Ocean current-based ship trial course planning method and system

Through the BP neural network model and current forecast model, multiple ship trial routes were generated and screened, solving the dependence and flexibility of pilot route planning in the existing technology, and realizing intelligent and flexible trial route planning.

CN119086003APending Publication Date: 2024-12-06JIANGNAN SHIPYARD (GRP) CO LTD
View PDF 0 Cites 2 Cited by

Patent Information

Application Number
CN202411361949.4
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-27
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The existing ship trial route planning methods rely on the experience of operators and limited current data, making it difficult to quickly find trial routes that meet user requirements, and it is difficult to flexibly plan routes according to user preferences.

Method used

The trial route planning method based on the BP neural network model is adopted, and the neural network model is constructed in combination with the optimization algorithm. The ocean current forecast model is used to obtain the ocean current field data in the trial space-time and space, and multiple planned trial routes are generated, and the comparison is selected according to the targets specified by the user. Optimize routes.

Benefits of technology

Based on the meteorological, marine environment and ship characteristic parameters in the trial space, multiple better trial routes are intelligently planned to display the ocean current situation and energy consumption, and improve the efficiency and flexibility of trial route planning.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119086003A_ABST
    Figure CN119086003A_ABST
Patent Text Reader

Abstract

The invention provides a ship trial flight route planning method based on ocean currents, and the method comprises the steps: building a target trial flight route planning model, obtaining a plurality of planned trial flight routes in trial flight time and space through the model, carrying out the target screening of the obtained plurality of planned trial flight routes, and obtaining an optimal planned trial flight route. Through the planning scheme of the invention, a plurality of better trial navigation routes and ocean current conditions and energy consumption thereof can be quickly obtained, so that energy reserve, carbon emission and the like of trial navigation ships can be measured and calculated. The trial flight route planning model is intelligent, flexible and upgradable. A user can decide training pool data of the neural network planning model, and the model can timely learn the preference of the user to select a pilot route and a real ship pilot route data rule, so that the ship pilot route planning efficiency is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the technical field of ship design and manufacturing, and in particular, to a method and system for planning a ship trial route based on ocean currents. Background Art

[0002] In the field of ship design and manufacturing, sea trials and speed measurements are the core steps to verify the design effectiveness and navigation capabilities of ships. This process requires the ship to sail in a mild ocean environment to obtain accurate and reliable speed and power data. However, operators face many challenges when selecting such sea trial routes.

[0003] Currently, route selection mainly depends on the operator's experience and ability to interpret nautical charts. They need to manually select routes with stable currents, little external interference, and suitable lengths in the vast sea area based on limited ocean current data and environmental information. This method is not only time-consuming and labor-intensive, but also due to the large scope of the ship trial area and the low spatiotemporal resolution of ocean current data, the method that relies on the operator's judgment and experience to find the trial route is difficult to quickly find a trial route that meets user requirements. In addition, the current trial route planning method is also difficult to flexibly plan the trial route according to user preferences. Summary of the invention

[0004] The purpose of this application is to provide a method and system for planning a ship trial route based on ocean currents, so as to obtain ocean current field data in the time and space of the ship trial, and provide multiple better trial routes for users to choose from.

[0005] In order to achieve the above objectives, this application adopts the following technical solutions:

[0006] A method for planning a ship trial voyage route based on ocean currents comprises: constructing a target trial voyage route planning model based on trial voyage route regulations and historical trial voyage route data; obtaining ship characteristic parameters and trial voyage related parameters and inputting them into the target trial voyage route planning model to obtain a plurality of planned trial voyage routes within the trial voyage time and space; the trial voyage related parameters include the ocean current field within the trial voyage time and space, and the ocean current field within the trial voyage time and space is predicted by an ocean current prediction model; and screening a plurality of the planned trial voyage routes according to the target parameters to obtain a target planned trial voyage route.

[0007] In some embodiments, the process of constructing the target trial voyage route planning model includes: constructing a neural network planning model based on a BP neural network model combined with an optimization algorithm; training the neural network planning model using a first training data set, and obtaining the neural network planning model when the termination standard is reached; loading a first test data set to test the trained neural network planning model, and obtaining the target trial voyage route planning model when the test accuracy is met. In some embodiments, the optimization algorithm may include a genetic algorithm, a particle swarm algorithm, a butterfly algorithm, or an ant colony algorithm, which is used to optimize the initial weights and initial thresholds of the BP neural network model.

