CN117371303A - Prediction method for effective wave height under sea wave - Google Patents
Prediction method for effective wave height under sea wave Download PDFInfo
- Publication number
- CN117371303A CN117371303A CN202311175859.1A CN202311175859A CN117371303A CN 117371303 A CN117371303 A CN 117371303A CN 202311175859 A CN202311175859 A CN 202311175859A CN 117371303 A CN117371303 A CN 117371303A
- Authority
- CN
- China
- Prior art keywords
- data
- wave height
- model
- effective wave
- sea
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000012360 testing method Methods 0.000 claims abstract description 35
- 238000012549 training Methods 0.000 claims abstract description 34
- 238000013528 artificial neural network Methods 0.000 claims abstract description 16
- 238000012545 processing Methods 0.000 claims abstract description 16
- 238000010606 normalization Methods 0.000 claims abstract description 13
- 230000007246 mechanism Effects 0.000 claims abstract description 10
- 238000011156 evaluation Methods 0.000 claims abstract description 8
- 230000002159 abnormal effect Effects 0.000 claims abstract description 6
- 238000007781 pre-processing Methods 0.000 claims abstract description 6
- 230000006870 function Effects 0.000 claims description 13
- 239000011159 matrix material Substances 0.000 claims description 9
- 238000004364 calculation method Methods 0.000 claims description 8
- 230000004913 activation Effects 0.000 claims description 6
- 238000013135 deep learning Methods 0.000 claims description 6
- 238000013178 mathematical model Methods 0.000 claims description 6
- 238000004140 cleaning Methods 0.000 claims description 4
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 3
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 3
- 230000008859 change Effects 0.000 claims description 3
- 238000009826 distribution Methods 0.000 claims description 3
- 230000008569 process Effects 0.000 claims description 3
- 238000002372 labelling Methods 0.000 claims 1
- 230000000694 effects Effects 0.000 description 13
- 238000010586 diagram Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 7
- 230000003321 amplification Effects 0.000 description 3
- 238000003199 nucleic acid amplification method Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 230000007547 defect Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000000284 extract Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000004088 simulation Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 238000013079 data visualisation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A90/00—Technologies having an indirect contribution to adaptation to climate change
- Y02A90/10—Information and communication technologies [ICT] supporting adaptation to climate change, e.g. for weather forecasting or climate simulation
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Geometry (AREA)
- Computer Hardware Design (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a prediction method of effective wave height under sea waves, which comprises the following steps: based on the historical NOAA buoy data, composing a dataset; preprocessing the data set to screen out missing value and abnormal value data, and dividing the data set into a training set and a testing set; step two: adopting a Tensorflow architecture, calling an LSTM layer from a keras library to build an LSTM model, and then adding an Attention layer to obtain a forecast model combining an LSTM neural network and an Attention mechanism; step three: performing iterative training on the forecasting model by using the training set, and inputting the testing set into the trained forecasting model for testing; verifying the accuracy of the forecasting model according to inverse normalization processing, and taking the forecasting model meeting the standard evaluation of the preset forecasting accuracy as an effective wave height forecasting model of the sea wave for determining the effective wave height of the sea wave to be measured point; the method is used in the field of marine hydrological weather forecast, improves the accuracy of the LSTM model for forecasting the effective wave height, and lays a foundation for researching the intelligent sea wave forecasting system.
Description
Technical Field
The invention relates to the field of marine hydrological weather forecast, in particular to a method for predicting effective wave height under sea waves.
Background
As one of the most important marine phenomena, the research of waves has a crucial meaning for guaranteeing navigation safety, coastal activities and climate systems, and the complex and random nature of waves brings great challenges to coastal and marine research works. Coastal and offshore engineers typically use different methods of field measurements, theoretical studies, and numerical simulations to identify wave climates and extreme wave characteristics, as well as the annual nature of the waves. In the field of navigation and fishery, sea wave forecasting plays an important role in resisting severe sea conditions and guaranteeing operation safety. Waves are mainly described in terms of wave height, wave period, wave direction, etc., where wave height dominates the wave parameters. At present, most of wave height prediction research is based on numerical simulation, but numerical model prediction has the problems of long calculation time, wide range, high precision requirement and the like.
The current commonly used numerical forecasting model is a numerical approximation model driven by a physical rule, and wave forecasting is realized by solving a physical equation through iterative calculation. The main forecasting tool of the intelligent forecasting system is a deep learning forecasting model, which is an intelligent forecasting model driven by big data, and the time-space evolution rule of sea waves is learned from historical stormy wave data by using a deep learning method, so that wave forecasting is realized.
