Intelligent ship formation navigation control method

Abstract

The invention discloses an intelligent ship formation navigation control method, and relates to the technical field of ship navigation control, and the control method comprises the specific steps: S100, multi-source data collection: installing various sensors and communication equipment on a ship, according to the invention, a ship AIS sensor, a meteorological radar echo sensor, an underwater sonar detection sensor, a satellite communication device and a VHF communication device are installed on a ship, so that comprehensive acquisition of various environmental information and states of the ship in the ship navigation process is realized; the multi-source heterogeneous data is subjected to cleaning, conversion and normalization processing of a data center and then is efficiently fused by a deep hybrid network model H-Net, and deep feature extraction is performed by using a convolution attention module and an improved gating cycle unit iGRU, so that the efficiency and accuracy of data processing are greatly improved, and the data processing efficiency and accuracy are greatly improved. And the key information in the navigation environment can be captured more accurately.

Description

Technical Field

[0001] The present invention relates to the technical field of ship navigation control, and in particular to a method for controlling the navigation of an intelligent ship formation. Background Art

[0002] With the vigorous development of global trade, maritime transport, as the main mode of transport for international trade, has continuously expanded its transport volume and scope. Coastal and inland shipping, as an important part of water transport, undertakes a large amount of cargo transportation tasks. However, coastal and inland shipping faces many complex environmental factors. On the one hand, the traffic in water cities is complex, the waterways are narrow and changeable, and the water depth, width, and bending radius parameters of different waterways vary greatly, which brings great challenges to ship navigation. On the other hand, the weather conditions are unpredictable, and heavy fog, heavy rain and severe weather occur from time to time. These weather conditions will not only affect the ship's line of sight, but also interfere with the ship's navigation equipment, increasing the risk of ship navigation. In this context, how to ensure the safe and efficient navigation of ships in complex environments has become an important issue that needs to be solved urgently.

[0003] Traditional ship navigation methods have obvious shortcomings in dealing with complex water city traffic and changeable weather. In route planning, traditional methods often lack sufficient consideration of the coordinated relationship between ships. They simply plan routes based on the performance and navigation tasks of individual ships, which makes it easy for ships to interfere with each other and increase the risk of collision during navigation. During navigation, there is a lack of effective real-time communication and coordination mechanisms between ships, and they cannot obtain the position and speed information of other ships in a timely manner, making it difficult to achieve coordinated navigation and formation maintenance. In addition, traditional methods lack the ability to respond quickly and dynamically adjust routes in the face of sudden weather changes, and often require manual intervention. This is not only inefficient, but may also lead to more serious consequences due to human errors. Moreover, there is room for improvement in energy saving and safety of traditional navigation methods, and they cannot well meet the needs of modern shipping for efficiency, safety and environmental protection.

[0004] Therefore, the development of an intelligent ship formation navigation control method effectively improves the safety, efficiency and flexibility of ship formation navigation, reduces fuel consumption, and brings innovative solutions to the field of intelligent shipping. Summary of the Invention

[0005] The purpose of the present invention is to make up for the shortcomings of the existing technology and provide an intelligent ship formation navigation control method. Through multi-source data fusion and deep feature extraction technology, it realizes comprehensive perception of the ship's navigation environment and its own status. In route planning, it uses game models and quantum evolution algorithms to plan the optimal route for the ship. In collaborative navigation control, it realizes collaborative navigation and dynamic route adjustment of ships through real-time communication and feedback mechanisms.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for controlling the navigation of an intelligent ship formation, wherein the specific steps of the control method are as follows:

[0007] S100, multi-source data acquisition: Various sensors and communication equipment are installed on ships to collect multi-source heterogeneous data such as ship AIS, weather radar echoes, underwater sonar detection, port scheduling, and ship position and speed, and transmit it to the control center for storage;

[0008] S200, Data Processing and Feature Extraction: Extract data from the data center, clean, transform, and normalize the data, fuse multi-source data, and use the deep hybrid network model H-Net for feature extraction;

[0009] S300, game-based route planning: A game model is constructed based on the collaborative relationships between ships. An objective function is set for each ship that integrates its own and the fleet's benefits. A strategy set is constructed that includes speed, heading, and path. A quantum evolutionary algorithm is used to find the Nash equilibrium solution. The control center then plans the route based on this strategy and combines it with the ship's performance parameters.

