Ship body local vibration prediction method based on SA-XGBOOST algorithm

Abstract

The invention discloses a ship body local vibration prediction method based on an SA-XGBOOST algorithm. The ship body local vibration prediction method comprises the following steps of obtaining a plurality of groups of historical vibration data of local vibration measurement points when a ship sails; performing dimension reduction processing on the historical vibration data to obtain dimension-reduced vibration data; the method comprises the following steps of: training and testing a ship body local vibration prediction model by utilizing dimension reduction vibration data based on an XGBOOST algorithm, optimizing XGBOOST parameters by utilizing a simulated annealing algorithm in an iterative calculation process of training and testing to obtain an optimal XGBOOST parameter combination, and substituting the optimal XGBOOST parameter combination into the model to carry out a next round of iterative calculation, so that a ship body local vibration prediction model is obtained. The accuracy of the ship body local vibration prediction model meets the requirement. And predicting local vibration data of the to-be-measured position by using the ship body local vibration prediction model. According to the method, the precision of the ship body local vibration prediction model is improved, the model is prevented from falling into local optimum, and the robustness of the algorithm is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of ship vibration monitoring, and in particular to a method for predicting local vibration of a hull based on a SA-XGBOOST algorithm. Background Art

[0002] In recent years, with the development of shipping and shipbuilding industry, related basic research has been continuously improved, and the requirements of shipowners, inspection agencies and related regulations and conventions on ship vibration performance have been continuously improved. However, since many ships do not consider vibration analysis during design, or the results of analysis and prediction are inaccurate, harmful vibrations often occur in actual operation of ships.

[0003] With the advancement of shipbuilding technology, the hull structure is lightened, and the structural stiffness is also reduced, which makes it easier to arouse larger hull vibration. Excessive hull vibration can cause damage to the hull structure, such as fatigue damage at stress concentration points, which will affect various instruments, equipment and mechanical devices on the ship, such as radar, sonar, periscope and various navigation. Therefore, during the sea trial, it is necessary to detect local vibrations in typical parts of the whole ship. And check with the safety value. However, for ultra-large liquefied natural gas ships, ultra-large container ships and other ultra-large ships, the manpower and time costs of testing are high, and for some areas that testers cannot reach, as well as confined spaces with greater dangers. Only relatively close test points can be selected. This undoubtedly affects the test accuracy. However, when harmful vibrations occur in unmeasured areas, the cost of remedial improvement is often very high, and sometimes it is necessary to adjust the structural form to change the natural frequency of the hull. Subsequent modifications not only greatly increase the construction cost, but also greatly reduce the ship's ship performance and competitiveness in the ship market. Therefore, the local vibration of the hull is effectively predicted, and the hull vibration prediction data is used as a result-oriented method to conduct safety verification of the existing hull, and optimize the design parameters for subsequent series of ships, so as to achieve the purpose of controlling local vibration within a safe range, reducing the negative impact of local vibration, and improving the safety and comfort of ship operation.

[0004] In the prior art, actual on-site measurements are often used. On-site measurements require high costs under specific conditions, and are particularly dangerous when measuring in certain confined spaces. Summary of the invention

[0005] In view of this, the present invention provides a method for predicting local vibration of a hull based on the SA-XGBOOST algorithm to solve the problems existing in the above-mentioned background technology.

[0006] A method for predicting local vibration of a hull based on the SA-XGBOOST algorithm specifically comprises the following steps:

[0007] S1, obtaining several groups of historical vibration data of local vibration measurement points when the ship is sailing and preprocessing the data;

[0008] Each set of historical vibration data includes multiple vibration characteristic parameters;

[0009] S2, performing dimension reduction processing on the historical vibration data of the local vibration measurement points when the ship is sailing to obtain dimension-reduced vibration data;

[0010] S3, divide the dimension-reduced vibration data into a training set and a test set, build a hull local vibration prediction model based on the XGBOOST algorithm, and use the training set and the test set to train and test the hull local vibration prediction model.

[0011] In the iterative calculation process of training and testing the hull local vibration prediction model, the simulated annealing algorithm is used to optimize the XGBOOST parameters to obtain the optimal XGBOOST parameter combination, and the optimal XGBOOST parameter combination is substituted into the model for the next round of iterative calculation until a hull local vibration prediction model with an accuracy rate that meets the requirements is obtained;

[0012] S4, inputting the three-dimensional position of the position to be measured from the main engine, the current speed, the wave height, the main engine rotation speed and the plate thickness of the position to be measured into the hull local vibration prediction model to obtain the local vibration data of the position.

[0013] Preferably, the vibration characteristic parameters in step S1 include the distance X from the local vibration measurement point to the main engine in the direction of the ship length. e , the distance Y between the local vibration measurement point and the main engine in the ship width direction e , the distance Z between the local vibration measurement point and the main engine in the ship height direction e , the distance from the propeller in the direction of the ship's length X p , the distance Y between the local vibration measurement point and the propeller in the ship width direction p , the distance Z between the local vibration measurement point and the propeller in the ship height direction p , the ship speed V when measuring at the local vibration measurement point, the wave height H when measuring at the local vibration measurement point, and the wind speed W when measuring at the local vibration measurement point s , the thickness of the plate at the local vibration measurement point B, the filling value of the vibration reduction equipment when measuring the local vibration measurement point H m , liquefaction level R of the reliquefaction equipment when measuring at the local vibration measurement point l , the rotation speed R of the ship's main engine.

[0014] Preferably, when the vibration reduction device is turned on, the filling value H of the vibration reduction device when measuring the local vibration measurement point is m =1; when the vibration reduction device is not turned on, the vibration reduction device fill value H when measuring the local vibration measurement pointm =0.

[0015] Preferably, the specific steps of performing dimensionality reduction processing on the historical vibration data of the local vibration measurement points when the ship is sailing in step S2 are:

[0016] According to several groups of historical vibration data of local vibration measurement points, the correlation coefficients between various vibration characteristics are calculated respectively;

[0017] According to the calculated correlation coefficients between the vibration features, vibration features with correlation coefficients greater than 0 are selected and the vibration data corresponding to these vibration features are used as dimension-reduced vibration data.

[0018] Preferably, the dimension-reduced vibration data includes the distance X of the local vibration measurement point from the main engine in the ship length direction. e , the distance Y between the local vibration measurement point and the main engine in the ship width direction e , the distance Z between the local vibration measurement point and the main engine in the ship height direction e , the ship speed V when measuring at the local vibration measurement point, the wave height H when measuring at the local vibration measurement point, the plate thickness B where the local vibration measurement point is located, and the main engine speed R.

[0019] Preferably, the calculation formula of the correlation coefficient between the vibration characteristics is:

[0020]

[0021] Among them, r represents the Pearson correlation coefficient between two vibration features x and y, x i ,y i represent the sample values ​​of vibration features x and y in the i-th group of historical vibration data, respectively. They represent the sample average values ​​of vibration features x and y respectively, and i is the number of groups of historical vibration data obtained.

[0022] Preferably, in step S3, the XGBOOST parameters are optimized using a simulated annealing algorithm, and the specific steps for obtaining the optimal XGBOOST parameter combination are:

[0023] The current value S1 of one of the XGBOOST parameters of the hull local vibration prediction model is randomly disturbed to generate a new value S2, and the new value S2 is used as the optimal value of the XGBOOST parameter;

[0024] Determine whether the new value S2 meets the set acceptance condition. If so, accept the new value S2 and use the new value S2 as the optimal value of the XGBOOST parameter. Otherwise, keep the current value S1.

[0025] Repeat the above steps to obtain the optimal value of each XGBOOST parameter in turn, thereby obtaining the optimal XGBOOST parameter combination.

[0026] Preferably, the specific steps of judging whether the new value S2 satisfies the set acceptance condition are:

[0027] If S2-S1<0, accept the new value S2;

[0028] If S2 ≥ S1, calculate the probability of accepting S2 exp(-dS / initial_t), and generate a random number r in (0,1). If exp(-dS / initial_t)>r, accept the new value S2 and take the new value S2 as the optimal value of the XGBOOST parameter. Otherwise, keep the current value S1. dS is the difference between the new value S2 and the current value S1. initial_t is the initial temperature set in the simulated annealing algorithm.

[0029] The beneficial effects of the present invention are:

[0030] 1. The SA-XGboost algorithm of the present invention integrates the advantages of machine learning and uses decision tree training samples. From the results, the SA-XGboost algorithm has a better prediction effect on the local vibration of the hull. While improving the accuracy of the local vibration prediction model of the hull, it prevents the model from falling into the local optimum, improves the robustness of the algorithm, and can be used for the local vibration prediction of the hull.

[0031] 2. For the inaccessible areas and confined spaces with high risks in subsequent ship models, the characteristic factors such as the three positions of the measuring point in the area from the main engine, the test conditions, etc. (i.e., the vibration characteristic parameters obtained after the dimensionality reduction processing in step S2 above) can be input into the SA-XGboost-based hull local vibration prediction model to obtain the local vibration data of the collection point, and perform safety verification on the international standard values ​​to determine whether the standard requirements are met, thereby improving the safety and comfort of ship operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0033] Figure 1 It is a flow chart of the present invention. DETAILED DESCRIPTION

[0034] In order to make the purpose, technical scheme and advantages of the present invention clearer, the present invention is described below by the specific embodiments shown in the accompanying drawings. However, it should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. In addition, in the following description, the description of well-known structures and technologies is omitted to avoid unnecessary confusion of the concept of the present invention.

[0035] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. The singular forms of "a", "said" and "the" used in this disclosure and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0036] In order to better understand the technical solution of the present invention, the present invention is described in detail below with reference to the accompanying drawings.

[0037] The present invention provides a method for predicting local vibration of a hull based on the SA-XGBOOST algorithm, which specifically comprises the following steps:

[0038] S1, obtaining several groups of historical vibration data of local vibration measurement points when the ship is sailing and preprocessing the data.

[0039] Specifically, when obtaining several groups of historical vibration data of local vibration measurement points when the ship is sailing, a certified triaxial vibration meter is placed at the local vibration measurement point and left to stand for 1 minute to collect the vibration data of the ship when sailing. At the same time, multiple vibration characteristic parameters of the local vibration measurement point are also collected. The collected vibration data and multiple vibration characteristic parameters are classified into a group of historical vibration data.

[0040] The vibration characteristic parameters include the distance X between the local vibration measurement point and the main engine in the ship length direction. e , the distance Y between the local vibration measurement point and the main engine in the ship width direction e , the distance Z between the local vibration measurement point and the main engine in the ship height direction e , the distance from the propeller in the direction of the ship's length X p , the distance Y between the local vibration measurement point and the propeller in the ship width direction p , the distance Z between the local vibration measurement point and the propeller in the ship height direction p , the ship speed V when measuring at the local vibration measurement point, the wave height H when measuring at the local vibration measurement point, and the wind speed W when measuring at the local vibration measurement point s , the thickness of the plate at the local vibration measurement point B, the filling value of the vibration reduction equipment when measuring the local vibration measurement point H m(When the vibration reduction device is turned on, the filling value H of the vibration reduction device is measured at the local vibration measurement point m =1; when the vibration reduction device is not turned on, the vibration reduction device fill value H when measuring the local vibration measurement point m = 0), liquefaction level R of the reliquefaction equipment when measuring at the local vibration measurement point l , the rotation speed R of the ship's main engine.

[0041] According to the above method, the historical vibration data of multiple local vibration measurement points in multiple navigation stages are respectively used, so that several groups of historical vibration data can be obtained.

[0042] The obtained sets of historical vibration data are preprocessed, including data cleaning (eliminating outliers, etc.), normalization processing, etc.

[0043] The data normalization process is as follows:

[0044]

[0045] Where: x i is the hull vibration data obtained through measurement, x min is the minimum value in the original data set of hull vibration, x max is the maximum value in the original data set of hull vibration, is the normalized data of hull vibration.

[0046] S2, performing dimension reduction processing on the historical vibration data of the local vibration measurement points when the ship is sailing to obtain dimension-reduced vibration data.

[0047] Specifically, first, the correlation coefficients between the various vibration features are calculated based on several groups of historical vibration data of the local vibration measurement points;

[0048] In this embodiment, the Pearson correlation coefficient is used to analyze the correlation between the vibration features.

[0049] The correlation coefficient between vibration characteristics is calculated as:

[0050]

[0051] Among them, r represents the Pearson correlation coefficient between two vibration features x and y, x i ,y i represent the sample values ​​of vibration features x and y in the i-th group of historical vibration data, respectively. They represent the sample averages of the vibration features x and y, respectively, and i is the number of groups of historical vibration data obtained. In the above formula, the numerator is the covariance of the two variables x and y. The covariance is the expectation of the overall error between the two variables, which is used to reflect whether the two variables deviate from their respective means at the same time; and the denominator represents the standard deviation of the two variables x and y, respectively. The standard deviation is a statistic that measures the degree of discreteness of the data distribution, reflecting the degree of deviation between the data and its mean.

[0052] The value range of the Pearson correlation coefficient r is [-1,1]. When the correlation coefficient is less than 0, it indicates that there is a linear negative correlation between the two characteristic parameters; when the correlation coefficient is greater than 0, it indicates that there is a linear positive correlation between the two characteristic parameters; when the correlation coefficient is equal to 0, it indicates that there is no obvious correlation between the two characteristic parameters.

[0053] Then, according to the calculated correlation coefficients between the vibration features, vibration features with correlation coefficients greater than 0 are selected and the vibration data corresponding to these vibration features are used as dimension-reduced vibration data.

[0054] In this application, according to the data dimension reduction processing method, the vibration characteristics of the local vibration data are reduced to 7, which are the distance X from the local vibration measurement point to the main engine in the direction of the ship length, e , the distance Y between the local vibration measurement point and the main engine in the ship width direction e , the distance Z between the local vibration measurement point and the main engine in the ship height direction e , the speed V when measuring the local vibration measurement point, the wave height H when measuring the local vibration measurement point, the plate thickness B where the local vibration measurement point is located, and the main engine speed R. These 7 vibration characteristics after dimensionality reduction are the most relevant and necessary features for the local vibration of the hull. Selecting them as sample feature data for subsequent machine learning will help improve the accuracy of predicting the local vibration size of the test point. It is worth noting that in addition to the feature parameters described in the implementation of this application, other feature parameters can also be selected or added as sample feature data for machine learning prediction.

[0055] S3, divide the dimension-reduced vibration data into a training set and a test set, build a hull local vibration prediction model based on the XGBOOST algorithm, and use the training set and the test set to train and test the hull local vibration prediction model.

[0056] Specifically, at least 70%-80% of the data in the dimension-reduced vibration data is used as training set data, and the remaining data is used as test set data.

[0057] In the XGBOOST algorithm, the GridSearchCV method is used for model training and the cross-validation method is used to adjust the parameter combination. Multiple decision trees are constructed through iteration. Each decision tree fits the residual of the previous tree as much as possible, and gradually approaches the true value of the target variable by accumulating the prediction results of each tree.

[0058] The parameters in the XGBOOST algorithm are shown in the following table:

[0059]

[0060] Since the XGBOOST algorithm will stay at the local optimal solution during the iteration process, which affects the accuracy of the prediction results, in order to solve this problem, the present application uses a simulated annealing algorithm to optimize the XGBOOST parameters during the iterative calculation process of training and testing the hull local vibration prediction model, obtains the optimal XGBOOST parameter combination, and substitutes the optimal XGBOOST parameter combination into the model for the next round of iterative calculation until a hull local vibration prediction model with an accuracy rate that meets the requirements is obtained.

[0061] Specifically, the simulated annealing algorithm is used to optimize the XGBOOST parameters, and the specific steps to obtain the optimal XGBOOST parameter combination are as follows: first, the current value S1 of one of the XGBOOST parameters of the hull local vibration prediction model is randomly disturbed to generate a new value S2, and the new value S2 is used as the optimal value of the XGBOOST parameter;

[0062] Then, if S2-S1<0, accept the new value S2;

[0063] If S2≥S1, calculate the probability of accepting S2 exp(-dS / initial_t), and generate a random number r in (0,1). If exp(-dS / initial_t)>r, accept the new value S2 and take the new value S2 as the optimal value of the XGBOOST parameter. Otherwise, keep the current value S1. dS is the difference between the new value S2 and the current value S1. initial_t is the initial temperature set in the simulated annealing algorithm.

[0064] Repeat the above steps to obtain the optimal value of each XGBOOST parameter in the above table in turn, so as to obtain the optimal XGBOOST parameter combination.

[0065] In this embodiment, the simulated annealing algorithm has the following parameters: the initial temperature is set to "initial_t=100", the final temperature is set to "final_t_t=1", and the annealing coefficient is w=0.97.

[0066] By integrating the XGBOOST algorithm with the simulated annealing algorithm, the simulated annealing algorithm is introduced into the XGBOOST algorithm. During the iteration process of the XGBOOST algorithm, on the one hand, the better solution is retained, and on the other hand, the inferior target solution is retained with a certain probability, so that the algorithm has a good balance in fast convergence and jumping out of the local optimal solution.

[0067] When the prediction accuracy of the hull local vibration prediction model meets the requirements, it means that the model is well trained.

[0068] The mean absolute error (MAE) and the coefficient of determination (R 2 ) These two parameters are used to determine the accuracy of model prediction. The calculation formulas for the mean absolute error and the coefficient of determination are as follows:

[0069]

[0070] Where y is the predicted value of the local vibration of the hull, is the true value of the local vibration of the hull. N is the number of samples, y i and are the true value and predicted value of the local vibration value at the i-th measurement point. is the mean of the true values ​​of the local vibration at N measurement points.

[0071] Mean Absolute Error (MAE) is an indicator used to measure the prediction accuracy of a prediction model on continuous data. It measures the average absolute difference between the predicted value and the true value, indicating the average degree of deviation between the predicted value and the true value. MAE is the average of the absolute values ​​of the difference between the predicted value and the true value. It does not consider the positive or negative value of the predicted value, but only focuses on the absolute value of the error.

[0072] Coefficient of Determination (R 2 ) is an indicator used to evaluate the goodness of fit of a regression model, which reflects the percentage of the dependent variable variation (i.e., data variance) that the model can explain. 2 The value range is between 0 and 1. 2 When R is 1, it means that the model perfectly predicts the data; when R 2 When it is 0, it means that the model cannot explain the data variance.

[0073] S4, inputting the three-dimensional position of the position to be measured from the main engine, the current speed, the wave height, the main engine rotation speed and the plate thickness of the position to be measured into the hull local vibration prediction model to obtain the local vibration data of the position.

[0074] The specific embodiments of the present invention are described in detail below through specific examples.

[0075] The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm in this embodiment includes the following steps:

[0076] S1, when the ship is sailing, a certified triaxial vibration meter is placed at a local vibration measurement point and left to stand for 1 minute to collect the vibration data of the ship during sailing. At the same time, multiple vibration characteristic parameters of the local vibration measurement point are also collected. The collected vibration data and multiple vibration characteristic parameters are classified into a set of historical vibration data.

[0077] According to the above method, the historical vibration data of multiple local vibration measurement points in multiple navigation stages are respectively used, so that several groups of historical vibration data can be obtained.

[0078] The obtained sets of historical vibration data are preprocessed, including data cleaning (eliminating outliers, duplicate data, etc.), normalization processing, etc.

[0079] S2, the historical vibration data of the local vibration measurement point during the ship's voyage is processed by dimensionality reduction, and the 12 vibration characteristic parameters of the historical vibration data are reduced to 7, including the distance X from the measurement point to the main engine in the direction of the ship's length. e , the distance Y between the measuring point and the main engine in the ship width direction e , the distance Z between the measuring point and the main engine in the ship height direction e , the ship speed V when measuring at the measuring point, the wave height H when measuring at the measuring point, the plate thickness B at the measuring point, and the main engine speed R, in order to obtain the reduced-dimensional vibration data.

[0080] S3, divide 70% of the reduced-dimensional vibration data into training set data, and the remaining 30% of the data into test set data, build a hull local vibration prediction model based on the XGBOOST algorithm, and use the training set and test set to train and test the hull local vibration prediction model.

[0081] In the iterative calculation process of training and testing the hull local vibration prediction model, the simulated annealing algorithm is used to optimize the XGBOOST parameters to obtain the optimal XGBOOST parameter combination, and the optimal XGBOOST parameter combination is substituted into the model for the next round of iterative calculation until a hull local vibration prediction model with accuracy that meets the requirements is obtained.

[0082] S4, inputting the three-dimensional position of the position to be measured from the main engine, the current speed, the wave height, the main engine rotation speed and the plate thickness of the position to be measured into the hull local vibration prediction model to obtain the local vibration data of the position.

[0083] Using the same sample data, the local vibration of the hull is predicted by this method and the SVM algorithm respectively. The accuracy of the two methods in predicting the local vibration of the hull is shown in the following table:

[0084]

[0085] According to the above table, the prediction accuracy of this method is significantly better than that of the SVM algorithm. The SA-XGboost algorithm of the present invention integrates the advantages of machine learning and uses decision tree training samples. From the results, the SA-XGboost algorithm has a better prediction effect on the local vibration of the hull. While improving the accuracy of the hull local vibration prediction model, it prevents the model from falling into the local optimum, improves the robustness of the algorithm, and can be used for the local vibration prediction of the hull.

[0086] In subsequent ship models, for inaccessible areas and confined spaces with high risks, the characteristic factors such as the three positions of the measurement point in the area from the main engine, test conditions, etc. (i.e., the vibration characteristic parameters obtained after the dimensionality reduction processing in step S2 above) can be input into the SA-XGboost hull local vibration prediction model to obtain the local vibration data of the collection point, and perform safety verification on the international standard values ​​to determine whether the standard requirements are met. This will improve the safety and comfort of ship operations.

[0087] It should be clear that the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

Claims

1. A method for predicting local vibration of a hull based on the SA-XGBOOST algorithm, characterized in that: The specific steps include: S1, obtaining several groups of historical vibration data of local vibration measurement points when the ship is sailing and preprocessing the data; Each set of historical vibration data includes multiple vibration characteristic parameters; S2, performing dimension reduction processing on the historical vibration data of the local vibration measurement points when the ship is sailing to obtain dimension-reduced vibration data; S3, divide the dimension-reduced vibration data into a training set and a test set, build a hull local vibration prediction model based on the XGBOOST algorithm, and use the training set and the test set to train and test the hull local vibration prediction model. In the iterative calculation process of training and testing the hull local vibration prediction model, the simulated annealing algorithm is used to optimize the XGBOOST parameters to obtain the optimal XGBOOST parameter combination, and the optimal XGBOOST parameter combination is substituted into the model for the next round of iterative calculation until a hull local vibration prediction model with an accuracy rate that meets the requirements is obtained; S4, inputting the three-dimensional position of the position to be measured from the main engine, the current speed, the wave height, the main engine rotation speed and the plate thickness of the position to be measured into the hull local vibration prediction model to obtain the local vibration data of the position.

2. The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm according to claim 1, characterized in that: The vibration characteristic parameters in step S1 include the distance X between the local vibration measurement point and the main engine in the direction of the ship length. e , the distance Y between the local vibration measurement point and the main engine in the ship width direction e , the distance Z between the local vibration measurement point and the main engine in the ship height direction e , the distance from the propeller in the direction of the ship's length X p , the distance Y between the local vibration measurement point and the propeller in the ship width direction p , the distance Z between the local vibration measurement point and the propeller in the ship height direction p , the ship speed V when measuring at the local vibration measurement point, the wave height H when measuring at the local vibration measurement point, and the wind speed W when measuring at the local vibration measurement point s , the thickness of the plate at the local vibration measurement point B, the filling value of the vibration reduction equipment when measuring the local vibration measurement point H m , liquefaction level R of the reliquefaction equipment when measuring at the local vibration measurement point l , the rotation speed R of the ship's main engine.

3. The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm according to claim 2, characterized in that: When the vibration reduction device is turned on, the vibration reduction device filling value H is measured at the local vibration measurement point m =1; when the vibration reduction device is not turned on, the vibration reduction device fill value H when measuring the local vibration measurement point m =0.

4. The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm according to claim 1, characterized in that: The specific steps of performing dimensionality reduction processing on the historical vibration data of the local vibration measurement points when the ship is sailing in step S2 are: According to several groups of historical vibration data of local vibration measurement points, the correlation coefficients between various vibration characteristics are calculated respectively; According to the calculated correlation coefficients between the vibration features, the vibration features with correlation coefficients greater than 0 are selected and the vibration data corresponding to these vibration features are used as the dimension-reduced vibration data.

5. The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm according to claim 4 is characterized in that: The dimension-reduced vibration data includes the distance X between the local vibration measurement point and the main engine in the ship length direction. e , the distance Y between the local vibration measurement point and the main engine in the ship width direction e , the distance Z between the local vibration measurement point and the main engine in the ship height direction e , the ship speed V when measuring at the local vibration measurement point, the wave height H when measuring at the local vibration measurement point, the plate thickness B where the local vibration measurement point is located, and the main engine speed R.

6. The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm according to claim 4, characterized in that: The correlation coefficient between vibration characteristics is calculated as: Among them, r represents the Pearson correlation coefficient between two vibration features x and y, x i ,y i represent the sample values ​​of vibration features x and y in the i-th group of historical vibration data, respectively. They represent the sample average values ​​of vibration features x and y respectively, and i is the number of groups of historical vibration data obtained.

7. The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm according to claim 1, characterized in that: In step S3, the XGBOOST parameters are optimized using the simulated annealing algorithm, and the specific steps for obtaining the optimal XGBOOST parameter combination are: The current value S1 of one of the XGBOOST parameters of the hull local vibration prediction model is randomly disturbed to generate a new value S2, and the new value S2 is used as the optimal value of the XGBOOST parameter; Determine whether the new value S2 meets the set acceptance condition. If so, accept the new value S2 and use the new value S2 as the optimal value of the XGBOOST parameter. Otherwise, keep the current value S1. Repeat the above steps to obtain the optimal value of each XGBOOST parameter in turn, thereby obtaining the optimal XGBOOST parameter combination.

8. The method for predicting local vibration of a hull based on the SA-XGBOOST algorithm according to claim 7, characterized in that: The specific steps to determine whether the new value S2 meets the set acceptance conditions are: If S2-S1<0, accept the new value S2; If S2 ≥ S1, calculate the probability of accepting S2 exp(-dS / initial_t), and generate a random number r in (0,1). If exp(-dS / initial_t)>r, accept the new value S2 and take the new value S2 as the optimal value of the XGBOOST parameter. Otherwise, keep the current value S1. dS is the difference between the new value S2 and the current value S1. initial_t is the initial temperature set in the simulated annealing algorithm.

Images