Ship shore-end intelligent management method, system and equipment and storage medium

Abstract

The invention discloses a ship shore-end intelligent management method, system and device and a storage medium, and the method comprises the steps: carrying out the ship position prediction based on an electronic chart, the current navigation state data of a ship, and the current environment data, and obtaining the information of a to-be-docked ship according to the future position information of the ship, constructing a port operation knowledge graph based on the information of the ships to arrive at the port, the equipment data of the ships approaching and leaving the port, the historical tug operation data and the port data, and predicting the tug workload demand based on the port operation knowledge graph; taking tug operation data as a state space, taking a tug scheduling strategy and a cargo stacking layout adjustment strategy as an action space, taking maximization of wharf operation efficiency as a reward target, and adopting a reinforcement learning method and a simulated annealing method to carry out wharf operation scheduling optimization to obtain an optimal wharf operation scheduling strategy. According to the method, the port working efficiency is improved, manual interference is not needed when the wharf operation scheduling strategy is obtained, and autonomous decision-making can be completely achieved.

Description

Technical Field

[0001] The present application relates to the field of ship management technology, and in particular to a method, system, device and storage medium for intelligent shore-side ship management. Background Art

[0002] Intelligent shore-based ship management is a key development direction for the modern shipping industry. It integrates sensors, communications, the Internet of Things, and artificial intelligence technologies to enable real-time monitoring of ship operating conditions, route optimization, and energy efficiency management. In recent years, with the booming shipping industry, an increasing number of container ports have become overcrowded. Tugboat pilotage and berthing assistance, a critical process for ships entering ports, directly impacts port operations and service quality. This presents a complex scheduling and management challenge for port operators. Summary of the Invention

[0003] The present application provides a method, system, device and storage medium for intelligent management of ships on shore, which are used to improve port work efficiency.

[0004] In view of this, the first aspect of the present application provides a method for intelligent shore-side management of ships, comprising:

[0005] Predict the ship's position based on electronic charts, the ship's current navigation status data, and current environmental data to obtain the ship's future position information;

[0006] Acquiring information about ships expected to arrive at the port based on the future position information of the ship, constructing a port operation knowledge graph based on the information about ships expected to arrive at the port, equipment data of ships that have recently departed the port, historical tugboat operation data, and port data, and forecasting tugboat workload demand based on the port operation knowledge graph;

[0007] Taking tugboat operation data as the state space, tugboat scheduling strategies and cargo stacking layout adjustment strategies as the action space, and maximizing terminal operation efficiency as the reward goal, the reinforcement learning method and simulated annealing method are used to optimize terminal operation scheduling and obtain the optimal terminal operation scheduling strategy. The tugboat operation data includes the tugboat workload demand forecast results, tugboat status, current port berth occupancy status, and current port cargo stacking situation.

[0008] Optionally, the predicting of the ship's position based on the electronic nautical chart, the ship's current navigation status data, and the current environmental data to obtain the ship's future position information includes:

[0009] The electronic nautical chart is constructed as a graph structure, where nodes are port facilities and waypoints, and edges represent the connection between nodes;

[0010] Construct the feature vector of each node based on the current navigation status data and current environmental data of the ship;

[0011] The feature vector of each node is updated through a graph neural network, and the updated feature vector of each node is mapped to the position coordinates of the ship at a future time to obtain the future position information of the ship.

[0012] Optionally, the ship position prediction is performed based on the electronic nautical chart, the current navigation status data of the ship, and the current environmental data to obtain the future position information of the ship, and then further includes:

[0013] Taking the ship's current navigation status, future position information and environmental data as the state space and the ship scheduling strategy as the action space, the Monte Carlo tree search method is used to calculate the value of each action based on the state reached after executing each action, starting from the current state.

[0014] The action with the highest value is selected to generate a ship dispatching strategy, and the ship is controlled to sail to the port according to the ship dispatching strategy.