[0010] S400, Collaborative Navigation Control: The control center sends formation, relative position, and speed instructions to ships according to the planned route. After receiving the instructions, the ships adjust their status to achieve collaborative navigation. Status information is fed back in real time during navigation. In case of environmental changes, deviation correction formulas are introduced to adjust the route in real time.

[0011] S500, Strategy Evaluation Optimization: The control center collects ship navigation data, calculates comprehensive evaluation index values based on the shipping company's needs and cost structure, and adjusts the parameters of the deep hybrid network model H-Net and the dynamic game algorithm strategy by analyzing the factors that affect the comprehensive evaluation index values.

[0012] Furthermore, in said S100, the sensors and communication equipment used in the multi-source data collection are: ship AIS sensors, weather radar echo sensors, underwater sonar detection sensors, port scheduling sensors, ship position and speed sensors, satellite communication equipment and VHF communication equipment.

[0013] Furthermore, in the S200, the deep hybrid network model H-Net is used for feature extraction in the data processing and feature extraction. For the input fusion data X, the key area features are focused by the convolutional attention module CAM(X), and the formula is: CAM(X) = σ(W2×ReLU(W1×X))×X, where CAM(X) is the result of focusing on the key area features after the convolutional attention module processes the input data X, W1 and W2 are learnable weight matrices for extracting the key area features of the input data, σ is the activation function, and is input to the improved gated recurrent unit iGRU to process the time series features: z t=σ(W z ×[h t-1 ,x t ]), r t =σ(W r ×[h t-1 ,x t ]), Among them, x t is the input at time t, h t -1 is the hidden state at the last moment, z t is the update gate, r t is the reset gate, ⊙ represents element-by-element multiplication, W z 、W r , W is the weight matrix, where W z Used to calculate the update gate, W r Used to calculate the reset gate, W is used to calculate the candidate hidden state, is the calculated candidate hidden state, h t It is the hidden state at time t, and the final output feature vector F = H-Net(X).

[0014] Furthermore, the above S300 is based on the construction of a game model in route planning based on dynamic game:

[0015] (1) Determine the game participants: Each ship in the fleet is set as a participant in the game model, and each ship needs to make a strategic choice in the navigation decision;

[0016] (2) Setting the objective function: constructing an objective function for each ship that integrates its own and the fleet's benefits;

[0017] (3) Constructing a strategy set: clarifying the strategy set of each ship, including speed, heading, and path decision variables;

[0018] (4) Establish an information exchange mechanism: determine the information exchange rules between ships. Based on the extracted feature information, each ship can know the scope and impact of the strategies adopted by other ships, providing a basis for its own strategy selection;

[0019] (5) Clarify the profit calculation rules: Based on the objective function and the strategy selected by each ship, calculate the profit of each ship under different strategy combinations. By comparing the profits brought by different strategies, evaluate the advantages and disadvantages of the strategies and select the optimal strategy;

[0020] (6) Set the game termination conditions: when the preset number of iterations is reached, the strategies of each ship do not change for many consecutive times, or the overall benefits reach a stable state, the game process ends and the results are output.

[0021] Furthermore, in the above S300, the objective function of the ship's own and fleet's benefits in the game planning route is set, and the benefit function of ship i is U i , the calculation formula is: Among them, T i is the sailing time of ship i, C i is the fuel consumption cost, S i is the contribution of ship i to the safety of the formation, which is quantified based on maintaining a safe distance with other ships and avoiding collisions. Its value range is 0-1. ω1, ω2, and ω3 are weight coefficients, and ω1+ω2+ω3=1. It is dynamically adjusted according to the mission type.

[0022] Furthermore, in the above S300, the strategy set in the game planning route is set:

[0023] Speed strategy: The ship's optional sailing speed v i Limited by its own performance, it must meet v i-min ≤v i ≤v i-max , where v i-min is the minimum speed at which the ship can sail stably, v i-max It is the maximum sailing speed of the ship under the current load and sea conditions;

[0024] Heading strategy: the ship's heading θ i The value range is 0-360°, and its determination takes into account the direction of the channel, the distribution of obstacles, and meteorological conditions;

[0025] Path strategy: Path selection is based on a comprehensive analysis of the waterway map, real-time obstacle distribution, and port scheduling information, and a decision is made from feasible paths.