Typically, the effective wave height at each location during wave propagation is affected by some other feature in addition to the wind, but it is necessary to consider how to choose the feature. Therefore, in order to more accurately recognize and master the method for intelligently forecasting the effective wave height of the sea wave based on deep learning, it is necessary to consider the influence of the feature weight on the forecasting, and the forecasting accuracy is improved.
Disclosure of Invention
The invention aims to provide a prediction method for effective wave height under sea waves, which is used for solving the defects of large calculated amount, high cost, incapability of fast prediction, dependence on characteristic engineering and the like of the existing numerical model and realizing fast and accurate prediction with low cost.
In order to achieve the above purpose, the present invention provides the following technical solutions:
a prediction method of effective wave height under sea waves comprises the following steps:
step one: based on the historical NOAA buoy data, composing a dataset; preprocessing the data set to screen out missing value and abnormal value data, and dividing the data set into a training set and a testing set;
step two: adopting a Tensorflow architecture, calling an LSTM layer from a keras library to build an LSTM model, and then adding an Attention layer to obtain a forecast model combining an LSTM neural network and an Attention mechanism;
step three: performing iterative training on the forecasting model by using the training set, and inputting the testing set into the trained forecasting model for testing; verifying the accuracy of the forecasting model according to inverse normalization processing, and taking the forecasting model meeting the standard evaluation of the preset forecasting accuracy as an effective wave height forecasting model of the sea wave;
step four: and after the buoy data of the sea wave to-be-measured point are input to the sea wave effective wave height forecasting model for outputting, determining the effective wave height of the sea wave to-be-measured point.
Further, the data set in the first step specifically includes wind direction, average wind speed, peak wind speed, main waveguide period, average waveguide period, waveguide direction, sea level pressure, air temperature, sea surface temperature and effective waveguide height.
Further, the input data of the training set and the testing set in the third step respectively comprise wind direction, average wind speed, peak wind speed, main wave period, average wave period, wave direction, sea level pressure, air temperature and sea surface temperature; the output data of the training set and the test set respectively comprise effective wave heights.
Furthermore, in the first step, the data set is specifically divided into a training set and a testing set according to the year data in a 3:1 manner based on the time sequence.
Further, the specific training method of the effective wave height forecasting model of the sea wave comprises the following steps:
inputting the input data of the training set into an LSTM neural network, predicting according to the change rule of the data at the previous moment, and outputting prediction information;
and inputting the LSTM prediction information into the added attribute layer, distributing probability weight to the output information of the LSTM neural network by the attribute layer, and improving generalization of the effective wave height prediction model by dynamically adjusting learning rate.
Further, the mathematical model of the LSTM neural network is:
I t ＝σ(X t W xi +H t-1 W hi +b i )
F t ＝σ(X t W xf +H t-1 W hf +b f )；
O t ＝σ(X t W xo +H t-1 W ho +b o )；
wherein: i t ，F t ，O t Respectively representing an input door, a forget door and an output door; x is X t Representing input data W ij A weight matrix representing output gates, H t-1 Representing the hidden state of the previous time step, b i Representing the bias term of the output gate, sigma represents the activation function,C t representing candidate memory cells and memory cells, respectively, wherein +.>Representing the multiplication of matrix elements.
Further, the Attention mechanism mathematical model is:
e t ＝σ(W e x t +b e )
wherein: e, e t ，W e ，b e Sigma respectively represents a weight coefficient combination, a trainable weight matrix, a bias vector and an activation function corresponding to each input data at the current moment;
normalizing the attention weight coefficients through a SOfimax function to obtain attention weights, wherein alpha m，t The attention weight value for the mth feature is expressed as:
will input the feature vector x t Recalculate as a weightVector, expressed as:
further, the normalization processing calculation formula is as follows:
wherein X is * Is normalized data, X is raw data,is the mean of the raw data, and delta is the standard deviation of the raw data. The processed characteristic data accords with standard normal distribution with the mean value of 0 and the standard deviation of 1.
Further, the method for verifying the accuracy of the forecasting model specifically comprises the following steps:
inputting the test set into the trained forecasting model, and obtaining the forecasting effective wave height through inverse normalization processing;
comparing the forecast wave height with the effective wave height in the test set, and calculating root mean square error, average absolute percentage error and fitting goodness as evaluation indexes to verify model accuracy; the calculation formula is as follows:
wherein,representing a deep learning forecast result; y represents the effective wave height in the NOAA dataset; />Representing the effective wave height average value in the NOAA data set; n represents the number of observations; RMSE represents root mean square error; MAE represents the mean absolute error; MAPE represents absolute percent error; r is R 2 Indicating the goodness of fit.
