A method for predicting the motion attitude of a ship

Abstract

The present application relates to the technical field of ship motion posture, and particularly relates to a ship motion posture prediction method. The method comprises the following steps: S1: normalizing and classifying historical detailed data of a ship motion posture; S2: determining variable parameters and adaptive functions of ship motion posture data by solving posture data of the ship motion posture; S3: establishing a three-dimensional coordinate system to simulate the running of the ship motion posture, so that simulation training data under the historical motion posture data parameters can be obtained; S4: changing the formula content of the ship motion posture model; S5: generating a ship historical motion posture data model under the classification category; and S6: putting the trained detection data into the ship historical motion posture data model under the classification category to predict the ship motion posture. The present application provides a ship motion posture prediction method which can classify and predict parameters of influencing factors and realize the prediction of multi-condition data.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of ship motion posture, in particular to a ship motion posture prediction method. BACKGROUND

[0002] The ship is a kind of man-made traffic tool mainly running in geographical water, in addition, the civilian ship is generally called ship, the military ship is called warship, and the small ship is called boat or boat, and the general term is warship or boat, the ship mainly includes accommodation space, support structure and drainage structure, has a propulsion system using external or self-contained energy, and the shape is generally a streamlined envelope to overcome fluid resistance, and the material is constantly updated with the progress of science and technology, and the early natural materials are wood, bamboo and hemp, and in modern times, steel and aluminum, glass fiber, acrylic and various composite materials are used.

[0003] The ship motion posture is different due to numerous influencing parameters during the motion process, and the ship is divided into the following according to the characteristics of the motion state: roll, that is, the ship shakes left and right, generally roll phenomenon; sway, that is, the object immersed in water oscillates horizontally along the vertical direction of the longest extension direction; pitch, that is, the ship body rotates around the horizontal axis; surge, that is, the object immersed in water oscillates horizontally along the longest extension direction; yaw, that is, the object immersed in water rotates around the vertical axis of the ship body; heave, that is, the object immersed in water oscillates along the vertical direction, in order to predict the ship motion posture, a certain prediction method is needed, and the existing methods mainly include Kalman filter prediction, autoregressive prediction, neural network prediction and the like, the above prediction methods do not realize classified management of model data in the use process, so that the predicted data lacks classification, which will cause large data error and unable to predict data under different conditions. SUMMARY

[0004] The technical problem to be solved by the present application is to provide a ship motion posture prediction method capable of classifying parameters of influencing factors to realize prediction of multi-condition data.

[0005] The technical scheme adopted by the present application is as follows: a ship motion posture prediction method, comprising the following steps:

[0006] S1: collecting the historical detailed data of the ship motion posture, removing the abnormal data of the ship historical motion posture data, normalizing and classifying the data without abnormality, so as to obtain the classified historical modeling data;

[0007] S2: The classified historical modeling data is calculated by using the EMD-PSO-LSTM combined model, and the ship motion posture data is solved to determine the variable parameters and adaptive functions of the ship motion posture data;

[0008] S3: A three-dimensional coordinate system of the ship is established, the load parameters are set, and the historical motion posture data of the similar ship is input to simulate the motion posture of the ship, so that the simulation training data under the historical motion posture data parameters can be obtained;

[0009] S4: The parameter proportion change of the ship motion posture model is back calculated according to the simulation training data, so that the formula content of the ship motion posture model can be changed;

[0010] S5: The ship historical motion posture data under the classification category is repeatedly subjected to the S3 step, until the training result value of the test simulation parameters can reach the predetermined error range, at this time, the ship historical motion posture data model under the classification category can be generated;

[0011] S6: The trained detection data is input into the ship historical motion posture data model under the classification category to predict the ship motion posture.

[0012] As a further scheme of the application: the classification categories in S1 are motion date, motion distance, sea area, ship type and navigation state, and the historical modeling data under the motion date, motion distance, sea area, ship type and navigation state categories are respectively stored separately.

[0013] As a further scheme of the application: the LSTM in S2 is a long short-term memory artificial neural network, and the gradient descent method is used to modify the weight of each time according to the error in view of the nonlinear and strong random characteristics of the ship motion posture historical modeling data.

[0014] As a further scheme of the application: the load parameters are dynamic coordinates, and the load parameters and the three-dimensional coordinate system of the ship are combined in value and volume, so that specific dynamic parameters can be generated as prediction parameters.

[0015] As a further scheme of the application: the S3 step is repeated in S5 to meet the time sequence back propagation algorithm of the EMD-PSO-LSTM combined model, so that the weight of each time can be modified according to the error, the data error can be calculated back, and the training error can be minimized.

[0016] As a further scheme of the present application: the predicted detection data of the ship motion posture under the categories of the motion date, the motion distance, the sea area passed, the ship type, the navigation state, the motion date, the motion distance, the sea area passed, the ship type, and the navigation state can obtain the predicted ship motion posture data.

[0017] Compared with the prior art, the present application has the following beneficial effects:

[0018] The present application is fully functional, and through the motion date, the motion distance, the sea area passed, the ship type, the navigation state, the classification categories of the motion date, the motion distance, the sea area passed, the ship type, and the navigation state, and through the motion date, the motion distance, the sea area passed, the ship type, and the navigation state, the predicted detection data of the ship motion posture under the categories of the motion date, the motion distance, the sea area passed, the ship type, and the navigation state can obtain the classification parameter prediction, thereby realizing the ship motion posture data of multi-condition data.

[0019] Overall, the present application can classify the influencing factors and predict the parameters, thereby realizing the prediction of multi-condition data. DETAILED DESCRIPTION

[0020] The present application will be further described below.

[0021] In order to realize the classification parameter prediction of influencing factors and multi-condition data for a ship motion posture prediction method, the present application provides a ship motion posture prediction method, which is characterized by comprising the following steps:

[0022] S1: collecting the historical detailed data of the ship motion posture, removing the abnormal data of the ship historical motion posture data, normalizing and classifying the data without abnormalities, thereby obtaining the classified historical modeling data;

[0023] S2: inputting the classified historical modeling data into the EMD-PSO-LSTM combined model for calculation, and determining the variable parameters and adaptive function of the ship motion posture data by solving the posture data of the ship motion posture;

[0024] S3: establishing a three-dimensional coordinate system for the ship, setting the load parameter, and simulating the ship motion posture by inputting the historical motion posture data of similar ships, thereby obtaining the simulation training data under the historical motion posture data parameters;

[0025] S4: inversely calculating the parameter proportion change of the ship motion posture model according to the simulation training data, thereby changing the formula content of the ship motion posture model;

[0026] S5: the ship historical motion posture data under the classification category is repeatedly performed S3 step, until the training result value of the test simulation parameter can reach the predetermined error range, at this time, the ship historical motion posture data model under the classification category can be generated;

[0027] S6: the trained detection data is put into the ship historical motion posture data model under the classification category to which the detection data belongs to, and the ship motion posture is predicted.

[0028] In the S1 of the application, the classification categories are motion date, motion distance, sea area, ship type and navigation state, and the historical modeling data under the motion date, motion distance, sea area, ship type and navigation state categories are stored separately.

[0029] In the S2 of the application, the LSTM is a long short-term memory artificial neural network, and the gradient descent method is used to modify the weight of each time according to the error, in view of the nonlinear and strong random characteristics of the ship motion posture historical modeling data.

[0030] In the application, the load parameter is a dynamic coordinate, and the load parameter and the three-dimensional coordinate system of the ship are combined in value and volume, so that the specific dynamic parameter can be generated as the prediction parameter.

[0031] In the S5 of the application, the S3 step is repeated to meet the EMD-PSO-LSTM combined model application time sequence back propagation algorithm, so as to modify the weight of each time according to the error, so that the data error can be calculated back, and the training error is minimized.

[0032] In the S6 of the application, the motion date, motion distance, sea area, ship type and navigation state are used to predict the detection data of the ship motion posture under the motion date, motion distance, sea area, ship type and navigation state categories, and the predicted ship motion posture data can be obtained.

[0033] EMD is a new method for processing non-stationary signals - an important part of Hilbert-Huang transform. The time-frequency analysis method based on EMD is suitable for the analysis of nonlinear, non-stationary signals, and is also suitable for the analysis of linear, stationary signals, and for the analysis of linear, stationary signals, it also reflects the physical meaning of the signal better than other time-frequency analysis methods, PSO is a particle swarm optimization algorithm, which is a population-based stochastic optimization technique, particle swarm optimization algorithm simulates the group behavior of insects, herds, flocks and schools, these groups search for food in a cooperative way, each member of the group changes its search pattern by learning its own experience and the experience of other members, LSTM is a kind of time recurrent neural network, which is specially designed to solve the long-term dependence problem of general RNN (recurrent neural network), all RNNs have a chain form of repeated neural network modules, using EMD-PSO-LSTM combined model can combine the prediction detection data of ship motion posture under the classification of motion date, motion distance, sea area, ship type and navigation state.

[0034] Through the classification of motion date, motion distance, sea area, ship type and navigation state, and through the prediction detection data of ship motion posture under the classification of motion date, motion distance, sea area, ship type and navigation state, the classification parameter prediction can be obtained to realize the ship motion posture data of multi-condition data

[0035] The above describes the present application and its embodiments, which is not limited. In general, if a person skilled in the art is inspired by it, without departing from the purpose of the present application, without creative design, similar structure and embodiments of the technical scheme can be designed, which should belong to the protection scope of the present application.

Claims

1. A method of predicting a motion attitude of a marine vessel, characterized by: Comprise the following steps: S1: collect the historical detailed data of the ship's motion posture, remove the abnormal data of the ship's historical motion posture data, normalize and classify the data without abnormalities, and thus obtain the classified historical modeling data; S2: the classified historical modeling data is calculated by using the EMD-PSO-LSTM combined model, and the posture data of the ship motion posture is solved to determine the variable parameters and adaptive function of the ship motion posture data; S3: establish a three-dimensional coordinate system for the ship, set the load parameters, and simulate the ship motion posture by inputting the historical motion posture data of similar ships, thereby obtaining the simulation training data under the historical motion posture data parameters; S4: According to the simulation training data, the parameter proportion change of the ship motion posture model is back calculated, so that the formula content of the ship motion posture model can be changed; S5: repeat the S3 step for the ship historical motion posture data under the same classification category several times, until the test simulation parameter training result value can reach the specified error range, at this time the ship historical motion posture data model under the classification category can be generated; S6: The trained detection data is input into the ship historical motion posture data model under the classification category to predict the ship motion posture.

2. A method of predicting the motion attitude of a marine vessel according to claim 1, characterized in that: The classification categories in S1 are motion date, motion distance, sea area, ship type, and navigation state, and the historical modeling data under the motion date, motion distance, sea area, ship type, and navigation state categories are stored separately.

3. A method of predicting the motion attitude of a ship according to claim 1, characterized in that: In S2, LSTM is a long short-term memory artificial neural network. For the characteristics of nonlinearity and strong randomness of ship motion posture historical modeling data, gradient descent method is used to modify the weight of each time according to the error.

4. A method of predicting the motion attitude of a ship according to claim 2, characterized in that: The load parameters are dynamic coordinates. By combining the load parameters with the three-dimensional coordinate system of the ship in terms of numerical value and volume, specific dynamic parameters can be generated as prediction parameters.

5. A method of predicting the motion attitude of a ship according to claim 1, characterized in that: In S5, the S3 step is repeated to satisfy the EMD-PSO-LSTM combined model application time sequence back propagation algorithm to modify the weight of each time according to the error, so that the data error can be calculated back, and the training error is minimized.

6. A method of predicting the motion attitude of a marine vessel according to claim 1, characterized in that: In S6, the prediction detection data of the ship motion posture under the motion date, motion distance, sea area, ship type, and navigation state categories can be obtained by the motion date, motion distance, sea area, ship type, and navigation state.