An intelligent combined fleet navigation status management method and system

Abstract

The present invention relates to the technical field of fleet navigation status management, and discloses an intelligent combined fleet navigation status management method and system, including the following steps: S101, collecting ship feature data and environmental feature data of P ships in a pure electric unmanned fleet during navigation, and representing them through a first feature sequence; S102, preprocessing the first feature sequence to obtain a second feature sequence; S103, constructing a fleet collaborative topology map according to the second feature sequence, and constructing a fleet energy consumption prediction model according to the fleet collaborative topology map, and outputting the power consumption of P ships in the fleet in the second preset time period; S104, optimizing the control parameter combination based on an improved genetic algorithm to obtain the optimal control parameter combination. The present invention comprehensively improves the navigation efficiency, green development capability and task execution effect of the unmanned fleet through dynamic collaborative optimization, intelligent management of battery life and carbon emissions, and deep integration of formation optimization and energy-saving design.

Description

Technical Field

[0001] The invention belongs to the technical field of fleet navigation status management, and in particular relates to an intelligent combined fleet navigation status management method and system. Background Art

[0002] The current management methods of unmanned ship fleets are gradually improving with the development of intelligent technology, but are still in the exploratory stage. As an important direction for future marine transportation, resource exploration and military applications, the management methods of unmanned fleets mainly rely on the automatic navigation and path planning technology of a single ship, and have obvious deficiencies in multi-ship coordination and overall fleet optimization.

[0003] At present, the main defects of the management method are reflected in the following aspects: 1. Lack of dynamic collaborative optimization capabilities: The existing technology mainly adopts preset task planning and fixed routes. This static management method cannot fully utilize the synergy between ships, resulting in a significant decrease in navigation efficiency and task execution capabilities; 2. Lack of intelligent management of battery life and indirect carbon emissions: Unmanned fleets mostly use pure electric drive, and battery life has become the core limiting factor for mission execution, but current technology has failed to achieve accurate prediction of battery status and energy consumption optimization, restricting the green development of the fleet; 3. Lack of formation management and energy-saving optimization: When ships are sailing, reasonable formation arrangements can reduce water resistance through the wake effect, thereby significantly reducing energy consumption, but existing technologies do not pay enough attention to formation management and fail to fully utilize the potential value of formation layout for energy consumption optimization and speed improvement. Summary of the invention

[0004] The present invention provides an intelligent combined fleet navigation status management method and system to solve the technical problems in related technologies of lack of dynamic collaborative optimization capability, intelligent management of battery life and carbon emissions, and effective combination of formation optimization and energy-saving design.

[0005] The present invention provides an intelligent combined fleet navigation status management method, comprising the following steps:

[0006] S101, collecting ship characteristic data and environmental characteristic data of P ships in the pure electric unmanned fleet during navigation at a first preset time interval within a first preset time period, and representing them through a first characteristic sequence, wherein the ship characteristic data includes: speed, heading, geographical location, relative position and remaining power, and the environmental characteristic data includes: wind speed, wind direction, tidal current speed and tidal current direction, and P is a custom parameter;

[0007] S102, preprocessing the first feature sequence to obtain a second feature sequence, wherein the second feature sequence includes N sequence units, the nth sequence unit includes P subunits, the pth subunit represents the ship feature data and environmental feature data of the pth ship at the nth time point after preprocessing, N represents the number of time points, n represents the index of the time point, and p represents the index of the ship;

[0008] S103, constructing a fleet coordination topology graph at N time points according to the second feature sequence, and constructing a fleet energy consumption prediction model based on the fleet coordination topology graph at the N time points based on the graph neural network model and the gated cyclic unit, and outputting the power consumption of P ships in the fleet in the second preset time period, wherein the fleet coordination topology graph is composed of nodes, edges and initial features of nodes, a mapping relationship is established between nodes and ships, and a mapping relationship is established between the initial features of nodes and subunits of the sequence unit of the second feature sequence of the ship corresponding to the node;

[0009] S104, constructing an objective function according to the fleet energy consumption prediction model, and optimizing the control parameter combination based on the improved genetic algorithm to obtain an optimal control parameter combination, where the control parameter combination represents the speeds and relative positions of the P ships.

[0010] Furthermore, the heading, wind direction and tide direction are represented by two-dimensional vector codes composed of cosine and sine values ​​of angles, the geographical location is represented by a two-dimensional vector code composed of longitude and latitude, and the relative position represents the two-dimensional coordinates of the ships in the fleet relative to the first preset reference point.

[0011] Further, the first feature sequence is preprocessed to obtain a second feature sequence, and the specific steps include:

[0012] S201, for missing values ​​in the first feature sequence, filling processing is performed by using feature values ​​of adjacent ships of the ship corresponding to the missing value;

[0013] S202, identifying outliers in the first feature sequence according to the physical laws of the unmanned ship's motion and environmental constraints, and performing mean replacement processing using the values ​​other than the outliers;

[0014] S203, extracting the wake drag reduction feature between ships according to the first feature sequence, and splicing it after the sub-unit of the sequence unit of the first sequence. The calculation formula of the feature is:

[0015] , where condition 1 is: , represents the drag reduction characteristics of wake flow between ships, represents the wake drag reduction proportional coefficient, and The horizontal and vertical coordinates represent the relative position of ship a respectively, and The horizontal and vertical coordinates represent the relative position of ship b, represents the first preset distance, represents the first weight parameter, and Respectively represent the headings of ship a and ship b, represents the maximum effective angle of the wake, and Represent the speeds of ship a and ship b respectively, represents a minimum positive value, It represents the absolute value of the difference between the heading angles of ship a and ship b;

[0016] S204, extracting a ship compactness feature according to the first feature sequence, and concatenating it after the subunit of the sequence unit of the first sequence. The calculation formula of the feature is:

[0017] , represents the compactness characteristic of the ship, c, d and f represent the ship index, and The horizontal and vertical coordinates represent the relative position of the ship c, and The horizontal and vertical coordinates represent the relative position of the ship d, and denote the speeds of ship c and ship d respectively, and P denotes the number of ships;

[0018] S205, extracting the ship power consumption rate feature according to the first feature sequence, and splicing it after the sub-unit of the sequence unit of the first sequence, the calculation formula of the feature is:

[0019] ,in, Indicates the characteristics of the ship's power consumption rate, and represents the remaining power of the ship at the n-1th time point and the nth time point respectively, t represents the first preset time interval, Indicates wind speed, Indicates the wind direction of the ship g, represents the heading of the ship g, represents the speed of the current, represents the tidal direction of the vessel g, represents the speed of the ship g;

[0020] S206, using the maximum and minimum normalization method to normalize the concatenated first feature sequence to obtain a second feature sequence.

[0021] Furthermore, the fleet collaboration topology graph consists of nodes, edges, and initial features of nodes. The edge construction methods include:

[0022] 301, according to S203, the wake drag reduction characteristic between the two ships is obtained, and when the characteristic value is not 0, an edge is constructed between the nodes corresponding to the two ships;

[0023] 302, when the wind direction angle between the two ships is less than a second preset angle, construct an edge between the nodes corresponding to the two ships;

[0024] 303 , calculating the cosine similarity value between the subunits corresponding to the two ships, and when the cosine similarity value is less than a first preset threshold, constructing an edge between the nodes corresponding to the two ships.

[0025] Furthermore, the structure of the fleet energy consumption prediction model includes: a first hidden layer, a combination layer, a second hidden layer and an output layer;

[0026] The first hidden layer includes N first units, the nth first unit inputs the nth fleet coordination topology map, and outputs an updated topology map, and the updated topology map includes P updated node features;

[0027] The combination layer includes N combination units, and the nth combination unit is used to splice the P update node features of the nth update topology graph into a combination feature;

[0028] The second hidden layer includes N second units, the nth second unit inputs the nth combined feature and outputs the hidden feature;

[0029] The output layer is used to input the hidden features output by the Nth second unit into the first classifier, and the classification space of the first classifier represents the power consumption of P ships in the fleet in the second preset time period.

[0030] Furthermore, the calculation formula of the nth first unit of the first hidden layer is:

[0031] ;

[0032] in, Indicates updating the topology map. represents the initial feature of the i-th node, pile represents the stacking operation, represents the first weight matrix, represents the degree of the i-th node, that is, the number of edges connected to the i-th node, including self-loop edges. i, j, and k represent the index of the node. represents the neighbor set of the i-th node, represents the degree of the jth node, represents the initial features of the jth node, represents the second weight matrix, represents the initial features of the kth node, represents the third weight matrix, Indicates the number of nodes contained in the neighbor set of the i-th node.

[0033] Furthermore, the second hidden layer includes N second units, the second hidden layer is constructed based on the gated recurrent unit, each second unit outputs the hidden feature corresponding to the unit, and finally the hidden feature output by the Nth second unit is used as the output of the entire second hidden layer.

[0034] Furthermore, the control parameter combination is optimized based on the improved genetic algorithm, and the specific steps include:

[0035] S401, generating an initialization population that meets the constraint conditions according to the preset ship formation and normal distribution disturbance combined with a random generation strategy, the initialization population includes Q individuals, where Q is a custom parameter;

[0036] Each individual represents a control parameter combination, and the individual is represented by a formation matrix. A row of the formation matrix represents the speed, x-axis coordinate of the relative position, and y-axis coordinate of the relative position of a ship. A column of the formation matrix represents an eigenvalue of the control parameters of P ships. The control parameters include: speed, x-axis coordinate of the relative position, and y-axis coordinate of the relative position.

[0037] S402, constructing an objective function according to the fleet energy consumption prediction model, the calculation formula of the objective function includes:

[0038] ;

[0039] ;

[0040] ;

[0041] Among them, F represents the fitness value of the individual, and They represent the power consumption weight parameter and speed penalty weight parameter respectively, s and r represent the ship index, represents the power consumption of the sth ship obtained according to the fleet energy consumption prediction model, Indicates the speed penalty value of a single ship. Indicates the speed consistency penalty value, and Represent the first penalty coefficient and the second penalty coefficient respectively, and Respectively represent the first preset speed and the second preset speed, and denote the speeds of the sth and rth ships respectively, represents the maximum value function, Indicates the maximum permissible speed difference;

[0042] S403, calculating the fitness values ​​of individuals in the population according to the objective function;

[0043] S404, divide the population into two parts, elite individuals and weak individuals, according to the fitness value of the individuals, and retain L% and R% respectively;

[0044] S405, randomly selecting two individuals to perform a crossover operation that satisfies the constraint conditions;

[0045] S406, randomly select an individual, and exchange the element values ​​of two different positions in any column of its formation matrix;

[0046] S407, when the current number of iterations reaches the maximum number of iterations, the individual with the largest fitness value in the current population is used as the optimal control parameter combination, otherwise return to S403, where the maximum number of iterations is a custom parameter.

[0047] Furthermore, the specific steps of S401 include:

[0048] S501, determining a speed range according to the performance of the ship, and randomly obtaining the speed of the ship;

[0049] S502, based on the preset ship formation, the x-axis coordinate and the y-axis coordinate of the relative position of each ship are calculated, and combined with the speed obtained in S501 into an individual;

[0050] S503, perturb an individual obtained by S501 and S502 using normal distribution, and obtain individuals that meet the constraints, and randomly generate Individuals that meet the constraints are combined to obtain the initialized population.

[0051] The present invention provides an intelligent combined fleet navigation status management system, comprising:

[0052] A data acquisition module is used to collect ship characteristic data and environmental characteristic data of P ships in the pure electric unmanned fleet during navigation at a first preset time interval within a first preset time period, and represent them through a first characteristic sequence, wherein the ship characteristic data includes: speed, heading, geographical location, relative position and remaining power, and the environmental characteristic data includes: wind speed, wind direction, tidal current speed and tidal current direction, and P is a custom parameter;

[0053] A data preprocessing module, used for preprocessing the first feature sequence to obtain a second feature sequence, wherein the second feature sequence includes N sequence units, the nth sequence unit includes P subunits, the pth subunit represents the ship feature data and environmental feature data of the pth ship at the nth time point after preprocessing, N represents the number of time points, n represents the index of the time point, and p represents the index of the ship;

[0054] A fleet energy consumption prediction module is used to construct a fleet collaborative topology map at N time points according to the second feature sequence, and to construct a fleet energy consumption prediction model based on the fleet collaborative topology map at N time points based on a graph neural network model and a gated loop unit, and output the power consumption of P ships in the fleet in the second preset time period, wherein the fleet collaborative topology map is composed of nodes, edges and initial features of nodes, a mapping relationship is established between nodes and ships, and a mapping relationship is established between the initial features of nodes and the subunits of the sequence unit of the second feature sequence of the ship corresponding to the node;

[0055] The control parameter combination optimization module is used to construct the objective function according to the fleet energy consumption prediction model, and optimize the control parameter combination based on the improved genetic algorithm to obtain the optimal control parameter combination. The control parameter combination represents the speed and relative position of P ships.

[0056] The beneficial effects of the present invention are as follows: the present invention constructs a ship energy consumption prediction model by combining a fleet collaborative topology map and a gated cycle unit, and the model can dynamically capture the collaborative relationship and state changes between ships, realize accurate prediction and optimization of fleet energy consumption, and improve the efficiency and accuracy of task execution;

[0057] The objective function of the present invention comprehensively considers the penalty items related to total energy consumption and speed, which not only minimizes the energy consumption, but also limits the excessive, insufficient and inconsistent speed, ensuring that the optimization result is both energy-saving and meets actual operation requirements, thereby improving physical feasibility and practical value. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] Figure 1 The present invention is a flowchart of an intelligent combined fleet navigation status management method. DETAILED DESCRIPTION

[0059] The subject matter described herein will now be discussed with reference to example embodiments. It should be understood that the discussion of these embodiments is only to enable those skilled in the art to better understand and implement the subject matter described herein, and the functions and arrangements of the elements discussed may be changed without departing from the scope of protection of the contents of this specification. Each example may omit, replace or add various processes or components as needed. In addition, the features described relative to some examples may also be combined in other examples.

[0060] like Figure 1 As shown, an intelligent combined fleet navigation status management method includes the following steps:

[0061] S101, collecting ship characteristic data and environmental characteristic data of P ships in the pure electric unmanned fleet during navigation at a first preset time interval within a first preset time period, and representing them through a first characteristic sequence, wherein the ship characteristic data includes: speed, heading, geographical location, relative position and remaining power, and the environmental characteristic data includes: wind speed, wind direction, tidal current speed and tidal current direction, and P is a custom parameter;

[0062] S102, preprocessing the first feature sequence to obtain a second feature sequence, wherein the second feature sequence includes N sequence units, the nth sequence unit includes P subunits, the pth subunit represents the ship feature data and environmental feature data of the pth ship at the nth time point after preprocessing, N represents the number of time points, n represents the index of the time point, and p represents the index of the ship;

[0063] S103, constructing a fleet coordination topology graph at N time points according to the second feature sequence, and constructing a fleet energy consumption prediction model based on the fleet coordination topology graph at the N time points based on the graph neural network model and the gated cyclic unit, and outputting the power consumption of P ships in the fleet in the second preset time period, wherein the fleet coordination topology graph is composed of nodes, edges and initial features of nodes, a mapping relationship is established between nodes and ships, and a mapping relationship is established between the initial features of nodes and subunits of the sequence unit of the second feature sequence of the ship corresponding to the node;

[0064] S104, constructing an objective function according to the fleet energy consumption prediction model, and optimizing the control parameter combination based on the improved genetic algorithm to obtain an optimal control parameter combination, where the control parameter combination represents the speeds and relative positions of the P ships.

[0065] In one embodiment of the present invention, the heading, wind direction and tidal direction are represented by a two-dimensional vector code consisting of the cosine and sine values ​​of the angle. For example, the heading angle is , the two-dimensional vector encoding of the heading is expressed as The geographical location is represented by a two-dimensional vector code consisting of longitude and latitude, and the relative position represents the two-dimensional coordinates of the ship in the fleet relative to the first preset reference point, specifically including: randomly selecting a ship in the fleet as the first preset reference point, taking its heading as the y-axis, the vertical direction of the heading as the x-axis, and the ship as the origin, establishing a rectangular coordinate system, and calculating the coordinates of other ships in the rectangular coordinate system.

[0066] In one embodiment of the present invention, before preprocessing the first feature sequence, in order to ensure data consistency, the units of the ship feature data and the environmental feature data need to be unified. Specifically, the unit of speed is knots, 1 knot means 1 nautical mile per hour, the unit of wind speed is knots, the unit of tidal speed is knots, and the unit of remaining power is percentage.

[0067] In one embodiment of the present invention, the first feature sequence is preprocessed to obtain the second feature sequence, and the specific steps include:

[0068] S201, for missing values ​​in the first feature sequence, filling processing is performed by using feature values ​​of adjacent ships of the ship corresponding to the missing value;

[0069] S202, identifying outliers in the first feature sequence according to the physical laws of the unmanned ship and environmental constraints, and using the mean replacement process of other values ​​except the outliers, wherein the physical laws of the ship refer to that the ship speed cannot exceed the maximum design speed or be a negative value, and the environmental constraints include: the wind speed and the tidal speed should be within the statistical range of the ocean observation data;

[0070] S203, extracting the wake drag reduction feature between ships according to the first feature sequence, and splicing it after the sub-unit of the sequence unit of the first sequence. The calculation formula of the feature is:

[0071] , where condition 1 is: , represents the drag reduction characteristics of wake flow between ships, It represents the wake drag reduction proportional coefficient, which is used to describe the contribution of the wake to the drag reduction of the following ship. It is determined by the shape of the ship, the speed of the ship, etc. and The horizontal and vertical coordinates represent the relative position of ship a respectively, and The horizontal and vertical coordinates represent the relative position of ship b, represents the first preset distance, i.e., the maximum effective distance of the wake area, and represents the maximum influence range of the wake generated by the leading ship. e is a natural constant. represents the first weight parameter, which is used to control the rapid attenuation of the wake effect on the angle. and Respectively represent the headings of ship a and ship b, represents the maximum effective angle of the wake, and Represent the speeds of ship a and ship b respectively, represents a minimum positive value, preferably, Pick , It represents the absolute value of the difference between the heading angles of ship a and ship b;

[0072] S204, extracting a ship compactness feature according to the first feature sequence, and concatenating it after the subunit of the sequence unit of the first sequence. The calculation formula of the feature is:

[0073] , represents the compactness characteristic of the ship, c, d and f represent the ship index, and The horizontal and vertical coordinates represent the relative position of the ship c, and The horizontal and vertical coordinates represent the relative position of the ship d, and denote the speeds of ship c and ship d respectively, and P denotes the number of ships;

[0074] S205, extracting the ship power consumption rate feature according to the first feature sequence, and splicing it after the sub-unit of the sequence unit of the first sequence, the calculation formula of the feature is:

[0075] ,in, Indicates the characteristics of the ship's power consumption rate, and represents the remaining power of the ship at the n-1th time point and the nth time point respectively, t represents the first preset time interval, Indicates wind speed, Indicates the wind direction of the ship g, represents the heading of the ship g, represents the speed of the current, represents the tidal direction of the vessel g, represents the speed of the ship g;

[0076] S206, using the maximum and minimum normalization method to normalize the concatenated first feature sequence to obtain a second feature sequence.

[0077] In one embodiment of the present invention, the fleet collaboration topology graph is composed of nodes, edges and initial features of the nodes, and the edge construction method includes:

[0078] 301, according to S203, the wake drag reduction characteristic between the two ships is obtained, and when the characteristic value is not 0, an edge is constructed between the nodes corresponding to the two ships;

[0079] 302, when the wind direction angle between the two ships is less than a second preset angle, construct an edge between the nodes corresponding to the two ships;

[0080] 303 , calculating the cosine similarity value between the subunits corresponding to the two ships, and when the cosine similarity value is less than a first preset threshold, constructing an edge between the nodes corresponding to the two ships.

[0081] In one embodiment of the present invention, the structure of the fleet energy consumption prediction model includes: a first hidden layer, a combination layer, a second hidden layer and an output layer;

[0082] The first hidden layer includes N first units, the nth first unit inputs the nth fleet coordination topology map, and outputs an updated topology map, and the updated topology map includes P updated node features;

[0083] The combination layer includes N combination units, and the nth combination unit is used to splice the P update node features of the nth update topology graph into a combination feature;

[0084] The second hidden layer includes N second units, the nth second unit inputs the nth combined feature and outputs the hidden feature;

[0085] The output layer is used to input the hidden features output by the Nth second unit into the first classifier, and the classification space of the first classifier represents the power consumption of P ships in the fleet in the second preset time period.

[0086] The fleet energy consumption prediction model can dynamically capture the collaborative relationship between ships and the time changes of their status. The first hidden layer captures the dynamic interaction between the fleet collaborative relationship and ships; the second hidden layer extracts deep features through nonlinear mapping of combined features, providing efficient input for energy consumption prediction. This design achieves an efficient combination of spatial collaboration and time series modeling, improving prediction accuracy, generalization ability and interpretability.

[0087] In one embodiment of the present invention, the calculation formula of the nth first unit of the first hidden layer is:

[0088] ;

[0089] in, Indicates updating the topology map. represents the initial feature of the i-th node, pile represents the stacking operation, represents the first weight matrix, represents the degree of the i-th node, that is, the number of edges connected to the i-th node, including self-loop edges. i, j, and k represent the index of the node. represents the neighbor set of the i-th node, represents the degree of the jth node, represents the initial features of the jth node, represents the second weight matrix, represents the initial features of the kth node, represents the third weight matrix, Indicates the number of nodes contained in the neighbor set of the i-th node.

[0090] In one embodiment of the present invention, the second hidden layer includes N second units, the second hidden layer is constructed based on the gated recurrent unit, each second unit outputs the hidden feature corresponding to the unit, and finally the hidden feature output by the Nth second unit is used as the output of the entire second hidden layer.

[0091] In one embodiment of the present invention, the control parameter combination is optimized based on an improved genetic algorithm, and the specific steps include:

[0092] S401, generating an initialization population that meets the constraint conditions according to the preset ship formation and normal distribution disturbance combined with a random generation strategy, the initialization population includes Q individuals, where Q is a custom parameter;

[0093] Each individual represents a control parameter combination, and the individual is represented by a formation matrix. A row of the formation matrix represents the speed, x-axis coordinate of the relative position, and y-axis coordinate of the relative position of a ship. A column of the formation matrix represents an eigenvalue of the control parameters of P ships. The control parameters include: speed, x-axis coordinate of the relative position, and y-axis coordinate of the relative position.

[0094] The constraints include:

[0095] The speed of each ship is between a first preset speed and a second preset speed, the first preset speed indicating the minimum speed required for the mission, and the second preset speed indicating the maximum speed allowed by the ship's performance;

[0096] The distance between the ships is calculated according to the relative positions of the ships, and the distance between the ships is between the second preset distance and the first preset distance, the first preset distance represents the maximum effective distance in the wake area, and the second preset distance represents the minimum safe distance to avoid collision;

[0097] The speed difference between any two ships cannot exceed a second preset threshold;

[0098] S402, constructing an objective function according to the fleet energy consumption prediction model, the calculation formula of the objective function includes:

[0099] ;

[0100] ;

[0101] ;

[0102] Among them, F represents the fitness value of the individual, and They represent the power consumption weight parameter and speed penalty weight parameter respectively, s and r represent the ship index, represents the power consumption of the sth ship obtained according to the fleet energy consumption prediction model, Indicates the speed penalty value of a single ship. Indicates the speed consistency penalty value, and Represent the first penalty coefficient and the second penalty coefficient respectively, and Respectively represent the first preset speed and the second preset speed, and denote the speeds of the sth and rth ships respectively, represents the maximum value function, Indicates the maximum permissible speed difference;

[0103] The objective function combines the total energy consumption and the speed penalty, so that the optimization not only minimizes the energy consumption, but also punishes excessive, too low and inconsistent speeds, avoiding the optimization from focusing too much on the minimum energy consumption at low speeds and ignoring the actual operating requirements, thereby improving the physical feasibility and actual application value.

[0104] S403, calculating the fitness values ​​of individuals in the population according to the objective function;

[0105] S404, dividing the population into two parts, elite individuals and weak individuals, according to the fitness value of the individuals, and retaining L% and R% respectively, preferably, L is set to 10 and R is set to 5;

[0106] S405, randomly selecting two individuals to perform a crossover operation that satisfies the constraint conditions, the crossover operation comprising: selecting element values ​​at the same position of the formation matrix of the two individuals to exchange, and determining whether the result after the exchange satisfies the constraint conditions, if yes, the crossover operation is completed, if not, the crossover operation is re-executed;

[0107] S406, randomly select an individual, and exchange the element values ​​of two different positions in any column of its formation matrix;

[0108] S407, when the current number of iterations reaches the maximum number of iterations, the individual with the largest fitness value in the current population is used as the optimal control parameter combination, otherwise return to S403, where the maximum number of iterations is a custom parameter, preferably, the maximum number of iterations is set to 200.

[0109] In one embodiment of the present invention, the specific steps of S401 include:

[0110] S501, determining a speed range according to the performance of the ship, and randomly obtaining the speed of the ship;

[0111] S502, based on a preset ship formation, calculating the x-axis coordinate and the y-axis coordinate of the relative position of each ship, and combining them with the speed obtained in S501 into an individual, wherein the preset ship formation may be a V formation or a trapezoidal formation;

[0112] S503, perturb an individual obtained by S501 and S502 using normal distribution, and obtain individuals that meet the constraints, and randomly generate Individuals that meet the constraints are combined to obtain an initialized population, where the individuals are perturbed using normal distribution, specifically including: adding a perturbation value to the x-axis coordinate and y-axis coordinate of the relative position of each ship of the individual, and the perturbation value conforms to the normal distribution. Similarly, adding a perturbation value to the speed of each ship of the individual, and the perturbation value conforms to the normal distribution.

[0113] In one embodiment of the present invention, individuals are generated by combining normal distribution perturbation and random generation to ensure the diversity of the population. Normal distribution perturbation makes individuals distributed near the preset formation, which not only retains the basic structure of the formation, but also introduces a certain degree of randomness to explore different optimization directions. Random generation provides more possible solution spaces, does not rely on the limitations of the initial formation, and further expands the search range. The combination of these two can enhance the global search ability of the algorithm and avoid falling into the local optimum too early.

[0114] The present invention provides an intelligent combined fleet navigation status management system, comprising:

[0115] A data acquisition module is used to collect ship characteristic data and environmental characteristic data of P ships in the pure electric unmanned fleet during navigation at a first preset time interval within a first preset time period, and represent them through a first characteristic sequence, wherein the ship characteristic data includes: speed, heading, geographical location, relative position and remaining power, and the environmental characteristic data includes: wind speed, wind direction, tidal current speed and tidal current direction, and P is a custom parameter;

[0116] A data preprocessing module, used for preprocessing the first feature sequence to obtain a second feature sequence, wherein the second feature sequence includes N sequence units, the nth sequence unit includes P subunits, the pth subunit represents the ship feature data and environmental feature data of the pth ship at the nth time point after preprocessing, N represents the number of time points, n represents the index of the time point, and p represents the index of the ship;

[0117] A fleet energy consumption prediction module is used to construct a fleet collaborative topology map at N time points according to the second feature sequence, and to construct a fleet energy consumption prediction model based on the fleet collaborative topology map at N time points based on a graph neural network model and a gated loop unit, and output the power consumption of P ships in the fleet in the second preset time period, wherein the fleet collaborative topology map is composed of nodes, edges and initial features of nodes, a mapping relationship is established between nodes and ships, and a mapping relationship is established between the initial features of nodes and the subunits of the sequence unit of the second feature sequence of the ship corresponding to the node;

[0118] The control parameter combination optimization module is used to construct the objective function according to the fleet energy consumption prediction model, and optimize the control parameter combination based on the improved genetic algorithm to obtain the optimal control parameter combination. The control parameter combination represents the speed and relative position of P ships.

[0119] The above describes an embodiment of the present invention, but this embodiment is not limited to the above specific implementation methods. The above specific implementation methods are merely illustrative and not restrictive. Under the guidance of this embodiment, ordinary technicians in this field can also make many forms, all of which are protected by this embodiment.

Claims

1. An intelligent combined fleet navigation status management method, characterized in that: The following steps are involved: S101, collecting ship characteristic data and environmental characteristic data of P ships in the pure electric unmanned fleet during navigation at a first preset time interval within a first preset time period, and representing them through a first characteristic sequence, wherein the ship characteristic data includes: speed, heading, geographical location, relative position and remaining power, and the environmental characteristic data includes: wind speed, wind direction, tidal current speed and tidal current direction, and P is a custom parameter; S102, preprocessing the first feature sequence to obtain a second feature sequence, wherein the second feature sequence includes N sequence units, the nth sequence unit includes P subunits, the pth subunit represents the ship feature data and environmental feature data of the pth ship at the nth time point after preprocessing, N represents the number of time points, n represents the index of the time point, and p represents the index of the ship; S103, constructing a fleet coordination topology graph at N time points according to the second feature sequence, and constructing a fleet energy consumption prediction model based on the fleet coordination topology graph at the N time points based on the graph neural network model and the gated cyclic unit, and outputting the power consumption of P ships in the fleet in the second preset time period, wherein the fleet coordination topology graph is composed of nodes, edges and initial features of nodes, a mapping relationship is established between nodes and ships, and a mapping relationship is established between the initial features of nodes and subunits of the sequence unit of the second feature sequence of the ship corresponding to the node; S104, constructing an objective function according to the fleet energy consumption prediction model, and optimizing the control parameter combination based on the improved genetic algorithm to obtain an optimal control parameter combination, where the control parameter combination represents the speeds and relative positions of the P ships.

2. The method for managing the navigation status of an intelligent combined fleet according to claim 1, characterized in that: The heading, wind direction and tide direction are represented by two-dimensional vector codes composed of cosine and sine values ​​of angles, the geographical location is represented by a two-dimensional vector code composed of longitude and latitude, and the relative position represents the two-dimensional coordinates of the ships in the fleet relative to the first preset reference point.

3. The method for managing the navigation status of an intelligent combined fleet according to claim 1, characterized in that: Preprocessing the first feature sequence to obtain a second feature sequence includes: S201, for missing values ​​in the first feature sequence, filling processing is performed by using feature values ​​of adjacent ships of the ship corresponding to the missing value; S202, identifying outliers in the first feature sequence according to the physical laws of the unmanned ship's motion and environmental constraints, and performing mean replacement processing using the values ​​other than the outliers; S203, extracting the wake drag reduction feature between ships according to the first feature sequence, and splicing it after the sub-unit of the sequence unit of the first sequence. The calculation formula of the feature is: , where condition 1 is: , represents the drag reduction characteristics of wake flow between ships, represents the wake drag reduction proportional coefficient, and The horizontal and vertical coordinates represent the relative position of ship a, and The horizontal and vertical coordinates represent the relative position of ship b, represents the first preset distance, represents the first weight parameter, and Respectively represent the headings of ship a and ship b, represents the maximum effective angle of the wake, and Represent the speeds of ship a and ship b respectively, represents a minimum positive value, Indicates the absolute value of the difference between the heading angles of ship a and ship b; S204, extracting a ship compactness feature according to the first feature sequence, and concatenating it after the subunit of the sequence unit of the first sequence. The calculation formula of the feature is: , represents the compactness characteristic of the ship, c, d and f represent the ship index, and The horizontal and vertical coordinates represent the relative position of the ship c, and The horizontal and vertical coordinates represent the relative position of the ship d, and denote the speeds of ship c and ship d respectively, and P denotes the number of ships; S205, extracting the ship power consumption rate feature according to the first feature sequence, and splicing it after the sub-unit of the sequence unit of the first sequence, the calculation formula of the feature is: ,in, Indicates the characteristics of the ship's power consumption rate, and represents the remaining power of the ship at the n-1th time point and the nth time point respectively, t represents the first preset time interval, Indicates wind speed, Indicates the wind direction of the ship g, represents the heading of the ship g, represents the speed of the current, represents the tidal direction of the vessel g, represents the speed of the ship g; S206, using the maximum and minimum normalization method to normalize the concatenated first feature sequence to obtain a second feature sequence.

4. The method for managing the navigation status of an intelligent combined fleet according to claim 3, characterized in that: The fleet collaboration topology graph consists of nodes, edges, and initial features of nodes. The edge construction methods include: 301, according to S203, a wake drag reduction feature between the two ships is obtained, and when the feature is not 0, an edge is constructed between the nodes corresponding to the two ships; 302, when the wind direction angle between the two ships is less than a second preset angle, construct an edge between the nodes corresponding to the two ships; 303 , calculating the cosine similarity value between the subunits corresponding to the two ships, and when the cosine similarity value is less than a first preset threshold, constructing an edge between the nodes corresponding to the two ships.

5. The method for managing the navigation status of an intelligent combined fleet according to claim 1, characterized in that: The structure of the fleet energy consumption prediction model includes: the first hidden layer, the combination layer, the second hidden layer and the output layer; The first hidden layer includes N first units, the nth first unit inputs the nth fleet coordination topology map, and outputs an updated topology map, and the updated topology map includes P updated node features; The combination layer includes N combination units, and the nth combination unit is used to splice the P update node features of the nth update topology graph into a combination feature; The second hidden layer includes N second units, the nth second unit inputs the nth combined feature and outputs the hidden feature; The output layer is used to input the hidden features output by the Nth second unit into the first classifier, and the classification space of the first classifier represents the power consumption of P ships in the fleet in the second preset time period.

6. The method for managing the navigation status of an intelligent combined fleet according to claim 5, characterized in that: The calculation formula for the nth first unit of the first hidden layer is: ； in, Indicates updating the topology map. represents the initial feature of the i-th node, pile represents the stacking operation, represents the first weight matrix, represents the degree of the i-th node, that is, the number of edges connected to the i-th node, including self-loop edges. i, j, and k represent the index of the node. represents the neighbor set of the i-th node, represents the degree of the jth node, represents the initial features of the jth node, represents the second weight matrix, represents the initial features of the kth node, represents the third weight matrix, Indicates the number of nodes contained in the neighbor set of the i-th node.

7. The method for managing the navigation status of an intelligent combined fleet according to claim 5, characterized in that: The second hidden layer includes N second units. The second hidden layer is constructed based on the gated recurrent unit. Each second unit outputs the hidden feature corresponding to the unit. Finally, the hidden feature output by the Nth second unit is used as the output of the entire second hidden layer.

8. The method for managing the navigation status of an intelligent combined fleet according to claim 1, characterized in that: The control parameter combination is optimized based on the improved genetic algorithm. The specific steps include: S401, generating an initialization population that meets the constraint conditions according to the preset ship formation and normal distribution disturbance combined with a random generation strategy, the initialization population includes Q individuals, where Q is a custom parameter; Each individual represents a control parameter combination, and the individual is represented by a formation matrix. A row of the formation matrix represents the speed, x-axis coordinate of the relative position, and y-axis coordinate of the relative position of a ship. A column of the formation matrix represents an eigenvalue of the control parameters of P ships. The control parameters include: speed, x-axis coordinate of the relative position, and y-axis coordinate of the relative position. S402, constructing an objective function according to the fleet energy consumption prediction model, the calculation formula of the objective function includes: ； ； ； Among them, F represents the fitness value of the individual, and They represent the power consumption weight parameter and speed penalty weight parameter respectively, s and r represent the ship index, represents the power consumption of the sth ship obtained according to the fleet energy consumption prediction model, Indicates the speed penalty value of a single ship. Indicates the speed consistency penalty value, and Represent the first penalty coefficient and the second penalty coefficient respectively, and Respectively represent the first preset speed and the second preset speed, and denote the speeds of the sth and rth ships respectively, represents the maximum value function, Indicates the maximum permissible speed difference; S403, calculating the fitness values ​​of individuals in the population according to the objective function; S404, divide the population into two parts, elite individuals and weak individuals, according to the fitness value of the individuals, and retain L% and R% respectively; S405, randomly selecting two individuals to perform a crossover operation that satisfies the constraint conditions; S406, randomly select an individual, and exchange the element values ​​of two different positions in any column of its formation matrix; S407, when the current number of iterations reaches the maximum number of iterations, the individual with the largest fitness value in the current population is used as the optimal control parameter combination, otherwise return to S403, where the maximum number of iterations is a custom parameter.

9. The method for managing the navigation status of an intelligent combined fleet according to claim 8, characterized in that: The specific steps of S401 include: S501, determining a speed range according to the performance of the ship, and randomly obtaining the speed of the ship; S502, based on the preset ship formation, the x-axis coordinate and the y-axis coordinate of the relative position of each ship are calculated, and combined with the speed obtained in S501 into an individual; S503, perturb an individual obtained by S501 and S502 using normal distribution, and obtain individuals that meet the constraints, and randomly generate Individuals that meet the constraints are combined to obtain the initialized population.

10. An intelligent combined fleet navigation status management system, characterized in that: A method for managing the navigation status of an intelligent combined fleet as claimed in any one of claims 1 to 9 is adopted, comprising: A data acquisition module is used to collect ship characteristic data and environmental characteristic data of P ships in the pure electric unmanned fleet during navigation at a first preset time interval within a first preset time period, and represent them through a first characteristic sequence, wherein the ship characteristic data includes: speed, heading, geographical location, relative position and remaining power, and the environmental characteristic data includes: wind speed, wind direction, tidal current speed and tidal current direction, and P is a custom parameter; A data preprocessing module, used for preprocessing the first feature sequence to obtain a second feature sequence, wherein the second feature sequence includes N sequence units, the nth sequence unit includes P subunits, the pth subunit represents the ship feature data and environmental feature data of the pth ship at the nth time point after preprocessing, N represents the number of time points, n represents the index of the time point, and p represents the index of the ship; A fleet energy consumption prediction module is used to construct a fleet collaborative topology map at N time points according to the second feature sequence, and to construct a fleet energy consumption prediction model based on the fleet collaborative topology map at N time points based on a graph neural network model and a gated loop unit, and output the power consumption of P ships in the fleet in the second preset time period, wherein the fleet collaborative topology map is composed of nodes, edges and initial features of nodes, a mapping relationship is established between nodes and ships, and a mapping relationship is established between the initial features of nodes and the subunits of the sequence unit of the second feature sequence of the ship corresponding to the node; The control parameter combination optimization module is used to construct the objective function according to the fleet energy consumption prediction model, and optimize the control parameter combination based on the improved genetic algorithm to obtain the optimal control parameter combination. The control parameter combination represents the speed and relative position of P ships.

Images