Typhoon azimuth estimation method, system and terminal based on single vector hydrophone

Abstract

The application discloses a typhoon direction estimation method and system based on a single vector hydrophone and a terminal, and relates to the field of typhoon direction estimation.The method comprises the following steps: strictly checking sound pressure and three-directional vibration velocity data collected by the single vector hydrophone to ensure data quality; then, pre-processing the data, including filtering to remove noise interference; then, normalizing the filtered data; then, calculating sound intensity in each direction within a certain azimuth and pitch search range, constructing a sound intensity matrix, and finding the azimuth and pitch corresponding to the maximum sound intensity as the real reference value; dividing the processed data into a training set and a test set according to a certain proportion; constructing a multilayer perception machine model; and through experiments of different learning rates and small batch sizes, optimizing the model, selecting the parameters with the minimum loss to train the best model; finally, using the trained model to predict the test set to obtain the estimated value of the typhoon azimuth and pitch angle.The application effectively improves the accuracy of typhoon direction estimation.

Description

Technical Field

[0001] This invention relates to the field of marine observation technology, and in particular to a method, system, terminal, and computer-readable storage medium for estimating typhoon azimuth based on a single-vector hydrophone. Background Technology

[0002] Typhoons are highly destructive natural disasters, and accurate monitoring is crucial for disaster prevention and mitigation. Traditional monitoring methods such as satellite remote sensing, weather radar, and buoy observation are widely used, but they have significant drawbacks: satellite remote sensing suffers from cloud cover and sea surface reflection, resulting in low data accuracy and high latency; weather radar has limited coverage (only a few hundred kilometers), is susceptible to terrain interference, and has high equipment deployment costs; while acoustic monitoring technology based on traditional hydrophones can capture low-frequency typhoon noise, it relies solely on sound pressure as a single parameter, lacking vibration velocity information fusion, leading to insufficient accuracy in azimuth estimation. Furthermore, traditional algorithms rely on complex signal processing, resulting in low computational efficiency and difficulty meeting real-time requirements. Single-vector hydrophones, capable of simultaneously acquiring sound pressure and three-dimensional vibration velocity information, have become an important tool in the field of marine monitoring, and their multi-parameter characteristics provide a new approach to typhoon azimuth estimation.

[0003] However, current typhoon location estimation techniques based on single-vector hydrophones are mostly limited to traditional physical models and simple sound intensity calculations, failing to fully utilize deep learning technology to explore the potential correlations of multi-parameter data. This results in problems such as single data dimension, poor algorithm real-time performance, and inability to adapt to the monitoring needs of dynamic typhoon changes.

[0004] Therefore, existing technologies still need to be improved and developed. Summary of the Invention

[0005] The main objective of this invention is to provide a typhoon azimuth estimation method, system, terminal, and computer-readable storage medium based on a single-vector hydrophone. This invention aims to address the problems in existing typhoon azimuth estimation technologies based on single-vector hydrophones, which fail to fully utilize deep learning techniques to uncover potential correlations in multi-parameter data, resulting in limited data dimensions, poor real-time performance of the algorithm, and inability to adapt to the monitoring needs of dynamic typhoon changes.

[0006] To achieve the above objectives, this invention provides a typhoon azimuth estimation method based on a single-vector hydrophone, which includes the following steps:

[0007] The target single-vector hydrophone is placed at a predetermined depth below the sea surface, and sound pressure data and particle velocity data are collected through the target single-vector hydrophone.

[0008] The sound pressure data and the particle velocity data are subjected to data inspection, filtering and normalization to obtain the target sound pressure data and the target particle velocity data.

[0009] Within the search range of the preset direction, the true direction of the typhoon is calculated based on the target sound pressure data and the target particle velocity data;

[0010] The target sound pressure data and the target particle velocity data are used as sample data, and the real location corresponding to the target sound pressure data and the target particle velocity data is used as a label. A dataset is constructed based on the sample data and the label.

[0011] Construct a typhoon orientation estimation model, and use the dataset to train and evaluate the typhoon orientation estimation model to obtain the target model;

[0012] The predicted sound pressure data and predicted particle velocity data are collected by the single-vector hydrophone, and then input into the target model to obtain the estimated typhoon location.

[0013] Optionally, the typhoon azimuth estimation method based on a single-vector hydrophone, wherein the step of deploying the target single-vector hydrophone at a predetermined depth below the sea surface and collecting sound pressure data and particle velocity data through the target single-vector hydrophone specifically includes:

[0014] Select a single-vector hydrophone whose sensitivity, frequency response range, working depth, and directionality meet the preset standards as the target single-vector hydrophone;

[0015] The target single-vector hydrophone is placed at a predetermined depth below the sea surface, and sound pressure data and particle velocity data in three orthogonal directions are collected through the target single-vector hydrophone.

[0016] Optionally, the typhoon azimuth estimation method based on a single-vector hydrophone, wherein the step of performing data checking, filtering, and normalization on the sound pressure data and the particle velocity data to obtain the target sound pressure data and target particle velocity data specifically includes:

[0017] Check whether the sound pressure data and the particle velocity data are vectors and whether the vector lengths are consistent. If not, determine that the sound pressure data and the particle velocity data were acquired incorrectly and discard the data.

[0018] If so, a fourth-order Butterworth filter is used to filter the sound pressure data and the particle velocity data to obtain filtered data. The filtered data is then mapped to a preset range and normalized to obtain the target sound pressure data and target particle velocity data.

[0019] Optionally, in the typhoon azimuth estimation method based on a single-vector hydrophone, the true azimuth includes the true azimuth angle and the true pitch angle.

[0020] Within the preset search range, the calculation of the typhoon's true location based on the target sound pressure data and the target particle velocity data specifically includes:

[0021] Within the search range of the first preset azimuth angle and the second preset pitch angle, multiple sound intensities in each direction are calculated based on the target sound pressure data and the target particle vibration velocity data.

[0022] A sound intensity matrix is ​​constructed based on the multiple sound intensities, and the true azimuth and true pitch angle of the typhoon are determined based on the sound intensity matrix.

[0023] Optionally, the typhoon azimuth estimation method based on a single-vector hydrophone, wherein constructing the dataset based on the sample data and the labels specifically includes:

[0024] A dataset is constructed based on the sample data and the corresponding labels of the sample data, and the dataset is divided into a training set and a test set according to a preset ratio;

[0025] The training set is used for training the model and tuning its parameters, while the test set is used to evaluate the model's performance.

[0026] Optionally, the typhoon azimuth estimation method based on a single-vector hydrophone, wherein the step of constructing a typhoon azimuth estimation model and training and evaluating the typhoon azimuth estimation model using the dataset to obtain a target model specifically includes:

[0027] A multilayer perceptron model is constructed, which includes an input layer, a hidden layer, and a regression output layer. The hidden layer contains multiple neurons, and each neuron processes the data through a weighted summation and a nonlinear activation function.

[0028] Increase the number of hidden layers and the number of neurons to obtain a typhoon orientation estimation model. Use the training set to train the typhoon orientation estimation model and calculate the loss of the typhoon orientation estimation model on the training set.

[0029] The parameter combination with the minimum loss is selected as the optimal parameter. The typhoon orientation estimation model is then trained using the optimal parameter until the typhoon orientation estimation model converges or reaches the preset number of training rounds, thus obtaining the trained model.

[0030] The trained model is evaluated using the test set to obtain the target model.

[0031] Optionally, the typhoon azimuth estimation method based on a single-vector hydrophone, wherein inputting the sound pressure data to be predicted and the particle velocity data to be predicted into the target model to obtain the estimated typhoon azimuth specifically includes:

[0032] The sound pressure data to be predicted and the particle velocity data to be predicted are input into the input layer of the target model for transformation to obtain feature vectors;

[0033] The feature vector is input into the hidden layer of the target model for feature extraction and summation to obtain fused features. The fused features are then mapped using a non-linear activation function to obtain the target features.

[0034] The target features are input into the regression output layer of the target model for prediction, and the estimated typhoon location is output.

[0035] Furthermore, to achieve the above objectives, the present invention also provides a typhoon azimuth estimation system based on a single-vector hydrophone, wherein the typhoon azimuth estimation system based on a single-vector hydrophone includes:

[0036] The data acquisition module is used to place the target single-vector hydrophone at a preset depth below the sea surface and acquire sound pressure data and particle velocity data through the target single-vector hydrophone.

[0037] The data preprocessing module is used to perform data inspection, filtering and normalization on the sound pressure data and the particle velocity data to obtain target sound pressure data and target particle velocity data.

[0038] The true location calculation module is used to calculate the true location of the typhoon within a preset search range, based on the target sound pressure data and the target particle velocity data.

[0039] The dataset construction module is used to use the target sound pressure data and the target particle velocity data as sample data, and the real location corresponding to the target sound pressure data and the target particle velocity data as labels, and construct a dataset based on the sample data and the labels;

[0040] The model training and evaluation module is used to construct a typhoon orientation estimation model, and to train and evaluate the typhoon orientation estimation model using the dataset to obtain the target model.

[0041] The typhoon azimuth estimation module is used to collect the sound pressure data and particle velocity data to be predicted through the single-vector hydrophone, and input the sound pressure data and particle velocity data to be predicted into the target model to obtain the estimated typhoon azimuth.

[0042] In addition, to achieve the above objectives, the present invention also provides a terminal, wherein the terminal includes: a memory, a processor, and a typhoon azimuth estimation program based on a single-vector hydrophone stored in the memory and executable on the processor. When the typhoon azimuth estimation program based on a single-vector hydrophone is executed by the processor, it implements the steps of the typhoon azimuth estimation method based on a single-vector hydrophone as described above.

[0043] Furthermore, to achieve the above objectives, the present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a typhoon azimuth estimation program based on a single-vector hydrophone, and when the typhoon azimuth estimation program based on a single-vector hydrophone is executed by a processor, it implements the steps of the typhoon azimuth estimation method based on a single-vector hydrophone as described above.

[0044] In this invention, the sound pressure and three-directional vibration velocity data collected by a single-vector hydrophone are rigorously checked to ensure data quality. Next, the data undergoes preprocessing, including filtering to remove noise interference, followed by normalization. Then, within a certain azimuth and elevation angle search range, the sound intensity in each direction is calculated to construct a sound intensity matrix. The azimuth and elevation angles corresponding to the maximum sound intensity are found as true reference values. The processed data is proportionally divided into training and testing sets, and a multilayer perceptron model is constructed. The model is optimized by experimenting with different learning rates and mini-batch sizes, selecting the parameters with the least loss to train the optimal model. Finally, the trained model is used to predict the test set, obtaining estimated values ​​for the typhoon azimuth and elevation angles. This invention effectively improves the accuracy of typhoon azimuth estimation. Attached Figure Description

[0045] Figure 1 This is a flowchart of a preferred embodiment of the typhoon azimuth estimation method based on a single-vector hydrophone of the present invention;

[0046] Figure 2 This is a schematic diagram of the data obtained after filtering in the typhoon azimuth estimation method based on a single-vector hydrophone of the present invention;

[0047] Figure 3 This is a schematic diagram illustrating the variation of sound intensity with azimuth angle in the typhoon azimuth estimation method based on a single-vector hydrophone of the present invention.

[0048] Figure 4 This is a loss curve diagram of model training in the typhoon azimuth estimation method based on a single-vector hydrophone of the present invention;

[0049] Figure 5 This is a structural diagram of a preferred embodiment of the typhoon azimuth estimation system based on a single-vector hydrophone of the present invention;

[0050] Figure 6This is a structural diagram of a preferred embodiment of the terminal of the present invention. Detailed Implementation

[0051] This application provides a typhoon azimuth estimation method, system, and terminal based on a single-vector hydrophone. To make the objectives, technical solutions, and effects of this application clearer and more explicit, the following detailed description is provided with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only for explaining this application and are not intended to limit this application.

[0052] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0053] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.

[0054] The preferred embodiment of the typhoon azimuth estimation method based on a single-vector hydrophone described in this invention, such as... Figure 1 As shown, the typhoon azimuth estimation method based on a single-vector hydrophone includes the following steps:

[0055] Step S10: Place the target single-vector hydrophone at a preset depth below the sea surface, and collect sound pressure data and particle velocity data through the target single-vector hydrophone.

[0056] Specifically, a single-vector hydrophone whose sensitivity, frequency response range, working depth, and directionality meet preset standards is selected as the target single-vector hydrophone; the target single-vector hydrophone is placed at a preset depth below the sea surface, and sound pressure data and particle velocity data in three orthogonal directions are collected through the target single-vector hydrophone.

[0057] Understandably, selecting a suitable single-vector hydrophone requires considering its performance indicators such as sensitivity, frequency response range, operating depth, and directivity. For typhoon monitoring, a vector hydrophone with a wide low-frequency response range (e.g., 0.1Hz-1kHz) is generally chosen as the target single-vector hydrophone (because the marine environmental noise generated by typhoons is mainly concentrated in the low-frequency band). This target single-vector hydrophone is deployed at a predetermined depth below the sea surface, such as 10-50 meters, to avoid interference from direct impacts from surface waves, while ensuring effective capture of the underwater acoustic signal propagation characteristics caused by the typhoon. The installation of the single-vector hydrophone should ensure its stability and reliability, avoiding any impact on the accuracy of signal acquisition due to changes in water flow or the marine environment.

[0058] It should be noted that, besides data acquisition based on single-vector hydrophones, other types of sensor combinations can also be considered in other implementations of this application. For example, multiple distributed hydrophone arrays can be used, and similar acoustic information can be obtained through signal fusion processing. Distributed hydrophone arrays can cover a larger area and acquire richer spatial information, potentially leading to more accurate estimations of typhoon azimuth and pitch angles. However, this approach requires more complex signal synchronization and fusion techniques, and the cost is relatively high.