Compared with the prior art, the technical scheme provided by the invention has the following technical effects:
the invention provides a prediction method of effective wave height under sea waves, which is based on historical NOAA buoy data to form a data set; preprocessing the data set to screen out missing value and abnormal value data, and dividing the data set into a training set and a testing set; adopting a Tensorflow architecture, calling an LSTM layer from a keras library to build an LSTM model, and then adding an Attention layer to obtain a forecast model combining an LSTM neural network and an Attention mechanism; inputting a training set to iteratively train the forecasting model, and inputting the testing set to the trained forecasting model for testing; verifying the accuracy of the forecasting model according to inverse normalization processing, and taking the forecasting model meeting the standard evaluation of the preset forecasting accuracy as an effective wave height forecasting model of the sea wave; the prediction method of the effective wave height can be used for obtaining:
(1) The invention is suitable for predicting the effective wave height data of sea waves, and the designed prediction model adopts a network structure combining an LSTM neural network and an Attention mechanism, and has the advantages that: LSTM fully extracts the characteristics of time sequence data information, improves the model prediction precision, and is suitable for the prediction of time sequence data; the Attention mechanism can give weight to the prediction result of the LSTM, and extract an important part in the prediction result.
(2) The method solves the defects of large calculated amount, high cost, incapability of fast prediction, dependence on characteristic engineering and the like of the existing numerical model, and realizes fast and accurate prediction with low cost.
Drawings
For a clearer description of embodiments of the invention or of the prior art, the drawings that are necessary for the embodiments will be briefly described, it will be apparent that the drawings in the following description are some embodiments of the invention, and that other drawings can be obtained from these drawings, without inventive faculty, for a person skilled in the art
FIG. 1 shows the implementation steps of a method for predicting effective wave height under sea waves;
FIG. 2 is a schematic diagram of the specific flow chart of FIG. 1;
FIG. 3 is a schematic diagram of the structure of an LSTM-Attention model of the effective wave height prediction model of sea waves;
FIG. 4 is an error analysis index of a second embodiment;
FIG. 5 is a graph showing the comparison of the effect of forecasting the effective wave height in the second embodiment.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Embodiment one:
referring to fig. 1 to 3, the invention provides a method for predicting effective wave height under sea waves, which comprises the following steps:
step one: based on the historical NOAA buoy data, composing a dataset; and preprocessing the data set to screen out the missing value and abnormal value data, and dividing the data set into a training set and a testing set. The data sets include wind direction, average wind speed, peak wind speed, main waveguide period, average waveguide period, wave direction, sea level pressure, air temperature, sea surface temperature, and effective wave height. The wind direction, average wind speed, peak wind speed, main wave period, average wave period, wave direction, sea level pressure, air temperature and sea level temperature data in the training set and the testing set are respectively used as input data of the training set and input data of the testing set. The effective wave heights in the training set and the test set are respectively used as output data.
When the data set is preprocessed to screen out missing value and abnormal value data, data marking, data cleaning, normalization and feature processing are needed. The data sets are divided into training sets and test sets in a 3:1 ratio of year data based on time series. I.e., data training as 2019-2021, data testing as 2022.
The normalization processing calculation formula is as follows:
wherein X is * Is normalized data, X is raw data,is the mean of the raw data, and delta is the standard deviation of the raw data. The processed characteristic data accords with standard normal distribution with the mean value of 0 and the standard deviation of 1.
Step two: and constructing a basic forecasting model. And calling the LSTM layer from the keras library to build an LSTM model by adopting a Tensorflow architecture, and then adding an Attention layer to obtain a prediction model combining the LSTM neural network and an Attention mechanism.
Step three: performing iterative training on the forecast model constructed in the second step by using the training set in the first step, and inputting the test set in the first step into the trained forecast model for testing; and verifying the accuracy of the forecasting model according to inverse normalization processing, and taking the forecasting model meeting the standard evaluation of the preset forecasting accuracy as an effective wave height forecasting model of the sea wave.
The specific training method of the sea wave effective wave height forecasting model comprises the following steps:
inputting the input data of the training set into an LSTM neural network, predicting according to the change rule of the data at the previous moment, and outputting LSTM prediction information; and in the LSTM neural network prediction process, the characteristics of time sequence data information are fully extracted.