[0015] Optionally, constructing a port operation knowledge graph based on the expected arrival ship information, the equipment data of the recently departed ships, the historical tugboat operation data, and the port data, and performing tugboat workload demand forecasting based on the port operation knowledge graph includes:

[0016] Extracting entities and entity relationships from the information on vessels expected to arrive at the port, the equipment data of vessels that are about to depart from the port, the historical tugboat operation data, and the port data; the entities include the type of vessels entering and leaving the port, the size of vessels entering and leaving the port, the expected berthing time, the type of tugboat, and the port facilities; the entity relationships include the operational association between the vessel and the tugboat, and the location relationship between the vessel and the port facilities;

[0017] A port operation knowledge graph is constructed with the entities as nodes and the entity relationships as edges, and the nodes and edges in the port operation knowledge graph are mapped to a low-dimensional vector space using knowledge graph embedding technology to obtain entity vector representations and relationship vector representations;

[0018] The relationship probabilities between different types of ships entering and leaving the port and tugboat workloads are calculated based on the entity vector representation and the relationship vector representation to obtain a tugboat workload demand prediction result.

[0019] Optionally, after obtaining the tugboat workload demand forecast result, the method further includes:

[0020] The tugboat workload demand forecast result is revised according to real-time port operation data.

[0021] Optionally, the method further includes:

[0022] If it is identified that the current terminal operation scenario involves multiple scenario types, the terminal operation scheduling strategy corresponding to each scenario type is obtained according to the scenario type of the current terminal operation scenario, and the weighted sum of each terminal operation scheduling strategy is performed through the fusion weight to obtain the optimal terminal operation scheduling strategy.

[0023] Optionally, the fusion weight is obtained as follows:

[0024] Calculating the scene similarity between the scene feature vector corresponding to each scene type in the scene type of the current terminal operation scene and the scene feature vector of the current terminal operation scene;

[0025] A fusion weight is calculated based on the scene similarity.

[0026] The second aspect of the present application provides a shore-side intelligent management system for ships, comprising:

[0027] The ship position prediction module is used to predict the ship's position based on the electronic nautical chart, the ship's current navigation status data and current environmental data, and obtain the ship's future position information;

[0028] A tugboat workload demand prediction module is used to obtain information about ships expected to arrive at the port based on the future position information of the ship, construct a port operation knowledge graph based on the information about ships expected to arrive at the port, equipment data of ships that have recently left the port, historical tugboat operation data, and port data, and perform tugboat workload demand prediction based on the port operation knowledge graph;

[0029] The terminal operation optimization module is used to optimize terminal operation scheduling using tugboat operation data, current port berth occupancy status, and current port cargo stacking status as the state space, tugboat scheduling strategy and cargo stacking layout adjustment strategy as the action space, and maximizing terminal operation efficiency as the reward goal. The module adopts reinforcement learning method and simulated annealing method to optimize terminal operation scheduling and obtain the optimal terminal operation scheduling strategy. The tugboat operation data includes the tugboat workload demand prediction result, tugboat status, current port berth occupancy status, and current port cargo stacking status.

[0030] A third aspect of the present application provides an electronic device, the device comprising a processor and a memory;

[0031] The memory is used to store program code and transmit the program code to the processor;

[0032] The processor is used to execute any one of the ship shore intelligent management methods described in the first aspect according to the instructions in the program code.

[0033] In a fourth aspect, the present application provides a computer-readable storage medium for storing program code, which, when executed by a processor, implements the ship shore intelligent management method described in any one of the first aspects.

[0034] It can be seen from the above technical solutions that this application has the following advantages:

[0035] The intelligent shore-side ship management method provided by this application obtains information about ships arriving at the port by predicting their future positions, and then predicts the workload demand of tugboats. Based on tugboat operation data and current port data, etc., terminal operation scheduling is optimized with the goal of maximizing terminal operation efficiency, and the optimal terminal operation scheduling strategy is obtained, thereby improving port operation efficiency. In addition, this application does not require human intervention when obtaining the terminal operation scheduling strategy, and can fully realize autonomous decision-making. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 A schematic flow chart of a method for intelligent shore-based management of ships provided in an embodiment of the present application;

[0038] Figure 2 A structural diagram of a ship shore-side intelligent management system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0039] In order to help those skilled in the art better understand the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0040] For easier understanding, please refer to Figure 1 , the embodiment of the present application provides a method for intelligent management of ships on shore, including:

[0041] Step 110: Predict the ship's position based on the electronic nautical chart, the ship's current navigation status data, and the current environmental data to obtain the ship's future position information;