[0008] In some embodiments, the BP neural network includes an input layer, a hidden layer and an output layer, wherein the input layer is one layer, and the number of nodes in the input layer is equal to the number of first input parameters, and the first input parameters include ship characteristic parameters, trial voyage area, trial voyage starting point, trial voyage start and end time, ocean current field, speed, seawater temperature, seawater salinity and air density; the number of layers and nodes in the hidden layer are adjusted according to the performance and training efficiency of the BP neural network model; the output layer is one layer, and the number of nodes in the output layer is equal to the number of first output parameters, and the first output parameters include the trial voyage route trajectory, ocean current conditions and energy consumption of the ship on the trial voyage route.

[0009] In some embodiments, the first training data set and the first test data set are composed of independent data sets of the historical trial route data; the termination criterion is set according to the absolute error, relative error, root mean square error, and determination coefficient, wherein the calculation formulas of the root mean square error and the determination coefficient are shown in formula (1) and formula (2).

[0010]

[0011] Where: X rmse represents the root mean square error; y i Represents the output value of the i-th group model; y i0 represents the i-th group of measurements; N represents the total number of data groups; R 2 represents the coefficient of determination; y m Indicates the average of the measured values.

[0012] In some embodiments, the process of constructing an ocean current forecast model includes: programming using MATLAB, Python or R language, and constructing an initial ocean current forecast model based on a random forest model, a neural network model, a support vector machine model or a multivariate regression model; using a second training data set to train the initial ocean current forecast model, and obtaining the initial ocean current forecast model when the termination criteria are met; loading a second test data set to test the trained initial ocean current forecast model, and obtaining the target ocean current forecast model when the test accuracy is met.

[0013] In some embodiments, the second training data set and the second test data set are derived from actual measurement data, numerical simulation data, and physical test data of ocean currents; the second training data set and the second test data set include second input parameters and second output parameters, and the second input parameters include latitude and longitude range, time range, wind field, temperature and salinity field, topography and coastline; the second output parameter is an ocean current field; the training termination criteria of the ocean current prediction model are set based on the absolute error, relative error, root mean square error, and determination coefficient between the model prediction value and the actual measurement value.

[0014] In some embodiments, the process of screening multiple planned trial voyages according to target parameters to obtain a target planned trial voyage route includes: the target parameters are the ocean current conditions and energy consumption of the ship on the planned trial voyage route, and the ocean current conditions include the magnitude of the current speed, the direction of the current, the ratio of downstream / upstream time, and the ratio of downstream / upstream intensity.

[0015] A second aspect of the present application provides a ship trial voyage route planning system based on ocean currents, comprising a route planning model construction module, a trial voyage route planning module and a trial voyage route screening module, wherein the route planning model construction module is used to construct a target trial voyage route planning model based on trial voyage route regulations and historical trial voyage route data; the trial voyage route planning module is used to obtain ship characteristic parameters and trial voyage related parameters and input them into the target trial voyage route planning model to obtain multiple planned trial voyage routes within the trial voyage time and space; the trial voyage related parameters include the ocean current field within the trial voyage time and space, and the ocean current field within the trial voyage time and space is obtained by predicting an ocean current forecasting model; the trial voyage route screening module is used to screen multiple planned trial voyage routes according to target parameters to obtain a target planned trial voyage route.

[0016] In some embodiments, the target planned trial voyage route obtained by screening is the current optimal planned trial voyage route, and the current optimal planned trial voyage route is added to the training data set of the current trial voyage route planning model for training to obtain a better target trial voyage route planning model.

[0017] Compared with the prior art, the present invention has the following advantages:

[0018] The present application provides a method and system for planning a ship trial voyage route based on ocean currents. Before the trial voyage, the ocean current field in the trial voyage time and space is obtained through an ocean current forecast model based on the weather, ocean environment, topography, etc. in the trial voyage time and space; based on the characteristics of the ship, the ocean current field in the trial voyage time and space, the preset speed and other parameters, by changing the trial voyage area, starting point and start and end time, a plurality of planned trial voyage routes are obtained with the help of a neural network planning model; according to the planning goal specified by the user, a plurality of better planned trial voyage routes are screened out, and the ocean current conditions and energy consumption of the better trial voyage routes are displayed for the user to choose; the user can specify a certain planned trial voyage route data to enter the neural network planning model training pool. During the trial voyage, the ocean current conditions, energy consumption and speed of the real-time trial voyage route are monitored and displayed; the user can specify a certain real-time trial voyage route data to enter the neural network planning model training pool. Based on the user-specified planned trial voyage route data and the user-specified real-time trial voyage route data in the neural network planning model training pool, the neural network planning model can timely learn the user's trial voyage route preferences and the rules of real ship trial voyage data, so as to plan the trial voyage route more flexibly and intelligently and predict the ocean current conditions and energy consumption of each trial voyage route. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 A schematic diagram showing a flow chart of a method for planning a ship trial route based on ocean currents according to an embodiment of the present application is shown;

[0021] Figure 2 A schematic diagram showing a flow chart of a method for planning a ship trial route based on ocean currents according to another embodiment of the present application;

[0022] Figure 3 A schematic diagram of the ship's trial route is shown;

[0023] Figure 4 A schematic diagram of the construction process of the target trial flight route planning model in an embodiment of the present application is shown;

[0024] Figure 5 A schematic diagram of the construction process of the ocean current forecast model in an embodiment of the present application is shown;

[0025] Figure 6 The interface display diagram shows the real-time ocean current conditions, energy consumption, speed and trial route;

[0026] Figure 7The interface display schematic diagram shows the optimal trial route and its ocean current conditions, energy consumption, and speed;

[0027] Figure 8 A schematic diagram of the process of upgrading the neural network planning model is shown.

[0028] Description of reference numerals:

[0029] 3-1 – Corrected Williamson Turn;

[0030] 3-2——Trial route;

[0031] 3-3——The end / start position of the outbound / return speed measurement voyage;

[0032] 3-4——Start / end position of the outbound / return speed measurement trip;

[0033] 6-1——Real-time trial route;

[0034] 6-2 - Hull;

[0035] 6-3 – Trial area micro-unit;

[0036] 6-4——Microcell color, indicating the impact of ocean currents on the ship's sea trials during the round trip;

[0037] 6-5——Real-time display of ocean current conditions, energy consumption and speed of the trial route;

[0038] 7-1——1# better trial route;

[0039] 7-2——2# better trial route;

[0040] 7-3——3# better trial route;

[0041] 7-4——Ocean current conditions, energy consumption and speed display of the preferred trial route. DETAILED DESCRIPTION

[0042] The technical solutions of the specific implementation methods of the present application are further described in detail below in conjunction with the accompanying drawings. These implementation methods are only used to illustrate the present application, but not to limit the present application.

[0043] The term "and / or" in this application is a description of the association relationship of associated objects, indicating that three relationships may exist, for example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. In the present invention, the symbol " / " indicates that the associated objects are in an "or" relationship, for example, "A / B" means "A or B".

[0044] The terms "first" and "second" in the specification and claims of this application are used to distinguish different objects rather than to describe a specific order of the objects. For example, "first response message" and "second response message" are used to distinguish different response messages rather than to describe a specific order of the response messages.

[0045] In the embodiments of the present application, words such as "exemplary" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary" or "for example" in the embodiments of the present application should not be considered as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary" or "for example" is intended to present related concepts in a specific way.

[0046] In the description of the embodiments of the present application, unless otherwise specified, "multiple" means two or more than two. For example, multiple processing units refer to two or more processing units, etc., and multiple elements refer to two or more elements, etc.

[0047] The following is a detailed description of a method and system for planning a ship trial route based on ocean currents in the present application in conjunction with the accompanying drawings.

[0048] Figure 1 , Figure 2 FIG. 1 is a flow chart showing a method for planning a ship trial route based on ocean currents according to an embodiment of the present application. Figure 1 , Figure 2 As shown, the method for planning a trial route for a ship based on ocean currents in an embodiment of the present application includes the following steps:

[0049] Based on the trial route regulations and historical trial route data, a target trial route model is constructed. Specifically, Figure 3 As shown, the ship trial route is stipulated as follows: the trial voyages shall be carried out in the same area, and for each cycle, each speed test voyage shall start and end at the same position. Between the round trips, the modified Williamson turn 3-1 or similar maneuvering method shall be used to return the ship to the opposite heading or parallel to the trial baseline, where 3-3 is the end / start position of the round / return speed test voyage, and 3-4 is the start / end position of the round / return speed test voyage. Parallel means that the deviation from the trial baseline does not exceed one ship length. During the trial, the main engine throttle, speed or pitch remains unchanged, and the duration of the speed test voyage should be the same and at least ten minutes. The speed test voyage should be carried out in the main wave direction into the waves or with the waves. Once the heading is determined, the selected track remains unchanged throughout the trial.