The mathematical model of LSTM neural network is:
I t ＝σ(X t W xi +H t-1 W hi +b i )
F t ＝σ(X t W xf +H t-1 W hf +b f )；
O t ＝σ(X t W xo +H t-1 W ho +b o )；
wherein: i t ，F t ，O t Respectively representing an input door, a forget door and an output door; x is X t Representing input data W ij A weight matrix representing output gates, H t-1 Representing the hidden state of the previous time step, b i Representing the bias term of the output gate, sigma represents the activation function,C t representing candidate memory cells and memory cells, respectively, wherein +.>Representing the multiplication of matrix elements.
And (3.2) inputting the LSTM prediction information into the added attribute layer, distributing probability weight to the output information of the LSTM neural network by the attribute layer, and improving generalization of the ocean wave effective wave height prediction model by adopting dynamic adjustment of learning rate.
The Attention mechanism mathematical model is:
e t ＝σ(W e x t +b e )
wherein: e, e t ，W e ，b e Sigma respectively represents weight coefficient combination and trainable corresponding to each input data at the current momentTraining a weight matrix, a bias vector and an activation function;
normalizing each attention weight coefficient through softmax function to obtain attention weight, wherein alpha m,t The attention weight value for the mth feature is expressed as:
will input the feature vector x t Recalculate as a weightVector, expressed as:
in the third step, the method for verifying the accuracy of the forecast model specifically comprises the following steps:
(1) Inputting the test set into a trained forecasting model, and obtaining the forecasting effective wave height through inverse normalization processing.
(2) Comparing the forecast wave height with the effective wave height in the test set, and calculating root mean square error, average absolute percentage error and fitting goodness as evaluation indexes to verify model accuracy; the calculation formula is as follows:
wherein,representing a deep learning forecast result; y represents the effective wave height in the NOAA dataset; />Representing the effective wave height average value in the NOAA data set; n represents the number of observations; RMSE represents root mean square error; MAE represents the mean absolute error; MAPE represents absolute percent error; r is R 2 Indicating the goodness of fit.
Example two
Referring to fig. 4 and 5, in this embodiment, the prediction method of the effective wave height under the ocean wave in the first embodiment is used, buoy data provided by NOAA is used as an experimental data set (https:// www.ndbc.noaa.gov), data buoys 44013 and 44014 are selected, and geographical positions of the two buoys are shown in the following table:
| sequence number | Site name | Longitude and latitude | Latitude of latitude |
| 1 | 44013 | 70.651°W | 42.346°N |
| 2 | 44014 | 74.842°W | 36.609°N |
The spatial resolution is 0.5 degrees multiplied by 0.5 degrees, and data of the buoys 2019-2022 of 44013 and 44014 are selected for carrying out effective wave height prediction experiments. The dataset in the experiment was partitioned as shown in the following table:
| sequence number | Site name | Training set length | Test set length |
| 1 | 44013 | 2019, 1 st to 2021, 12 nd 31 st | 2022, 1/12/31 |
| 2 | 44014 | 2019, 1 st to 2021, 12 nd 31 st | 2022, 1 month 1 day to 10 months 3 days |
44013. 44014 site training set is 2019 1, 0 to 2021, 12, 31, 23, 44013 site test set is 2022, 1, 0 to 12, 31, 23; since the 44014 site was essentially totally missing after day 16 of 10, 2022, the test set span for 44014 site was 2022, 1, 0 to 10, 3, 16. Because the data of the NOAA buoy station has a certain period of loss, the data set of the buoy station is firstly subjected to data cleaning, and then model training and verification are carried out. After the cleaning, the data is normalized and then is input into the effective wave height forecasting model of the sea wave.
