Absolute domain wave parameter inversion method and system under ship speed condition

Abstract

The invention belongs to the technical field of ship and ocean engineering, and discloses an absolute domain wave parameter inversion method and system under the condition that a ship has a navigational speed. The method comprises the following steps: carrying out segmentation processing on obtained six-degree-of-freedom motion time-calendar data and wave height time-calendar data; wave direction inversion preprocessing and wave height time calendar inversion preprocessing are carried out; dividing a data set; wave height time calendar estimation and wave direction estimation; and taking the obtained wave height time calendar information, the wave direction information and the navigational speed information as priori knowledge input of a Doppler frequency shift correction method, correcting an encountering domain wave spectrum to an absolute domain under the condition that the ship has the navigational speed, and completing inversion of wave statistical parameters in the absolute domain. According to the method, the encountering wave spectrum under the navigational speed condition is converted into the absolute wave spectrum, so that the accurate inversion of the wave statistical parameters in the absolute domain is realized.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of ship and ocean engineering, and particularly relates to a ship absolute domain wave parameter inversion method and system under a sailing speed condition. BACKGROUND

[0002] In the field of ocean engineering and environmental monitoring, the acquisition of wave direction, significant wave height and characteristic period has important theoretical and practical significance. First, as an important indicator of wave energy, the significant wave height can provide key parameters for the design of marine structures, ensuring their safety and stability under the action of waves. Second, the characteristic period reflects the periodicity of waves, which is crucial for ship navigation safety. It can guide ships to develop more reasonable navigation routes to avoid potential navigation risks. Therefore, systematic research and accurate acquisition of the significant wave height and characteristic period of sea waves can provide solid data support for the in-depth development of marine science.

[0003] The current mainstream wave monitoring methods include crew experience judgment, remote sensing satellites, shipborne radars and wave buoys, etc. However, they have a series of problems and challenges in the application of actual wave environment perception, such as subjective errors in crew experience and judgment, negative impact of atmospheric conditions on satellite capture quality, and information lag of several hours due to long processing time. Wave radar can accurately report wave parameters other than wave height, but its installation and calibration cost is high, which is not cost-effective in wave monitoring. Wave buoys are relatively reliable tools for obtaining wave information and are widely used today. Considering the loss and cost, it is not feasible to use wave buoys to cover a wide geographical area, so wave buoys are mostly used in coastal areas.

[0004] Most ships are equipped with sensors to record data such as heave and roll, so the ship can be regarded as a wave buoy with specific geometric shape, and the on-site sea conditions can be estimated by processing the ship's motion response. With the continuous development of intelligent technology, data-driven methods based on neural networks have obvious advantages in solving a series of nonlinear, non-stationary complex mapping problems. Therefore, wave inversion technology based on the idea of ship analogy buoy gradually transitions from traditional hydrodynamic physical models to intelligent models.

[0005] When the ship sails with a sailing speed in the downwind direction, the characteristic period shows a significant decrease. This is due to the Doppler effect caused by the moving platform, which causes the wave observation to be converted from the fixed spatial domain to the encounter frequency domain, resulting in distortion of the wave energy spectrum shape: the compression or expansion of the spectrum on the frequency axis causes the spectrum peak to shift, and the significant change in the spectrum shape significantly affects the calculation of the second-order spectral moment sensitive to frequency distribution, ultimately leading to distortion of the evaluation of the characteristic period based on the ratio of spectral moments.

[0006] In the prior art, the invention patent "Non-coherent radar phase analysis wave time history inversion method based on deep learning" (publication number CN117647808A, publication date 20240430) discloses a non-coherent radar phase analysis wave time history inversion method based on deep learning, which relates to the field of ocean remote sensing detection, and includes the following steps: preprocessing radar sea clutter images, converting the gray data of radar sea clutter images in the region to radar image spectral data through Fourier transform; constructing a deep learning spectral mapping model, inputting the radar image spectral data into the constructed deep learning spectral mapping model, the deep learning spectral mapping model including a three-dimensional convolution module, a position coding, an attention module, a residual module and a three-dimensional transpose convolution module; and finally data inverse processing, converting the spectral data calculated by the model into actual phase analysis sea wave time history data. The present invention can reduce the phenomenon of spectral data missing and improve the accuracy of inversion data.

[0007] Disadvantages: This invention mainly uses non-coherent radar to solve wave time history, and ordinary unmanned ships are limited by size and economy and cannot carry non-coherent radar, so the universality is limited.

[0008] The invention patent "Wave inversion method and system based on ship oscillation motion" (publication number CN117104452A, publication date 20231124) belongs to the field of ship and ocean engineering technology, and discloses a wave inversion method and system based on ship oscillation motion based on artificial neural network. This method uses data-driven method to invert the significant wave height, characteristic period and wave direction of the wave statistical characteristics based on ship motion, and acquires ship motion data through the arrangement of pose sensors on the ship, and further acquires the surrounding sea wave information of the ship. This invention has good robustness for different sea conditions. At the same time, this method can realize the acquisition of ship motion data through the arrangement of pose sensors on the ship, which has simple hardware requirements, high cost-effectiveness, and can realize ship wave monitoring. This method provides a new solution for current acquisition of surrounding sea wave information, and can provide wave environment data support for ship navigation and operation decision-making.

[0009] Disadvantages: This invention mainly focuses on statistical value inversion using ship motion, does not involve wave time history inversion, and can only give statistical values in the absolute domain without considering the inversion of absolute domain wave parameters with ship speed. SUMMARY

[0010] To overcome the problems in the related art, the embodiment of the present application provides a ship absolute domain wave parameter inversion method and system under the condition of ship speed. The existing invention mainly uses radar to invert wave time history, and does not use ship motion to invert wave time history; the existing invention does not consider the absolute domain wave parameter inversion of the ship under the condition of ship speed, the present application can dynamically calculate the direction, significant wave height and characteristic period of the actual wave in the current sea area by installing a motion sensor on the ship to collect ship attitude data in real time, in combination with ship speed information.

[0011] The technical solution is as follows: a ship absolute domain wave parameter inversion method under the condition of ship speed, the method comprising the following steps:

[0012] S1, obtaining ship six-degree-of-freedom motion time history data from an attitude sensor, and obtaining wave height time history data from a wave buoy or a wave height instrument;

[0013] S2, performing segmentation processing on the obtained six-degree-of-freedom motion time history data and wave height time history data;

[0014] S3, performing wave direction inversion preprocessing and wave height time history inversion preprocessing on the segmented six-degree-of-freedom motion time history data and wave height time history data;

[0015] S4, performing data set division on the samples after wave direction inversion preprocessing and wave height time history inversion preprocessing;

[0016] S5, using a time convolution neural network to perform wave height time history estimation and wave direction estimation on the data after data set division;

[0017] S6, inputting the obtained wave height time history information, wave direction information and speed information as prior knowledge of a Doppler frequency shift correction method, correcting the encounter domain wave spectrum to the absolute domain under the condition of ship speed, and completing the inversion of the wave statistical parameters in the absolute domain;

[0018] S7, comparing and verifying the obtained corrected absolute domain spectrum with the original absolute domain spectrum and the uncorrected encounter spectrum.

[0019] In step S1, the ship six-degree-of-freedom motion time history data obtained from the attitude sensor is:

[0020]

[0021] In the formula, are respectively time ship motion attitude data of each degree of freedom n is the length of the data, m is the number of degrees of freedom, v is the speed, and d is the current wave direction angle;

[0022] The six degrees of freedom of a ship are: sway, surge, heave, roll, pitch, and bow pitch;

[0023] The wave height history data obtained from the wave buoy or wave height meter is:

[0024] W n =[w1,w2…w n-1 ,w n ]

[0025] Where w1, w2…w n-1 ,w n Time The motion data under time is the wave height history data corresponding to time.

[0026] In step S2, the obtained six-degree-of-freedom motion history data is segmented, including: Data corresponding to time Right now Divided into window _ size The data window is of window size and slides on each time series at a fixed interval of slip_size;

[0027] The first set of times is: The first set of ship motion posture data for each degree of freedom is: The second set of times is: The second set of ship motion posture data for each degree of freedom is: And so on, until the end of the time series, the same operation is performed on each degree of freedom;

[0028] The first set of wave height history data is: [w1,w2…w window_size ], the second set of wave height history data is: [w 1+slip_size ,w 2+slip_size …w window_size+slip_size ].

[0029] In step S3, the wave direction inversion preprocessing includes: performing preprocessing work for wave direction estimation, dividing the segmented For time series segments, fill in the wave direction angle d corresponding to the current data as the classification label; then randomly shuffle the data and labels of each segment to make the continuous wave directions randomly distributed;

[0030] The pre-processing of wave height history inversion includes: Movement history data and [w1,w2…w window_size ] The wave height history data correspond one to one in time, and a total of several one-to-one corresponding sample points of the specified window size are generated.

[0031] In step S5, the time convolutional neural network is used to estimate the wave height and direction, and the input is the ship motion data The output is [w1,w2…w window_size ] and the wave direction angle d of the data in this section.

[0032] In step S6, the relationship between the absolute domain frequency and the encounter domain frequency is as follows:

[0033]

[0034] Where w e is the encounter frequency, w0 is the actual wave frequency, U is the ship speed, g is the acceleration of gravity, β is the wave direction angle, and τ is the intensity of the Doppler shift;

[0035] The Doppler shift correction method is used to correct the encounter domain wave spectrum to the absolute domain under the condition of ship speed. The inversion of wave statistical parameters in the absolute domain is completed, including:

[0036] S601, perform frequency domain statistics on the wave height history obtained by time convolutional neural network inversion, and calculate the wave energy spectrum S in the encounter domain. e (w e ), obtain the significant wave height Hse and characteristic period Tse of the encounter domain;

[0037] S602, the obtained S e (w e ) Press w e Discretize into N e , recorded as: S e (w e (i)),i∈1:N e , when w e When ≤1 / 4τ, the encountered frequency is one-to-three with the actual frequency, and w e (i) Substitute into w in three steps 01 (i),w 02 (i),w 03 (i) In the 01 (i),w 02 (i),w 03 (i) The actual frequencies corresponding to the three-to-one encounter frequencies;

[0038] S603, the obtained w 01 (i),w 02 (i),w 03 (i) and the significant wave height Hse, characteristic period Tse are respectively brought into S 01 ,S 02 ,S 03 , where S 01 ,S 02 ,S03 w 01 (i),w 02 (i),w 03 (i) respectively

[0039] S604, w e (i) is divided into three regions, and the absolute frequency w 0,j and the absolute wave spectrum S 0,j are calculated for each region respectively.

[0040] S605, the modified absolute frequency w 0,j and the absolute wave spectrum S 0,j are sorted from small to large according to w 0,j , and the modified absolute wave spectrum is obtained.

[0041] S606, the modified absolute wave spectrum is used to calculate the absolute domain wave parameters of the ship at a speed, i.e. significant wave height and characteristic period.

[0042] In step S602, when w e ≤1 / 4τ, one encounter frequency corresponds to three actual frequencies, w e (i) is divided into three segments and brought in

[0043] In step S603, S 01 =ITTC(w 01 (i),Hse,Tse), S 02 =ITTC(w 02 (i),Hse,Tse), S 03 =ITTC(w 03 (i),Hse,Tse).

[0044]

[0045] In the formula, a1, a2, a3 are the proportions of the theoretical wave spectrum values of each segment respectively.

[0046] In step S604, for region I, w 0,j = w 01 (i), S 0,j =a1·S e (w e (i))·(1-2w 01 (i)·τ), where w0,j is the jth frequency in the original frequency, S 0,j is w 0,j The actual wave energy spectrum value of the corresponding absolute domain;

[0047] For interval II, w 0,j = w 02 (i), S 0,j = a2·S e (w e (i))·(1-2w 02 (i)·τ);

[0048] For interval III, w 0,j = w 03 (i), S 0,j = a3·S e (w e (i))·(1-2w 03 (i)·τ);

[0049] For the case of w e >1 / 4τ, S 0,j =S e (w e (i))·(1-2w 0,j ·τ), where j is increased by 1 each time.

[0050] Another object of the present application is to provide a ship absolute domain wave parameter inversion system under sailing speed conditions, which implements the ship absolute domain wave parameter inversion method under sailing speed conditions, and the system comprises:

[0051] A time history data acquisition module is configured to acquire ship six-degree-of-freedom motion time history data from a posture sensor and wave height time history data from a wave buoy or wave height meter;

[0052] A data preprocessing module is configured to perform segmentation processing on the acquired six-degree-of-freedom motion time history data and wave height time history data;

[0053] A wave inversion data preprocessing module is configured to perform wave direction inversion preprocessing and wave height time history inversion preprocessing on the segmented six-degree-of-freedom motion time history data and wave height time history data;

[0054] A data set division module is configured to perform data set division on the samples after wave direction inversion preprocessing and wave height time history inversion preprocessing;

[0055] A model training module is configured to use a time convolution neural network to perform wave height time history estimation and wave direction estimation on the data after data set division;

[0056] A Doppler frequency shift correction module is used to input the obtained wave height time history information, wave direction information and speed information as prior knowledge of a Doppler frequency shift correction method, correct the encountered domain wave spectrum to the absolute domain under the condition of the ship speed, and complete the inversion of the wave statistical parameters in the absolute domain.

[0057] A verification module is used to compare and verify the corrected absolute domain spectrum, the original absolute domain spectrum and the uncorrected encountered spectrum.

[0058] In combination with all the technical solutions described above, the present application has the following beneficial effects:

[0059] Firstly, the present application is used to solve the problem that the wave parameters in the absolute domain cannot be accurately obtained under the condition of the ship speed during the wave inversion process, and a wave parameter inversion method in the absolute domain under the condition of the ship speed is proposed, the intelligent wave height time history and wave direction inversion model based on the multi-degree-of-freedom motion information is constructed by using the ship motion data, the wave direction and wave height time history information are accurately obtained, and the time history information and wave direction information obtained by inversion are input as prior knowledge of the Doppler frequency shift correction method, so that the encountered domain wave spectrum is effectively corrected to the absolute domain under the condition of the ship speed, and the encountered wave spectrum under the condition of the ship speed is corrected to the absolute wave spectrum, so as to realize the accurate inversion of the wave statistical parameters in the absolute domain.

[0060] Secondly, the current wave inversion research can only obtain the wave information in the encountered domain when the ship is sailing at a speed, but it is difficult to obtain the absolute domain wave information under the objective condition, the encountered domain information is used as prior knowledge to complete the correction of the absolute domain information, and the accurate acquisition of the absolute domain wave parameter information is realized. BRIEF DESCRIPTION OF DRAWINGS

[0061] The accompanying drawings, which are incorporated into and form part of the specification, illustrate embodiments consistent with the present disclosure and, together with the specification, serve to explain the principles of the present disclosure;

[0062] Figure 1 is a principle diagram of the wave parameter inversion method in the absolute domain under the condition of the ship speed provided by the embodiment of the present application;

[0063] Figure 2 is a flowchart of the wave parameter inversion method in the absolute domain under the condition of the ship speed provided by the embodiment of the present application;

[0064] Figure 3 is an inversion effect diagram of the wave direction by the time convolution neural network at 0 knots speed provided by the embodiment of the present application;

[0065] Figure 4 is a correlation diagram of the wave direction by the time convolution neural network at 0 knots speed provided by the embodiment of the present application;

[0066] Figure 5 is a wave direction inversion effect map provided by the embodiment of the present application in the 3 knot speed time convolution neural network inversion;

[0067] Figure 6 is a wave direction correlation map provided by the embodiment of the present application in the 3 knot speed time convolution neural network inversion;

[0068] Figure 7 is a wave height time history inversion result comparison effect map provided by the embodiment of the present application under 0 knot speed;

[0069] Figure 8 is a wave height time history inversion result comparison effect map provided by the embodiment of the present application under 3 knots speed;

[0070] Figure 9 is a typical Doppler frequency shift curve graph in the downwind time provided by the embodiment of the present application;

[0071] Figure 10 is a 0 degree wave direction angle spectrum comparison graph in the 3 knot speed each wave direction angle spectrum comparison;

[0072] Figure 11 is a 20 degree wave direction angle spectrum comparison graph in the 3 knot speed each wave direction angle spectrum comparison;

[0073] Figure 12 is a 40 degree wave direction angle spectrum comparison graph in the 3 knot speed each wave direction angle spectrum comparison;

[0074] Figure 13 is a 60 degree wave direction angle spectrum comparison graph in the 3 knot speed each wave direction angle spectrum comparison;

[0075] Figure 14 is a 0 degree wave direction angle spectrum comparison graph in the 5 knot speed each wave direction angle spectrum comparison;

[0076] Figure 15 is a 20 degree wave direction angle spectrum comparison graph in the 5 knot speed each wave direction angle spectrum comparison;

[0077] Figure 16 is a 40 degree wave direction angle spectrum comparison graph in the 5 knot speed each wave direction angle spectrum comparison;

[0078] Figure 17 is a 60 degree wave direction angle spectrum comparison graph in the 5 knot speed each wave direction angle spectrum comparison;

[0079] Figure 18 is a 0 degree wave direction angle spectrum comparison graph in the 10 knot speed each wave direction angle spectrum comparison;

[0080] Figure 19 is a 20 degree wave direction angle spectrum comparison graph in the 10 knot speed each wave direction angle spectrum comparison;

[0081] Figure 20 This is the comparison chart of the wave angle spectrum of 40 degrees in the comparison of wave angle spectrum at 10 knots speed;

[0082] Figure 21 This is the comparison chart of the wave angle spectrum of 60 degrees in the comparison of wave angle spectrum at 10 knots speed;

[0083] Figure 22 The error comparison curve of each working condition is shown in Figure 2. DETAILED DESCRIPTION

[0084] To make the above-mentioned objects, features, and advantages of the present invention more readily apparent, specific embodiments of the present invention are described in detail below with reference to the accompanying drawings. The following description sets forth numerous specific details to facilitate a full understanding of the present invention. However, the present invention can be implemented in many other ways than those described herein, and those skilled in the art may make similar modifications without departing from the scope of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0085] The innovation of the present invention lies in: in order to solve the problem that Doppler frequency shift will be generated when the ship has a certain speed, and the wave parameters in the absolute domain cannot be accurately obtained, the present invention proposes for the first time a wave parameter inversion method in the absolute domain that is suitable for the ship under the condition of a certain speed. The wave height time history information and wave direction information in the encounter domain are obtained by inversion. This information is combined with the speed information obtained by GPS as a priori knowledge to complete the correction of the absolute domain wave spectrum, so as to achieve accurate acquisition of wave parameters in the absolute domain.

[0086] Example 1, as Figure 1 As shown, the embodiment of the present invention provides an absolute domain wave parameter inversion principle under the condition of ship speed.

[0087] Specifically, such as Figure 2 As shown, the absolute domain wave parameter inversion method provided by the embodiment of the present invention under the condition of ship speed includes:

[0088] S1, obtains the ship's six-degree-of-freedom motion history data from the attitude sensor, and obtains the wave height history data from the wave buoy or wave height meter;

[0089] S2, segmenting the acquired six-degree-of-freedom motion time history data and wave height time history data;

[0090] S3, performing wave direction inversion preprocessing and wave height time history inversion preprocessing on the segmented six-degree-of-freedom motion time history data and wave height time history data;

[0091] S4, dividing the data set of the samples after wave direction inversion preprocessing and wave height time history inversion preprocessing;

[0092] S5, using time convolution neural network, wave height time history estimation and wave direction estimation are performed on the divided data set;

[0093] S6, the obtained wave height time history information and wave direction information are used as prior knowledge of Doppler frequency shift correction method, and the wave spectrum in encounter domain is corrected to absolute domain under the condition of ship speed, and the inversion of wave statistical parameters in absolute domain is completed;

[0094] S7, the corrected absolute domain spectrum, the original absolute domain spectrum and the uncorrected encounter spectrum are compared and verified.

[0095] Exemplarily, in step S1, a set of ship six degree of freedom motion time history data obtained by the attitude sensor is:

[0096]

[0097] In the formula, respectively, time under the ship each degree of freedom motion attitude data n is the length of the data, m is the number of degrees of freedom, v is the speed, and d is the current data angle of wave direction;

[0098] The six degrees of freedom motion of the ship are ship sway, ship surge, ship heave, ship roll, ship pitch and bow yaw, respectively, and are marked as 1-6. The wave height time history data obtained by the wave buoy or wave height instrument is:

[0099] The wave height time history data obtained by the wave buoy or wave height instrument is:

[0100] W n =[w1,w2…w n-1 ,w n ]

[0101] In the formula, w1, w2…w n-1 ,w n are the wave height time history data corresponding to the motion data at time

[0102] Exemplarily, in step S2, data preprocessing, the six degree of freedom motion time history data and wave height time history data W n =[w1,w2…w n-1 ,w n ] corresponding to each wave direction angle file are divided and processed, and the data corresponding to time ​​​​The data window is divided into data windows with a window size of window_size, and is slid on each time series at a fixed interval slip_size;

[0103] The first group of time is: The motion attitude data of each degree of freedom of the first group of ships is: The second group of time is: The motion attitude data of each degree of freedom of the second group of ships is: By analogy, the same operation is performed on each degree of freedom until the end of the time series;

[0104] The first group of wave height time history data is: [w1, w2…w window_size ] and the second group of wave height time history data is: [w 1+slip_size ,w 2+slip_size …w window_size+slip_size ].

[0105] It can be understood that the present application divides and processes to construct a one-to-one mapping relationship between ship motion and wave height time history, and the purpose is to obtain the current wave height by inputting the current motion.

[0106] Exemplarily, in step S3, wave inversion data preprocessing: wave direction inversion preprocessing and wave height time history inversion preprocessing are performed respectively.

[0107] The wave direction inversion preprocessing includes: first, performing wave direction estimation preprocessing on the divided time series segment, filling the wave direction angle d corresponding to the current data as the classification label; and then randomly shuffling the data and the label of each segment to randomly distribute the continuous wave direction.

[0108] Exemplarily, the data under the wave direction angle file of 30 degrees, each segment after division is filled with the wave direction angle label d=30, and other wave direction angles are sequentially processed until all segments are labeled with the corresponding wave direction.

[0109] Finally, the data and the label of each segment are randomly shuffled to randomly distribute the continuous wave direction. In other words, after the above processing operation, the first several are 0-degree wave direction angles, followed by several 10-degree wave direction angles, and so on, and after shuffling, the label becomes a random distribution of possible wave direction angles such as 20, 70, and 10. This step of operation enables the model to learn more features and improves the estimation accuracy.

[0110] The wave height time history inversion preprocessing includes: after data preprocessing, the motion time history data and [w1, w2…w window_sizeThe wave height time history data is made one-to-one corresponding in time, and a plurality of sample points of a specified window size are generated for the model to train.

[0111] For example, in step S4, the data set is divided: the samples processed in the previous step are divided into a data set, the training set train is 70%, the test set test is 20%, and the verification set prediction is 10%. The data and labels in the samples are divided in a ratio of 7:2:1.

[0112] For example, in step S5, the model is trained: the present application uses a time convolutional neural network to estimate the wave height time history and the wave direction: the input is the ship motion data The output is [w1, w2…w window_size ] and the wave direction angle d of the data in this paragraph.

[0113] The inversion results are as follows: wave direction estimation results: input the training set into the time convolutional neural network model, update the model using the verification set results and save, after training 500 rounds at 0 knots and 3 knots using the above model and hyperparameters respectively, extract the best model on the verification set, and verify it on the test set to obtain the prediction results, as shown in Figure 3 Wave direction inversion effect of time convolutional neural network at 0 knots, Figure 4 Wave direction inversion correlation of time convolutional neural network at 0 knots, Figure 5 Wave direction inversion effect of time convolutional neural network at 3 knots, Figure 6 Wave direction inversion correlation of time convolutional neural network at 3 knots,

[0114] Wave height time history results output: input the training set into the model, update the model using the verification set results and save, after training 500 rounds at 0 knots and 3 knots using the above model and hyperparameters respectively, extract the best model on the verification set, and verify it on the test set to obtain the prediction results, as shown in Figure 7 Comparison effect diagram of wave height time history inversion results at 0 knots, Figure 8 Comparison effect diagram of wave height time history inversion results at 3 knots,

[0115] Exemplary, in step S6, Doppler shift correction: the inversion of the time history information, the wave direction information and the speed information as the prior knowledge input of the Doppler shift correction method, so as to effectively correct the encounter domain wave spectrum to the absolute domain under the condition of ship speed, so as to realize the accurate inversion of the wave statistical parameters in the absolute domain. The specific principle is as follows:

[0116] The relationship between the absolute domain frequency and the encounter domain frequency is as follows:

[0117]

[0118] In the formula, w e is the encounter frequency, w0 is the actual frequency of the wave, U is the ship speed, g is the gravity acceleration, β is the wave direction angle, and τ is the intensity of the Doppler shift;

[0119] The influence of the wave direction angle β on the relationship between the encounter frequency and the actual frequency is analyzed. As shown in the typical Doppler shift curve when sailing with the wave, Figure 9 When β∈[0, 90], that is, in the case of sailing with the wave, there is a one-to-many relationship between the encounter frequency domain and the actual frequency, and a typical curve is taken for analysis;

[0120] If you want to deduce the actual frequency w0 of the wave from the encounter frequency w e , the influence of τ needs to be considered. When w e >1 / 4τ, the encounter frequency and the actual frequency are also one-to-one, but when w e ≤1 / 4τ, the encounter frequency and the actual frequency may have a one-to-two or one-to-three relationship, which directly leads to changes in the shape of the spectrum, resulting in errors in the statistical value results. Therefore, it is necessary to correct the errors. Therefore, the algorithm is divided into two sections: w e >1 / 4τ and w e ≤1 / 4τ.

[0121] S601, the wave height time history obtained by the time convolution neural network inversion is statistically processed in the frequency domain, and the encounter domain wave energy spectrum S e (w e ) is calculated, and the significant wave height Hse and the characteristic period Tse in the encounter domain are obtained;

[0122] S602, the obtained S e (w e ) is discretized according to w e to N e , denoted as: S e (w e (i)), i∈1:N e , when w e ≤1 / 4τ, when the encounter frequency and the actual frequency are one-to-three, w e (i) is divided into three sections and brought into w01 (i),w 02 (i),w 03 (i),w 01 (i),w 02 (i),w 03 (i) respectively are the actual frequencies corresponding to the encounter frequency of one to three respectively; when w e ≤1 / 4τ, w e (i) is divided into three segments

[0123] S603, the obtained w 01 (i),w 02 (i),w 03 (i) and the significant wave height Hse, the characteristic period Tse are respectively brought into S 01 ,S 02 ,S 03 , wherein S 01 ,S 02 ,S 03 are w 01 (i),w 02 (i),w 03 (i) should theoretically correspond to the wave energy spectrum value, ITTC(·) function is ITTC two-parameter wave energy spectrum function, after obtaining the theoretical absolute domain wave energy spectrum value, the proportion of each theoretical spectrum value a1, a2, a3 is calculated; S 01 =ITTC(w 01 (i), Hse, Tse), S 02 =ITTC(w 02 (i), Hse, Tse), S 03 =ITTC(w 03 (i), Hse, Tse);

[0124]

[0125] In the formula, a1, a2, a3 are respectively the proportion of the theoretical wave spectrum value of each segment.

[0126] S604, w e ≤1 / 4τ is divided into three regions I, II, III, and the absolute frequency w 0,j and the absolute wave spectrum S 0,j are calculated for each interval respectively;

[0127] For interval I, w 0,j = w 01 (i), S 0,j =a1·S e (w e(i))·(1-2w 01 (i)·τ), where w 0,j is the jth frequency in the original frequency, S 0,j w 0,j The corresponding actual wave energy spectrum value in the absolute domain;

[0128] For interval II, w 0,j =w 02 (i),S 0,j =a2·S e (w e (i))·(1-2w 02 (i)·τ);

[0129] For interval III, w 0,j =w 03 (i),S 0,j =a3·S e (w e (i))·(1-2w 03 (i)·τ);

[0130] For w e For the case of >1 / 4τ, S 0,j =S e (w e (i))·(1-2w 0,j ·τ), where j increases by 1 each time.

[0131] S605, the corrected absolute frequency w 0,j and absolute spectrum S 0,j Press w 0,j Sort from small to large to obtain the corrected absolute spectrum;

[0132] S606, using the corrected absolute wave spectrum statistics, calculate the absolute domain wave parameters under the ship's speed, that is, the significant wave height and the characteristic period.

[0133] For example, in step S7, the results are compared using the ITTC two-parameter wave energy spectrum to construct an absolute domain spectrum. By inputting the significant wave height and characteristic period, the corresponding spectral shape can be directly obtained. The encounter spectrum calculated using the encounter time history and the corrected absolute domain spectrum obtained after Doppler shift correction are compared and verified. Level 5 sea conditions are selected for verification, and the spectrum corrections are compared for various wave angles at speeds of 3 knots, 5 knots, and 10 knots.

[0134] Among them, such as Figure 10 The following is a comparison chart of the wave angle spectrum at 0 degrees in the comparison of wave angle spectrum at 3 knots speed, as shown in the figure below: Figure 11 This is the comparison chart of the wave angle spectrum of 20 degrees in the comparison of wave angle spectrum at 3 knots speed.Figure 12 is a 40-degree wave direction spectrum comparison chart in the 3-knot speed spectrum comparison of each wave direction spectrum, Figure 13 is a 60-degree wave direction spectrum comparison chart in the 3-knot speed spectrum comparison of each wave direction spectrum;

[0135] Figure 14 is a 0-degree wave direction spectrum comparison chart in the 5-knot speed spectrum comparison of each wave direction spectrum, Figure 15 is a 20-degree wave direction spectrum comparison chart in the 5-knot speed spectrum comparison of each wave direction spectrum, Figure 16 is a 40-degree wave direction spectrum comparison chart in the 5-knot speed spectrum comparison of each wave direction spectrum, Figure 17 is a 60-degree wave direction spectrum comparison chart in the 5-knot speed spectrum comparison of each wave direction spectrum; Figure 18 is a 0-degree wave direction spectrum comparison chart in the 10-knot speed spectrum comparison of each wave direction spectrum, Figure 19 is a 20-degree wave direction spectrum comparison chart in the 10-knot speed spectrum comparison of each wave direction spectrum, Figure 20 is a 40-degree wave direction spectrum comparison chart in the 10-knot speed spectrum comparison of each wave direction spectrum, Figure 21 is a 60-degree wave direction spectrum comparison chart in the 10-knot speed spectrum comparison of each wave direction spectrum;

[0136] The present application takes significant wave height and characteristic period as the correction determination standard. In all working conditions, the absolute wave spectrum has a significant wave height Hs of 3.45 m and a characteristic period Ts of 7.86 s. In order to quantify the deviation of each correction result, the present application selects relative error as an evaluation index, as shown in Table 1, so as to systematically evaluate the effectiveness of the correction scheme under different speed and wave direction conditions.

[0137] Table 1 Error comparison of spectrum correction methods

[0138]

[0139] The intuitive error is as shown in the error comparison curve of each working condition. Figure 22

[0140] The present application selects the test results of different wave direction angles under 3 knots, 5 knots and 10 knots to compare. The error analysis shows that whether the significant wave height calculated by the encounter spectrum or the modified spectrum by the Doppler algorithm remains basically unchanged, which shows that there is almost no energy loss between the absolute domain wave energy spectrum, the encounter spectrum and the modified spectrum. Then, the results of the characteristic period are verified, and the results show that under the condition of the encounter spectrum, the characteristic period increases obviously, and with the increase of the speed, the Doppler frequency shift phenomenon is more and more obvious, and the phenomenon is closely related to the wave direction angle, and the closer the wave direction angle is to the downwind (0 degree), the greater the frequency shift error is. In comparison, the spectrum after the algorithm modification shows high consistency, and the relative error of each working condition remains at a low level.

[0141] ​It can be known from the above embodiment that the present application can obtain the wave direction information of the current area, the wave height time history information, the significant wave height and the characteristic period of the absolute domain by real-time inversion of the motion data collected by the ship in real time, the present application can accurately invert the wave parameters during the ship navigation process, i.e., under the condition of the navigation speed, the present application is simple and fast in operation, does not occupy too much memory, and can be quickly run on most computers.

[0142] Exemplarily, the verified wave spectrum of the present application adopts the ITTC two-parameter wave energy spectrum, and other wave spectra such as the JONSWAP wave energy spectrum can also be selected for verification.

[0143] In embodiment 2, the present application provides a ship absolute domain wave parameter inversion system under the condition of navigation speed, which comprises:

[0144] A time history data acquisition module is used to acquire the six-degree-of-freedom motion time history data of the ship by the attitude sensor, and to obtain the wave height time history data by the wave buoy or the wave height instrument;

[0145] A data preprocessing module is used to perform segmentation processing on the obtained six-degree-of-freedom motion time history data and wave height time history data;

[0146] A wave inversion data preprocessing module is used to perform wave direction inversion preprocessing and wave height time history inversion preprocessing on the six-degree-of-freedom motion time history data and wave height time history data after segmentation processing;

[0147] A data set division module is used to divide the samples after wave direction inversion preprocessing and wave height time history inversion preprocessing into data sets;

[0148] A model training module is used to use a time convolution neural network to estimate the wave height time history and the wave direction of the data after data set division;

[0149] A Doppler frequency shift correction module is used to input the obtained wave height time history information and wave direction information and the navigation speed information as prior knowledge of a Doppler frequency shift correction method, correct the encounter domain wave spectrum to the absolute domain under the condition of the navigation speed of the ship, and complete the inversion of the wave statistical parameters in the absolute domain;

[0150] A verification module is used to compare and verify the corrected absolute domain spectrum, the original absolute domain spectrum and the uncorrected encounter spectrum.

[0151] The above is only a preferred specific embodiment of the present application, but the protection scope of the present application is not limited thereto, any modification, equivalent replacement and improvement made by any person skilled in the art within the technical range disclosed by the present application, as long as it is within the spirit and principles of the present application, should be covered within the protection scope of the present application.

Claims

1. A method for inverting absolute domain wave parameters under ship speed conditions, characterized in that: The method comprises the following steps: S1, obtains the ship's six-degree-of-freedom motion history data from the attitude sensor, and obtains the wave height history data from the wave buoy or wave height meter; S2, segmenting the acquired six-degree-of-freedom motion time history data and wave height time history data; S3, performing wave direction inversion preprocessing and wave height time history inversion preprocessing on the segmented six-degree-of-freedom motion time history data and wave height time history data; S4, dividing the data set of the samples after wave direction inversion preprocessing and wave height time history inversion preprocessing; S5, using a temporal convolutional neural network to estimate the wave height and direction of the data after the data set is divided; S6, using the obtained wave height history information, wave direction information, and ship speed information as the prior knowledge input of the Doppler shift correction method, correcting the encounter domain wave spectrum to the absolute domain under the condition of the ship's speed, and completing the inversion of wave statistical parameters in the absolute domain; S7, comparing and verifying the obtained corrected absolute domain spectrum, the original absolute domain spectrum, and the uncorrected encounter spectrum.

2. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 1 is characterized in that: In step S1, the six-degree-of-freedom motion history data of the ship is obtained by the attitude sensor as follows: Where, They are Motion posture data of each degree of freedom of the ship under time n is the length of the data, m is the number of the degree of freedom, v is the ship speed, and d is the wave direction angle of the current data; The six degrees of freedom of a ship are: sway, surge, heave, roll, pitch, and bow pitch; The wave height history data obtained from the wave buoy or wave height meter is: IN n =[w1,w2…w n-1 ,In n ] Where w1, w2…w n-1 ,w n Time The motion data under time is the wave height history data corresponding to time.

3. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 2 is characterized in that: In step S2, the obtained six-degree-of-freedom motion history data is segmented, including: Data corresponding to time Right now Divide into data windows with window_size as the window size, and slide on each time series at a fixed interval slip_size; The first set of times is: The first set of ship motion posture data for each degree of freedom is: The second set of times is: The second set of ship motion posture data for each degree of freedom is: And so on, until the end of the time series, the same operation is performed on each degree of freedom; The first set of wave height history data is: [w1,w2…w window_size ], the second set of wave height history data is: [w 1+slip_size ,w 2+slip_size …w window_size+slip_size ].

4. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 1 is characterized in that: In step S3, the wave direction inversion preprocessing includes: performing preprocessing work for wave direction estimation, dividing the segmented For time series segments, fill in the wave direction angle d corresponding to the current data as the classification label; then randomly shuffle the data and labels of each segment to make the continuous wave directions randomly distributed; The pre-processing of wave height history inversion includes: Movement history data and [w1,w2…w window_size ] The wave height history data correspond one to one in time, and a total of several one-to-one corresponding sample points of the specified window size are generated.

5. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 1 is characterized in that: In step S5, the time convolutional neural network is used to estimate the wave height and direction, and the input is the ship motion data The output is [w1,w2…w window_size ] and the wave direction angle d of the data in this section.

6. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 1 is characterized in that: In step S6, the relationship between the absolute domain frequency and the encounter domain frequency is as follows: Where w e is the encounter frequency, w0 is the actual wave frequency, U is the ship speed, g is the acceleration of gravity, β is the wave direction angle, and τ is the intensity of the Doppler shift; The Doppler shift correction method is used to correct the encounter domain wave spectrum to the absolute domain under the condition of ship speed. The inversion of wave statistical parameters in the absolute domain is completed, including: S601, perform frequency domain statistics on the wave height history obtained by time convolutional neural network inversion, and calculate the wave energy spectrum S in the encounter domain. e (w e ), obtain the significant wave height Hse and characteristic period Tse of the encounter domain; S602, the obtained S e (w e ) Press w e Discretize into N e , recorded as: S e (w e (i)),i∈1:N e , when w e When ≤1 / 4τ, the encountered frequency is one-to-three with the actual frequency, and w e (i) Substitute into w in three steps 01 (i),w 02 (i),w 03 (i) In the 01 (i),w 02 (i),w 03 (i) The actual frequencies corresponding to the three-to-one encounter frequencies; S603, the obtained w 01 (i),w 02 (i),w 03 (i) and the significant wave height Hse, characteristic period Tse are respectively brought into S 01 ,S 02 ,S 03 , where w 01 ,S 02 ,S 03 w 01 (i),w 02 (i),w 03 (i) The theoretically corresponding wave energy spectrum value. The ITTC(·) function is an ITTC two-parameter wave energy spectrum function. After obtaining the theoretical absolute domain wave energy spectrum value, the proportions a1, a2, and a3 of each theoretical spectrum value are calculated. S604, w e The one-to-three interval of ≤1 / 4τ is divided into three regions Ⅰ, Ⅱ, and Ⅲ, and the absolute frequency w is calculated for each interval. 0,j and absolute spectrum S 0,j calculate; S605, the corrected absolute frequency w 0,j and absolute spectrum S 0,j Press w 0,j Sort from small to large to obtain the corrected absolute spectrum; S606, using the corrected absolute wave spectrum statistics, calculate the absolute domain wave parameters under the ship's speed, that is, the significant wave height and the characteristic period.

7. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 6 is characterized in that: In step S602, when w e When ≤1 / 4τ, that is, one encounter frequency corresponds to three actual frequencies, w e (i) Bring in three stages 8. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 6 is characterized in that: In step S603, S 01 =ITTC(w 01 (i),Hse,Tse),S 02 =ITTC(w 02 (i),Hse,Tse),S 03 =ITTC(w03(i),Hse,Tse); Where a1, a2, and a3 are the corresponding proportions of the theoretical spectrum values ​​of each segment.

9. The absolute domain wave parameter inversion method under the condition of ship speed according to claim 4 is characterized in that: In step S604, for interval I, w 0,j =w 01 (i),S 0,j =a1·S e (w e (i))·(1-2w 01 (i)·τ), where w 0,j is the jth frequency in the original frequency, S 0,j w 0,j The corresponding actual wave energy spectrum value in the absolute domain; For interval II, w 0,j =w 02 (i),S 0,j =a2·S e (w e (i))·(1-2w 02 (i)·τ); For interval III, w 0,j =w 03 (i),S 0,j =a3·S e (w e (i))·(1-2w 03 (i)·τ); For w e >1 / 4τ, S 0,j =S e (w e (i))·(1-2w 0,j ·τ), where j increases by 1 each time.

10. An absolute domain wave parameter inversion system under ship speed conditions, characterized in that: The system implements the absolute domain wave parameter inversion method under the condition of ship speed as described in any one of claims 1 to 9, and the system includes: The time history data acquisition module is used to obtain the ship's six-degree-of-freedom motion time history data from the attitude sensor, and obtain the wave height time history data from the wave buoy or wave height meter; Data pre-processing module, used for segmenting the acquired six-degree-of-freedom motion time history data and wave height time history data; The wave inversion data preprocessing module is used to perform wave direction inversion preprocessing and wave height time history inversion preprocessing on the six-degree-of-freedom motion time history data and wave height time history data after segmentation processing; The data set division module is used to divide the data sets of samples after wave direction inversion preprocessing and wave height time history inversion preprocessing; The model training module is used to use a temporal convolutional neural network to estimate the wave height and direction of the data after the data set is divided; The Doppler shift correction module is used to input the obtained wave height history information, wave direction information and ship speed information as the prior knowledge of the Doppler shift correction method. Under the condition of the ship's speed, the encounter domain wave spectrum is corrected to the absolute domain, and the wave statistical parameters in the absolute domain are inverted. The verification module is used to compare and verify the obtained corrected absolute domain spectrum, the original absolute domain spectrum, and the uncorrected encounter spectrum.

Images