Ship berthing scheme generation method and system, electronic device and storage medium

Abstract

The present application relates to the ship technology field, propose a kind of ship berthing scheme generation method and system, electronic equipment and storage medium.The ship berthing scheme generation method includes: collecting port structure parameter information, the environmental data information of ship, the dynamic information of adjacent ship and the state information of ship itself;According to the first port model determined according to port structure parameter information;According to the first port model, respectively determine port agent, tugboat agent and the ship agent;According to the three-party game matrix constructed according to port agent, tugboat agent and the ship agent;According to the first ship berthing scheme determined according to port structure parameter information, the environmental data information of ship, the dynamic information of adjacent ship, the state information of ship itself and three-party game matrix;According to the second ship berthing scheme determined according to first ship berthing scheme and neural network model.The present application improves the accuracy of ship berthing.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ships, in particular to a ship berthing scheme generation method and system, an electronic device and a storage medium. BACKGROUND

[0002] In the related art, ship berthing mainly focuses on single-dimensional data analysis or static risk assessment, and representative technologies include: environment monitoring oriented, berth matching oriented, and risk assessment oriented, etc. However, due to the use of single-dimensional data analysis or static risk assessment, the decision information is one-sided, resulting in low accuracy of the final ship berthing scheme. SUMMARY

[0003] The present application aims to at least solve one of the problems in the prior art or related art.

[0004] To this end, a first aspect of the present application provides a ship berthing scheme generation method.

[0005] A second aspect of the present application provides a ship berthing scheme generation system.

[0006] A third aspect of the present application provides an electronic device.

[0007] A fourth aspect of the present application provides a storage medium.

[0008] Therefore, according to a first aspect of the present application, a ship berthing scheme generation method is provided, comprising: collecting port structure parameter information, ship environmental data information, adjacent ship dynamic information and ship state information; determining a first port model according to the port structure parameter information; determining a port agent, a tug agent and a ship agent according to the first port model, wherein the port agent aims to balance the load, the tug agent aims to minimize the sum of fuel consumption and path cost, and the ship agent aims to shorten the berthing time and reduce the safety risk; constructing a three-party game matrix according to the port agent, the tug agent and the ship agent; determining a first ship berthing scheme according to the port structure parameter information, the ship environmental data information, the adjacent ship dynamic information, the ship state information and the three-party game matrix; and determining a second ship berthing scheme according to the first ship berthing scheme and a neural network model.

[0009] The application provides a ship berthing scheme generation method, mainly comprising the following steps: firstly, collecting port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information, wherein the port structure parameter information can comprise: port fender coordinates, port mooring pile layout and berth water depth; the ship environment data information can comprise: wind flow pressure difference, wave height and visibility; the adjacent ship dynamic information refers to the dynamic information of a ship located within a preset range of the ship, and can comprise: speed, heading and draft; and the ship state information can comprise: ship attitude, ship propeller power and rudder angle. Further, the above data information can be collected in real time by a laser radar (LiDAR, Light Detection and Ranging), a millimeter wave radar, an AIS (Automatic Identification System, ship automatic identification system) receiver and a weather station. After obtaining the above data information, a first port model is determined according to the port structure parameters, that is, a three-dimensional model of the port where the ship is to be berthed is determined according to the port structure parameters. Then, a port agent, a tugboat agent and a ship agent are determined according to the first port model, wherein the agent is an entity capable of perceiving the environment, making autonomous decisions and performing actions to achieve a specific goal, therefore, the specific goal of the port agent is load balancing, the specific goal of the tugboat agent is to minimize the sum of fuel consumption and path cost, and the specific goal of the ship agent is to shorten the berthing time and reduce the safety risk. By determining the port agent, the tugboat agent and the ship agent by using the first port model, the interests of each party can be clearly expressed, which lays a foundation for game deduction, that is, the load, fuel consumption and path are considered in multiple aspects. After obtaining the port agent, the tugboat agent and the ship agent, a three-party game matrix is constructed according to the port agent, the tugboat agent and the ship agent, and the three-party game matrix is used to analyze the strategy interaction behavior and result of three independent decision-making subjects (the port agent, the tugboat agent and the ship agent). After obtaining the three-party game matrix, the collected port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information are input into the three-party game matrix, so as to obtain a first ship berthing scheme that can meet the goals of the port agent, the tugboat agent and the ship agent. Finally, a neural network model is used to optimize the first ship berthing scheme, wherein the neural network model is trained based on historical accident data, and can balance some target contradictions in the first ship berthing scheme, for example, the contradictions among safety, efficiency and fuel saving, so as to output an optimal ship berthing scheme, that is, a second ship berthing scheme.The application breaks through the limitation of related technologies relying only on wind speed or berth state by collecting port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information, realizes port-environment-ship full-element data fusion, provides global input for game deduction, further, realizes global optimization of ship berthing scheme through determining multiple intelligent agents and realizing ship berthing scheme global optimization through multi-agent cooperation and dynamic game mechanism, thereby improving the accuracy of the ship berthing scheme.

[0010] In some technical solutions, the step of determining the first port model according to the port structure parameter information comprises: acquiring a port structure feature database, wherein the port structure feature database comprises a plurality of second port models and first port structure feature data corresponding to the second port models; performing geometric feature extraction on the port structure parameter information to obtain a plurality of second data information; performing coordinate mapping on the plurality of second data information to obtain a plurality of second port structure feature data; and determining the first port model from the plurality of second port models according to the plurality of second port structure feature data and the port structure feature database.

[0011] In the technical solution, the step of determining the first port model according to the port structure parameter information comprises: first acquiring a port structure feature database, wherein the port structure feature database comprises a plurality of second port models and first port structure feature data corresponding to the second port models. Then, the virtual anchor algorithm (VAA) is used to process the collected port structure parameter information, that is, first performing geometric feature extraction on the port structure parameter information to obtain a plurality of first data information, then performing coordinate mapping on the plurality of first data information to obtain a plurality of second port structure feature data, wherein the geometric feature can be coordinates and / or intensity, etc. The second data information can be converted into standardized virtual anchor points through coordinate mapping. Then, the first port model is determined from the plurality of second port models according to the plurality of second port structure feature data and the port structure feature database, that is, the second port structure feature data is first matched with the first port structure feature information in the port structure feature database, if there is similar first port structure feature information to the second port structure feature data, the second port model corresponding to the first port structure feature information is taken as the first port model. The virtual anchor algorithm is used to realize the digitization of the port structure parameter information, and then combined with the adaptive template library, that is, the port structure feature database, thereby solving the problem of high cross-port deployment cost in related technologies.

[0012] In some embodiments, the step of constructing the three-party game matrix according to the port agent, the tugboat agent and the ship agent comprises: determining a port strategy space according to the port agent; generating a first utility function according to the port strategy space; determining a tugboat strategy space according to the tugboat agent; generating a second utility function according to the tugboat strategy space; defining a ship strategy space according to the ship agent; generating a third utility function according to the ship strategy space; and constructing the three-party game matrix according to the first utility function, the second utility function and the third utility function.