[0050] The construction process of the above-mentioned target trial route planning model includes the following steps: Figure 4As shown, a neural network planning model is constructed based on a BP (back propagation) neural network model combined with an optimization algorithm using computer programming languages ​​such as MATLAB, Python, and R. BP neural network has good nonlinear mapping capabilities and exhibits unique advantages when processing complex and uncertain marine environmental data. In order to improve the accuracy of the forecast results, the present application embodiment further combines the optimization algorithm with the BP neural network to optimize the initial weights and initial thresholds of the neural network, which can effectively avoid the local optimal problem of the BP neural network model, thereby improving the forecast accuracy of the model. In one embodiment, the aforementioned optimization algorithm may include a genetic algorithm, a particle swarm algorithm, a butterfly algorithm, or an ant colony algorithm. These algorithms search for optimal solutions or approximate optimal solutions in the search space by simulating the optimization process of nature (such as genetic evolution, swarm intelligence, etc.), thereby effectively avoiding the local optimal problem.

[0051] The BP neural network model usually includes an input layer, a hidden layer and an output layer, each layer is composed of nodes, wherein the input layer and the output layer are usually one layer, the number of nodes is equal to the number of the first input parameter and the number of the first output parameter respectively, and the number of layers and nodes of the hidden layer are usually determined by debugging the performance and training efficiency of the BP neural network model. The optimization algorithm is used to improve the planning accuracy of the BP neural network model, which includes but is not limited to the genetic algorithm, particle swarm algorithm, butterfly algorithm, ant colony algorithm, and can optimize the initial weights and initial thresholds of the BP neural network model.

[0052] This application uses the first training data set to train the neural network planning model to debug the model settings, and then verifies the accuracy of the model through the first test data set. For example, the historical trial route data set can be divided into a first training data set and a first test data set, which are independent of each other and have the same format, and both contain first input parameters and first output parameters. Among them, the first input parameters include ship characteristic parameters, trial area, trial starting point, trial start and end time, ocean current field, speed and other parameters, and the first output parameters include trial route trajectory, ocean current conditions and energy consumption of the ship on the trial route. Among them, the ship's characteristic parameters include waterline, length between perpendiculars, breadth, draft, trim, square coefficient, main engine power, engine capacity, etc.; the input trial area and trial starting point are often expressed by longitude and latitude; the input ocean current field usually includes the magnitude and direction of the ocean current speed within the trial time and space; the input speed includes the ship's ground speed, water speed, etc.; other input parameters include seawater temperature, seawater salinity, air density, etc.; the output ocean current conditions of the ship on the trial route include the ratio of downstream / upstream time, the ratio of downstream / upstream intensity, etc.

[0053] The training termination criteria of the neural network planning model can be determined based on the absolute error, relative error, root mean square error, determination coefficient, etc., where the calculation formulas of the root mean square error and determination coefficient are shown in formula (1) and formula (2). The training algorithms of the model include but are not limited to the gradient descent method, Levenberg-Marquardt method, Powell-Beale conjugate gradient method, and adaptive learning rate momentum gradient descent method.

[0054]

[0055] Where: X rmse represents the root mean square error; y i Represents the output value of the model of group i; y i0 represents the i-th group of measurements; N represents the total number of data groups; R 2 represents the coefficient of determination; y m Indicates the average of the measured values.

[0056] On the programming platform, a neural network planning model is constructed according to the model type, number of model layers, number of nodes in each layer, termination criteria, and training algorithm, and then the first training data set is loaded to train the model. The model that meets the termination criteria can plan the trial route. In the model that can plan the trial route, the first test data set is loaded to verify the planning accuracy of the model. The steps of verifying the accuracy of the model with the first test data set are: loading the input parameters of the first test data set into the trained model, and obtaining the planned trial route data through the model; calculating the error between the trial route data of the first test data set and the planned trial route data, and verifying the planning accuracy of the model. Furthermore, in the model that can plan the trial route, the planning accuracy and calculation time of the model are compared, and then the optimal trial route planning model, that is, the target trial route planning model, is selected for subsequent trial route planning.

[0057] Input the ship characteristic parameters and trial-related parameters into the target trial route planning model obtained in the above steps to obtain multiple planned trial routes. The above trial-related parameters can be, for example, the trial area, the trial starting point, the trial start and end time, the ocean current field in the trial time and space, the preset speed and other parameters. The ship characteristic parameters, the ocean current field in the trial time and space and other parameters are obtained through actual measurement, model prediction, theoretical calculation, etc.; the trial area, trial starting point, trial start and end time, and preset speed are usually set manually.

[0058] The ocean current field within the trial voyage time and space can be obtained by inputting the longitude and latitude range, time range, wind field, temperature and salinity field, topography and coastline data within the trial voyage time and space into the ocean current forecast model. Among them, the longitude and latitude range and time range of the trial voyage time and space are usually set artificially. The wind field, temperature and salinity field, topography and coastline within the trial voyage time and space are usually obtained through actual measurement, model prediction, theoretical calculation and other methods. By changing the input parameters of the model, multiple planned trial voyage routes can be obtained based on the target trial voyage route planning model. For example, set i trial voyage areas, j trial voyage starting points, and k trial voyage start and end times, that is, the planning model has i*j*k input situations, and i*j*k trial voyage routes can be planned.

[0059] According to the multiple (referred to as z) planned trial routes obtained in the above steps, m better planned trial routes are selected from the z planned trial routes according to the planning objectives specified by the user, where z≥m≥1. The m planned trial routes are sorted by ascending or descending order of the target parameters. The above target parameters include the ocean current conditions and energy consumption of the ship on the planned trial route. The ocean current conditions include the ocean current speed, ocean current direction, downstream / upstream duration ratio, and downstream / upstream intensity ratio. The above z value and m value can be determined according to user needs, equipment conditions, etc.

[0060] In one embodiment, Figure 5 As shown in FIG. 1 , the construction process of the above-mentioned ocean current forecast model includes the following steps: using computer programming languages ​​such as MATLAB, Python, and R to construct an initial ocean current forecast model based on a random forest model, a neural network model, a support vector machine model, or a multivariate regression model. Taking the random forest model as an example, Figure 5 As shown in FIG. 1 , the process of constructing the initial ocean current forecast model includes: determining a data set, randomly sampling n samples from the data set, where n ≥ 1, and establishing decision tree models for the n samples to obtain n forecast results. The final forecast result is determined according to a voting mechanism. Among them, random sampling can generally adopt a self-service sampling method, the voting mechanism can take the mode or mean, and the value of n can be determined according to user needs, equipment conditions, etc.

[0061] Further, a second training data set is set to train the initial ocean current prediction model, and the initial ocean current prediction model is obtained when the termination standard is reached; a second test data set is set to test the trained ocean current prediction model, and the target ocean current prediction model is obtained when the test accuracy is met. The aforementioned second training data set and the second test data set are usually composed of actual measurement data, numerical simulation data, and physical test data of ocean currents. The second training data set and the second test data set are independent of each other and have the same format. They both contain second input parameters and second output parameters. The second training data set is used to train the model so that it can learn the relationship between the second input parameters and the second output parameters; the second test data set is used to evaluate the performance of the trained model on an independent data set. Among them, the second input parameter includes at least a latitude and longitude range, a time range, a wind field, a temperature and salinity field, a topography and a coastline, etc. The second output parameter is an ocean current field, and the resolution of the ocean current field is usually set by the user. The accuracy of the ocean current prediction model can be set based on the absolute error, relative error, root mean square error, and determination coefficient between the model prediction value and the actual measurement value, wherein the calculation formulas of the root mean square error and the determination coefficient are shown in formula (1) and formula (2) respectively. Generally speaking, the smaller the error and the higher the determination coefficient, the higher the prediction accuracy of the model.

[0062]

[0063] Where: X rmse represents the root mean square error; y i Represents the output value of the i-th model; y i0 represents the i-th group of measurements; N represents the total number of data groups; R 2 represents the coefficient of determination; y m Indicates the average of the measured values.