[0026] Furthermore, in the aforementioned S300, the Nash equilibrium solution obtained by the quantum evolutionary algorithm in the game planning route is as follows: Among them, Δθ is the angle value used for adjustment in the quantum evolution algorithm, f best is the optimal fitness value of the current group, f i is the fitness value of individual i, f avg is the average fitness value of the group, k1 and k2 are preset adjustment coefficients, which are determined by historical data regression analysis. k1 is the angle adjustment coefficient when the individual fitness is lower than the optimal value, reflecting the optimization intensity for the poorer individuals. k2 is the angle adjustment coefficient when the individual fitness is higher than or equal to the optimal value, reflecting the tendency to retain the better individuals.

[0027] Furthermore, in the collaborative navigation control, the route is adjusted in real time by introducing a deviation correction formula. Assuming that the original planned route is L0, the adjusted route is L1, and the current position of the ship is P, the route deviation angle θ and the distance deviation d are calculated to make corrections. The calculation formula is: d = |P-L1|, where and are the direction vectors of the original route and the adjusted route, respectively. According to θ and d, the correction function R(θ, d) is used to further optimize the adjusted route. The formula is: R(θ, d) = γ1×θ+γ2×d, where γ1 and γ2 are correction coefficients determined by historical navigation data. γ1 is the correction weight of the route deviation angle. A larger value indicates a higher priority for correcting the heading deviation. γ2 is the correction weight of the distance deviation. A larger value indicates a higher priority for correcting the position offset. Fine-tune L1 according to the value of R(θ, d).

[0028] Furthermore, the calculation of the comprehensive evaluation index value in the strategy evaluation optimization in S500 is as follows: Among them, I is the comprehensive evaluation index, T actual and T expeeted The actual sailing time and the expected sailing time are respectively, actual and C expected are the actual fuel consumption cost and the expected fuel consumption cost, S accident-free is the accident-free mileage, S total is the total voyage mileage, λ1, λ2, and λ3 are weight coefficients and λ1+λ2+λ3=1, which is determined according to shipping demand and cost structure.

[0029] Compared with the existing technology, this intelligent ship formation navigation control method has the following beneficial effects:

[0030] 1. The present invention realizes the comprehensive collection of various environmental information and the ship's own status during navigation by installing multiple sensors such as ship AIS sensors, weather radar echo sensors, underwater sonar detection sensors, as well as satellite communication equipment and VHF communication equipment on ships. These multi-source heterogeneous data are cleaned, converted and normalized by the data center, and then efficiently integrated by the deep hybrid network model H-Net. The convolutional attention module and the improved gated recurrent unit iGRU are used for deep feature extraction, which not only greatly improves the efficiency and accuracy of data processing, but also enables the present invention to more accurately capture key information in the navigation environment, providing a data basis for subsequent route planning and collaborative navigation control.

[0031] 2. In the route planning stage, the present invention constructs a game model based on the collaborative relationship of ships, sets an objective function for each ship that integrates its own benefits and the benefits of the formation, and solves the Nash equilibrium solution through the quantum evolution algorithm, thereby achieving a comprehensive consideration of ship performance parameters and navigation environment, and planning the optimal route. In the collaborative navigation control stage, the control center sends formation, relative position and speed instructions to the ships according to the planned route. After receiving the instructions, the ships adjust their status to achieve collaborative navigation. At the same time, real-time status information is fed back during navigation. When encountering environmental changes, the deviation correction formula can be quickly introduced to adjust the route in real time to ensure the continuity and safety of navigation. This not only improves the flexibility and adaptability of ship formation navigation, but also effectively reduces navigation risks and improves shipping efficiency through real-time feedback and dynamic adjustment mechanisms.

[0032] Other advantages, objects and features of the present invention will be described in part in the following description and, in part, will be apparent to those skilled in the art based on an examination of the following or may be learned from the practice of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] To more clearly illustrate the embodiments of the present invention 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 invention. Those skilled in the art can also derive other drawings based on these drawings without inventive effort.