In order to further show the prediction effect of the effective wave height prediction model of the sea wave, the same data set is respectively input into the existing LSTM model and the applied effective wave height prediction model of the sea wave, and the effect comparison of the two models is carried out. The effective wave height forecasting model of the sea wave in the application is marked as LSTM-attribute (because of the length problem in the chart, the effective wave height forecasting model is abbreviated as LSTM-ATT in the chart); model accuracy verification is carried out on the predicted results of the two models, root mean square error, average absolute percentage error and fitting goodness are calculated, and as shown in the result figure 4, the prediction conditions of the applied wave effective wave height prediction model, namely LSTM-Attention at 44013 and 44014 sites, are better than those of the existing LSTM model, wherein the root mean square error can be reduced by 10.72%, the absolute percentage error can be reduced by 7.58%, the average absolute error can be reduced by 8.05%, and the fitting goodness can be improved by 5.91%. In order to intuitively see the difference between the two models, data visualization drawing is performed, as shown in fig. 4, in order to better compare the prediction effects of the two models and select a certain period of time for amplification, the first row is a comparison graph of the prediction effects of the two models of two stations, the second row is an amplification graph of the selected period of time, in fig. 5 (a), the prediction effect of the station 44013 is shown, and in combination with the amplification graph of fig. 5 (a), the effect of LSTM-Attention fitting is better, and the existing LSTM model shows the effect that the prediction value is smaller than the true value; FIG. 5 (b) shows the predicted effect of 44014 sites, and the same enlarged view can be combined to see that the LSTM-Attention fitting is better, and the existing LSTM model shows the effect that the predicted value is larger than the true value. In conclusion, the applied wave effective wave height prediction model shows better effect than the existing LSTM model.
It will be appreciated by those skilled in the art that embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (9)
1. A method of predicting effective wave height under ocean waves, comprising:
step one: based on the historical NOAA buoy data, composing a dataset; preprocessing the data set to screen out missing value and abnormal value data, and dividing the data set into a training set and a testing set;
step two: adopting a Tensorflow architecture, calling an LSTM layer from a keras library to build an LSTM model, and then adding an Attention layer to obtain a forecast model combining an LSTM neural network and an Attention mechanism;
step three: performing iterative training on the forecasting model by using the training set, and inputting the testing set into the trained forecasting model for testing; verifying the accuracy of the forecasting model according to inverse normalization processing, and taking the forecasting model meeting the standard evaluation of the preset forecasting accuracy as an effective wave height forecasting model of the sea wave;
step four: and after the buoy data of the sea wave to-be-measured point are input to the sea wave effective wave height forecasting model for outputting, determining the effective wave height of the sea wave to-be-measured point.
2. A method of predicting effective wave height under ocean waves as claimed in claim 1, wherein the data set comprises wind direction, average wind speed, peak wind speed, main waveguide period, average waveguide period, wave direction, sea level pressure, air temperature, sea surface temperature and effective wave height; the input data of the training set and the testing set respectively comprise wind direction, average wind speed, peak wind speed, main wave guide period, average wave guide period, wave direction, sea level pressure, air temperature and sea surface temperature; the output data of the training set and the test set respectively comprise effective wave heights.
3. A method for predicting effective wave height under sea waves according to claim 1, wherein the preprocessing of the data set in the step one specifically comprises: data labeling, data cleaning, normalization and feature processing.
4. A method of predicting effective wave height under ocean waves as claimed in claim 1, wherein the dataset is based on a time series and divided into the training set and test set in a 3:1 ratio according to year data.
5. The method for predicting the effective wave height under the sea wave according to claim 2, wherein the specific training method of the effective wave height prediction model of the sea wave comprises the following steps:
inputting the input data of the training set into an LSTM neural network, predicting according to the change rule of the data at the previous moment, and outputting LSTM prediction information;
and inputting the LSTM prediction information into the added attribute layer, distributing probability weight to the output information of the LSTM neural network by the attribute layer, and improving generalization of the effective wave height prediction model by dynamically adjusting learning rate.
6. A method of predicting effective wave height under ocean waves as claimed in claim 2, wherein the mathematical model of the LSTM neural network is:
I t ＝σ(X t W xi +H t-1 W hi +b i )
F t ＝σ(X t W xf +H t-1 W hf +b f )；
O t ＝σ(X t W xo +H t-1 W ho +b o )；
wherein: i t ，F t ，O t Respectively representing an input door, a forget door and an output door; x is X t Representing input data W ij A weight matrix representing output gates, H t-1 Representing the hidden state of the previous time step, b i Representing the bias term of the output gate, sigma represents the activation function,C t representing candidate memory cells and memory cells, respectively, wherein +.>Representing the multiplication of matrix elements.
7. The method for predicting the effective wave height under the sea waves according to claim 2, wherein the Attention mechanism mathematical model is:
e t ＝σ(W e x t +b e )
wherein: e, e t ，W e ，b e Sigma respectively represents a weight coefficient combination, a trainable weight matrix, a bias vector and an activation function corresponding to each input data at the current moment;
normalizing each attention weight coefficient through softmax function to obtain attention weight, wherein alpha m,t The attention weight value for the mth feature is expressed as:
will input the feature vector x t Recalculate as a weightVector, expressed as:
8. a method of predicting effective wave height under ocean waves according to claim 3, wherein the normalization process calculation formula is:
wherein X is * Is normalized data, X is raw data,is the mean of the raw data, and delta is the standard deviation of the raw data. The processed characteristic data accords with standard normal distribution with the mean value of 0 and the standard deviation of 1.