[0042] Based on the large model technology, the standard electronic nautical chart and the real-time navigation status data of the tugboat fleet and the ship can be integrated and analyzed. According to the historical navigation data, the current navigation status data of the ship (current position, current speed, current heading) and the port's tides, water flow and other environmental factors, the future position coordinates of the ship can be predicted. The electronic nautical chart is constructed as a graph structure, in which the nodes represent geographical elements such as port facilities and key points of the waterway, and the edges represent the connection relationship between the nodes. The current position, current speed, current heading and current environmental factors of the ship are used as the feature vectors of the nodes, and the graph structure is set as follows: , is a node set, is the edge set. Each node With characteristic vector , including the ship’s current position, current speed, current course, current tide, current current, etc. Using the message passing mechanism of the graph neural network, the nodes The eigenvector of To update:

[0043]

[0044] Where, For nodes Updated feature vector; For nodes The set of neighbor nodes of For nodes Neighbor nodes The eigenvector of 、 is the weight matrix, is the bias term, is the activation function (such as ReLU function). Then the fully connected layer (FC) in the graph neural network is used to update the feature vectors of all nodes Mapped to the future position coordinates of the ship , get the future position information of the ship (such as the position coordinates of the ship at the next moment), that is .

[0045] Furthermore, after obtaining the future position information of the ship, the ship scheduling strategy can be generated based on the Monte Carlo tree search method, combining the current navigation state and future position information of the ship. The state space S is composed of the ship's navigation state (position, speed, heading, etc.) and environmental data (tide, current, other ship positions), and the ship scheduling strategy (changing heading, speed adjustment, avoidance decision, etc.) is used as the action space A. The Monte Carlo tree search method is used to perform different actions starting from the current state s. After the state transition, calculate the value of each action , select the action with the highest value as the generated ship scheduling strategy, and the value calculation formula is:

[0046]

[0047] Where, From the state Execute an action The number of simulations, From the state in the simulation Execute an action Transfer to state The rewards earned can be set based on whether navigation conflicts are reduced, whether congestion is reduced, and the degree of congestion reduction. During navigation, the generated ship scheduling strategy is used to control the ship to the port, thereby avoiding navigation conflicts in complex intersection areas and reducing congestion.

[0048] Furthermore, the embodiment of the present application is also based on a self-supervised learning method to learn the unsupervised features of multi-source heterogeneous data of ships, mine the hidden patterns and relationships in the data, construct a feature representation of the ship's working status, and combine historical data such as the ship's online status and working status with real-time sensor data to predict the tugboat's standby, navigation, towing, pushing and other working states, thereby realizing equipment management perception, ship safe navigation guarantee, personnel management and deployment, etc.

[0049] The proximal strategy optimization algorithm in reinforcement learning is introduced. The accuracy of the ship working status prediction results and the effectiveness of advance planning of maintenance resources are used as reward goals. The parameters of the ship working status prediction model are dynamically adjusted to enable the ship working status prediction model to better adapt to changes in the ship's operating environment and working mode, and to achieve maintenance decision-making strategies based on status to reduce downtime, risk warnings, and reasonable task allocation.

[0050] The ship working status feature learning model based on self-supervised learning is as follows:

[0051] The ship's operating frequency is measured using an automatic encoder (AE). , Equipment operating status indicators , Maintenance record times , crew working hours Encode and decode multiple source data. Assume the input data vector , the encoder maps it to a low-dimensional feature space , the decoder then Reconstruction to obtain reconstructed features By minimizing the reconstruction error Train the model.

[0052] The parameters of the prediction model based on proximal strategy optimization are adjusted as follows:

[0053] Establish a strategic network , where As parameters, is the working state of the ship (the feature vector obtained by the above self-supervised learning express), The prediction model outputs the ship working status prediction results (such as the probability of the tugboat entering the maintenance state and the maintenance time window). By maximizing the objective function Update parameters , where is the importance sampling ratio, is the advantage function, which is obtained by comparing the predicted results of ship working status with the actual ship working status. The function is used to limit the range of importance sampling ratio. is a hyperparameter.