[0034] Figure 1 This is a flow chart of a method for controlling the navigation of an intelligent ship formation;

[0035] Figure 2 This is a structural diagram of an intelligent ship formation navigation control method. DETAILED DESCRIPTION

[0036] In order to further illustrate the technical means and effects adopted by the present invention to achieve the predetermined purpose of the invention, the specific implementation methods, structures, features and effects of the present invention are described in detail below in conjunction with the accompanying drawings and preferred embodiments.

[0037] Example 1:

[0038] Large container ships sailing in formation in the waters near a busy port

[0039] Multi-source data collection: Each large container ship participating in the formation is equipped with ship AIS sensors, weather radar echo sensors, underwater sonar detection sensors, port scheduling sensors, ship position and speed sensors, and is also equipped with satellite communication equipment and VHF communication equipment. The ship AIS sensor collects the ship's own identification code, position, speed, and heading information; the weather radar echo sensor collects weather data of nearby sea areas, such as storm and fog weather information; the underwater sonar detection sensor detects underwater obstacles and water depth data; the port scheduling sensor receives scheduling instructions issued by the port, such as berthing order and channel occupancy information; the ship position and speed sensor monitors the ship's position and speed information in real time. These multi-source heterogeneous data are continuously transmitted to the control center through satellite communication equipment and VHF communication equipment for storage, providing a data basis for subsequent processing, such as Figure 1 shown.

[0040] Data processing and feature extraction: The control center extracts stored data from the data center and first cleans the data to remove erroneous data and duplicate redundant data caused by equipment failure and signal interference. Then, it converts data of different formats and dimensions into a unified usable format, and then normalizes the data to make it comparable. After that, the processed data is fused and recorded as X. The fused data is input into the deep hybrid network model H-Net. The convolutional attention module CAM(X) is used to focus on key area features. The calculation formula is: CAM(X) = σ(W2×ReLU(W1×X))×X, where CAM(X) is the result of the convolutional attention module focusing on key area features after processing the input data X. W1 and W2 are learnable weight matrices, and σ is the activation function. After processing by the convolutional attention module, the data highlights areas with potential collision risks around the ship and information on key nodes in busy waterways. The data is then input into the improved gated recurrent unit iGRU to process time series features. At time t, the update gate is calculated using the formula z t =σ(W z ×[h t-1 ,x t ]), r t =σ(W r ×[h t-1 ,x t ]), Among them, x t is the input at time t, h t -1 is the hidden state at the last moment, z t is the update gate, r t is the reset gate, ⊙ represents element-by-element multiplication, W z 、W r , W is the weight matrix, where W zUsed to calculate the update gate, W r Used to calculate the reset gate, W is used to calculate the candidate hidden state, is the calculated candidate hidden state, h t It is the hidden state at time t, and the final output feature vector F = H-Net(X) contains the key information of the ship's navigation status and the changes of the surrounding environment over time.

[0041] Game-based route planning: Each container ship in the fleet is set as a game participant, and an objective function is constructed for each ship that combines its own benefits with those of the fleet. The calculation formula is: Among them, T i is the sailing time of ship i, including the travel time from the departure port to the destination port and the time spent at the port; C i is the fuel consumption cost, which is related to the ship's speed, sailing distance, and engine performance; S i S is the contribution of ship i to the safety of the formation, which is quantified based on the situation of maintaining a safe distance with other ships and avoiding collisions. The value range is 0-1. For example, when a ship strictly follows the safe distance and responds to the avoidance command in time, S i The value is relatively high, ω1, ω2, and ω3 are weight coefficients, and ω1+ω2+ω3=1, which is determined according to the urgency of the container transportation task and the fuel price cost structure of the shipping market.

[0042] Construct a strategy set. In terms of speed strategy, the ship can choose the sailing speed v i Limited by its own performance, it needs to meet v i-min ≤v i ≤v i-max ,Near a busy port, considering the congestion of the waterway, the conditions of other ships and ,port regulations, the actual sailing speed of a ship is often lower than the maximum ,speed, and the speed needs to be flexibly adjusted according to the dynamic ,situation of the heading strategy, θ i The value range is 0-360°. When determining the heading, the direction of the channel, the direction of the port entrance, the trajectory of other ships and the distribution of obstacles are comprehensively considered. For example, to avoid other ships entering and leaving the port, the prescribed heading of the channel must be followed. The path strategy is based on a comprehensive analysis of the channel map near the port, real-time obstacle distribution, and port scheduling information (such as berthing order and temporary no-fly zones). The optimal path is selected from the feasible paths. For example, the path with the shortest distance and avoiding congested and dangerous areas is given priority.

