A method for optimizing and forecasting ocean routes for ships

Abstract

The present invention provides a method for optimizing and forecasting ocean routes for ships, comprising the following steps: S1, establishing an initial route according to a voyage mission, a turning point and a route type; S2, establishing a mathematical model of the ship's motion response to the environment at sea; S3, obtaining the forecast result of the marine environment of the sea area through which the ship passes; S4, calculating the motion response of the ship passing through the sea area; S5, collecting other required conditions in the ship's voyage mission as other constraints; S6, optimizing and calculating to obtain the optimized route and optimized propeller speed that the ship should take under the optimization target; S8, repeating S3 to S7, and executing the latest route and speed optimization scheme. The present invention optimizes the ship's navigation plan elements including the route and propeller speed, and can obtain a navigation plan that meets the user's optimization needs while avoiding severe sea conditions to ensure safety. The method is conducive to the intelligent construction of ship navigation.

Description

Technical Field

[0001] The present invention relates to the technical field of route planning, and in particular to a method for optimizing and forecasting ocean routes for ships. Background Art

[0002] Ocean routes refer to the safe and economical routes chosen by ships when sailing in the ocean. Since ocean voyages are long, the water is deep, there are few obstacles, and the hydrological and meteorological changes are large, there is a greater degree of selectivity in the routes. The selection of routes should take into account the following factors: the season, weather, waves, and currents of the ocean; positioning and avoidance conditions; the technical performance of the ship, the cargo loaded, the technical level of the crew, etc., and the voyage time should be shortened as much as possible under the premise of safety.

[0003] The current common method for designing ocean routes for ships calculates the response of ship motions based on weather forecasts and empirical formulas, which has the problem of large calculation errors; invention CN107246871A provides a green ocean route design method based on ECDIS, which is affected by multicollinearity in high-dimensional calculations, affecting the generalization accuracy. When optimizing, it is often optimized based on a single forecast, and adjustments are made too frequently. Due to excessive reliance on forecast information, slow updates of forecasts will affect route optimization and route and speed adjustments, which can easily delay ship countermeasures and timing in severe sea conditions. Summary of the invention

[0004] In view of this, the purpose of the present invention is to propose a method for optimizing and predicting ship ocean routes to solve the technical problem of low generalization accuracy of existing ship ocean route design methods.

[0005] The technical means adopted by the present invention are as follows:

[0006] A method for optimizing and forecasting ocean routes for ships comprises the following steps:

[0007] S1. Establish the initial route according to the voyage mission, turning point and route type;

[0008] S2. Establish a mathematical model of the ship's motion response to the environment at sea;

[0009] S3. Establish a mathematical forecast model for the wind field, wave field and flow field in the sea area where the ship passes through, use the spatiotemporal generative adversarial neural network model with an attention mechanism to achieve the assimilation and fusion of observation data and high-resolution model forecast data, and dynamically update the forecast of external wind, wave and flow fields to obtain the forecast results of the marine environment in the sea area where the ship passes through;

[0010] S4, inputting the marine environment forecast results of the sea area where the ship passes through the initial route into the motion response mathematical model, using the relative motion between the ship and the marine environmental elements, the marine environmental element data and the ship status data as inputs, and using them as inputs of the neural network to calculate the motion response of the ship passing through the sea area;

[0011] S5. Collect other requirements of the ship's voyage mission as other constraints;

[0012] S6. Input the motion response and other constraints of the ship passing through the sea area into the intelligent optimization algorithm, add time constraints and fuel consumption constraints, and optimize the calculation to obtain the optimized route and optimized propeller speed that the ship should take under the optimization target;

[0013] S8. After obtaining the latest ocean hydrological and meteorological forecast, based on the latest ocean hydrological and meteorological forecast, repeat S3 to S7 and execute the latest route and speed optimization plan.

[0014] Furthermore, S2 specifically includes the following steps:

[0015] S21, setting the model input and model output of the training model, wherein the model input includes ship element information, environmental element information and dynamic information, and the model output is the speed and motion posture of the ship under the influence of wind and waves;

[0016] S22. Based on the model input and model output, an input-output training sample set of the model is established by a ship hydrodynamic calculation method to calculate the response of a specific ship to environmental factors in terms of ship speed and motion posture; assuming that there are m samples in the sample set, the input vector x has n independent variables, the input X is an m×n dimensional matrix, the output vector y has l dependent variables, and the output Y is an m×l dimensional matrix;

[0017] S23. According to the input and output data of the training samples, the neural network is trained so that the neural network accurately approximates the input-output mapping, and a neural network mapping model of the greenhouse gas emission model of ship navigation is established.

[0018] Furthermore, the ship element information includes ship size and superstructure parameters, ship draft and draft difference, and propeller size; the environmental element information includes wind and waves; and the dynamic information includes the relative position and movement relationship between the ship and environmental elements, propeller speed, and rudder angle.

[0019] Furthermore, S23 specifically includes the following steps:

[0020] S231, according to the uniform distribution sampling method, generate random numbers from the input and output sample sets X and Y to perform sampling, and extract the first group of data x in step 2 1 and 1, the second step is to extract the second set of data x 2 and 2 , and so on, extract data x in the kth step k and k , construct a training sample set U∈R k×n and V∈R k×l ;

[0021] S232, construct a radial basis function neural network using the input and output training sample sets U and V, and perform regression calculation of the hidden layer response matrix and the output using the partial least squares method:

[0022] In the RBF network, the Gaussian function is used as the radial basis function, and the vector u j (j=1,…,k) becomes the hidden node center; the hidden layer response matrix A∈R is obtained by radial basis function calculation. k×k , where the element in row i and column j is:

[0023]

[0024] Among them, u i is the i-th sample, where ||·|| represents the Euclidean distance, c j and σ j are the center and width of the jth basis function respectively;

[0025] S233, performing PLS principal component extraction and regression operation between the hidden layer response matrix A and the output matrix V;

[0026] After extracting the principal component T, project A and V onto the principal component matrix respectively, and we get:

[0027] V=TR+F=AWR+F

[0028] Among them, T is the principal component matrix of A, which is k×n T dimensional matrix; W is the transformation matrix of A, which is k×n T dimensional matrix; R is n T ×l-dimensional regression coefficient matrix; F is the k×l-dimensional residual matrix;

[0029] Get the generalized model:

[0030]

[0031] Among them A p1 and W 1 are k×k and k×n respectively. T dimensional matrix, R 1 n T ×l-dimensional matrix, generalize the obtained network to the samples in the set U, and set the accuracy threshold to determine the number of principal component extraction times nT , when the change of the generalized root mean square error is less than the set accuracy change threshold e 1 Stop extracting principal components when:

[0032]

[0033] S234, generalize the obtained network to the training data in the entire data set X, and calculate the generalization error for the entire data set:

[0034]

[0035] in:

[0036]

[0037] S235, if RMSE O Greater than the set accuracy threshold e 2 , then repeat steps S231 to S234;

[0038] S236, if RMSE O Less than the set accuracy threshold e 2 , then it ends;

[0039] S237, the number of principal components obtained n T The number of loop steps k is the number of principal component extractions and the number of hidden nodes of the final radial basis function network, and the vectors in the set U are the hidden nodes of the obtained radial basis function network.

[0040] Furthermore, in S3:

[0041] The spatiotemporal generative adversarial neural network model includes two modules: the generator G(Z) and the discriminator D(x). The generator module G(Z) assimilates and fuses the multimodal data of observation data and model forecast fields, where Z represents the observation point information and the meteorological field; the meteorological field map generated by the generator module G and the real map decoded in the reanalysis data are fed back to the discriminator D, and the input of the discriminator D is the generated meteorological field data and the reanalysis meteorological field data; the output of D is a scalar, which is used to judge the probability of whether the output map comes from the real input map; the stopping criterion for the training of the spatiotemporal generative adversarial network model is that the discriminator cannot distinguish whether the generated meteorological field forecast map conforms to the real map;

[0042] The loss function of the spatiotemporal generative adversarial network model is as follows:

[0043]

[0044] Among them, L D is the discriminator loss, L G is the generator loss function;

[0045] The parameters of the mathematical forecast model of the external marine meteorological environment field are trained and updated based on the following minimization and maximization formulas:

[0046]

[0047] Among them, P data (x) is the distribution of reanalysis meteorological field mapping data, P z (z) is the distribution of observed and numerically simulated meteorological field data input to the model, and E represents the expectation operator;

[0048] The number of features of the time series sequential tensor of the multimodal meteorological factors input to the spatiotemporal generative adversarial network model is C, H is the height of the model input tensor, and W is the width of the model input tensor. Represents the input tensor of the model. The attention mechanism module uses dot-product to further calculate the three matrices K, Q and V of the input tensor to obtain three The corresponding mapping weight feature matrix value;

[0049] The dimensions of the K and Q feature matrices should be consistent, and the i-th query eigenvalue in the iterative operation is and the jth key eigenvalue The similarity between them is determined by the regularization function Calculate acquisition;

[0050] The shipboard instruments are set to measure external meteorological elements such as wind, current and wave, respectively. wind ,S current ,S wave , the ERA5 high-resolution numerical model predicts the two-dimensional field elements of wind, current and wave respectively. wind ,ERA current ,ERA wave ,Dynamically updated sea condition forecast model for ships sailing in the sea area:

[0051] P wind,t =STGAN(S wind,t ,ERA wind,t )

[0052] P current,t =STGAN(S curren,t ,ERA current,t )

[0053] P wave,t =STGAN(S wave,t ,ERA wave,t )

[0054] Among them, STGAN is a spatiotemporal generative adversarial neural network model, P*,t represents the external wind and wave forecast field information of the sea area where the ship is sailing at time t;

[0055] Based on the generative adversarial ocean environment forecast model, accurate forecast results of the ocean environment such as wind field, wave field and flow field in the sea area where the ship passes are obtained in real time and dynamically, providing real-time guidance for ship route planning.

[0056] Furthermore, S5 specifically includes the following steps: on both sides of the planned route of the ship, a geographical boundary is set between the departure port and the arrival port and grid division is performed within the boundary, grid nodes are forward connected in the direction from the ship's position to the destination port to establish alternative routes, and routes that do not meet geographical restrictions such as obstacles and restricted areas are eliminated in advance from the alternative routes, and the alternative routes are preliminarily screened.

[0057] Furthermore, S6 specifically includes the following steps:

[0058] Using intelligent optimization algorithms, under the constraints of time, fuel consumption and safety, optimization is used to calculate the recommended route and propeller speed that should be taken to minimize the ship's sailing time or fuel consumption under the wind, wave and current conditions encountered by the ship;

[0059] The optimization function of the intelligent optimization algorithm is designed as follows:

[0060] Among them, the optimization target is set according to the ship response forecast results calculated based on the real-time marine environment forecast according to the set T1, T2, T3, T4, and T5 forecasts at different forecast time intervals; in the absence of long-term forecasts, climate data are used as a substitute;

[0061] For the optimization in the shortest time, the optimization function is set to:

[0062] J t =min(C 1 t 1 +C 2 t 2 +C 3 t 3 +C 4 t 4 +C 5 t 5 )

[0063] Among them, t 1 is the ship sailing time corresponding to the 1-6h forecast, t 2 ,t 3 ,t 4 The corresponding ship sailing time for the ocean hydrological and meteorological forecasts for the 6th to 12th hour, 12th to 24th hour, 1st to 3rd day, and 3rd to 10th day, respectively; C k(k=1, 2, ..., 5) is the proportional coefficient corresponding to each flight segment.

[0064] For the optimization with minimum fuel consumption, the optimization function is set as:

[0065] J f =min(C 1 f 1 +C 2 f 2 +C 3 f 3 +C 4 f 4 +C 5 f 5 )

[0066] Among them, t 1 is the ship sailing time corresponding to the 1-6h forecast, t 2 ,t 3 ,t 4 The corresponding ship sailing time for the ocean hydrological and meteorological forecasts for the 6th to 12th hour, 12th to 24th hour, 1st to 3rd day, and 3rd to 10th day, respectively; C k (k=1, 2, ..., 5) is the proportional coefficient corresponding to each flight segment.

[0067] C k In this scheme, the attenuation coefficient is adopted, C k The value of is:

[0068] C k =0.5 k-1 (k=1,2,...,5).

[0069] The present invention also provides a storage medium, which includes a stored program, wherein when the program is run, any of the above-mentioned ship ocean route optimization forecasting methods is executed.

[0070] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes any one of the above-mentioned methods for optimizing and forecasting ocean routes for ships through the operation of the computer program.

[0071] Compared with the prior art, the present invention has the following advantages:

[0072] By assigning different weighting coefficients to forecasts of different durations, optimization indicators can be constructed. On the one hand, more accurate and reliable routes can be constructed. On the other hand, ocean routes can be smoothed to avoid large changes in routes in a short period of time, reduce ship voyages and increase navigation safety.

[0073] The motion response of the ship to the environment is calculated using the offline neural network fitting model to increase the speed of the calculation of the ship's motion response. In the establishment of the neural network, the partial least squares method is used to solve the problem of too high input dimension, and a simple network structure is selected by gradually increasing the network nodes to improve the generalization ability of the obtained neural network.

[0074] In the marine hydrological and meteorological forecasting method, compared with the traditional generative adversarial model, the present invention introduces a dual attention mechanism module to map the high-dimensional feature information containing multiple meteorological factors to the multi-dimensional feature matrix of the encoder and decoder. And the basic skeleton of the generative adversarial model is replaced with a two-dimensional convolutional neural network module for directly inputting two-dimensional meteorological and ocean element field information. A 2D deep CNN layer is used to build a deep encoder and decoder of the generative adversarial model. The 2D convolution kernel is used to mine the deep coupling interaction influence features between various meteorological factors; the deep convolution block of this method can provide a method to improve the prediction efficiency and performance, while reducing its computational and memory requirements, and can better perform feature extraction and aggregation. The encoder part extracts the high-dimensional feature map of the tensor sequence with low-order nonlinear information, and the decoder module derives high-dimensional semantic information; improves the forecast accuracy of data fusion field elements. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0076] Figure 1 The figure is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0077] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0078] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0079] like Figure 1 As shown, the present invention provides a method for optimizing and predicting ocean routes for ships. The optimization method takes ship routes and ship propeller speeds as optimization objects, takes the safety of ships in severe sea conditions as constraints, takes the shortest sailing time or the lowest fuel consumption as optimization indicators, establishes a ship navigation optimization model, and performs optimization solutions. This optimization algorithm optimizes the ship's navigation plan elements including routes and propeller speeds, and can obtain a navigation plan that meets the user's optimization needs while avoiding severe sea conditions to ensure safety. This method contributes to the intelligent construction of ship navigation, and the specific steps are as follows:

[0080] (1) Based on the voyage mission and route design reference materials such as "World Ocean Routes" and "Route Design Map", the initial route for the ship's ocean voyage is generated according to the turning point and route type;

[0081] (2) Establish a mathematical model of the ship's motion response to the environment at sea. First, set the training model input and model output. Set the model input as ① information on ship elements such as ship size and superstructure parameters, ship draft and draft difference, propeller size, etc.; ② information on environmental elements such as wind and waves; ③ dynamic information such as the relative position and motion relationship between the ship and environmental elements, propeller speed, and rudder angle. Set the training model output to the ship's speed and motion posture under the influence of wind and waves. Secondly, establish the model input-output training samples through ship hydrodynamic calculations, and calculate the response of a specific ship to environmental elements in terms of ship speed and motion posture. Finally, based on the input and output data of the obtained samples, train the neural network so that the neural network accurately approximates the input-output mapping, and establish a neural network mapping model for the greenhouse gas emission model of ship navigation. Specifically,

[0082] Suppose there are m samples in the sample set, the input vector x has n independent variables, the input X is an m×n dimensional matrix, the output vector y has l dependent variables, and the output Y is an m×l dimensional matrix.

[0083] The algorithm steps for establishing the neural network calculation model of ship marine motion are as follows:

[0084] I. According to the uniform distribution sampling method, random numbers are generated from the input and output sample sets X and Y to implement sampling. In step 2, the first set of data x is extracted. 1 and 1 , the second step is to extract the second set of data x 2 and 2 , and so on, extract data x in the kth step k and k , construct a training sample set U∈R k×n and V∈R k×l ;

[0085] II. Use the input and output training sample sets U and V to construct a radial basis function neural network, and use the partial least squares method to perform regression calculations on the hidden layer response matrix and output:

[0086] In the radial basis function network, the Gaussian function is used as the radial basis function, and the vector u j (j=1,…,k) becomes the hidden node center. The hidden layer response matrix A∈R is obtained by radial basis function calculation. k×k . The element in row i and column j is:

[0087]

[0088] where u i is the i-th sample, where ||·|| represents the Euclidean distance, c j and σ j are the center and width of the jth basis function respectively.

[0089] III. Perform PLS principal component extraction and regression operations between the response matrix A and the output matrix V. After extracting the principal component T, project A and V onto the principal component matrix respectively to obtain:

[0090] V=TR+F=AWR+F

[0091] Where T is the principal component matrix of A, which is k×n T dimensional matrix; W is the transformation matrix of A, which is k×n T dimensional matrix; R is n T ×l-dimensional regression coefficient matrix; F is the k×l-dimensional residual matrix.

[0092] Get the generalized model:

[0093]

[0094] Among them A p1 and W 1 are k×k and k×n respectively.T dimensional matrix. 1 n T ×l-dimensional matrix. The obtained network is generalized to the samples in the set U, and an accuracy threshold is set to determine the number of principal component extractions n T , when the change of the generalized root mean square error (RMSE) is less than the set accuracy change threshold e 1 Stop extracting principal components when:

[0095]

[0096] IV. Generalize the obtained network to the training data in the entire dataset X and calculate the generalization error for the entire dataset:

[0097]

[0098] in

[0099] V. If RMSE O Greater than the accuracy threshold e 2 , then repeat steps I-IV.

[0100] VI. Algorithm in RMSE O Less than the accuracy threshold e 2 When it ends.

[0101] The number of principal components obtained is n T The number of loop steps k is the number of principal component extractions and the number of hidden nodes of the final radial basis function network, and the vectors in the set U are the hidden nodes of the obtained radial basis function network.

[0102] (3) Based on the selected ocean hydrological and weather forecast, a mathematical forecast model of the wind field, wave field and current field in the sea area through which the ship passes is established.

[0103] According to the coverage of wind field, flow field and wave field selected in the sea area where the ship's planned route passes, the high-resolution ERA5 ocean environment forecast model in the European medium- and long-term weather forecast system is used as an example, and the data assimilation is carried out in combination with the ship's real-time observation of the external marine meteorological environment data information. The spatiotemporal generative adversarial neural network with the introduction of the attention mechanism is used to realize the assimilation and fusion of the observation data and the high-resolution model forecast data, thereby improving the accuracy of the weather and sea condition forecast in the navigation area. According to the practicality and timeliness of the weather forecast information required by the ship during navigation, the dynamic update forecast of the external wind, wave and flow field is carried out with reference to the real-time speed of the ship to provide more accurate external marine environment forecast information.

[0104] ① The spatiotemporal generative adversarial model STGAN consists of two modules: the generator G(Z) and the discriminator D(x). The generator module G(Z) assimilates and fuses the multimodal data of observation data and model forecast fields, where Z represents the observation point information and the meteorological field. The meteorological field map generated by G and the real map decoded from the reanalysis data are fed back into the discriminator D. The input of D is the generated meteorological field data and the reanalysis meteorological field data; the output of D is a scalar, which is used to determine the probability of whether the output map comes from the real input map (reanalysis data information). The stopping criterion for model training is that the discriminator cannot distinguish whether the generated meteorological field forecast map conforms to the real map.

[0105] The loss function of the model is:

[0106]

[0107] Where L D is the discriminator loss, L G is the generator loss function. The prediction model parameters are trained and updated based on the following minimization and maximization methods:

[0108]

[0109] Among them, P data (x) is the distribution of reanalysis meteorological field mapping data. z (z) is the distribution of observed and numerically simulated meteorological field data input to the model. E represents the expectation operator.

[0110] ② The number of features of the time series sequential tensor of the multimodal meteorological factors input to the generator model is C. H and W are the height and width of the model input tensor. Represents the input tensor of the model. The attention mechanism module uses dot-product to further calculate the three matrices K, Q and V of the input tensor to obtain three The corresponding mapping weight feature matrix value.

[0111] The dimensions of the K and Q feature matrices should be consistent. and the jth key eigenvalue The similarity between them is determined by the regularization function Calculate and obtain. Due to the similarity calculation between different convolutional network layers, such as different positions and The similarity calculation is different. The linear attention mechanism in the high-dimensional feature extraction calculation of multimodal meteorological factors can reduce the consumption of computing resources and improve the model calculation efficiency.

[0112] ③ Setting, the shipboard instruments measured external wind, current and wave meteorological elements are S wind ,Scurrent ,S wave , the ERA5 high-resolution numerical model predicts the two-dimensional field elements of wind, current and wave respectively. wind ,ERA current ,ERA wave ,Dynamically updated sea condition forecast model for ships sailing in the sea area:

[0113] P wind,t =STGAN(S wind,t ,ERA wind,t )

[0114] P current,t =STGAN(S curren,t ,ERA current,t )

[0115] P wave,t =STGAN(S wave,t ,ERA wave,t )

[0116] Among them, STGAN is a spatiotemporal generative adversarial neural network model, P *,t It indicates the external wind and wave forecast field information of the sea area where the ship is sailing at time t.

[0117] (4) Calculate the motion response of the ship passing through the sea area according to the sea area through which the ship's planned route passes. The input of the calculation model is ① information on ship elements such as ship size and superstructure parameters, ship draft and draft difference, propeller size, etc.; ② information on environmental elements such as wind, waves, and currents; ③ dynamic information such as the relative position and motion relationship between the ship and environmental elements, propeller speed, and rudder angle. The output of the calculation model is the speed and motion posture of the ship under the influence of wind, waves, and currents.

[0118] (5) Establish alternative ship routes along the ship's departure port, and conduct preliminary elimination and screening. On both sides of the ship's planned route, set a geographical boundary between the departure port and the arrival port and divide the grid within the boundary. Connect the grid nodes forward from the ship's position to the destination port to establish alternative routes. Eliminate routes that do not meet geographical restrictions such as obstacles and restricted areas in advance from the alternative routes, and conduct preliminary screening of the alternative routes.

[0119] (6) Optimize the selection of ship routes and speeds. Using optimization methods such as dynamic programming and variational methods or intelligent optimization algorithms such as genetic algorithms and particle swarm algorithms, under the constraints of time, fuel consumption, safety, etc., using the established neural network calculation model of speed and ship motion posture, optimize the calculation of the recommended route and propeller speed that should be taken under the optimization objectives of minimizing the ship's sailing time or minimizing fuel consumption under the wind, wave, and current conditions encountered by the ship.

[0120] Among them, the optimization target is set according to the ship response forecast results calculated based on the real-time marine environment forecast according to different forecast time intervals such as T1 (6h), T2 (12h), T3 (1 day), T4 (3 days), and T5 (10 days). In the absence of long-term forecasts, climate data are used as a substitute.

[0121] For the optimization in the shortest time, the optimization function is set to:

[0122] J t =min(C 1 t 1 +C 2 t 2 +C 3 t 3 +C 4 t 4 +C 5 t 5 )

[0123] Among them, t 1 is the ship sailing time corresponding to the 1-6h forecast, t 2 ,t 3 ,t 4 The corresponding ship sailing time for the ocean hydrological and meteorological forecasts for the 6th to 12th hour, 12th to 24th hour, 1st to 3rd day, and 3rd to 10th day, respectively; C k (k=1, 2, ..., 5) is the proportional coefficient corresponding to each flight segment.

[0124] For the optimization with minimum fuel consumption, the optimization function is set as:

[0125] J f =min(C 1 f 1 +C 2 f 2 +C 3 f 3 +C 4 f 4 +C 5 f 5 )

[0126] Among them, t 1 is the ship sailing time corresponding to the 1-6h forecast, t 2 ,t 3 ,t 4 The corresponding ship sailing time for the ocean hydrological and meteorological forecasts for the 6th to 12th hour, 12th to 24th hour, 1st to 3rd day, and 3rd to 10th day, respectively; C k (k=1, 2, ..., 5) is the proportional coefficient corresponding to each flight segment.

[0127] C kIn this scheme, the attenuation coefficient is adopted, C k The value of is:

[0128] C k =0.5 k-1 (k=1,2,...,5)

[0129] (7) Optimize the route and speed.

[0130] It displays the turning points and speeds corresponding to the route, and displays options such as using the original route and speed, using the optimized route and speed with the shortest sailing time, and using the optimized route and speed with the shortest fuel consumption for the ship drivers to choose from. It also displays the time series values ​​and curves of the ship's speed changes and ship's motion posture changes under different options for the ship drivers' decision-making reference.

[0131] (8) Carry out rolling optimization of routes.

[0132] After obtaining the latest ocean hydrological and meteorological forecast, repeat steps (3)-(7) and execute the latest route and speed optimization plan.

[0133] The "forecast" in this invention is reflected in two aspects: first, modeling and time interpolation based on hydrological and meteorological forecasts are forecasts to a certain extent; second, modeling based on ship status and marine environmental elements to calculate the motion response of the ship when it passes a certain location in the future is also a forecast of the ship's motion response. Route design is based on the optimization of forecast results.

[0134] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing and forecasting ocean routes for ships, characterized in that: The steps include: S1. Establish the initial route according to the voyage mission, turning point and route type; S2. Establish a mathematical model of the ship's motion response to the environment at sea; S3. Establish a mathematical forecast model for the wind field, wave field and flow field in the sea area where the ship passes through, use the spatiotemporal generative adversarial neural network model with an attention mechanism to achieve the assimilation and fusion of observation data and high-resolution model forecast data, and dynamically update the forecast of external wind, wave and flow fields to obtain the forecast results of the marine environment in the sea area where the ship passes through; S4, inputting the marine environment forecast results of the sea area where the ship passes through the initial route into the motion response mathematical model, using the relative motion between the ship and the marine environmental elements, the marine environmental element data and the ship status data as inputs, and using them as inputs of the neural network to calculate the motion response of the ship passing through the sea area; S5. Collect other requirements of the ship's voyage mission as other constraints; S6. Input the motion response and other constraints of the ship passing through the sea area into the intelligent optimization algorithm, add time constraints and fuel consumption constraints, and optimize the calculation to obtain the optimized route and optimized propeller speed that the ship should take under the optimization target; S8. After obtaining the latest ocean hydrological and meteorological forecast, based on the latest ocean hydrological and meteorological forecast, repeat S3 to S7 and execute the latest route and speed optimization plan.

2. The method for optimizing and forecasting ocean routes of ships according to claim 1, characterized in that: S2 specifically includes the following steps: S21, setting the model input and model output of the training model, wherein the model input includes ship element information, environmental element information and dynamic information, and the model output is the speed and motion posture of the ship under the influence of wind and waves; S22. Based on the model input and model output, an input-output training sample set of the model is established by a ship hydrodynamic calculation method to calculate the response of a specific ship to environmental factors in terms of ship speed and motion posture; assuming that there are m samples in the sample set, the input vector x has n independent variables, the input X is an m×n dimensional matrix, the output vector y has l dependent variables, and the output Y is an m×l dimensional matrix; S23. According to the input and output data of the training samples, the neural network is trained so that the neural network accurately approximates the input-output mapping, and a neural network mapping model of the greenhouse gas emission model of ship navigation is established.

3. The method for optimizing and forecasting ocean routes of ships according to claim 2, characterized in that: The ship element information includes ship size and superstructure parameters, ship draft and draft difference, and propeller size; the environmental element information includes wind and waves; and the dynamic information includes the relative position and movement relationship between the ship and environmental elements, propeller speed, and rudder angle.

4. The method for optimizing and forecasting ocean routes of ships according to claim 2, characterized in that: S23 specifically includes the following steps: S231, according to the uniform distribution sampling method, generate random numbers from the input and output sample sets X and Y to perform sampling, extract the first group of data x1 and y1 in the second step, extract the second group of data x2 and y2 in the second step, and so on, extract data x in the kth step. k and k , construct a training sample set U∈R k×n and V∈R k×l ; S232, construct a radial basis function neural network using the input and output training sample sets U and V, and perform regression calculation of the hidden layer response matrix and the output using the partial least squares method: In the RBF network, the Gaussian function is used as the radial basis function, and the vector u j Become the hidden node center, j = 1, ..., k; the hidden layer response matrix A∈R is obtained by radial basis function calculation k×k , where the element in row i and column j is: Among them, u i is the i-th sample, where ||·|| represents the Euclidean distance, c j and σ j are the center and width of the jth basis function respectively; S233, performing PLS principal component extraction and regression operation between the hidden layer response matrix A and the output matrix V; After extracting the principal component T, project A and V onto the principal component matrix respectively, and we get: V=TR+F=AWR+F Among them, T is the principal component matrix of A, which is k×n T dimensional matrix; W is the transformation matrix of A, which is k×n T dimensional matrix; R is n T ×l-dimensional regression coefficient matrix; F is the k×l-dimensional residual matrix; Get the generalized model: Among them A p1 and W1 are k×k and k×n respectively. T dimensional matrix, R1 is n T ×l-dimensional matrix, generalize the obtained network to the samples in the set U, and set the accuracy threshold to determine the number of principal component extraction times n T , stop extracting the principal components when the change in the generalized root mean square error is less than the set accuracy change threshold e1: S234, generalize the obtained network to the training data in the entire data set X, and calculate the generalization error for the entire data set: in: S235, if RMSE O If it is greater than the set accuracy threshold e2, steps S231 to S234 are repeated; S236, if RMSE O If it is less than the set accuracy threshold e2, then the process ends; S237, the number of principal component extractions n obtained T The number of loop steps k is the number of principal component extractions and the number of hidden nodes of the final radial basis function network, and the vectors in the set U are the hidden nodes of the obtained radial basis function network.

5. The method for optimizing and forecasting ocean routes of ships according to claim 1, characterized in that: In S3: The spatiotemporal generative adversarial neural network model includes two modules: the generator G(Z) and the discriminator D(x). The generator module G(Z) assimilates and fuses the multimodal data of observation data and model forecast fields, where Z represents the observation point information and the meteorological field; the meteorological field map generated by the generator module G and the real map decoded in the reanalysis data are fed back to the discriminator D, and the input of the discriminator D is the generated meteorological field data and the reanalysis meteorological field data; the output of D is a scalar, which is used to judge the probability of whether the output map comes from the real input map; the stopping criterion for the training of the spatiotemporal generative adversarial network model is that the discriminator cannot distinguish whether the generated meteorological field forecast map conforms to the real map; The loss function of the spatiotemporal generative adversarial network model is as follows: Among them, L D is the discriminator loss, L G is the generator loss function; The parameters of the mathematical forecast model of the external marine meteorological environment field are trained and updated based on the following minimization and maximization formulas: Among them, P data (x) is the distribution of reanalysis meteorological field mapping data, P z (z) is the distribution of observed and numerically simulated meteorological field data input to the model, and E represents the expectation operator; The number of features of the time series sequential tensor of the multimodal meteorological factors input to the spatiotemporal generative adversarial network model is C, H is the height of the model input tensor, and W is the width of the model input tensor. Represents the input tensor of the model. The attention mechanism module uses dot-product to further calculate the three matrices K, Q and V of the input tensor to obtain three The corresponding mapping weight feature matrix value; The dimensions of the K and Q feature matrices should be consistent, and the i-th query eigenvalue in the iterative operation is and the jth key eigenvalue The similarity between them is determined by the regularization function Calculate acquisition; The external wind, current and wave meteorological elements measured by the shipboard instrument are set to S wind ,S current ,S wave , the ERA5 high-resolution numerical model predicts the two-dimensional field elements of wind, current and wave respectively. wind ,ERA current ,ERA wave ,Dynamically updated sea condition forecast model for ships sailing in the sea area: P wind,t =STGAN(S wind,t ,ERA wind,t ) P current,t =STGAN(S curren,t ,ERA current,t ) P wave,t =STGAN(S wave,t ,ERA wave,t ) Among them, STGAN is a spatiotemporal generative adversarial neural network model, P*,t represents the external wind and wave forecast field information of the sea area where the ship is sailing at time t; Based on the generative adversarial ocean environment forecast model, accurate forecast results of the wind field, wave field and flow field ocean environment of the sea area through which the ship passes are obtained in real time and dynamically, providing real-time guidance for ship route planning.

6. The method for optimizing and forecasting ocean routes of ships according to claim 1, characterized in that: S5 specifically includes the following steps: on both sides of the planned route of the ship, a geographical boundary is set between the departure port and the arrival port and grid division is performed within the boundary, grid nodes are connected forward from the ship's position to the destination port to establish alternative routes, and routes that do not meet the geographical restrictions of obstacles and restricted areas are eliminated in advance from the alternative routes, and the alternative routes are preliminarily screened.

7. The method for optimizing and forecasting ocean routes of ships according to claim 1, characterized in that: S6 specifically includes the following steps: Using intelligent optimization algorithms, under the constraints of time, fuel consumption and safety, optimization is used to calculate the recommended route and propeller speed that should be taken to minimize the ship's sailing time or fuel consumption under the wind, wave and current conditions encountered by the ship; The optimization function of the intelligent optimization algorithm is designed as follows: Among them, the optimization target is set according to the ship response forecast results calculated based on the real-time marine environment forecast according to the set T1, T2, T3, T4, and T5 forecasts at different forecast time intervals; in the absence of long-term forecasts, climate data are used as a substitute; For the optimization in the shortest time, the optimization function is set to: J t =min(C1t1+C2t2+C3t3+C4t4+C5t5) Among them, t1 is the ship sailing time corresponding to the 1-6h forecast, t2, t3, t4, and t5 are the ship sailing times corresponding to the 6-12h, 12-24h, 1-3 days, and 3-10 days ocean hydrological and meteorological forecasts respectively; C k is the proportional coefficient corresponding to each flight segment, k = 1, 2, ..., 5; For the optimization with minimum fuel consumption, the optimization function is set as: J f =min(C1f1+C2f2+C3f3+C4f4+C5f5) Among them, f1 is the ship sailing time corresponding to the 1-6h forecast, f2, f3, f4, and f5 are the ship sailing times corresponding to the 6-12h, 12-24h, 1-3 days, and 3-10 days ocean hydrological and meteorological forecasts respectively; C k is the proportional coefficient corresponding to each flight segment, k = 1, 2, ..., 5; C k is the attenuation coefficient, C k The value of is: C k =0.5 k-1 (k=1,2,...,5)。 8. A storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is run, the method for optimizing and forecasting ocean routes for ships as described in any one of claims 1 to 7 is executed.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: The processor executes the ship ocean route optimization forecasting method according to any one of claims 1 to 7 by running the computer program.

Images