[0054] Step 120: Obtain information about ships expected to arrive at the port based on the future position information of the ship, construct a port operation knowledge graph based on the information about ships expected to arrive at the port, equipment data of ships that have recently departed the port, historical tugboat operation data, and port data, and perform tugboat workload demand forecasting based on the port operation knowledge graph;

[0055] Based on a ship's future position, information about ships expected to arrive at the port, such as each ship's estimated berthing time, can be obtained. Knowledge graph technology can be used to construct a port operations knowledge graph based on ship arrival and departure plans, information about expected arrivals, historical tugboat operation data, and multi-source knowledge such as port traffic and weather. By leveraging large models and knowledge graph reasoning, the relationship between different types of incoming and outgoing ships and tugboat workload can be analyzed to predict tugboat workload demand, thereby improving the accuracy of forecasts for tugboat operations such as assisting ships with course adjustment, speed control, and berthing.

[0056] The tugboat workload demand prediction process based on knowledge graph reasoning is as follows:

[0057] Constructing a knowledge graph for port operations , where It is a node set, which contains knowledge entities such as ship type, size, expected berthing time, tugboat type, port facilities, etc. is a set of edges, representing the relationship between entities (such as the operational association between ships and tugboats, the location relationship between ships and port facilities, etc.). Using knowledge graph embedding technology, the nodes and edges are mapped to a low-dimensional vector space to obtain the entity vector representation. and relation vector representation , and then calculate the relationship probability between different types of ships entering and leaving the port and the tugboat operation volume based on the entity vector representation and the relationship vector representation. Specifically, the relationship prediction model (such as the TransE model) can be used to calculate the relationship probability between different types of ships entering and leaving the port and the tugboat operation volume, that is, the type vector of the required tugboat and the number of tugboat operations The probability distribution of tugboat workload demand forecast results are obtained, including the type of tugboat required and the number of tugboat operations.

[0058] Furthermore, in order to improve the accuracy of the tugboat workload demand forecast results, real-time sensing technology (such as data collection by IoT devices) can be combined to obtain the actual port operation status (real-time ship arrival time, berth occupancy status, etc.) in real time, and the tugboat workload demand forecast results can be dynamically corrected. The prediction result correction model based on real-time sensing data R (including real-time ship arrival time, berth occupancy status, etc.) is as follows:

[0059] Correction functions can be learned based on historical port operation data , and then the tugboat workload demand prediction result is corrected according to the current perception data R, that is, the corrected tugboat workload demand prediction result is .

[0060] Step 130: Using the tugboat operation data, the current port berth occupancy status, and the current port cargo stacking status as the state space, the tugboat scheduling strategy and the cargo stacking layout adjustment strategy as the action space, and maximizing the terminal operation efficiency as the reward goal, the reinforcement learning method and the simulated annealing method are used to optimize the terminal operation scheduling to obtain the optimal terminal operation scheduling strategy.

[0061] In one embodiment, a method combining reinforcement learning and simulated annealing is used to optimize terminal operation scheduling based on tugboat workload demand forecasts, tugboat status, current berth occupancy at each port, and historical tugboat operation data. This optimization is achieved by exploring different tugboat scheduling schemes based on reinforcement learning, with maximizing terminal operation efficiency as the reward objective. Simulated annealing is used to search for optimal solutions, avoiding local optima and improving search efficiency.

[0062] The optimization process of terminal operation scheduling based on reinforcement learning and simulated annealing algorithm is as follows:

[0063] Assume the state space The current port terminal operation status (berth occupancy status, tugboat workload demand forecast results, tugboat status, cargo stacking status, etc.), action space Tugboat scheduling strategy and cargo stacking layout adjustment strategy; with maximizing terminal operation efficiency as the reward goal, the reinforcement learning method and simulated annealing method are used to optimize terminal operation scheduling and obtain the optimal terminal operation scheduling strategy. Among them, the reinforcement learning strategy network By maximizing long-term cumulative rewards To update the parameters , where In state Execute an action Rewards received, is the discount factor. In each iteration, the simulated annealing algorithm Accept the inferior solution, where is the difference between the objective function value of the new solution and the current solution, is the current temperature, and the temperature gradually decreases with each iteration. Finally, the action with the highest cumulative reward is obtained as the optimal terminal operation scheduling strategy.

[0064] In another embodiment, terminal operation scheduling is optimized for different tugboat operation scenarios. For different tugboat operation scenarios, such as port routine scheduling, high-density ship scheduling, ship emergency rescue, collaborative operations, etc., a large model is used to analyze tugboat operation data in various scenarios to build a training data set. ,in The scene feature vector after unified representation (such as ship arrival time, terminal operation task arrangement, etc.) is the scene category label.