[0043] An information exchange mechanism is established between ships. Each ship, based on the extracted feature information, knows the scope and impact of the strategies adopted by other ships through communication equipment, providing a basis for its own strategy selection. The profit calculation rules are clarified. Based on the objective function and the strategy selected by each ship, the profit of each ship under different strategy combinations is calculated. The profits brought by different strategies are compared, the advantages and disadvantages of the strategies are evaluated, the optimal strategy is selected, and the game termination conditions are set. When the preset number of iterations is reached, the strategies of each ship do not change for many consecutive times, or the overall profit reaches a stable state, the game process ends and the results are output. The Nash equilibrium solution is obtained through the quantum evolutionary algorithm. The formula is: Among them, Δθ is the angle value used for adjustment in the quantum evolution algorithm, f best is the optimal fitness value of the current group, f i is the fitness value of individual i, f avg is the average fitness value of the group, k1 and k2 are preset adjustment coefficients, which are determined through historical data regression analysis. The control center plans the route based on this coefficient in combination with the ship performance parameters.

[0044] Collaborative navigation control: The control center sends formation, relative position and speed instructions to ships according to the planned route. After receiving the instructions, the ships adjust their status to achieve collaborative navigation. During navigation, the ships use sensors to feedback status information in real time, such as position, speed, and heading. When the route needs to be adjusted due to temporary changes in port scheduling, sudden bad weather or new obstacles, the deviation correction formula is introduced to adjust the route in real time, such as Figure 2 As shown in the figure, let the original planned route be L0, the adjusted route be L1, and the current position of the ship be P. The correction is made by calculating the route deviation angle θ and the distance deviation d. The calculation formula is: d = |P-L1|, where and are the direction vectors of the original route and the adjusted route, respectively. According to θ and d, the adjusted route is further optimized using a correction function. The formula is: R(θ, d) = γ1×θ+γ2×d, where γ1 and γ2 are correction coefficients determined by historical navigation data and are used to adjust the weights of the route deviation angle and distance deviation in the correction function. L1 is fine-tuned according to the value of R(θ, d) to ensure that the ship can sail to the target port safely and efficiently.

[0045] Strategy evaluation and optimization: The control center collects ship navigation data and calculates comprehensive evaluation index values based on shipping companies' needs for transportation efficiency and cost control, as well as the current cost structure of the shipping market. The formula is: Among them, I is the comprehensive evaluation index, T actual and T expected The actual sailing time and the expected sailing time are respectively, actual and C expectedare the actual fuel consumption cost and the expected fuel consumption cost, S accident-free is the accident-free mileage, S t otal is the total voyage mileage, λ1, λ2, and λ3 are weight coefficients, and λ1+λ2+λ3=1. By analyzing the factors affecting the value of this indicator, such as the increase in sailing time caused by port congestion and the impact of fuel price fluctuations on costs, the parameters of the deep learning model H-Net and the dynamic game algorithm strategy are adjusted to continuously optimize the navigation control strategy of the ship formation.

[0046] In summary, in the navigation scenario of large container ship formations in the waters near busy ports, the intelligent ship formation navigation control method demonstrates strong effectiveness. Multi-source data collection provides comprehensive information for navigation decisions, data processing and feature extraction mine key values, and routes are planned through game theory. Strategies are determined by comprehensively considering the ship itself and the benefits of the formation. Collaborative navigation control ensures that ships move forward safely and collaboratively according to instructions, and routes can be corrected in time when changes occur. Strategy evaluation and optimization calculates indicator values based on the needs of shipping companies, and continuously adjusts models and algorithms. This method effectively improves the navigation efficiency and safety of large container ship formations in complex port environments, reduces costs, and provides strong support for the development of the shipping industry.