9. The method for predicting the effective wave height under the sea waves according to claim 1, wherein the method for verifying the accuracy of the prediction model in the third step specifically comprises the following steps:
inputting the test set into the trained forecasting model, and obtaining the forecasting effective wave height through inverse normalization processing;
comparing the forecast wave height with the effective wave height in the test set, and calculating root mean square error, average absolute percentage error and fitting goodness as evaluation indexes to verify model accuracy;
the calculation formula is as follows:
wherein,representing a deep learning forecast result; y represents the effective wave height in the NOAA dataset; />Representing the effective wave height average value in the NOAA data set; n represents the number of observations; RMSE represents root mean square error; MAE represents the mean absolute error; MAPE represents absolute percent error; r is R 2 Indicating the goodness of fit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311175859.1A CN117371303A (en) | 2023-09-13 | 2023-09-13 | Prediction method for effective wave height under sea wave |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311175859.1A CN117371303A (en) | 2023-09-13 | 2023-09-13 | Prediction method for effective wave height under sea wave |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117371303A true CN117371303A (en) | 2024-01-09 |
Family
ID=89391878
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311175859.1A Pending CN117371303A (en) | 2023-09-13 | 2023-09-13 | Prediction method for effective wave height under sea wave |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117371303A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117471575A (en) * | 2023-12-28 | 2024-01-30 | 河海大学 | Typhoon wave height forecasting method based on BO-LSTM neural network model |
-
2023
- 2023-09-13 CN CN202311175859.1A patent/CN117371303A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117471575A (en) * | 2023-12-28 | 2024-01-30 | 河海大学 | Typhoon wave height forecasting method based on BO-LSTM neural network model |
| CN117471575B (en) * | 2023-12-28 | 2024-03-08 | 河海大学 | Typhoon wave height forecasting method based on BO-LSTM neural network model |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110942194A (en) | Wind power prediction error interval evaluation method based on TCN | |
| CN113704693B (en) | High-precision effective wave height data estimation method | |
| CN107480781A (en) | The nuclear accident Source Term Inversion method of neutral net adaptive Kalman filter | |
| CN111797573A (en) | Ionized layer electron concentration total content time sequence prediction method based on deep learning technology | |
| MacPherson et al. | A stochastic extreme sea level model for the German Baltic Sea coast | |
| CN114493052B (en) | Multi-model fusion self-adaptive new energy power prediction method and system | |
| CN117371303A (en) | Prediction method for effective wave height under sea wave | |
| CN114912077B (en) | Sea wave forecasting method integrating random search and mixed decomposition error correction | |
| Lewis et al. | Bay of Bengal cyclone extreme water level estimate uncertainty | |
| Jung et al. | Modelling monthly near‐surface maximum daily gust speed distributions in Southwest Germany | |
| CN116307291B (en) | Distributed photovoltaic power generation prediction method and prediction terminal based on wavelet decomposition | |
| Yang et al. | A rapid forecasting and mapping system of storm surge and coastal flooding | |
| CN116933621A (en) | Urban waterlogging simulation method based on terrain feature deep learning | |
| CN115758876A (en) | Method, system and computer equipment for forecasting accuracy of wind speed and wind direction | |
| CN116029419A (en) | Deep learning-based long-term new energy daily average generation power prediction method and system | |
| Qiu et al. | Selection optimal method of evaporation duct model based on sensitivity analysis | |
| CN110852415B (en) | Vegetation index prediction method, system and equipment based on neural network algorithm | |
| CN117114190A (en) | River runoff prediction method and device based on mixed deep learning | |
| CN117113828A (en) | Numerical forecast correction method based on ship-based navigation observation | |
| CN116757321A (en) | Solar direct radiation quantity prediction method, system, equipment and storage medium | |
| CN116796649A (en) | SPEI coarse resolution data space downscaling method and device based on machine learning | |
| Hadihardaja et al. | Decision support system for predicting tsunami characteristics along coastline areas based on database modelling development | |
| CN115062526B (en) | Three-dimensional ionosphere electron concentration distribution model training method based on deep learning | |
| CN116467933A (en) | Storm surge water increasing prediction method and system based on deep learning | |
| Chang et al. | Neural network with multi-trend simulating transfer function for forecasting typhoon wave |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |