A method for predicting ship motion response based on Gaussian interpolation

Abstract

The present invention discloses a ship motion response prediction method based on Gaussian interpolation, which relates to the technical field of ship motion response prediction, and includes the following steps: S1, data acquisition and preprocessing, including data cleaning, coordinate system conversion, data standardization and data set division; S2, constructing a Gaussian interpolation model, including setting an initial σ value and selecting interpolation points; S3, optimizing model parameters by a random search algorithm, including calculating the weight of each point in a training set, model verification, and model parameter optimization; S4, ship motion response prediction, substituting the ship to be predicted and sea state data into the Gaussian interpolation model to predict the ship motion response. The present invention uses a Gaussian interpolation model to accurately simulate the motion response of a ship under complex sea conditions. Compared with traditional methods, the Gaussian interpolation model can more accurately depict the complex relationship between sea state parameters and the ship motion state with its excellent fitting ability for multidimensional and nonlinear data.

Description

Technical Field

[0001] The invention relates to the technical field of ship motion response prediction, and in particular to a ship motion response prediction method based on Gaussian interpolation. Background Art

[0002] As a core topic of multidisciplinary cross-study, the prediction of ship motion response covers the fields of marine engineering, navigation science, ship design and manufacturing, and navigation safety and optimization. Against the backdrop of global trade prosperity and increasingly active marine resource development activities, the safety, economy and environmental protection of various types of ships operating in complex marine environments have become the focus of the industry. Accurately predicting the motion response of ships under various sea conditions, such as ship mooring force, lateral movement, roll, pitch, heave and other dynamic characteristics, is not only the key to ensuring navigation safety, improving the rationality of route planning, optimizing energy efficiency, and reducing environmental pressure, but also has important guiding significance for the design and upgrade of ship control systems. With the development of marine engineering and shipping industry, higher requirements are placed on the accurate prediction of the motion characteristics of ships in complex marine environments.

[0003] Traditionally, the prediction of ship motion response mainly relies on direct numerical simulation of physical models. This type of method is based on Newtonian mechanics and combines the principles of fluid dynamics to describe the interaction between the ship and the surrounding medium by establishing a ship dynamics model and a marine environment model. Typical ship dynamics models include six-degree-of-freedom motion models, three-dimensional hydrodynamic models, ship maneuvering models, etc., while the marine environment model covers the generation, propagation and interaction mechanisms of multiple factors such as waves, wind, currents, tides, salinity, and temperature. By solving these model equations through numerical integration or finite element methods, the motion state of the ship under specific sea conditions can be predicted. However, although the traditional direct numerical simulation method based on physical models has high accuracy in theory, it has the problems of high computational cost, low efficiency, and poor adaptability to changes in environmental parameters. It is often difficult to fully consider the comprehensive effects of all relevant sea state parameters and cannot effectively cope with the high complexity and uncertainty of the marine environment. Especially under the coupling effect of wind, waves and currents, the prediction ability is limited. Therefore, it is urgent to develop a new method that can efficiently and accurately simulate the motion response of ships under complex sea conditions to improve the prediction accuracy and adaptability to practical applications.

[0004] The closest existing technology is a direct numerical simulation method of ship motion response based on a physical model. This method establishes detailed formulas to describe the dynamic behavior of the ship and the dynamics of the ocean environment, and then uses numerical integration and other methods to solve the model equations to predict the motion response of the ship under given sea conditions. Although this method has high accuracy in theory, it requires processing complex nonlinear partial differential equations, resulting in large amounts of calculation and time consumption. In addition, when sea condition parameters change or involve non-uniform, non-stationary wave fields and wind-wave-current coupling effects, the adjustment and calibration process of the model is cumbersome and lacks adaptability to changes in environmental parameters.

[0005] In the prior art, the prediction of ship motion response mainly relies on direct numerical simulation of physical models, but such methods have high computational cost, low efficiency, poor real-time performance, and poor adaptability to changes in environmental parameters. In addition, the marine environment is highly complex and uncertain, and multiple factors such as waves, wind, and currents are intertwined and jointly affect the motion state of the ship. Traditional prediction methods often find it difficult to fully consider the comprehensive effects of all relevant sea state parameters, especially when dealing with non-uniform, non-stationary wave fields and wind-wave-current coupling effects, the prediction ability is limited. This application introduces a Gaussian interpolation model, which can flexibly handle multi-dimensional, nonlinear sea state parameter space, accurately characterize the impact of various complex sea conditions on ship motion, thereby enhancing the adaptability of the prediction method in practical application scenarios. The performance of the Gaussian interpolation model is highly dependent on the selection of its kernel function parameters, which determine the model's degree of fit to the data, smoothness, and the generalization ability and accuracy of the prediction. However, manual adjustment or empirical selection of these parameters is often time-consuming and difficult to achieve the best results, which may lead to overfitting or underfitting of the model. This application uses a random search algorithm to automatically optimize the parameters of the Gaussian kernel function. By efficiently exploring the parameter space, it finds the parameter combination that can maximize the model's prediction performance, overcomes the limitations of manual parameter tuning, and ensures the model's prediction accuracy and stability under complex sea conditions. Summary of the invention

[0006] The purpose of the present invention is to solve the above technical problems and to propose a ship motion response prediction method based on Gaussian interpolation.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] A ship motion response prediction method based on Gaussian interpolation comprises the following steps:

[0009] S1, data acquisition and preprocessing, including data cleaning, coordinate system conversion, data standardization, and data set division;

[0010] S2, build a Gaussian interpolation model, including setting the initial value, select interpolation points;

[0011] S3, random search algorithm optimizes model parameters, including calculating the weight of each point in the training set, model verification, and model parameter optimization;

[0012] S4, ship motion response prediction, substitutes the ship and sea condition data to be predicted into the Gaussian interpolation model to predict the ship motion response.

[0013] Preferably, the data cleaning includes: detecting and filling missing values, identifying and processing outliers, and eliminating duplicate values ​​for the input ship and sea condition data.

[0014] Preferably, the coordinate system conversion: according to the correspondence between the coordinate system used by the model and the coordinate system adopted by each ship and sea condition parameter, the wind direction, wave direction and flow direction data are converted accordingly.

[0015] Preferably, the data standardization: performing Z-score standardization processing on all ship and sea condition parameters, that is, converting the original data into a standard normal distribution form with a mean of 0 and a standard deviation of 1.

[0016] Preferably, the data set is divided into: an existing ship motion response data set is divided into a training set and a test set, the training set is used to construct a Gaussian interpolation model, and the test set is used to verify the prediction performance of the model.

[0017] Preferably, the setting of the initial σ value is as follows: according to experience or simple trial calculation, the initial bandwidth parameter of the Gaussian kernel function is set: .

[0018] Preferably, the interpolation point is selected by selecting a ship and sea condition vector to be interpolated from the test set. , which contains nine ship and sea state parameters: ship type, load, wave height h4, spectrum peak period, wave direction, wind speed, wind direction, current direction and current speed.

[0019] Preferably, calculating the weight of each point in the training set includes calculating the Euclidean distance and calculating the weight.

[0020] Preferably, the model verification includes interpolation calculation and model verification.

[0021] Preferably, the model parameter optimization includes defining a search range and a sampling number and sampling.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] (1) Improving prediction accuracy: By adopting the Gaussian interpolation model, it is possible to flexibly handle the multi-dimensional and nonlinear sea condition parameter space and accurately characterize the impact of various complex sea conditions on ship motion, thereby significantly improving the prediction accuracy.

[0024] (2) Enhanced adaptability: The Gaussian interpolation model can effectively cope with the complexity and uncertainty of the marine environment, especially when dealing with non-uniform and non-stationary wave fields and wind-wave-current coupling effects, showing a predictive ability that is superior to traditional methods.

[0025] (3) Improve efficiency: The random search algorithm is used to automatically optimize the kernel function parameters of the Gaussian interpolation model, replacing the time-consuming manual parameter tuning process and greatly improving the efficiency of model construction and optimization.

[0026] (4) Ensuring stability: The automated parameter optimization method can avoid overfitting or underfitting of the model and ensure the prediction accuracy and stability of the model under complex sea conditions.

[0027] The present invention adopts the Gaussian interpolation model to accurately simulate the motion response of the ship under complex sea conditions. Compared with the traditional method, the Gaussian interpolation model can more accurately characterize the complex relationship between sea condition parameters and the ship's motion state with its excellent fitting ability for multi-dimensional and nonlinear data. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 The present invention provides a flow chart of a method for predicting ship motion response based on Gaussian interpolation. DETAILED DESCRIPTION

[0029] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments;

[0030] Gaussian interpolation method: an interpolation method based on Gaussian kernel function. The core is to interpolate neighboring points by weight. For a certain interpolation point, the farther the neighboring point is from it, the smaller its weight is, and vice versa. It is suitable for processing multidimensional and nonlinear data space. By assigning distance-based weights to data points for weighted averaging, the value of unknown points can be estimated.

[0031] Sea condition parameters: refers to the marine environmental factors that affect the movement of ships, such as wave height, peak period, wave direction, wave speed, wind speed, wind direction, wave direction, etc.

[0032] Z-score standardization: A data preprocessing method that converts the original data into a standard normal distribution with a mean of 0 and a standard deviation of 1 by subtracting its mean and dividing it by the standard deviation, so as to facilitate comparison and analysis between data of different dimensions and numerical ranges.

[0033] Euclidean distance: A measure of the difference between two multidimensional vectors, calculated as ,in and They are two vectors in Dimensional weight.

[0034] Gaussian kernel function: A kernel function with Gaussian distribution as the weight distribution function, usually expressed as ,in is the bandwidth parameter, which determines the smoothness of the function; is the mean parameter, which is usually set to 0. is the distance variable.

[0035] Random search algorithm: A global optimization method that randomly samples and evaluates the objective function value in the parameter space to find the parameter combination that makes the objective function optimal. It does not rely on gradient information and is suitable for optimization problems that are non-convex, non-continuous, or have a large number of local extreme values.

[0036] Mean square error (MSE): An indicator used to evaluate the performance of a prediction model. The calculation formula is: ,in For the The true response value of samples, is the corresponding predicted value, is the sample size.

[0037] Reference Figure 1 , a ship motion response prediction method based on Gaussian interpolation, comprising the following steps:

[0038] S1, data acquisition and preprocessing, including data cleaning, coordinate system conversion, data standardization, and data set division;

[0039] S2, build a Gaussian interpolation model, including setting the initial value, select interpolation points;

[0040] S3, random search algorithm optimizes model parameters, including calculating the weight of each point in the training set, model verification, and model parameter optimization;

[0041] S4, ship motion response prediction, substitutes the ship and sea condition data to be predicted into the Gaussian interpolation model to predict the ship motion response.

[0042] In the present invention, the data cleaning includes: detecting and filling missing values, identifying and processing abnormal values, and eliminating duplicate values ​​of the input ship and sea condition data.

[0043] In the present invention, the coordinate system conversion is as follows: based on the correspondence between the coordinate system used by the model and the coordinate system adopted by each ship and sea condition parameter, the wind direction, wave direction and flow direction data are converted accordingly.

[0044] In the present invention, the data standardization is as follows: since the dimensions and numerical ranges of the ship and sea condition parameters vary greatly, in order to avoid the unbalanced influence between the parameters, all ship and sea condition parameters are subjected to Z-score standardization, that is, the original data is converted into a standard normal distribution form with a mean of 0 and a standard deviation of 1.

[0045] In the present invention, the data set is divided into: the existing ship motion response data set is divided into a training set and a test set, the training set is used to build a Gaussian interpolation model, and the test set is used to verify the prediction performance of the model.

[0046] In the present invention, the initial setting Value: Set the initial bandwidth parameter of the Gaussian kernel function based on experience or simple trial calculations .

[0047] In the present invention, the interpolation point is selected: a ship and sea condition vector to be interpolated is selected from the test set , which contains nine ship and sea state parameters: ship type, load, wave height h4, spectrum peak period, wave direction, wind speed, wind direction, current direction and current speed.

[0048] In the present invention, the calculation of the weight of each point in the training set includes calculating the Euclidean distance and calculating the weight; wherein the calculation of the Euclidean distance is: for each ship and sea condition vector in the training set (including nine independent variables), calculate its interpolation point Euclidean distance ;

[0049] Calculate weights: The calculated Euclidean distance Substitute into the Gaussian function The weights corresponding to each point in the training set are obtained ;

[0050] In the present invention, the model verification includes interpolation calculation and model verification;

[0051] Interpolation calculation: According to the interpolation formula , the weight of each point in the training set The corresponding ship motion response value Multiply and sum to get the interpolation point The predicted response value of .

[0052] Model validation: Repeat the above steps on the test set, make predictions for multiple interpolation points, and calculate the mean square error (MSE) between the predicted results and the true response values ​​to evaluate the prediction performance of the model.

[0053] In the present invention, the model parameter optimization includes defining the search range and the number of samplings and sampling;

[0054] Define the search range and sampling times: Set a reasonable search interval to ensure that it covers the possible prediction results. The value range is set, and the maximum sampling number N is set as the number of search iterations.

[0055] Sampling: For each iteration, a random number is generated within the search range using normal distribution sampling. value, use this The values ​​are interpolated by Gaussian to predict the ship motion response and calculate the mean square error (MSE). The value is recorded as the optimal , using the optimal Update the Gaussian interpolation model with the given value.

[0056] The core technology of the present invention is to use the Gaussian interpolation model to accurately simulate the motion response of the ship under complex sea conditions. Compared with the traditional method, the Gaussian interpolation model can more accurately depict the complex relationship between sea state parameters and the ship's motion state with its excellent fitting ability for multi-dimensional and nonlinear data. This key technology is reflected in the construction of the model, parameter selection, training process and practical application under different sea conditions, ensuring the high accuracy and wide applicability of the prediction. In addition, the introduction of a random search algorithm realizes the automated and intelligent optimization of the core parameters of the Gaussian interpolation model. Unlike traditional algorithms that rely on manual experience to set parameters or adopt a fixed parameter strategy, this algorithm effectively optimizes the kernel function parameter configuration of the model through a systematic random exploration method. This algorithm not only significantly improves the efficiency of parameter selection, but also ensures the scientific nature of parameter setting, which helps to improve the overall prediction performance.

[0057] The specific implementation method of the present invention using the Gaussian interpolation model to predict the ship motion response includes but is not limited to the model structure, kernel function form, multi-dimensional representation of ship and sea state parameters, and a method for determining model parameters.

[0058] The present invention provides a random search optimization method for Gaussian interpolation model parameters, including but not limited to search range setting, search step definition, initial seed point selection, iterative strategy design, and use of performance evaluation criteria.

[0059] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A ship motion response prediction method based on Gaussian interpolation, characterized in that: The following steps are involved: S1, data acquisition and preprocessing, including data cleaning, coordinate system conversion, data standardization, and data set division; S2, constructing a Gaussian interpolation model, including setting an initial bandwidth value and selecting interpolation points; wherein the selecting interpolation points specifically includes: selecting a ship and sea condition vector to be interpolated from a test set, the vector including nine ship and sea condition parameters: ship type, load, wave height, spectrum peak period, wave direction, wind speed, wind direction, flow direction and flow speed; S3, random search algorithm optimizes model parameters, including calculating the weight of each point in the training set, model verification, and model parameter optimization; wherein, for each ship and sea condition vector in the training set, calculate its Euclidean distance with the interpolation point; substitute the calculated Euclidean distance into the Gaussian function to obtain the weight corresponding to each point in the training set; multiply the weight of each point in the training set by its corresponding ship motion response value and sum them to obtain the predicted response value of the interpolation point; repeat the above steps on the test set, predict multiple interpolation points, and calculate the mean square error between the predicted result and the true response value to evaluate the prediction performance of the model; set the search interval and the maximum number of sampling times as the number of search iterations; for each iteration, use normal distribution sampling to randomly generate a bandwidth value within the search range, use the bandwidth value for Gaussian interpolation, predict the ship motion response, and calculate the mean square error, record the bandwidth value corresponding to the minimum mean square error value as the optimal bandwidth value, and use the optimal bandwidth value to update the Gaussian interpolation model; S4, ship motion response prediction, substitutes the ship and sea condition data to be predicted into the Gaussian interpolation model to predict the ship motion response.

2. A ship motion response prediction method based on Gaussian interpolation according to claim 1, characterized in that: The data cleaning includes: detecting and filling missing values, identifying and processing abnormal values, and eliminating duplicate values ​​for the input ship and sea condition data.

3. A ship motion response prediction method based on Gaussian interpolation according to claim 1, characterized in that: The data standardization: Z-score standardization is performed on all ship and sea condition parameters, that is, the original data is converted into a standard normal distribution form with a mean of 0 and a standard deviation of 1.

Images