[0047] Example 2:

[0048] Multi-source data collection: Ship AIS sensors, weather radar echo sensors, underwater sonar detection sensors, ship position and speed sensors, as well as satellite communication equipment and VHF communication equipment are installed on each ship in the small scientific research vessel formation. The ship AIS sensor is used to obtain basic information about itself and surrounding ships, facilitating avoidance and coordination during navigation; the weather radar echo sensor monitors meteorological changes in complex sea areas, such as severe convective weather and sudden storms, and provides early warning for safe navigation of ships; the underwater sonar detection sensor detects information on seabed topography and marine life distribution, providing key data support for scientific research; the ship position and speed sensor monitors the ship's position and speed in real time to ensure that the ship sails as planned, and the collected data is transmitted to the control center through communication equipment.

[0049] Data processing and feature extraction: The control center cleans, converts, and normalizes the collected data, removes erroneous data, unifies the data format and dimension, and then fuses the data. The fused data is recorded as X. Feature extraction is performed using the deep hybrid network model H-Net. The convolutional attention module CAM(X) focuses on key area features using the formula: CAM(X) = σ(W2×ReLU(W1×X))×X, highlighting information related to the scientific expedition mission, such as special marine terrain areas and bio-dense areas. The improved gated recurrent unit iGRU is then input to process time series features. The formula is calculated to capture the change pattern of data over time. The formula is: zt =σ(W z ×[h t-1 ,x t ]), r t =σ(W r ×[h t-1 ,x t ]), The final output feature vector F = H-Net(X) contains information about changes in the ocean environment over time, which is crucial for scientific research and navigation decisions.

[0050] Game planning route: Set each scientific research ship as a game participant, and construct an objective function and a benefit function for each ship Where T i is the sailing time of ship i. In scientific research missions, it is critical to reach the designated scientific research area as soon as possible and complete the research mission; C i In order to reduce fuel consumption, small scientific research vessels have limited fuel reserves and need to strictly control costs to ensure that there is enough fuel throughout the scientific research trip; i is the contribution of ship i to the safety of the fleet and the completion of the scientific expedition mission. For example, the contribution increases when an important scientific expedition target area is discovered. The value range is 0-1. ω1, ω2, and ω3 are weight coefficients, and ω1+ω2+ω3=1. It is determined according to the focus and budget of the scientific expedition mission.

[0051] Construct a strategy set, speed strategy, ship speed v i Limited by its own performance and complex sea conditions, such as in areas with strong winds and waves, the speed needs to be reduced to ensure safety, meeting v i-min ≤v i ≤v i-max , heading strategy, heading θ i The value range is 0-360°, and it is determined by comprehensively considering the location of the scientific research area, the direction of ocean currents, and meteorological conditions. For example, sailing in the direction of the current can save fuel, while avoiding areas with bad weather. The path strategy is based on analyzing the seabed topography map and real-time sea conditions information (such as undercurrents and whirlpools). It selects a path from feasible paths that can better complete the scientific research mission and ensure safety. For example, it chooses a path that passes through more potential scientific research target areas and avoids dangerous sea conditions. An information exchange mechanism is established to share scientific research data and navigation strategy information between ships for better coordinated actions. The profit calculation rules are clarified, and the profit is calculated according to the objective function and ship strategy. The advantages and disadvantages of the strategy are evaluated, and the game termination conditions are set. For example, when the preset number of iterations is reached, the strategy is stable, or the key scientific research mission point is completed, the game ends and the results are output. The Nash equilibrium solution is obtained through the quantum evolution algorithm. The formula is: The control center plans the route accordingly.

[0052] Collaborative navigation control: The control center sends navigation instructions to the ship, and the ship adjusts its status to coordinate navigation. Status information is fed back in real time during navigation. When encountering sudden sea conditions (such as storms, strong currents) or new scientific research discoveries that require route adjustment, the route is adjusted using the deviation correction formula. Let the original planned route be L0, the adjusted route be L1, and the current position of the ship be P. The route deviation angle θ and distance deviation d are calculated using the following formula: d = |P-L1|, and the correction function R(θ,d) is used to optimize the adjusted route. The formula is: R(θ,d) = γ1×θ+γ2×d, where γ1 and γ2 are correction coefficients determined by historical navigation data. L1 is fine-tuned according to the value of R(θ,d) to ensure that the ship can avoid dangerous areas and get as close as possible to the scientific research target area.