[0013] In the embodiments, the step of constructing the three-party game matrix according to the port agent, the tugboat agent and the ship agent comprises: firstly, determining a port strategy space according to the port agent, wherein the strategy space refers to a set of all possible strategies in a specific problem, and since the port agent aims to balance the load, the port strategy space includes all possible combinations of the load and the bollard in the port; then, generating a first utility function according to the port strategy space, wherein the first utility function aims to balance the load of the bollard and can minimize the load variance; simultaneously, determining a tugboat strategy space according to the tugboat agent, wherein the tugboat agent aims to minimize the sum of the fuel consumption and the path cost, and thus the tugboat strategy space includes all paths and the thrust of each path; then, generating a second utility function according to the tugboat strategy space, wherein the second utility function aims to minimize the sum of the fuel consumption and the path cost and can minimize the sum of the fuel consumption and the path cost; simultaneously, determining a ship strategy space according to the ship agent, wherein the ship agent aims to shorten the berthing time and reduce the safety risk, and thus the ship strategy space includes all combinations of the berthing speed and the rudder angle; then, generating a third utility function according to the ship agent, wherein the third utility function aims to shorten the berthing time and reduce the safety risk and can shorten the berthing time and reduce the safety risk; and finally, constructing the three-party game matrix according to the first utility function, the second utility function and the third utility function. By defining the utility functions of the three agents, the port utility function (i.e., the first utility function) aims to balance the load, the tugboat utility function (i.e., the second utility function) aims to minimize the sum of the fuel consumption and the path cost, and the ship utility function (i.e., the third utility function) aims to shorten the berthing time and reduce the safety risk, and then constructing the three-party game matrix according to the utility functions of the three agents, a collaborative berthing scheme acceptable to all parties can be output, and the efficiency loss caused by the resource contention can be avoided.

[0014] In some embodiments, the step of determining the first ship berthing scheme according to the port structure parameter information, the ship environment data information, the adjacent ship dynamic information, the ship state information and the three-party game matrix comprises: inputting the port structure parameter information, the ship environment data information and the adjacent ship dynamic information into the three-party game matrix respectively to obtain a first data matrix; and solving the first data matrix by using a Nash equilibrium solver to obtain the first ship berthing scheme.

[0015] In this embodiment, the step of determining the first ship berthing scheme according to the port structure parameter information, the ship environment data information, the adjacent ship dynamic information, the ship state information and the three-party game matrix comprises: firstly inputting the port structure parameter information, the ship environment data information and the adjacent ship dynamic information into the three-party game matrix to obtain a first data matrix; inputting the port structure parameter information, the ship environment data information and the adjacent ship dynamic information into the three-party game matrix, so that the three-party game matrix can conform to the current ship state; and then solving the first data matrix by using a Nash equilibrium solver to obtain the first ship berthing scheme. The Nash equilibrium solver is a tool for solving the Nash equilibrium in game theory, and its core is to find the optimal strategy combination of all participants by algorithm, so that any party cannot increase the income by changing the strategy unilaterally. The present application solves the first data matrix by using the Nash equilibrium solver, and outputs a collaborative berthing scheme acceptable to multiple parties, thereby avoiding the efficiency loss caused by resource contention.

[0016] In some embodiments, the step of determining the first ship berthing scheme according to the port structure parameter information, the ship environment data information, the adjacent ship dynamic information, the ship state information and the three-party game matrix comprises: based on the ship being in a crosswind environment, adjusting the three-party game matrix by using a conflict-compromise algorithm to obtain a second data matrix; inputting the port structure parameter information, the ship environment data information and the adjacent ship dynamic information into the second data matrix to obtain a third data matrix; and solving the third data matrix by using a Nash equilibrium solver to obtain the first ship berthing scheme.

[0017] In the technical solution, the step of determining the first ship berthing scheme according to the port structure parameter information, the ship environmental data information, the adjacent ship dynamic information, the ship state information and the three-party game matrix includes: when the ship is in a crosswind environment, a conflict-compromise algorithm is used to adjust the weight in the three-party game matrix to obtain a second data matrix, the conflict-compromise algorithm can realize real-time trade-off between safety and efficiency, then the port structure parameter information, the ship environmental data information and the adjacent ship dynamic information are input into the second data matrix to obtain a third data matrix, and finally a Nash equilibrium solver is used to solve the third data matrix to obtain the first ship berthing scheme. By dynamically adjusting the game weight in the sudden crosswind scene, the safety of the ship is prioritized, so that the ship berthing scheme can be more accurate.

[0018] In some technical solutions, the step of determining the second ship berthing scheme according to the first ship berthing scheme and the neural network model includes: obtaining the neural network model; and inputting the first ship berthing scheme into the neural network model to obtain the second ship berthing scheme.

[0019] In the technical solution, the step of determining the second ship berthing scheme according to the first ship berthing scheme and the neural network model includes: first obtaining the neural network model, and then inputting the first ship berthing scheme into the neural network model to obtain the second ship berthing scheme. The neural network model is a multi-objective reinforcement learning (MORL) model. When a ship is berthing, multiple “fighting” objectives need to be considered, such as safety (not hitting the port), efficiency (stopping quickly), and fuel saving (tugboat burning less oil). However, these objectives often contradict each other, for example, if safety is considered, the ship needs to be berthed slowly and leave enough safety distance, but it takes longer; if efficiency is considered, the ship needs to be berthed faster, but it may increase the risk of collision; if fuel saving is considered, the tugboat needs to be less powerful, but it may affect the control accuracy. Therefore, in order to balance these contradictions, the neural network model needs to be obtained, and the neural network model is used to dynamically balance these contradictions to give the optimal scheme.

[0020] In some technical solutions, before the step of obtaining the neural network model, the method further includes: determining a safety score, an efficiency score and an energy efficiency score; constructing a reward function according to the safety score, the efficiency score, the energy efficiency score, a safety weight, an efficiency weight and an energy efficiency weight; obtaining a plurality of ship historical berthing data; and determining the neural network model according to the plurality of ship historical berthing data and the reward function.

[0021] In the technical solution, before the step of obtaining the neural network model, the following steps are included: firstly, determining the safety score, the efficiency score and the energy efficiency score, wherein the safety score is negatively correlated with the deviation of the ship from the port, the wind speed and the tugboat braking response time, the efficiency score is negatively correlated with the berthing time, and the energy efficiency score is negatively correlated with the oil consumption of the tugboat. Then, constructing a reward function according to the safety score, the efficiency score, the energy efficiency score, the safety weight, the efficiency weight and the energy efficiency weight, specifically, the reward function can be the safety score multiplied by the safety weight plus the efficiency score multiplied by the efficiency weight plus the energy efficiency score multiplied by the energy efficiency weight. Then, obtaining a plurality of ship historical berthing data, training the reward function by using the plurality of ship historical berthing data, and adjusting the weights in real time to balance the target conflicts, so as to obtain the final neural network model. Through multi-objective reinforcement learning of the neural network model, the contradiction between safety, efficiency and energy efficiency can be balanced, so as to generate a more optimal ship berthing scheme.

[0022] According to a second aspect of the present application, a ship berthing scheme generation system is provided, comprising a collection module, the collection module is used to collect port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information; a first determination module, the first determination module is used to determine a first port model according to the port structure parameter information; a second determination module, the second determination module is used to determine a port agent, a tugboat agent and a ship agent according to the first port model, respectively, wherein the port agent takes load balancing as the target, the tugboat agent takes the sum of the oil consumption and the path cost as the minimum as the target, and the ship agent takes the short berthing time and the low safety risk as the target; a calculation module, the calculation module is used to construct a three-party game matrix according to the port agent, the tugboat agent and the ship agent; a third determination module, the third determination module is used to determine a first ship berthing scheme according to the port structure parameter information, the ship environment data information, the adjacent ship dynamic information, the ship state information and the three-party game matrix; and an optimization module, the optimization module is used to determine a second ship berthing scheme according to the first ship berthing scheme and a neural network model.