[0065] The deep neural network (DNN) is used as the basic model architecture, with a uniformly represented scene feature vector as input and output scene category. During the model training process, the cross entropy loss function is used. To optimize:

[0066]

[0067] Where, is the number of training samples, is the number of scene categories, Indicates the The samples belong to The true label of the class scene (0 or 1), The model predicts The samples belong to The probability of the scenario.

[0068] Use stochastic gradient descent and its variants (such as Adagrad, Adadelta, Adam, etc.) to update the model parameters. Let the model parameters be , the learning rate is , and are the first-order moment estimate and the second-order moment estimate of the gradient at the current time t, is the gradient at the current moment t, then the parameter The update formula is:

[0069]

[0070] in, and is the hyperparameter of the Adam optimizer, usually set to 0.9 and 0.999, is a very small constant used to prevent the denominator from being zero. The deep neural network is trained through the above process, and the trained deep neural network is used as the scene classifier.

[0071] In practical applications, the feature vector of the current scene is obtained in real time , based on the trained scene classifier, determine the current scene type. Suppose the probability distribution of each scene category is , the scene category set is ,but:

[0072]

[0073] Where, For scene classifiers, The output score of Judge and obtain the predicted category of the current scene :

[0074]

[0075] When a scene switch or mixed scene is detected, the appropriate basic strategy is selected from the pre-trained strategy library based on the scene recognition results. The strategy library stores the decision strategies trained for different scenarios (i.e., terminal operation scheduling strategies for different types of operation scenarios). For mixed scenarios, multiple basic strategies can be fused in a linear combination. Assuming that the current operation scenario involves the scene 、 , and the corresponding basic strategies are 、 , the fusion weights are 、 , then the adjusted strategy for:

[0076]

[0077] According to the proportion or importance of different scene features in the current mixed scene, the cosine similarity between the feature vectors of different operation scenes can be calculated to measure their proportion and obtain the fusion weight.

[0078] set up 、 For work scenes 、 The scene feature vector of the current scene The cosine similarities are 、 , then the fusion weight calculation formula is:

[0079]

[0080] When the current operation scene involves three types, the fusion weights are 、 、 The specific calculation process is similar to the above process, which is obtained by calculating the proportion of the similarity between the scene feature vectors of various scene types and the current scene type in all similarities, that is:

[0081]

[0082] Where, is the fusion weight corresponding to the i-th scene type, n is the number of scene types involved in the current terminal operation scene, is the similarity between the scene feature vector of the i-th scene type and the current terminal operation scene.

[0083] The intelligent shore-side ship management method provided by this application obtains information about ships arriving at the port by predicting their future positions, and then predicts the workload demand of tugboats. Based on tugboat operation data and current port data, etc., terminal operation scheduling is optimized with the goal of maximizing terminal operation efficiency, and the optimal terminal operation scheduling strategy is obtained, thereby improving port operation efficiency. In addition, this application does not require human intervention when obtaining the terminal operation scheduling strategy, and can fully realize autonomous decision-making.

[0084] This application uses large model technology to create a decision-making model for complex situations such as sudden changes, changes in navigation rules, or sudden ship failures in real time. Through massive data training, the model can quickly identify abnormal situations, generate the optimal response strategy in a short period of time, and reduce the frequency of manual intervention. Integrate multi-domain knowledge such as ship navigation knowledge and port operation rules into the large model, and possess deep reasoning capabilities. When faced with edge cases and complex problems, autonomous analysis is performed based on the integrated knowledge, and scientific and reasonable decisions are independently generated to reduce dependence on expert experience and manual judgment. Introduce a reinforcement learning mechanism to autonomously explore the results of different decisions, optimize decision-making strategies through continuous trial and error, enhance the model's ability and confidence in autonomous decision-making in complex scenarios, and independently and effectively deal with various problems in actual applications. Possess a general understanding and response capability across scenarios, and quickly adjust strategies when switching between different scenarios or when mixed scenarios appear, to provide coherent and effective solutions.

[0085] Please refer to Figure 2 The embodiment of the present application also provides a shore-side intelligent management system for ships, including:

[0086] The ship position prediction module is used to predict the ship's position based on the electronic nautical chart, the ship's current navigation status data and current environmental data, and obtain the ship's future position information;

[0087] The tugboat workload demand forecasting module is used to obtain information about ships expected to arrive at the port based on the future position information of the ships, build a port operation knowledge graph based on the information about ships expected to arrive at the port, equipment data of ships that have recently departed the port, historical tugboat operation data, and port data, and then predict the tugboat workload demand based on the port operation knowledge graph;

[0088] The terminal operation optimization module is used to optimize terminal operation scheduling using tugboat operation data, current port berth occupancy status, and current port cargo stacking status as the state space, tugboat scheduling strategy and cargo stacking layout adjustment strategy as the action space, and maximizing terminal operation efficiency as the reward goal. It uses reinforcement learning and simulated annealing methods to optimize terminal operation scheduling and obtain the optimal terminal operation scheduling strategy.

[0089] As a further improvement, the system further includes: a ship scheduling strategy generation module for:

[0090] Taking the ship's current navigation status, future position information and environmental data as the state space and the ship scheduling strategy as the action space, the Monte Carlo tree search method is used to calculate the value of each action based on the state reached after executing each action, starting from the current state.

[0091] The action with the highest value is selected to generate a ship scheduling strategy, and the ship is controlled to sail to the port according to the ship scheduling strategy.

[0092] As a further improvement, the system further includes: a correction module for correcting the tugboat workload demand prediction result according to real-time port operation data.

[0093] As a further improvement, the system also includes: a fusion module, which is used to obtain the terminal operation scheduling strategy corresponding to each scene type according to the scene type of the current terminal operation scene if it is identified that the current terminal operation scene involves multiple scene types, and to perform weighted summation of each terminal operation scheduling strategy through fusion weights to obtain the optimal terminal operation scheduling strategy.

[0094] An embodiment of the present application further provides an electronic device, the device including a processor and a memory;

[0095] The memory is used to store program codes and transmit the program codes to the processor;

[0096] The processor is used to execute the ship shore intelligent management method in the aforementioned method embodiment according to the instructions in the program code.

[0097] An embodiment of the present application also provides a computer-readable storage medium, which is used to store program code. When the program code is executed by a processor, the ship shore intelligent management method in the aforementioned method embodiment is implemented.

[0098] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0099] In the specification of this application and the above-mentioned drawings, the terms "first," "second," "third," "fourth," etc. (if any) are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. In addition, the terms "including" and "having," and any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or apparatus that includes a series of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not explicitly listed or inherent to such process, method, product, or apparatus.

[0100] It should be understood that in this application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships can exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or plural.

[0101] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely schematic. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0102] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0103] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.

[0104] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the method described in each embodiment of the present application through a computer device (which can be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (full name: Read-Only Memory, English abbreviation: ROM), random access memory (full name: Random Access Memory, English abbreviation: RAM), disk or optical disk, and other media that can store program code.

[0105] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for intelligent management of ships on shore, characterized in that: include: Predict the ship's position based on electronic charts, the ship's current navigation status data, and current environmental data to obtain the ship's future position information; Acquiring information about ships expected to arrive at the port based on the future position information of the ship, constructing a port operation knowledge graph based on the information about ships expected to arrive at the port, equipment data of ships that have recently departed the port, historical tugboat operation data, and port data, and forecasting tugboat workload demand based on the port operation knowledge graph; Taking tugboat operation data as the state space, tugboat scheduling strategies and cargo stacking layout adjustment strategies as the action space, and maximizing terminal operation efficiency as the reward goal, the reinforcement learning method and simulated annealing method are used to optimize terminal operation scheduling and obtain the optimal terminal operation scheduling strategy. The tugboat operation data includes the tugboat workload demand forecast results, tugboat status, current port berth occupancy status, and current port cargo stacking situation.

2. The method for intelligent ship shore management according to claim 1, characterized in that: The ship position prediction based on the electronic nautical chart, the current navigation status data of the ship and the current environmental data to obtain the future position information of the ship includes: The electronic nautical chart is constructed as a graph structure, where nodes are port facilities and waypoints, and edges represent the connection between nodes; Construct the feature vector of each node based on the current navigation status data and current environmental data of the ship; The feature vector of each node is updated through a graph neural network, and the updated feature vector of each node is mapped to the position coordinates of the ship at a future time to obtain the future position information of the ship.

3. The method for intelligent ship shore management according to claim 1, characterized in that: The method of predicting the ship's position based on the electronic nautical chart, the ship's current navigation status data, and the current environmental data to obtain the ship's future position information further includes: Taking the ship's current navigation status, future position information and environmental data as the state space and the ship scheduling strategy as the action space, the Monte Carlo tree search method is used to calculate the value of each action based on the state reached after executing each action, starting from the current state. The action with the highest value is selected to generate a ship dispatching strategy, and the ship is controlled to sail to the port according to the ship dispatching strategy.

4. The method for intelligent ship shore management according to claim 1, characterized in that: The method of constructing a port operation knowledge graph based on the information of the expected arriving ships, the equipment data of the ships that have recently departed the port, the historical tugboat operation data, and the port data, and predicting the tugboat workload demand based on the port operation knowledge graph includes: Extracting entities and entity relationships from the information on vessels expected to arrive at the port, the equipment data of vessels that are about to depart from the port, the historical tugboat operation data, and the port data; the entities include the type of vessels entering and leaving the port, the size of vessels entering and leaving the port, the expected berthing time, the type of tugboat, and the port facilities; the entity relationships include the operational association between the vessel and the tugboat, and the location relationship between the vessel and the port facilities; A port operation knowledge graph is constructed with the entities as nodes and the entity relationships as edges, and the nodes and edges in the port operation knowledge graph are mapped to a low-dimensional vector space using knowledge graph embedding technology to obtain entity vector representations and relationship vector representations; The relationship probabilities between different types of ships entering and leaving the port and tugboat workloads are calculated based on the entity vector representation and the relationship vector representation to obtain a tugboat workload demand prediction result.

5. The method for intelligent ship shore management according to claim 4, characterized in that: After obtaining the tugboat workload demand forecast result, the following steps are further included: The tugboat workload demand forecast result is revised according to real-time port operation data.

6. The method for intelligent ship shore management according to claim 1, characterized in that: The method further comprises: If it is identified that the current terminal operation scenario involves multiple scenario types, the terminal operation scheduling strategy corresponding to each scenario type is obtained according to the scenario type of the current terminal operation scenario, and the weighted sum of each terminal operation scheduling strategy is performed through the fusion weight to obtain the optimal terminal operation scheduling strategy.

7. The method for intelligent ship shore management according to claim 6, characterized in that: The process of obtaining the fusion weight is as follows: Calculating the scene similarity between the scene feature vector corresponding to each scene type in the scene type of the current terminal operation scene and the scene feature vector of the current terminal operation scene; A fusion weight is calculated based on the scene similarity.

8. A ship shore intelligent management system, characterized in that: include: The ship position prediction module is used to predict the ship's position based on the electronic nautical chart, the ship's current navigation status data and current environmental data, and obtain the ship's future position information; A tugboat workload demand prediction module is used to obtain information about ships expected to arrive at the port based on the future position information of the ship, construct a port operation knowledge graph based on the information about ships expected to arrive at the port, equipment data of ships that have recently left the port, historical tugboat operation data, and port data, and perform tugboat workload demand prediction based on the port operation knowledge graph; The terminal operation optimization module is used to optimize terminal operation scheduling using tugboat operation data, current port berth occupancy status, and current port cargo stacking status as the state space, tugboat scheduling strategy and cargo stacking layout adjustment strategy as the action space, and maximizing terminal operation efficiency as the reward goal. The module adopts reinforcement learning method and simulated annealing method to optimize terminal operation scheduling and obtain the optimal terminal operation scheduling strategy. The tugboat operation data includes the tugboat workload demand prediction result, tugboat status, current port berth occupancy status, and current port cargo stacking status.

9. An electronic device, characterized in that: The device includes a processor and a memory; The memory is used to store program code and transmit the program code to the processor; The processor is used to execute the ship shore intelligent management method according to any one of claims 1 to 7 according to the instructions in the program code.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium is used to store program code, and when the program code is executed by the processor, the ship shore intelligent management method according to any one of claims 1 to 7 is implemented.

Images