redundant path points are removed to obtain the best static planning route.

[0062] It should be noted that the statically planned route of the target large merchant ship in the current waters is divided to generate several route segments as path nodes, and the selection weight is set according to the ship density in the area where the path node is located. The larger the selection weight, the more collision test points are selected; wave dispersion data is obtained according to the hydrological and meteorological environmental data of the collision test points of different path nodes, and the wave dispersion data sequence and meteorological data sequence of the collision test points are constructed, and the wave dispersion data sequence and meteorological data sequence are imported into the Bi-LSTM network; the attention mechanism is introduced into the Bi-LSTM network to extract the time series characteristics of wave dispersion and meteorology, and the attention weight is used to highlight the time series characteristics that are too different from other characteristics. The water area change data is obtained according to the weighted wave dispersion time series characteristics and meteorological time series characteristics. The water area change data reflects the characteristic information of abnormal wave dispersion data. Irregular abnormal wave dispersion data can easily cause the hull of large merchant ships to sway, the movement to become unstable, and deviate from the existing statically planned route. A retrieval tag is constructed according to the water area change data of the current water area, and the retrieval tag is used to perform data retrieval in a preset data space to obtain a ship navigation data set, and abnormal navigation data is marked in the ship navigation data set; the abnormal navigation probability of the target large commercial ship is obtained according to the data volume ratio of the abnormal navigation data and the ship navigation data set.

[0063] It should be noted that the abnormal navigation probability is compared with the preset probability threshold. If the abnormal navigation probability is less than or equal to the preset probability threshold, a dynamic ship domain is generated according to the semi-ellipse corresponding to the change in the bow collision avoidance size (course change) and the semi-circle corresponding to the change in the stern collision avoidance size; the dynamic ship domain is related to the ship length, the relative position between the two ships and the relative speed; if the abnormal navigation probability is greater than the preset probability threshold, a dynamic ship domain gain is generated, and the dynamic ship domain of the target large merchant ship is adjusted using the dynamic ship domain gain. The instability of the swaying motion of a large merchant ship can easily cause the ship to deviate, resulting in an increase in the dynamic ship domain.

[0064] Obtain the minimum encounter distance based on the dynamic ship area of ​​the target large merchant ship and other ships , minimum encounter time and relative distance, Indicates the location of the target large merchant ship and other ships. It represents the speed of the target large merchant ship and other ships; and uses the attention mechanism to obtain the index weights of the minimum encounter distance, minimum encounter time and relative distance, calculates the collision risk between the target large merchant ship and other ships, and marks the targets with collision danger. When the collision risk is greater than the preset risk threshold, it enters the collision avoidance state.

[0065] The encounter characteristics of the marked ship are obtained, and the encounter situations agreed upon by the collision avoidance rules are identified according to the encounter characteristics, including the encounter situation, the cross encounter situation and the overtaking situation, and the size of the dynamic ship area is adjusted according to the encounter situation. The minimum encounter distance, the minimum encounter time and the relative distance are obtained based on the dynamic ship area of ​​the target large merchant ship and other ships for analysis, and the encounter characteristics are obtained by comparing with the safe encounter distance. The navigation situation of other ships' positions, relative azimuths, relative speeds and other motion parameters are predicted to determine the encounter situation, and the encounter situation agreed upon by the collision avoidance rules is identified according to the encounter characteristics and encounter situation.

[0066] Figure 3 A flow chart of constructing a ship collision avoidance model according to the present invention is shown.

[0067] According to an embodiment of the present invention, a ship collision avoidance model is constructed, specifically:

[0068] S302: obtaining international collision avoidance rules for knowledge extraction, performing structured processing on the extracted collision avoidance rule knowledge, obtaining historical collision avoidance data of ships, dividing the historical collision avoidance data by using the collision avoidance rule knowledge, and generating collision avoidance experience data subsets for different encounter situations;

[0069] S304, constructing a ship collision avoidance model based on deep learning, using a subset of collision avoidance experience data corresponding to the collision avoidance rule knowledge to perform model training, generating model classification constraints, and obtaining encounter attributes of the marked ship, such as an overtaking ship or a overtaken ship, according to the model classification constraints;

[0070] S306, extracting change features according to the dynamic ship domain change time series of the marked ship, using the change features to predict the movement trend of the marked ship, and judging the avoidance attribute of the target large merchant ship to the marked ship, such as giving way to the left or giving way to the right, according to the movement trend and the encounter attribute;

[0071] S308, determining the avoidance direction based on the avoidance attribute and the relative angle between the target large merchant ship and the marked ship, and judging whether the dynamic ship domain of the marked ship is greater than the safe encounter distance, and if so, adjusting according to the safe encounter distance;

[0072] S310, selecting a target path point in the statically planned route, and performing obstacle avoidance path planning outside the dynamic ship domain of the marked ship according to the current position of the target large merchant ship and the target path point.

[0073] It should be noted that a ship collision avoidance model is constructed based on deep learning methods such as deep convolutional neural networks and deep Q networks, and a local obstacle avoidance path optimization is performed using genetic algorithms and other methods based on the current position of the target large commercial ship and the target path point in the waters outside the dynamic ship domain of the marked ship. When there are multiple dangerous ships, an interactive communication network of multiple ships is constructed, and the avoidance attributes of the ships are shared through the interactive communication network. The corresponding ships are classified according to the avoidance attributes, and the collision avoidance conflict ships are screened according to the classification results. The collision avoidance conflict ships are selected for coordination, and the navigation speed, length and collision risk of the collision avoidance conflict ships are obtained to determine the collision avoidance priority. The collision avoidance priority is sorted, and the collision avoidance heading of the collision avoidance conflict ship with the highest priority is obtained. The remaining collision avoidance conflict ships make coordinated collision avoidance decisions based on the collision avoidance heading. On the premise of ensuring that the collision avoidance conflict ship with the highest priority successfully avoids collision, the remaining ships achieve collaborative collision avoidance through a single turn. During the voyage of a ship, since frequent changes in speed will cause damage to the engine, the crew often avoids collisions by changing the course rather than the speed. The higher the speed, the worse the braking, so the collision avoidance priority is the highest. In addition, the longer the length of the ship and the greater the change in course, the higher the priority of collision avoidance.

[0074] Figure 4 A block diagram of a large commercial ship autonomous navigation and collision avoidance system based on collision avoidance rule constraints of the present invention is shown.

[0075] The second aspect of the present invention further provides a large merchant ship autonomous navigation collision avoidance system 4 based on collision avoidance rule constraints, the system comprising: a memory 41, a processor 42, the memory comprising a large merchant ship autonomous navigation collision avoidance method program based on collision avoidance rule constraints, the large merchant ship autonomous navigation collision avoidance method program based on collision avoidance rule constraints when executed by the processor implements the following steps:

[0076] Obtain the navigation data of the target large commercial ship and the hydrological and meteorological environment data of the waters where it is located, plan the route in the current waters based on the navigation data, and generate a static and collision-free static planning route;

[0077] Divide the path nodes according to the static planned route, select collision test points in the path nodes, obtain water area change data according to the hydrological and meteorological environment data of the collision test points, and calculate the abnormal navigation probability of the target large commercial ship;

[0078] Acquire a dynamic ship area according to the abnormal navigation probability, evaluate the collision risk according to the dynamic ship area, and judge the encounter situation;

[0079] A ship collision avoidance model is constructed, model constraints are established according to collision avoidance rules, geometric information in the encounter situation is obtained, and the geometric information is imported into the ship collision avoidance model to obtain a collision avoidance strategy.

[0080] The third aspect of the present invention also provides a computer-readable storage medium, which includes a program for the method for autonomous navigation and collision avoidance of large merchant ships based on collision avoidance rule constraints. When the program for the method for autonomous navigation and collision avoidance of large merchant ships based on collision avoidance rule constraints is executed by a processor, the steps of the method for autonomous navigation and collision avoidance of large merchant ships based on collision avoidance rule constraints as described in any one of the above items are implemented.

[0081] In the several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as: multiple units or components can be combined, or can be integrated into another system, or some features can be ignored or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed can be through some interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0082] The units described above as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units; they may be located in one place or distributed on multiple network units; some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0083] In addition, all functional units in the embodiments of the present invention may be integrated into one processing unit, or each unit may be separately used as a unit, or two or more units may be integrated into one unit; the above-mentioned integrated units may be implemented in the form of hardware or in the form of hardware plus software functional units.

[0084] A person of ordinary skill in the art can understand that: all or part of the steps of implementing the above method embodiment can be completed by hardware related to program instructions, and the aforementioned program can be stored in a computer-readable storage medium. When the program is executed, it executes the steps of the above method embodiment; and the aforementioned storage medium includes: a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and other media that can store program codes.

[0085] Alternatively, if the above-mentioned integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the embodiment of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, server or network device, etc.) to execute all or part of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as mobile storage devices, ROM, RAM, magnetic disks or optical disks.

[0086] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A large commercial ship autonomous navigation and collision avoidance method based on collision avoidance rule constraints, characterized in that: The following steps are involved: Obtain the navigation data of the target large commercial ship and the hydrological and meteorological environment data of the waters where it is located, plan the route in the current waters based on the navigation data, and generate a static and collision-free static planning route; Divide the path nodes according to the static planned route, select collision test points in the path nodes, obtain water area change data according to the hydrological and meteorological environment data of the collision test points, and calculate the abnormal navigation probability of the target large commercial ship; Acquire a dynamic ship area according to the abnormal navigation probability, evaluate the collision risk according to the dynamic ship area, and judge the encounter situation; Constructing a ship collision avoidance model, establishing model constraints according to collision avoidance rules, obtaining geometric information in the encounter situation, and importing the geometric information into the ship collision avoidance model to obtain a collision avoidance strategy; Based on the navigation data, the route planning in the current waters is performed to generate a static collision-free static planning route, specifically: Using big data to obtain historical navigation data of the current waters, statistically analyzing the historical navigation data to obtain underwater and surface landmark information of the current waters, and marking and displaying the underwater and surface landmark information as static obstacles on a two-dimensional grid map of the current waters; The genetic algorithm is used to optimize the whale algorithm to build a route planning model. The navigation data of the target large commercial ship and the annotated two-dimensional grid map are encoded as the model input, and the length and smoothness of the planned path are used as the fitness function. Mark the optimal whale position in the route planning model, update the whale position through genetic algorithm selection, mutation and crossover combined with adaptive weights, and preset safe navigation constraints in the neighborhood of static obstacles during the whale position update process; The heading is guided according to the safety navigation constraints, and when the iteration is satisfied, an updated optimal whale position is obtained, and a statically planned route is generated through the path points corresponding to the optimal whale position.

2. According to claim 1, a large commercial ship autonomous navigation and collision avoidance method based on collision avoidance rule constraints is characterized in that: Obtain the navigation data of the target large commercial ship and the hydrological and meteorological environmental data of the waters where it is located, specifically: Obtain the initial heading, speed and basic ship maneuvering parameters of the target large merchant ship according to the preset sensors, obtain the AIS data of the waters where the target large merchant ship is currently located, and perform data standardization and data cleaning on the AIS data; Select ship dynamic information and ship static information according to the pre-processed AIS data, divide the current waters into several grid areas, obtain a two-dimensional grid map, and extract the geographical elements of ship dynamic information and ship static information; Matching the geographic elements with the two-dimensional grid map to obtain a ship density heat map, presetting a ship density threshold to filter high-flow areas in the ship density heat map, and setting a preset area label in the two-dimensional grid map; The hydrological and meteorological environmental data of the current waters are obtained, the two-dimensional grid map is filled with data, and the safe encounter distance in the corresponding area is initialized based on the hydrological and meteorological environmental data and ship flow in different areas.

3. The method for autonomous navigation and collision avoidance of large commercial ships based on collision avoidance rule constraints according to claim 1 is characterized in that: In the annotated two-dimensional grid map corresponding to the current waters, the navigation direction of the ship in the vicinity of the static obstacle is preset, and the weight information is set according to the hydrological and meteorological environmental data of different areas, the ship density thermal value and the captain parameters; The weight information is used to weight the initialized safe encounter distance to obtain a safe distance, and a safe navigation constraint of a static obstacle neighborhood is generated according to the navigation direction of the ship and the safe distance.

4. The method for autonomous navigation and collision avoidance of large commercial ships based on collision avoidance rule constraints according to claim 1 is characterized in that: A collision test point is selected in the path node, and water area change data is obtained according to the hydrological and meteorological environment data of the collision test point, specifically: The statically planned route of the target large commercial ship in the current waters is divided to generate path nodes, and the selection weight is set according to the ship density in the area where the path node is located. The larger the selection weight, the more collision test points are selected; Acquire wave dispersion data according to the hydrological and meteorological environmental data of the collision test points of different path nodes, construct a wave dispersion data sequence and a meteorological data sequence of the collision test points, and import the wave dispersion data sequence and the meteorological data sequence into the Bi-LSTM network; The attention mechanism is introduced into the Bi-LSTM network to extract the time series features of wave dispersion and meteorology. The attention weight is used to highlight the time series features that are too different from other features. The water area change data is obtained based on the weighted wave dispersion time series features and meteorological time series features. Constructing a search tag according to the water area change data of the current water area, using the search tag to perform data search in a preset data space to obtain a ship navigation data set, and marking abnormal navigation data in the ship navigation data set; The abnormal navigation probability of the target large commercial ship is obtained according to the data volume ratio of the abnormal navigation data and the ship navigation data set.

5. The method for autonomous navigation and collision avoidance of large commercial ships based on collision avoidance rule constraints according to claim 1 is characterized in that: The dynamic ship area is obtained according to the abnormal navigation probability, the collision risk is evaluated according to the dynamic ship area, and the encounter situation is judged, specifically: The abnormal navigation probability is compared with a preset probability threshold, and if the abnormal navigation probability is less than or equal to the preset probability threshold, a dynamic ship area is generated according to the collision avoidance size changes of the bow and the stern; If the abnormal navigation probability is greater than a preset probability threshold, a dynamic ship domain gain is generated, and the dynamic ship domain of the target large commercial ship is adjusted using the dynamic ship domain gain; According to the dynamic ship domain of the target large merchant ship and other ships, the minimum encounter distance, minimum encounter time and relative distance are obtained, and the index weights of the minimum encounter distance, minimum encounter time and relative distance are obtained to calculate the collision risk between the target large merchant ship and other ships, and the targets with collision danger are marked; The encounter characteristics of the marked ships are obtained, the encounter situation stipulated by the collision avoidance rules is identified according to the encounter characteristics, and the size of the dynamic ship area is adjusted according to the encounter situation.

6. The method for autonomous navigation and collision avoidance of large commercial ships based on collision avoidance rule constraints according to claim 5 is characterized in that: Based on the dynamic ship domain of the target large merchant ship and other ships, the minimum encounter distance, minimum encounter time and relative distance are obtained for analysis, and the encounter characteristics are obtained by comparing with the safe encounter distance. The navigation situation of other ships' positions and motion parameters is predicted to judge the encounter situation, and the encounter situation agreed upon by the collision avoidance rules is identified based on the encounter characteristics and encounter situation.

7. The method for autonomous navigation and collision avoidance of large commercial ships based on collision avoidance rule constraints according to claim 1 is characterized in that: Construct a ship collision avoidance model, specifically: Acquire the international collision avoidance rules for knowledge extraction, perform structured processing on the extracted collision avoidance rules knowledge, acquire historical collision avoidance data of ships, divide the historical collision avoidance data by using the collision avoidance rules knowledge, and generate collision avoidance experience data subsets for different encounter situations; Building a ship collision avoidance model based on deep learning, using a subset of collision avoidance experience data corresponding to collision avoidance rule knowledge to train the model, generating model classification constraints, and obtaining the encounter attributes of the marked ships according to the model classification constraints; Extracting change features according to the time series of the dynamic ship domain changes of the marked ship, using the change features to predict the movement trend of the marked ship, and judging the avoidance attribute of the target large merchant ship to the marked ship through the movement trend and the encounter attribute; Determine the avoidance direction based on the avoidance attribute and the relative angle between the target large merchant ship and the marked ship, and judge whether the dynamic ship domain of the marked ship is greater than the safe encounter distance. If less than, adjust according to the safe encounter distance; The target path point is selected in the static planning route, and the obstacle avoidance path planning is performed outside the dynamic ship domain of the marked ship according to the current position of the target large merchant ship and the target path point.

Images