[0064] The second training data set is loaded to train the aforementioned initial ocean current forecast model, and the model settings are debugged according to the training results. The model that meets the termination criteria can be used for ocean current forecasting. The second test data set is loaded in the model that can predict ocean currents, and the prediction accuracy of the model is calculated. Furthermore, in the model that can predict ocean currents, the prediction accuracy, calculation time, etc. of the model are compared, and the optimal ocean current forecast model is selected for subsequent ocean current forecasting.

[0065] In one embodiment, in order to further improve the planning efficiency and accuracy of the ship trial route planning based on ocean currents, as Figure 2 As shown, the embodiment of the present application can also train the current trial voyage route planning model based on the planned trial voyage route and the real-time trial voyage route to upgrade the current trial voyage route planning model. In one embodiment, the planned trial voyage route and the real-time trial voyage route can be the planned trial voyage route specified by the user and the real-time trial voyage route specified by the user, respectively, and they are added to the original training data set to form a new training data set, such as Figure 8As shown, the new training data set may include, for example, an existing training data set, and may also include user-specified planned trial voyage route data and user-specified real-time trial voyage route data. The new training data set is used to train the current trial voyage route planning model, and the existing test data set may be used to test the current trial voyage route planning model. Compare the planning accuracy and calculation time of the model, and select p better trial voyage route planning models. Based on the given input parameters and planning objectives, each better trial voyage route planning model can plan q planned trial voyage routes. The user can select the target route according to demand every o days, and select the model with the most target routes, i.e., the optimal planning model, for subsequent trial voyage route planning. The above-mentioned o value, p value, and q value are determined according to user needs, equipment conditions, etc.

[0066] The aforementioned real-time trial voyage route data can be obtained by monitoring the real-time ocean current conditions, energy consumption, speed and trial voyage route. The real-time ocean current conditions include but are not limited to the magnitude of the ocean current speed, the direction of the ocean current, the ratio of downstream / upstream duration, and the ratio of downstream / upstream intensity. The real-time ocean current conditions, energy consumption, speed and trial voyage route can all be determined by shipborne measuring equipment. Shipborne measuring equipment includes but is not limited to current profilers, flow meters, speed meters, and ship navigation systems. The real-time measurement data of the shipborne measuring equipment is read every s seconds, and the real-time ocean current conditions, energy consumption, speed and trial voyage route displayed on the interface are updated every t seconds. The s value and t value are determined according to user needs, equipment conditions, etc.

[0067] The second aspect of the present application provides a ship trial route planning system based on ocean currents, including a route planning model construction module, a trial route planning module and a trial route screening module, wherein the route planning model construction module is used to construct a target trial route planning model based on the trial route regulations and historical trial route data. A neural network planning model is constructed based on a BP (back propagation) neural network model combined with an optimization algorithm using computer programming languages ​​such as MATLAB, Python, and R. The historical trial route data set can be divided into a first training data set and a first test data set, which are independent of each other and have the same format, and both contain a first input parameter and a first output parameter. Among them, the first input parameter includes ship characteristic parameters, trial area, trial starting point, trial start and end time, ocean current field, speed and other parameters, and the first output parameter includes the trial route trajectory, the current conditions of the ship on the trial route, and energy consumption.

[0068] The training termination criteria of the neural network planning model can be determined based on absolute error, relative error, root mean square error, determination coefficient, etc., wherein the calculation formulas of the root mean square error and determination coefficient refer to formula (1) and formula (2). On the programming platform, the neural network planning model is constructed according to the model type, number of model layers, number of nodes in each layer, termination criteria, and training algorithm, and then the first training data set is loaded to train the model. The model that meets the termination criteria can plan the trial route. Furthermore, in the model of the trial route that can be planned, the first test data set is loaded to test the model of the trial route that can be planned, and the planning accuracy and calculation time of the model are compared, and then the optimal trial route planning model, that is, the target trial route planning model, is selected for subsequent trial route planning.

[0069] The trial voyage route planning module is used to obtain ship characteristic parameters and trial voyage related parameters and input them into the target trial voyage route planning model to obtain multiple planned trial voyage routes within the trial voyage time and space; the trial voyage related parameters include the ocean current field within the trial voyage time and space, and the ocean current field within the trial voyage time and space is predicted by the ocean current prediction model. The initial ocean current prediction model is constructed based on a random forest model, a neural network model, a support vector machine model or a multivariate regression model using computer programming languages ​​such as MATLAB, Python, and R.

[0070] A second training data set is set to train the initial ocean current prediction model, and the initial ocean current prediction model is obtained when the termination standard is reached; a second test data set is set to test the trained ocean current prediction model, and the target ocean current prediction model is obtained when the test accuracy is met. The aforementioned second training data set and second test data set are usually composed of actual measurement data, numerical simulation data, and physical test data of ocean currents. The accuracy of the ocean current prediction model can be set based on the absolute error, relative error, root mean square error, and determination coefficient between the model prediction value and the actual measurement value, wherein the calculation formulas of the root mean square error and the determination coefficient are shown in formula (1) and formula (2), respectively. Generally speaking, the smaller the error and the higher the determination coefficient, the higher the prediction accuracy of the model.

[0071] The second training data set is loaded to train the aforementioned initial ocean current forecast model, and the model settings are debugged according to the training results. The model that meets the termination standard can be used for ocean current forecasting. Among the models that can predict ocean currents, the second test data set is loaded to test the models that can predict ocean currents, and the optimal ocean current forecast model is selected according to the test results for subsequent ocean current forecasting.

[0072] The trial voyage route screening module is used to screen multiple planned trial voyage routes according to target parameters to obtain a target planned trial voyage route. The multiple planned trial voyage routes are sorted by ascending or descending order of the target parameters. The aforementioned target parameters include the ocean current conditions and energy consumption of the ship on the planned trial voyage route. The ocean current conditions include the ocean current speed, ocean current direction, downstream / upstream duration ratio, and downstream / upstream intensity ratio.

[0073] In one embodiment, Figure 6 , Figure 7 As shown, the aforementioned trial voyage route planning system also includes a display module, which is used to display the preferred planned trial voyage route obtained by the aforementioned module and its ocean current conditions and energy consumption in the display interface. In addition, the plane of the trial voyage area can also be divided into multiple micro units, and the micro unit can be a polygon such as a square, rectangle, hexagon or a circle, and a single micro unit should be able to accommodate the entire hull. The color of the micro unit is displayed to indicate the impact of the ocean current in the unit on the ship's trial voyage. For example, colorless means that the impact of the ocean current on the ship's trial voyage can be offset during a round trip; green means that during a round trip, the ship has more downstream conditions than upstream conditions; red means that during a round trip, the ship has less downstream conditions than upstream conditions; the darker the green / red, the more significant the downstream / upstream conditions. According to the user's choice, the displayed data of a certain planned trial voyage route enters or does not enter the training pool of the neural network planning model. The user can customize and save the trial voyage route planning results.

[0074] In summary, the technical solution of the present application can accurately obtain the ocean current field data within the trial time and space through the ocean current forecast model, and quickly obtain multiple optimal trial routes and their ocean current conditions and energy consumption through the neural network planning model, so as to calculate the energy reserves, carbon emissions, etc. of the trial ship. The trial route planning model is intelligent, flexible and upgradeable. The user can determine the training pool data of the neural network planning model, and the model can timely learn the user's preference for trial routes and the laws of actual ship trial data, thereby improving the efficiency of ship trial route planning.

[0075] It should be understood that the specific process of each module executing the above corresponding steps has been described in detail in the above method embodiment, and for the sake of brevity, it will not be repeated here.

[0076] The above is only a preferred implementation of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and substitutions can be made without departing from the technical principles of the present application. These improvements and substitutions should also be regarded as the scope of protection of the present application.

Claims

1. A method for planning a ship trial route based on ocean currents, characterized in that: include: Based on the trial route regulations and historical trial route data, a target trial route planning model is constructed; Obtaining ship characteristic parameters and trial-related parameters and inputting them into the target trial-route planning model to obtain a plurality of planned trial-route within the trial-route time and space; the trial-related parameters include the ocean current field within the trial-route time and space, and the ocean current field within the trial-route time and space is predicted by the ocean current prediction model; A plurality of the planned trial voyage routes are screened according to target parameters to obtain a target planned trial voyage route.

2. The method for planning a ship trial route based on ocean currents as claimed in claim 1, characterized in that: The process of constructing the target trial flight route planning model includes: Construct a neural network planning model based on the BP neural network model combined with the optimization algorithm; The neural network planning model is trained using a first training data set, and the neural network planning model is obtained when a termination standard is reached; the trained neural network planning model is tested by loading a first test data set, and the target trial flight route planning model is obtained when the test accuracy is met.

3. The method for planning a ship trial route based on ocean currents as claimed in claim 2, characterized in that: The optimization algorithm includes a genetic algorithm, a particle swarm algorithm, a butterfly algorithm or an ant colony algorithm, which is used to optimize the initial weights and initial thresholds of the BP neural network model.

4. The method for planning a ship trial route based on ocean currents as claimed in claim 2, characterized in that: The BP neural network comprises an input layer, a hidden layer and an output layer, wherein: The input layer is one layer, the number of nodes in the input layer is equal to the number of first input parameters, and the first input parameters include ship characteristic parameters, trial voyage area, trial voyage starting point, trial voyage start and end time, ocean current field, speed, seawater temperature, seawater salinity and air density; The number of layers and nodes of the hidden layer are adjusted according to the performance and training efficiency of the BP neural network model; The output layer is one layer, and the number of nodes in the output layer is equal to the number of first output parameters. The first output parameters include the trial route trajectory, the ocean current conditions and energy consumption of the ship on the trial route.

5. The method for planning a ship trial route based on ocean currents as claimed in claim 2, characterized in that: The first training data set and the first test data set are composed of mutually independent data sets of the historical trial voyage route data; The termination criteria are set according to the absolute error, relative error, root mean square error, and determination coefficient, wherein the calculation formulas of the root mean square error and the determination coefficient are shown in formula (1) and formula (2). Where: X rmse represents the root mean square error; y i Represents the output value of the model of group i; y i0 represents the i-th group of measurements; N represents the total number of data groups; R 2 represents the coefficient of determination; y m Indicates the average of the measured values.

6. The method for planning a ship trial route based on ocean currents as claimed in claim 1, characterized in that: The process of building an ocean current forecast model includes: Use MATLAB, Python or R programming language to construct an initial ocean current forecast model based on random forest model, neural network model, support vector machine model or multivariate regression model; The initial ocean current forecast model is trained using the second training data set, and the initial ocean current forecast model is obtained when the termination standard is reached. The second test data set is loaded to test the trained initial ocean current forecast model, and the target ocean current forecast model is obtained when the test accuracy is met.

7. The method for planning a ship trial route based on ocean currents as claimed in claim 6, characterized in that: The second training data set and the second test data set are from actual measurement data, numerical simulation data and physical test data of ocean currents; The second training data set and the second test data set include a second input parameter and a second output parameter, wherein the second input parameter includes a latitude and longitude range, a time range, a wind field, a temperature and salinity field, and a topographic coastline; The second output parameter is an ocean current field; The training termination criteria of the ocean current forecast model are set based on the absolute error, relative error, root mean square error, and determination coefficient between the model forecast value and the actual measurement value.

8. The method for planning a ship trial route based on ocean currents as claimed in claim 6, characterized in that: The process of screening the plurality of planned trial voyage routes according to the target parameters to obtain the target planned trial voyage route includes: The target parameters are the ocean current conditions and energy consumption of the ship on the planned trial voyage route, and the ocean current conditions include the ocean current speed, ocean current direction, downstream / upstream duration ratio, and downstream / upstream intensity ratio.

9. A ship trial route planning system based on ocean currents, characterized in that: It includes a route planning model building module, a trial route planning module and a trial route screening module, among which: The route planning model building module is used to build a target trial voyage route planning model based on the trial voyage route regulations and historical trial voyage route data; The trial voyage route planning module is used to obtain ship characteristic parameters and trial voyage related parameters and input them into the target trial voyage route planning model to obtain multiple planned trial voyage routes within the trial voyage time and space; the trial voyage related parameters include the ocean current field within the trial voyage time and space, and the ocean current field within the trial voyage time and space is predicted by the ocean current prediction model; The trial voyage route screening module is used to screen a plurality of the planned trial voyage routes according to target parameters to obtain a target planned trial voyage route.

10. A ship trial route planning system based on ocean currents, characterized in that: The target planned trial voyage route obtained by screening is the current optimal planned trial voyage route. The current optimal planned trial voyage route is added to the training data set of the current trial voyage route planning model for training to obtain a better target trial voyage route planning model.

Citation Information

Cited By

  • Ship course prediction system and ship course prediction method

    CN119863950A

  • Cold and hot shock resistance test system for marine engine

    CN120948062A