[0053] Strategy evaluation and optimization: The control center collects ship navigation and scientific research data and calculates the comprehensive evaluation index value I. The formula is: According to the completion status of the scientific expedition mission and the cost control effect factors, the factors affecting the index value are analyzed. For example, if encountering severe sea conditions leads to increased sailing time and fuel consumption, the deep learning model parameters and dynamic game algorithm strategies are adjusted, and the subsequent navigation control strategy is optimized to better complete the scientific expedition mission.

[0054] In summary, the intelligent ship formation navigation control method plays a key role in the scientific expedition navigation of small scientific expedition ship formations in complex waters. From multi-source data collection to obtain the data required for scientific expedition and navigation, to the use of deep hybrid network models to process data and extract features, it lays the foundation for subsequent decision-making. Game planning routes take into account both scientific expedition tasks and ship safety and cost, reasonably set strategies, and coordinated navigation control so that ships can flexibly adjust according to instructions and can sail stably in complex sea conditions. Strategy evaluation optimization optimizes navigation strategies based on scientific expedition results and cost consumption. This method helps small scientific expedition ship formations to efficiently complete survey tasks, ensure ship safety, and promote the development of marine scientific expeditions.

[0055] The above description is merely a preferred embodiment of the present invention and does not constitute any form of limitation to the present invention. Although the present invention has been disclosed as above in terms of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can, without departing from the scope of the technical solution of the present invention, make some changes or modifications to equivalent embodiments using the technical contents disclosed above. However, any brief modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the content of the technical solution of the present invention are still within the scope of the technical solution of the present invention.

Claims

1. A method for controlling navigation of an intelligent ship formation, characterized in that: The specific steps of this control method are: S100, multi-source data acquisition: Various sensors and communication equipment are installed on ships to collect multi-source heterogeneous data such as ship AIS, weather radar echoes, underwater sonar detection, port scheduling, and ship position and speed, and transmit it to the control center for storage; S200, Data Processing and Feature Extraction: Extract data from the data center, clean, transform, and normalize the data, fuse multi-source data, and use the deep hybrid network model H-Net for feature extraction; S300, game-based route planning: A game model is constructed based on the collaborative relationships between ships. An objective function is set for each ship that integrates its own and the fleet's benefits. A strategy set is constructed that includes speed, heading, and path. A quantum evolutionary algorithm is used to find the Nash equilibrium solution. The control center then plans the route based on this strategy and combines it with the ship's performance parameters. S400, Collaborative Navigation Control: The control center sends formation, relative position, and speed instructions to ships according to the planned route. After receiving the instructions, the ships adjust their status to achieve collaborative navigation. Status information is fed back in real time during navigation. In case of environmental changes, deviation correction formulas are introduced to adjust the route in real time. S500, Strategy Evaluation Optimization: The control center collects ship navigation data, calculates comprehensive evaluation index values based on the shipping company's needs and cost structure, and adjusts the parameters of the deep hybrid network model H-Net and the dynamic game algorithm strategy by analyzing the factors that affect the comprehensive evaluation index values.

2. The intelligent ship formation navigation control method according to claim 1, characterized in that: The sensors and communication equipment used in the multi-source data collection in S100 are: ship AIS sensors, weather radar echo sensors, underwater sonar detection sensors, port scheduling sensors, ship position and speed sensors, satellite communication equipment and VHF communication equipment.

3. The intelligent ship formation navigation control method according to claim 1, characterized in that: In the S200, the deep hybrid network model H-Net is used for feature extraction in data processing and feature extraction. For the input fusion data X, the key area features are focused by the convolutional attention module CAM(X). The formula is: CAM(X) = σ(W2×ReLU(W1×X))×X, where CAM(X) is the result of focusing on the key area features after the convolutional attention module processes the input data X. W1 and W2 are learnable weight matrices used to extract the key area features of the input data. σ is an activation function, which is input to the improved gated recurrent unit iGRU to process the time series features: z t =σ(W z ×[h t-1 ,x t ]), r t =σ(W r ×[h t-1 ,x t ]), Among them, x t is the input at time t, h t -1 is the hidden state at the last moment, z t is the update gate, r t is the reset gate, ⊙ represents element-by-element multiplication, W z 、W r , W is the weight matrix, where W z Used to calculate the update gate, W r Used to calculate the reset gate, W is used to calculate the candidate hidden state, is the calculated candidate hidden state, h t It is the hidden state at time t, and the final output feature vector F = H-Net(X).

4. The intelligent ship formation navigation control method according to claim 1, characterized in that: S300, construction of a game model in route planning based on dynamic game: (1) Determine the game participants: Each ship in the fleet is set as a participant in the game model, and each ship needs to make a strategic choice in the navigation decision; (2) Setting the objective function: constructing an objective function for each ship that integrates its own and the fleet's benefits; (3) Constructing a strategy set: clarifying the strategy set of each ship, including speed, heading, and path decision variables; (4) Establish an information exchange mechanism: determine the information exchange rules between ships. Based on the extracted feature information, each ship can know the scope and impact of the strategies adopted by other ships, providing a basis for its own strategy selection; (5) Clarify the profit calculation rules: Based on the objective function and the strategy selected by each ship, calculate the profit of each ship under different strategy combinations. By comparing the profits brought by different strategies, evaluate the advantages and disadvantages of the strategies and select the optimal strategy; (6) Set the game termination conditions: when the preset number of iterations is reached, the strategies of each ship do not change for many consecutive times, or the overall benefits reach a stable state, the game process ends and the results are output.

5. The intelligent ship formation navigation control method according to claim 4, characterized in that: In the above S300, the objective function of the ship's own and fleet's benefits in the game planning route is set. Assume that the benefit function of ship i is U i , the calculation formula is: Among them, T i is the sailing time of ship i, C i is the fuel consumption cost, S i is the contribution of ship i to the safety of the formation, which is quantified based on maintaining a safe distance with other ships and avoiding collisions. Its value range is 0-1. ω1, ω2, and ω3 are weight coefficients, and ω1+ω2+ω3=1. It is dynamically adjusted according to the mission type.

6. The intelligent ship formation navigation control method according to claim 4, characterized in that: S300, setting of strategy set in game planning route: Speed strategy: The ship's optional sailing speed v i Limited by its own performance, it must meet v i-min ≤v i ≤v i-max , where v i-min is the minimum speed at which the ship can sail stably, v i-max It is the maximum sailing speed of the ship under the current load and sea conditions; Heading strategy: the ship's heading θ i The value range is 0-360°, and its determination takes into account the direction of the channel, the distribution of obstacles, and meteorological conditions; Path strategy: Path selection is based on a comprehensive analysis of the waterway map, real-time obstacle distribution, and port scheduling information, and a decision is made from feasible paths.

7. The intelligent ship formation navigation control method according to claim 4, characterized in that: In the aforementioned S300, the Nash equilibrium solution obtained by the quantum evolutionary algorithm in the game planning route is as follows: Among them, Δθ is the angle value used for adjustment in the quantum evolution algorithm, f best is the optimal fitness value of the current group, f i is the fitness value of individual i, f avg is the average fitness value of the group, k1 and k2 are preset adjustment coefficients, which are determined by regression analysis of historical data.

8. The intelligent ship formation navigation control method according to claim 1, characterized in that: In the above-mentioned S400, the route is adjusted in real time by introducing a deviation correction formula in the collaborative navigation control. Assuming that the original planned route is L0, the adjusted route is L1, and the current position of the ship is P, the route deviation angle θ and the distance deviation d are calculated to make corrections. The calculation formula is: d = |P-L1|, where and are the direction vectors of the original route and the adjusted route, respectively. According to θ and d, the correction function R(θ, d) is used to further optimize the adjusted route. The formula is: R(θ, d) = γ1×θ+γ2×d, where γ1 and γ2 are correction coefficients determined by historical navigation data. They are used to adjust the weights of the route deviation angle and distance deviation in the correction function. L1 is fine-tuned according to the value of R(θ, d).

9. The intelligent ship formation navigation control method according to claim 1, characterized in that: The calculation of the comprehensive evaluation index value in the strategy evaluation optimization in S500 is as follows: Among them, I is the comprehensive evaluation index, T actual and T expected The actual sailing time and the expected sailing time are respectively, actual and C expected are the actual fuel consumption cost and the expected fuel consumption cost, S accident-free is the accident-free mileage, S total is the total voyage mileage, λ1, λ2, and λ3 are weight coefficients and λ1+λ2+λ3=1, which is determined according to shipping demand and cost structure.

Images