[0023] The ship berthing scheme generation system provided by the application mainly comprises a collection module, a first determination module, a second determination module, a calculation module, a third determination module and an optimization module. The collection module can collect port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information. The port structure parameter information can include port fender coordinates, port mooring pile layout and berth water depth. The ship environment data information can include wind flow pressure difference, wave height and visibility. The adjacent ship dynamic information refers to the dynamic information of a ship within a preset range of the ship, which can include speed, heading and draft. The ship state information can include ship attitude, ship propeller power and rudder angle. Further, the above data information can be collected in real time by a laser radar (LiDAR, Light Detection and Ranging), a millimeter wave radar, an AIS (Automatic Identification System, ship automatic identification system) receiver and a weather station. After obtaining the above data information, the first determination module determines a first port model according to the port structure parameters, that is, a three-dimensional model of the port where the ship is to be berthed is determined according to the port structure parameters. Then the second determination module determines a port agent, a tug agent and a ship agent according to the first port model, wherein the agent is an entity that can perceive the environment, make autonomous decisions and perform actions to achieve a specific goal. Therefore, the specific goal of the port agent is load balancing, the specific goal of the tug agent is to minimize the sum of fuel consumption and path cost, and the specific goal of the ship agent is to shorten the berthing time and reduce the safety risk. By determining the port agent, the tug agent and the ship agent using the first port model, the interests of each party can be clearly expressed, which lays a foundation for game deduction, that is, the load, fuel consumption and path are considered in multiple aspects. After obtaining the port agent, the tug agent and the ship agent, the calculation module constructs a three-party game matrix according to the port agent, the tug agent and the ship agent. The three-party game matrix is used to analyze the strategy interaction behavior and result of three independent decision-making subjects (the port agent, the tug agent and the ship agent).After obtaining the three-party game matrix, the third determination module inputs the collected port structure parameter information, ship environment data information, adjacent ship dynamic information and ship self-state information into the three-party game matrix, so as to obtain a first ship berthing scheme that can meet the goals of the port agent, the tugboat agent and the ship agent, and finally the optimization module optimizes the first ship berthing scheme by using a neural network model, wherein the neural network model is trained based on historical accident data and can balance some contradictory goals in the first ship berthing scheme, for example, the contradiction among safety, efficiency and fuel saving, so as to output an optimal ship berthing scheme, that is, a second ship berthing scheme. The port structure parameter information, ship environment data information, adjacent ship dynamic information and ship self-state information are collected to break through the limitation of related technologies that only rely on wind speed or berth state, realize port-environment-ship full-element data fusion, provide global input for game deduction, further, by determining multiple agents and realizing global optimization of the ship berthing scheme through multi-agent cooperation and dynamic game mechanism, the accuracy of the ship berthing scheme is improved.

[0024] According to a third aspect of the present application, an electronic device is provided, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for generating a ship berthing scheme according to any one of the above aspects when executing the computer program.

[0025] The electronic device provided by the present application can implement the technical effects of any one of the above technical solutions, and the steps of the method for generating a ship berthing scheme are implemented when the processor executes the computer program, which will not be described here again.

[0026] According to a fourth aspect of the present application, a storage medium is provided, which stores a computer program, wherein the computer program is executable on a processor to implement the steps of the method for generating a ship berthing scheme according to any one of the above aspects.

[0027] The storage medium provided by the present application can implement the technical effects of any one of the above technical solutions, and the steps of the method for generating a ship berthing scheme are implemented when the computer program is executed on the processor, which will not be described here again.

[0028] Additional aspects and advantages of the present application will become apparent from the following description with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0029] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description, taken in conjunction with the accompanying drawings, in which:

[0030] Figure 1A flowchart of a method for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0031] Figure 2 A flowchart of a step of determining a first port model according to port structure parameter information in a method for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0032] Figure 3 A flowchart of a step of constructing a three-party game matrix according to a port agent, a tugboat agent and a ship agent in a method for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0033] Figure 4 A flowchart of a step of determining a first ship berthing scheme according to port structure parameter information, environmental data information of a ship, dynamic information of adjacent ships, state information of the ship itself and a three-party game matrix in a method for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0034] Figure 5 A flowchart of a step of determining a first ship berthing scheme according to port structure parameter information, environmental data information of a ship, dynamic information of adjacent ships, state information of the ship itself and a three-party game matrix in a method for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0035] Figure 6 A flowchart of a step of determining a second ship berthing scheme according to a first ship berthing scheme and a neural network model in a method for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0036] Figure 7 A flowchart of a step before obtaining a neural network model in a method for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0037] Figure 8 A schematic block diagram of a system for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0038] Figure 9 A schematic block diagram of a system for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0039] Figure 10 A schematic diagram of a modeling process of a virtual anchor point in a system for generating a ship berthing scheme according to an embodiment of the present application is shown;

[0040] Figure 11 A schematic block diagram of an electronic device according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0041] In order to enable a more clear understanding of the above-mentioned objects, features and advantages of the present application, the present application will be further described in detail below with reference to 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.

[0042] In the following description, a large number of specific details are set forth in order to facilitate a thorough understanding of the present application, however, the present application can also be implemented in other manners different from those described herein, and therefore, the protective scope of the present application is not limited to the specific embodiments disclosed below.

[0043] Figure 1 A flowchart of a method for generating a ship berthing scheme of an embodiment of the present application is shown. The method comprises:

[0044] Step 102: Collecting port structure parameter information, ship environmental data information, adjacent ship dynamic information and ship self-state information;

[0045] Step 104: Determining a first port model according to the port structure parameter information;

[0046] Step 106: Determining a port agent, a tugboat agent and a ship agent according to the first port model, wherein the port agent aims to balance load, the tugboat agent aims to minimize the sum of fuel consumption and path cost, and the ship agent aims to shorten berthing time and reduce safety risk;

[0047] Step 108: Constructing a three-party game matrix according to the port agent, the tugboat agent and the ship agent;

[0048] Step 110: Determining a first ship berthing scheme according to the port structure parameter information, the ship environmental data information, the adjacent ship dynamic information, the ship self-state information and the three-party game matrix;

[0049] Step 112: Determining a second ship berthing scheme according to the first ship berthing scheme and a neural network model.

[0050] The application provides a ship berthing scheme generation method, mainly comprising the following steps: firstly, collecting port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information, wherein the port structure parameter information can comprise: port fender coordinates, port mooring pile layout and berth water depth; the ship environment data information can comprise: wind flow pressure difference, wave height and visibility; the adjacent ship dynamic information refers to the dynamic information of a ship located within a preset range of the ship, and can comprise: speed, heading and draft; and the ship state information can comprise: ship attitude, ship propeller power and rudder angle. Further, the above data information can be collected in real time by a laser radar (LiDAR, Light Detection and Ranging), a millimeter wave radar, an AIS (Automatic Identification System, ship automatic identification system) receiver and a weather station. After obtaining the above data information, a first port model is determined according to the port structure parameters, that is, a three-dimensional model of the port where the ship is to be berthed is determined according to the port structure parameters. Then, a port agent, a tugboat agent and a ship agent are determined according to the first port model, wherein the agent is an entity capable of perceiving the environment, making autonomous decisions and performing actions to achieve a specific goal, therefore, the specific goal of the port agent is load balancing, the specific goal of the tugboat agent is to minimize the sum of fuel consumption and path cost, and the specific goal of the ship agent is to shorten the berthing time and reduce the safety risk. By determining the port agent, the tugboat agent and the ship agent by using the first port model, the interests of each party can be clearly expressed, which lays a foundation for game deduction, that is, the load, fuel consumption and path are considered in multiple aspects. After obtaining the port agent, the tugboat agent and the ship agent, a three-party game matrix is constructed according to the port agent, the tugboat agent and the ship agent, and the three-party game matrix is used to analyze the strategy interaction behavior and result of three independent decision-making subjects (the port agent, the tugboat agent and the ship agent). After obtaining the three-party game matrix, the collected port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information are input into the three-party game matrix, so as to obtain a first ship berthing scheme that can meet the goals of the port agent, the tugboat agent and the ship agent. Finally, a neural network model is used to optimize the first ship berthing scheme, wherein the neural network model is trained based on historical accident data, and can balance some target contradictions in the first ship berthing scheme, for example, the contradictions among safety, efficiency and fuel saving, so as to output an optimal ship berthing scheme, that is, a second ship berthing scheme.The application breaks through the limitation of related technologies relying only on wind speed or berth state by collecting port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information, realizes port-environment-ship full-element data fusion, provides global input for game deduction, further, determines multiple intelligent agents, and realizes global optimization of the ship berthing scheme through multi-agent cooperation and dynamic game mechanism, thereby improving the accuracy of the ship berthing scheme.

[0051] Figure 2 A flowchart of a step of determining a first port model according to port structure parameter information in a ship berthing scheme generation method of an embodiment of the application is shown; wherein the step of determining the first port model according to the port structure parameter information comprises:

[0052] Step 202: obtaining a port structure feature database, wherein the port structure feature database comprises multiple second port models and first port structure feature data corresponding to the second port models;

[0053] Step 204: performing geometric feature extraction on the port structure parameter information to obtain multiple second data information;

[0054] Step 206: performing coordinate mapping on the multiple second data information to obtain multiple second port structure feature data;

[0055] Step 208: determining the first port model from the multiple second port models according to the multiple second port structure feature data and the port structure feature database.

[0056] In the embodiment, the step of determining the first port model according to the port structure parameter information comprises: firstly acquiring a port structure feature database, wherein the port structure feature database comprises a plurality of second port models and first port structure feature data corresponding to the second port models. Then, the virtual anchor algorithm (VAA) is used to process the collected port structure parameter information, that is, a plurality of first data information is obtained by extracting the geometric features of the port structure parameter information, and then a plurality of second port structure feature data is obtained by respectively mapping the coordinates of the plurality of first data information, wherein the geometric features can be coordinates and / or intensity, etc. The second data information can be converted into standardized virtual anchor points by coordinate mapping. Then, the first port model is determined in the plurality of second port models according to the plurality of second port structure feature data and the port structure feature database, that is, the second port structure feature data is matched with the first port structure feature information in the port structure feature database, if there is similar first port structure feature information to the second port structure feature data, the second port model corresponding to the first port structure feature information is taken as the first port model. By using the virtual anchor algorithm, the digitization of the port structure parameter information is realized, and then combined with the adaptive template library, that is, the port structure feature database, thereby solving the problem of high cross-port deployment cost in the related art.

[0057] Figure 3 A flowchart of a method for generating a ship berthing scheme according to an embodiment of the present application is shown, and a step of constructing a three-party game matrix according to a port agent, a tugboat agent and a ship agent is shown; wherein the step of constructing the three-party game matrix according to the port agent, the tugboat agent and the ship agent comprises:

[0058] Step 302: determining a port strategy space according to the port agent;

[0059] Step 304: generating a first utility function according to the port strategy space;

[0060] Step 306: determining a tugboat strategy space according to the tugboat agent;

[0061] Step 308: generating a second utility function according to the tugboat strategy space;

[0062] Step 310: defining a ship strategy space according to the ship agent;

[0063] Step 312: generating a third utility function according to the ship strategy space;

[0064] Step 314: constructing a three-party game matrix according to the first utility function, the second utility function and the third utility function.

[0065] In this embodiment, the steps of constructing a three-party game matrix according to the port agent, the tugboat agent and the ship agent include: first, determining the port strategy space according to the port agent, wherein the strategy space refers to the set of all possible strategies in a specific problem, and since the port agent aims to balance the load, the port strategy space includes all possible combinations of the load and the bollard in the port; then generating a first utility function according to the port strategy space, wherein the first utility function aims to balance the load of the bollard and can minimize the load variance. At the same time, determining the tugboat strategy space according to the tugboat agent, wherein the tugboat agent aims to minimize the sum of fuel consumption and path cost, so the tugboat strategy space includes all paths and the thrust of each path; then generating a second utility function according to the tugboat strategy space, wherein the second utility function aims to minimize the sum of fuel consumption and path cost and can minimize the sum of fuel consumption and path cost. At the same time, determining the ship strategy space according to the ship agent, wherein the ship agent aims to shorten the berthing time and reduce the safety risk, so the ship strategy space includes all combinations of berthing speed and rudder angle; then generating a third utility function according to the ship agent, wherein the third utility function aims to shorten the berthing time and reduce the safety risk. Finally, constructing a three-party game matrix according to the first utility function, the second utility function and the third utility function. By defining the utility functions of the three agents, wherein the port utility function, i.e. the first utility function, aims to balance the load, the tugboat utility function, i.e. the second utility function, aims to minimize the sum of fuel consumption and path cost, and the ship utility function, i.e. the third utility function, aims to shorten the berthing time and reduce the safety risk, and then constructing a three-party game matrix according to the utility functions of the three agents, a collaborative berthing scheme acceptable to all parties can be output, avoiding the efficiency loss caused by resource contention.

[0066] Figure 4 Fig. 1 shows one of the flowcharts of the step of determining the first ship berthing scheme according to the port structure parameter information, the environmental data information of the ship, the dynamic information of the adjacent ship, the state information of the ship itself and the three-party game matrix in the method of generating the ship berthing scheme of one embodiment of the present application; wherein the step of determining the first ship berthing scheme according to the port structure parameter information, the environmental data information of the ship, the dynamic information of the adjacent ship, the state information of the ship itself and the three-party game matrix includes:

[0067] Step 402: inputting the port structure parameter information, the environmental data information of the ship and the dynamic information of the adjacent ship into the three-party game matrix respectively to obtain a first data matrix;

[0068] Step 404: solving the first data matrix by using a Nash equilibrium solver to obtain the first ship berthing scheme.

[0069] In this embodiment, the step of determining the first ship berthing scheme according to the port structure parameter information, the ship's environmental data information, the dynamic information of the adjacent ship, the state information of the ship itself and the three-party game matrix includes: first, input the port structure parameter information, the ship's environmental data information and the dynamic information of the adjacent ship into the three-party game matrix to obtain a first data matrix; by inputting the port structure parameter information, the ship's environmental data information and the dynamic information of the adjacent ship into the three-party game matrix, so that the three-party game matrix can conform to the current ship state. Then a Nash equilibrium solver is used to solve the first data matrix to obtain the first ship berthing scheme. The Nash equilibrium solver is a tool for solving the Nash equilibrium in game theory, and its core is to find the optimal strategy combination of all participants through an algorithm, so that any party cannot increase the income by changing the strategy unilaterally. The present application solves the first data matrix by the Nash equilibrium solver, and outputs a collaborative berthing scheme acceptable to multiple parties, avoiding the efficiency loss caused by resource contention.

[0070] Figure 5 Fig. 2 shows a flowchart of the step of determining the first ship berthing scheme according to the port structure parameter information, the ship's environmental data information, the dynamic information of the adjacent ship, the state information of the ship itself and the three-party game matrix in the method for generating a ship berthing scheme according to one embodiment of the present application; wherein the step of determining the first ship berthing scheme according to the port structure parameter information, the ship's environmental data information, the dynamic information of the adjacent ship, the state information of the ship itself and the three-party game matrix includes:

[0071] Step 502: based on the ship being in a crosswind environment, a conflict-compromise algorithm is used to adjust the three-party game matrix to obtain a second data matrix;

[0072] Step 504: input the port structure parameter information, the ship's environmental data information and the dynamic information of the adjacent ship into the second data matrix to obtain a third data matrix;

[0073] Step 506: a Nash equilibrium solver is used to solve the third data matrix to obtain the first ship berthing scheme.

[0074] In this embodiment, the step of determining the first ship berthing scheme according to the port structure parameter information, the environmental data information of the ship, the dynamic information of the adjacent ship, the state information of the ship itself and the three-party game matrix includes: when the ship is in a cross-wind environment, a conflict-compromise algorithm is used to adjust the weight in the three-party game matrix to obtain a second data matrix, the conflict-compromise algorithm can realize real-time trade-off between safety and efficiency, then the port structure parameter information, the environmental data information of the ship and the dynamic information of the adjacent ship are input into the second data matrix to obtain a third data matrix, and finally a Nash equilibrium solver is used to solve the third data matrix to obtain the first ship berthing scheme. Through dynamic adjustment of the game weight in the sudden cross-wind scene, the safety of the ship is prioritized, so that the ship berthing scheme can be more accurate.

[0075] Figure 6 A flowchart of the step of determining the second ship berthing scheme according to the first ship berthing scheme and the neural network model in the ship berthing scheme generation method of one embodiment of the present application is shown; wherein the step of determining the second ship berthing scheme according to the first ship berthing scheme and the neural network model includes:

[0076] Step 602: obtaining a neural network model;

[0077] Step 604: inputting the first ship berthing scheme into the neural network model to obtain the second ship berthing scheme.

[0078] In this embodiment, the step of determining the second ship berthing scheme according to the first ship berthing scheme and the neural network model includes: first obtaining a neural network model, and then inputting the first ship berthing scheme into the neural network model to obtain the second ship berthing scheme. The neural network model is a Multi-Objective Reinforcement Learning (MORL) model. When the ship is berthing, multiple “fighting” targets need to be considered at the same time, such as safety (not hitting the port), efficiency (stopping quickly), and fuel saving (tugboat burning less oil). However, these targets often contradict each other, for example, if safety is considered, the ship needs to be berthed slowly and a safe distance needs to be left, but it takes longer; if efficiency is considered, the ship needs to be berthed faster, but the risk of collision may increase; if fuel saving is considered, the tugboat needs to be less powered, but the control accuracy may be affected. Therefore, in order to balance these contradictions, a neural network model needs to be obtained, and the neural network model is used to dynamically balance these contradictions to give an optimal scheme.

[0079] Figure 7 A flowchart before the step of obtaining a neural network model in the ship berthing scheme generation method of one embodiment of the present application is shown; wherein before the step of obtaining a neural network model, it includes:

[0080] Step 702: determining a safety score, an efficiency score and an energy efficiency score;

[0081] Step 704: constructing a reward function according to the safety score, the efficiency score, the energy efficiency score, a safety weight, an efficiency weight and an energy efficiency weight;

[0082] Step 706: obtaining a plurality of ship historical berthing data;

[0083] Step 708: determining a neural network model according to the plurality of ship historical berthing data and the reward function.

[0084] In this embodiment, before the step of obtaining the neural network model, first, the safety score, the efficiency score and the energy efficiency score are determined, wherein the safety score is negatively correlated with the deviation of the ship from the port, the wind speed and the tugboat braking response time, the efficiency score is negatively correlated with the berthing time, and the energy efficiency score is negatively correlated with the oil consumption of the tugboat. Then, the reward function is constructed according to the safety score, the efficiency score, the energy efficiency score, the safety weight, the efficiency weight and the energy efficiency weight. Specifically, the reward function can be the safety score multiplied by the safety weight plus the efficiency score multiplied by the efficiency weight plus the energy efficiency score multiplied by the energy efficiency weight. Then, the plurality of ship historical berthing data is obtained, the plurality of ship historical berthing data is used to train the reward function, the weights are adjusted in real time to balance the target conflicts, and thus the final neural network model is obtained. Through the multi-objective reinforcement learning of the neural network model, the contradiction between safety, efficiency and energy efficiency can be balanced, and thus a more optimal ship berthing scheme can be generated.

[0085] Figure 8 One of the schematic block diagrams of the ship berthing scheme generation system of one embodiment of the application is shown; wherein the ship berthing scheme generation system 80 comprises:

[0086] The acquisition module 802 is configured to acquire the port structure parameter information, the environmental data information of the ship, the dynamic information of the adjacent ship and the state information of the ship itself.

[0087] The first determination module 804 is configured to determine a first port model according to the port structure parameter information.

[0088] The second determination module 806 is configured to determine a port agent, a tugboat agent and a ship agent according to the first port model, respectively, wherein the port agent aims to balance the load, the tugboat agent aims to minimize the sum of the oil consumption and the path cost, and the ship agent aims to shorten the berthing time and reduce the safety risk.

[0089] The calculation module 808 is configured to construct a three-party game matrix according to the port agent, the tugboat agent and the ship agent.

[0090] The third determination module 810 is configured to determine a first ship berthing scheme according to the port structure parameter information, the environment data information of the ship, the dynamic information of the adjacent ship, the state information of the ship itself, and a three-party game matrix.

[0091] The optimization module 812 is configured to determine a second ship berthing scheme according to the first ship berthing scheme and a neural network model.

[0092] The ship berthing scheme generation system 80 provided by the application mainly comprises a collection module 802, a first determination module 804, a second determination module 806, a calculation module 808, a third determination module 810 and an optimization module 812. The collection module 802 can collect port structure parameter information, ship environment data information, adjacent ship dynamic information and ship state information. The port structure parameter information can include port fender coordinates, port mooring pile layout and berth water depth. The ship environment data information can include wind flow pressure difference, wave height and visibility. The adjacent ship dynamic information refers to the dynamic information of a ship within a preset range of the ship, which can include speed, heading and draft. The ship state information can include ship attitude, ship propeller power and rudder angle. Further, the above data information can be collected in real time by a laser radar (LiDAR, Light Detection and Ranging), a millimeter wave radar, an AIS (Automatic Identification System, ship automatic identification system) receiver and a weather station. After obtaining the above data information, the first determination module 804 determines a first port model according to the port structure parameters, i.e., determines a three-dimensional model of the port where the ship is to be berthed according to the port structure parameters. Then the second determination module 806 determines a port agent, a tug agent and a ship agent according to the first port model, wherein the agent is an entity that can perceive the environment, make autonomous decisions and perform actions to achieve a specific goal. Therefore, the specific goal of the port agent is load balancing, the specific goal of the tug agent is to minimize the sum of fuel consumption and path cost, and the specific goal of the ship agent is to shorten the berthing time and reduce the safety risk. By determining the port agent, the tug agent and the ship agent using the first port model, the interests of each party can be clearly expressed, which lays a foundation for game deduction, i.e., considering load, fuel consumption and path in multiple aspects. After obtaining the port agent, the tug agent and the ship agent, the calculation module 808 constructs a three-party game matrix according to the port agent, the tug agent and the ship agent. The three-party game matrix is used to analyze the strategy interaction behavior and result of three independent decision-making subjects (the port agent, the tug agent and the ship agent).After obtaining the three-party game matrix, the third determination module 810 inputs the collected port structure parameter information, ship environmental data information, adjacent ship dynamic information and ship state information of the ship itself into the three-party game matrix, so that the first ship berthing scheme meeting the goals of the port agent, the tugboat agent and the ship agent is obtained, and finally the optimization module 812 optimizes the first ship berthing scheme by using a neural network model, wherein the neural network model is trained based on historical accident data and can balance some contradictory goals in the first ship berthing scheme, for example, the contradiction among safety, efficiency and fuel saving, so as to output an optimal ship berthing scheme, i.e., a second ship berthing scheme. The present application breaks through the limitation of related technologies relying only on wind speed or berth state by collecting port structure parameter information, ship environmental data information, adjacent ship dynamic information and ship state information, realizes port-environment-ship full-element data fusion, provides global input for game deduction, further, realizes global optimization of the ship berthing scheme by determining multiple agents and realizing multi-agent cooperation and dynamic game mechanism, thereby improving the accuracy of the ship berthing scheme.

[0093] Figure 9 A schematic block diagram of a ship berthing scheme generation system of one embodiment of the present application is shown; wherein the ship berthing scheme generation system comprises four core modules: data acquisition layer, agent modeling layer, game deduction layer and optimization decision layer.

[0094] The data acquisition layer is mainly used for multi-source heterogeneous data fusion, and the technical means can be to use a multi-modal sensor array, i.e., to deploy a laser radar (LiDAR, Light Detection and Ranging), a millimeter wave radar, an AIS (Automatic Identification System, ship automatic identification system) receiver and a weather station to collect the following data in real time: port structure parameters, including fender coordinates, mooring bitt layout and berth depth (obtained by laser radar three-dimensional modeling); environmental dynamic data, including wind flow pressure difference (obtained by fusing weather station and ship inertial measurement unit), wave height and visibility; adjacent ship dynamics, including speed, heading and draft (obtained by fusing AIS and radar data); and ship state, including ship attitude, propeller power and rudder angle. The role of the data acquisition layer is: first, to break through the limitation of related technologies relying only on wind speed or berth state, and to realize port-environment-ship full-element data fusion to provide global input for game deduction. Second, to combine laser radar and image recognition technology to replace traditional manual entry of port parameters, and to improve data acquisition efficiency by 90%.

[0095] The agent modeling layer mainly focuses on the virtual anchor algorithm and multi-agent definition. The process of the Virtual Anchor Algorithm (VAA) is as follows: Figure 10 As shown, firstly, the port's fenders, mooring bollards, and other facilities are scanned using lidar to extract geometric features (such as coordinates and strength), which are then mapped to standardized virtual anchor points. Figure 10 The following mooring bollards are selected: A1, B1(X1, Y1), A2, B2(X2, Y2), A3, B3(X3, Y3), and A4. These are then input into a port template library for matching. The port template library includes templates for multiple ports, such as the first port template and the second port template. If a match is successful, a port berthing model is generated. After obtaining the port model, multiple agents are defined, including: the port agent (Wharf Agent): aiming for load balancing, with a utility function that minimizes the variance of mooring bollard force; the tug agent: aiming for minimum fuel consumption and shortest path, with a utility function that minimizes the product of fuel consumption and distance; and the ship agent: aiming for optimal berthing time and safe distance, with a utility function that maximizes the inverse of time multiplied by the risk coefficient. The role of the intelligent agent modeling layer is as follows: First, to realize the automatic digitization of port parameters through the virtual anchor point algorithm, thereby solving the problem of high deployment costs of related technologies across ports; Second, to clarify the interests and demands of each party through the definition of multiple intelligent agents, laying the foundation for game theory simulation.

[0096] The game theory deduction layer primarily employs a non-cooperative game framework and a conflict-compromise algorithm. The non-cooperative game framework mainly involves: first, constructing a three-party game matrix and defining the strategy space; specifically, assuming the set of game participants is... Their policy spaces are as follows:

[0097] Port strategic space :

[0098] The port needs to allocate a combination of mooring bollards. Let the possible sets of mooring bollards be: Each of the mooring bollards The strength is Load capacity is Port strategy To select a combination of k mooring bollards from a set, and specify the load distribution weights. ,satisfy Therefore, port strategy The mathematical expression is:

[0099] ;

[0100] in, Port agent's strategy, Port strategy space, Bollard, Load distribution weight, Distribution weight of the jth load, k represents the number of bollards.

[0101] Tug strategy space :

[0102] The tug needs to choose a cooperative braking path, and the set of optional paths is , the voyage of each path is , and the fuel consumption rate is . The strategy of the tug is to select the path and distribute the thrust , where qq is the number of tugs. Therefore, the strategy of the tug is mathematically expressed as:

[0103] ;

[0104] where, represents the strategy of the tug agent, represents the tug strategy space, represents the jth path, represents the thrust, q is the number of tugs, represents the thrust of the Ith tug, I represents the Ith tug.

[0105] Ship strategy space :

[0106] The ship needs to decide on the berthing speed v and the rudder angle θ, and its strategy satisfies the physical constraints:

[0107] ;

[0108] where, represents the strategy of the ship agent, the ship strategy space, represents the berthing speed, θ represents the rudder angle, represents the minimum berthing speed, represents the maximum berthing speed, represents the minimum rudder angle, represents the maximum rudder angle.

[0109] After defining the strategy space, the utility function is designed. Specifically, the utility function of each participant reflects its core objective: where the port utility function is the target of mooring bitts load balancing, and the minimum of load variance, so the formula is:

[0110] ;

[0111] where, is the dynamic load generated by the ship berthing, is the mooring bitts strength coefficient, represents the strategy of the ship agent, represents the strategy of the tugboat agent, represents the strategy of the port agent, represents the allocation weight of the jth load.

[0112] Tugboat utility function is the target of joint minimization of fuel consumption and path cost, so the formula is:

[0113] ;

[0114] where, represents the strategy of the tugboat agent, represents the thrust of the Ith tugboat, and q is the number of tugboats, represents the strategy of the ship agent, represents fuel consumption, represents the jth path, and I represents the Ith tugboat.

[0115] Ship utility function is the target of balancing berthing time T and safety risk R:

[0116] ;

[0117] where, is the weight coefficient, and R is the safety risk function (such as distance from the port, wind speed correlation), represents the strategy of the ship agent, represents the strategy of the tugboat agent, represents the strategy of the port agent, represents the time of the ship berthing.

[0118] Then the Nash Equilibrium (Nash Equilibrium) solver is used to output the stable state solution that all three parties have no incentive to change their strategies unilaterally.

[0119] where, the Nash equilibrium solution of the three-party game is the strategy combination , which satisfies:

[0120] ;

[0121] where, a strategy of the port agent, an optimal strategy of the port agent, a strategy of the tugboat agent, an optimal strategy of the tugboat agent, a strategy of the own ship agent, an optimal strategy of the own ship agent.

[0122] The conflict-compromise algorithm (CCA) is mainly used in the sudden crosswind scene to dynamically adjust the game weight, and to give priority to guarantee the safety of the own ship. At the same time, it also generates an emergency strategy in real time: tugboat cooperative braking path (such as fan-shaped encirclement) and port bollard load balancing scheme. The role of the game deduction layer is to solve the existing technology emergency strategy generation lag problem, and to shorten the response time of the sudden working condition to 2 seconds (the traditional system needs more than 10 seconds) through dynamic game deduction. At the same time, the conflict-compromise algorithm is adopted, so as to realize the real-time trade-off between safety and efficiency.

[0123] The optimization decision layer mainly adopts multi-objective reinforcement learning (MORL) and Pareto frontier. Specifically, when the ship is berthing, the system needs to consider multiple “fighting” targets at the same time, such as safety (not hitting the port), efficiency (stopping well and quickly), and energy saving (tugboat burning less oil). But these goals often contradict each other: if you want to be safe, you have to slow down and leave enough safety distance, but it will take longer; if you want to be efficient, you have to speed up, but it may increase the risk of collision; if you want to save energy, you have to make the tugboat work less, but it may affect the control accuracy. Therefore, the role of the optimization decision layer is to dynamically balance these contradictions like an “intelligent referee” and give the optimal solution. Specifically, it is divided into two steps: first, multi-objective reinforcement learning (MORL): first, define the reward function: R = α × Safety + β × Efficiency + γ × Energy, where α is the safety weight coefficient, β is the efficiency weight coefficient, γ is the energy saving weight coefficient, Safety represents the safety score, Efficiency represents the efficiency score, and Energy represents the energy efficiency score. Among them, Safety (safety score) is negatively related to the deviation of the ship from the port, wind speed, and tugboat braking response time, so its formula is:

[0124] ;

[0125] Efficiency (efficiency score) is negatively related to the berthing time, so its formula is:

[0126] ;

[0127] Energy (Energy Score) is negatively correlated with the energy consumption of the tug, so its formula is:

[0128] ;

[0129] Then, the neural network is trained based on historical accident data, and the weight is adjusted in real time to balance the target conflict (such as automatically increasing the alpha weight to 0.8 when sudden crosswind occurs). Wherein, alpha, beta, gamma are dynamic weights: for example, normal condition: alpha = 0.5, beta = 0.3, gamma = 0.2; but sudden crosswind condition: alpha = 0.8, beta = 0.1, gamma = 0.1 (system automatically adjusts).

[0130] Second, the Pareto frontier is generated: by using NSGA-II (Non-dominated Sorting Genetic Algorithm II, Non-dominated Sorting Genetic Algorithm) to calculate the Pareto optimal solution set of safety-efficiency-energy efficiency, specifically, the objective function can be:

[0131] ;

[0132] Further, the optimal solution boundary can also be visualized to provide the crew with a preferred solution, that is, to display multiple ship docking schemes on the ship display device in front of the crew, so that the crew can make their own choices.

[0133] As shown in Table 1, the scheme provided by the present application has the following core differences from the related art:

[0134] Table 1

[0135]

[0136] The present application realizes real-time collaborative deduction and cross-port low-cost deployment of all factors of ship berthing through three core technologies of multi-agent game architecture, virtual anchor point algorithm and dynamic Pareto optimization, and provides a disruptive solution for intelligent ports.

[0137] In some embodiments, the first determining module 804 is specifically configured to obtain a port structure feature database, wherein the port structure feature database includes a plurality of second port models and first port structure feature data corresponding to the second port models; geometric feature extraction is performed on the port structure parameter information to obtain a plurality of second data information; coordinate mapping is performed on the plurality of second data information to obtain a plurality of second port structure feature data; and the first port model is determined in the plurality of second port models according to the plurality of second port structure feature data and the port structure feature database.

[0138] In this embodiment, the first determining module 804 is specifically configured to acquire a port structure feature database, wherein the port structure feature database includes a plurality of second port models and first port structure feature data corresponding to the second port models. Then, the virtual anchor algorithm (VAA) is used to process the collected port structure parameter information, that is, the first data information is obtained by performing geometric feature extraction on the port structure parameter information, and then the plurality of first data information is respectively mapped to obtain a plurality of second port structure feature data. The geometric feature can be coordinates and / or intensity, etc. The second data information can be converted into standardized virtual anchor points by coordinate mapping. Then, the first port model is determined in the plurality of second port models according to the plurality of second port structure feature data and the port structure feature database, that is, the second port structure feature data is matched with the first port structure feature information in the port structure feature database, if there is similar first port structure feature information to the second port structure feature data, the second port model corresponding to the first port structure feature information is taken as the first port model. By using the virtual anchor algorithm, the digitization of the port structure parameter information is realized, and then combined with the adaptive template library, that is, the port structure feature database, thereby solving the problem of high cross-port deployment cost in the related art.

[0139] In some embodiments, optionally, the computing module 808 is specifically configured to determine a port strategy space according to the port agent; generate a first utility function according to the port strategy space; determine a tug strategy space according to the tug agent; generate a second utility function according to the tug strategy space; define a ship strategy space according to the ship agent; generate a third utility function according to the ship strategy space; and construct a three-party game matrix according to the first utility function, the second utility function and the third utility function.

[0140] In this embodiment, the calculation module 808 is specifically configured to determine a port strategy space according to the port agent, wherein the strategy space refers to a set of all possible strategies in a specific problem, and since the port agent aims to balance the load, the port strategy space includes all possible combinations of the load and the bollard in the port. Then, a first utility function is generated according to the port strategy space, the first utility function aims to balance the load of the bollard and can minimize the load variance. At the same time, a tug strategy space is determined according to the tug agent, wherein the tug agent aims to minimize the sum of the fuel consumption and the path cost, and thus the tug strategy space includes all paths and the thrust of each path. Then, a second utility function is generated according to the tug strategy space, the second utility function aims to minimize the sum of the fuel consumption and the path cost and can minimize the sum of the fuel consumption and the path cost. At the same time, a ship strategy space is determined according to the ship agent, wherein the ship agent aims to shorten the berthing time and reduce the safety risk, and thus the ship strategy space includes all combinations of the berthing speed and the rudder angle. Then, a third utility function is generated according to the ship agent, wherein the third utility function aims to shorten the berthing time and reduce the safety risk and can shorten the berthing time and reduce the safety risk. Finally, a three-party game matrix is constructed according to the first utility function, the second utility function and the third utility function. By defining the utility functions of the three agents, the port utility function (i.e., the first utility function) aims to balance the load, the tug utility function (i.e., the second utility function) aims to minimize the sum of the fuel consumption and the path cost, and the ship utility function (i.e., the third utility function) aims to shorten the berthing time and reduce the safety risk. Then, the three-party game matrix is constructed according to the utility functions of the three agents, so as to output a collaborative berthing scheme acceptable to all parties and avoid efficiency loss caused by resource contention.

[0141] In some embodiments, optionally, the third determination module 810 is specifically configured to input the port structure parameter information, the environmental data information of the ship and the dynamic information of the adjacent ship into the three-party game matrix respectively to obtain a first data matrix; and a Nash equilibrium solver is used to solve the first data matrix to obtain a first ship berthing scheme.

[0142] In this embodiment, the third determining module 810 is specifically configured to input the port structure parameter information, the environmental data information of the ship, and the dynamic information of the adjacent ship into a three-party game matrix respectively to obtain a first data matrix; the port structure parameter information, the environmental data information of the ship, and the dynamic information of the adjacent ship are input into the three-party game matrix, so that the three-party game matrix can conform to the current ship state. Then, a Nash equilibrium solver is used to solve the first data matrix to obtain a first ship berthing scheme. The Nash equilibrium solver is a tool for solving the Nash equilibrium in game theory, and its core is to find the optimal strategy combination of all participants through an algorithm, so that any party cannot increase the income by changing the strategy unilaterally. The present application solves the first data matrix by the Nash equilibrium solver, and outputs a collaborative berthing scheme acceptable by multiple parties, thereby avoiding the efficiency loss caused by resource contention.

[0143] In some embodiments, optionally, the third determining module 810 is further specifically configured to, based on the ship being in a crosswind environment, adjust the three-party game matrix by using a conflict-compromise algorithm to obtain a second data matrix; input the port structure parameter information, the environmental data information of the ship, and the dynamic information of the adjacent ship into the second data matrix to obtain a third data matrix; and solve the third data matrix by using the Nash equilibrium solver to obtain the first ship berthing scheme.

[0144] In this embodiment, the third determining module 810 is further specifically configured to, when the ship is in a crosswind environment, adjust the weight in the three-party game matrix by using a conflict-compromise algorithm to obtain a second data matrix, the conflict-compromise algorithm can realize real-time trade-off between safety and efficiency, then input the port structure parameter information, the environmental data information of the ship, and the dynamic information of the adjacent ship into the second data matrix to obtain a third data matrix, and finally solve the third data matrix by using the Nash equilibrium solver to obtain the first ship berthing scheme. By dynamically adjusting the game weight in the sudden crosswind scene, the safety of the ship is prioritized, so that the ship berthing scheme can be more accurate.

[0145] In some embodiments, optionally, the optimization module 812 is specifically configured to obtain a neural network model; and input the first ship berthing scheme into the neural network model to obtain a second ship berthing scheme.

[0146] In this embodiment, the optimization module 812 is specifically configured to obtain a neural network model, and then input the first ship berthing scheme into the neural network model to obtain the second ship berthing scheme. The neural network model is a Multi-Objective Reinforcement Learning (MORL) model. When a ship is berthing, multiple “fighting” objectives need to be considered at the same time, such as safety (not to collide with the port), efficiency (to stop quickly), and fuel saving (tugboat to burn less oil). However, these objectives often conflict with each other. For example, if safety is considered, the ship needs to be berthed at a slow speed and with a sufficient safety distance, but it takes a longer time; if efficiency is considered, the ship needs to be berthed at a high speed, but it may increase the risk of collision; and if fuel saving is considered, the tugboat needs to be less powered, but it may affect the control accuracy. Therefore, in order to balance these contradictions, the neural network model needs to be obtained, and the neural network model is used to dynamically balance these contradictions to give an optimal scheme.

[0147] In some embodiments, optionally, the optimization module 812 is further specifically configured to determine a safety score, an efficiency score, and an energy efficiency score; construct a reward function according to the safety score, the efficiency score, the energy efficiency score, a safety weight, an efficiency weight, and an energy efficiency weight; obtain a plurality of historical ship berthing data; and determine the neural network model according to the plurality of historical ship berthing data and the reward function.

[0148] In this embodiment, the optimization module 812 is further specifically configured to determine a safety score, an efficiency score, and an energy efficiency score, wherein the safety score is negatively correlated with a deviation of the ship from the port, a wind speed, and a braking response time of the tugboat, the efficiency score is negatively correlated with a berthing time, and the energy efficiency score is negatively correlated with an oil consumption of the tugboat. Then, a reward function is constructed according to the safety score, the efficiency score, the energy efficiency score, the safety weight, the efficiency weight, and the energy efficiency weight. Specifically, the reward function can be the safety score multiplied by the safety weight plus the efficiency score multiplied by the efficiency weight plus the energy efficiency score multiplied by the energy efficiency weight. Then, a plurality of historical ship berthing data is obtained, the plurality of historical ship berthing data is used to train the reward function, the weights are adjusted in real time to balance the objective conflicts, and thus a final neural network model is obtained. Through the multi-objective reinforcement learning of the neural network model, the conflicts among safety, efficiency, and energy efficiency can be balanced, and thus a more optimal ship berthing scheme can be generated.

[0149] Figure 11 A schematic block diagram of an electronic device of one embodiment of the present application is shown; wherein the electronic device 1100 comprises a memory 1102, a processor 1104, and a computer program stored on the memory 1102 and executable on the processor 1104, and the processor 1104 implements the steps of the ship berthing scheme generation method of any one of the above when executing the computer program.

[0150] The electronic device 1100 provided by the present application, when the processor 1104 executes the computer program, the steps of the ship berthing scheme generation method are realized, and the technical effects of any one of the above embodiments can be realized, which will not be repeated.

[0151] One embodiment of the present application provides a storage medium having a computer program stored thereon, and the computer program, when executed by a processor, realizes the steps of the ship berthing scheme generation method according to any one of the above embodiments.

[0152] The storage medium provided by the present application, when the computer program is executed by the processor, realizes the steps of the ship berthing scheme generation method, and the technical effects of any one of the above embodiments can be realized, which will not be repeated.

[0153] In the description of the present application, the terms "first", "second" are only used for the purpose of description, and cannot be understood as indicating or implying relative importance, unless otherwise explicitly specified and limited; the terms "connection", "installation", "fixation" and the like should be understood in a broad sense, for example, "connection" can be fixed connection, can also be detachable connection, or integral connection; can be directly connected, or indirectly connected through an intermediate medium. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0154] In the description of the present application, the terms "one embodiment", "some embodiments", "a specific embodiment" and the like mean that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present application, the illustrative description of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner.

[0155] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. For those skilled in the art, the present application can have various modifications and changes. Any modification, equivalent replacement, improvement, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for generating a ship berthing scheme, characterized in that, include: Collect port structural parameter information, ship environmental data information, dynamic information of adjacent ships, and ship status information; The first port model is determined based on the port structure parameter information; Based on the first port model, port agents, tugboat agents, and ship agents are determined respectively. The port agent aims at load balancing, the tugboat agent aims at minimizing the sum of fuel consumption and route cost, and the ship agent aims at short berthing time and low safety risk. Construct a three-way game matrix based on the port agent, the tugboat agent, and the ship agent; The first ship berthing scheme is determined based on the port structure parameter information, the ship's environmental data information, the dynamic information of adjacent ships, the ship's own status information, and the three-party game matrix. The second ship berthing plan is determined based on the first ship berthing plan and the neural network model.

2. The method for generating a ship berthing scheme according to claim 1, characterized in that, The step of determining the first port model based on the port structure parameter information includes: A port structure feature database is obtained, wherein the port structure feature database includes: multiple second port models and first port structure feature data corresponding to the second port models; Geometric feature extraction is performed on the port structure parameter information to obtain multiple first data information; Multiple second port structure feature data are obtained by performing coordinate mapping on multiple first data information; The first port model is determined from multiple second port structure feature data and the port structure feature database.

3. The method for generating a ship berthing scheme according to claim 1, characterized in that, The step of constructing a three-way game matrix based on the port agent, the tugboat agent, and the ship's agent includes: The port strategy space is determined based on the port agent. A first utility function is generated based on the port strategy space; The tugboat strategy space is determined based on the tugboat agent; A second utility function is generated based on the tugboat strategy space; The ship's strategy space is defined according to the ship's intelligent agent; A third utility function is generated based on the ship's strategy space; The three-way game matrix is ​​constructed based on the first utility function, the second utility function, and the third utility function.

4. The method for generating a ship berthing scheme according to claim 1, characterized in that, The step of determining the first ship berthing scheme based on the port structure parameter information, the ship's environmental data information, the dynamic information of adjacent ships, the ship's own state information, and the three-party game matrix includes: The port structure parameter information, the ship's environmental data information, and the dynamic information of the adjacent ships are respectively input into the three-party game matrix to obtain the first data matrix; The first ship berthing scheme is obtained by solving the first data matrix using a Nash equilibrium solver.

5. The method for generating a ship berthing scheme according to claim 1, characterized in that, The step of determining the first ship berthing scheme based on the port structure parameter information, the ship's environmental data information, the dynamic information of adjacent ships, the ship's own state information, and the three-party game matrix includes: When the ship is in a crosswind environment, the conflict-compromise algorithm is used to adjust the three-party game matrix to obtain the second data matrix. The port structure parameter information, the ship's environmental data information, and the dynamic information of adjacent ships are input into the second data matrix to obtain the third data matrix; The first ship berthing scheme is obtained by solving the third data matrix using a Nash equilibrium solver.

6. The method for generating a ship berthing scheme according to claim 1, characterized in that, The step of determining the second ship berthing plan based on the first ship berthing plan and the neural network model includes: Obtain the neural network model; The first ship berthing scheme is input into the neural network model to obtain the second ship berthing scheme.

7. The method for generating a ship berthing scheme according to claim 6, characterized in that, Before the step of obtaining the neural network model, the method for generating the ship berthing plan further includes: Determine the safety score, efficiency score, and energy efficiency score; A reward function is constructed based on the safety score, the efficiency score, the energy efficiency score, the safety weight, the efficiency weight, and the energy efficiency weight. Obtain historical berthing data for multiple vessels; The neural network model is determined based on multiple historical ship berthing data and the reward function.

8. A system for generating ship berthing plans, characterized in that, include: The acquisition module is used to acquire port structural parameter information, ship environmental data information, dynamic information of adjacent ships, and ship status information. The first determining module is used to determine the first port model based on the port structure parameter information; The second determining module is used to determine the port agent, tugboat agent and ship agent respectively according to the first port model. The port agent aims at load balancing, the tugboat agent aims at minimizing the sum of fuel consumption and route cost, and the ship agent aims at short berthing time and low safety risk. The calculation module is used to construct a three-way game matrix based on the port agent, the tugboat agent, and the ship agent. The third determining module is used to determine the first ship berthing scheme based on the port structure parameter information, the ship's environmental data information, the dynamic information of adjacent ships, the ship's own state information, and the three-party game matrix. An optimization module is used to determine a second ship berthing scheme based on the first ship berthing scheme and the neural network model.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method for generating a ship berthing scheme as described in any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for generating a ship berthing scheme as described in any one of claims 1 to 7.

Images