[0059] The target single-vector hydrophone collects sound pressure data and particle velocity data. The single-vector hydrophone continuously collects sound signals within time t, and simultaneously measures the sound pressure signal p(t) and the particle velocity signal v. x (t), v y (t), v z (t). The sound pressure signal p(t) reflects the pressure fluctuation in the medium, and the particle velocity signal v x (t), v y (t), v z (t) represents the velocity of the medium particles in three orthogonal directions, typically the horizontal x-axis, y-axis, and vertical z-axis. These signals provide multi-dimensional data for subsequent analysis.

[0060] In addition to combining the aforementioned acoustic data for estimation, other implementations of this application can also integrate other types of data, such as optical and meteorological data. For example, macroscopic location information of the typhoon can be obtained using satellite optical images, and then combined with acoustic data from a single-vector hydrophone for refined azimuth and elevation angle estimation. Furthermore, this application can also perform azimuth estimation without relying on sound pressure and vibration velocity data characteristics. Instead, it can perform time-frequency analysis on the acoustic signal to extract spectral features or time-frequency distribution features, and use these features for model training and azimuth and elevation angle estimation, thus achieving the purpose of typhoon azimuth estimation.

[0061] Step S20: Perform data inspection, filtering and normalization on the sound pressure data and the particle velocity data to obtain target sound pressure data and target particle velocity data.

[0062] In this embodiment, it is checked whether the sound pressure data and the particle velocity data are vectors and whether the vector lengths are consistent. If not, it is determined that the sound pressure data and the particle velocity data were acquired incorrectly and the error is discarded.

[0063] Understandably, the input sound pressure and vibration velocity data are checked before data processing to ensure that the data are vectors with consistent length. This technical feature can avoid calculation errors caused by incorrect data format or inconsistent length, thus improving the stability and reliability of the system.

[0064] If so, a fourth-order Butterworth filter is used to filter the sound pressure data and the particle velocity data to obtain filtered data. The filtered data is then normalized by mapping it to a preset range to obtain the target sound pressure data and target particle velocity data (e.g., ...). Figure 2 (As shown).

[0065] In this embodiment, a fourth-order Butterworth filter is used to filter the data before processing. The Butterworth filter has flat passband and stopband characteristics, effectively removing noise interference and making the data smoother. Filtering improves data quality and reduces the impact of noise on subsequent sound intensity calculations and model training, thereby improving estimation accuracy. Finally, the filtered data is normalized, mapping it to a specific range. Normalization eliminates dimensional differences between different dimensions of data, making the model more likely to converge and improving training efficiency and generalization ability.

[0066] It is understandable that, in addition to the fourth-order Butterworth filter, other types of filters, such as Chebyshev filters or Elliptic filters, can also be used in this application. Chebyshev filters have steeper attenuation characteristics in the passband or stopband, and can more effectively suppress noise at specific frequencies; Elliptic filters have faster attenuation rates in both the passband and stopband, but may introduce some phase distortion. Different types of filters differ in performance such as passband ripple and stopband attenuation, and may be more suitable for certain specific marine environments or typhoon signal characteristics. Changing the filter type may bring better signal processing results depending on different noise characteristics and application scenarios. Adaptive filtering technology is used to dynamically adjust filter parameters according to real-time signal noise levels and characteristics to adapt to the noise characteristics of different stages of typhoons or different marine environments, making signal preprocessing more flexible and intelligent.

[0067] Step S30: Within the search range of the preset direction, calculate the true direction of the typhoon based on the target sound pressure data and the target particle vibration velocity data.

[0068] In this embodiment, the true location includes the true azimuth angle and the true pitch angle. The process of calculating the true location of the typhoon based on the target sound pressure data and the target particle velocity data is as follows: within the search range of the first preset azimuth angle and the second preset pitch angle, multiple sound intensities in each direction are calculated based on the target sound pressure data and the target particle velocity data; a sound intensity matrix is ​​constructed based on the multiple sound intensities, and the true azimuth angle and the true pitch angle of the typhoon are determined based on the sound intensity matrix.

[0069] Understandably, within a certain azimuth and elevation angle search range, the sound intensity in each direction is calculated to construct a sound intensity matrix. By comparing the sound intensity in different directions, the approximate direction of the typhoon can be preliminarily determined (e.g., ...). Figure 3 (As shown). The accuracy of sound intensity calculation directly affects the label accuracy of subsequent model training, thus affecting the accuracy of the entire estimation system.

[0070] Furthermore, the principle of sound intensity calculation is as follows: based on a certain sampling frequency f s Data acquisition is performed continuously for a period of time T to obtain discrete time series data p[n] and v. x [n]、v y [n]、v z [n], where n = 0, 1, ..., n-1, N, N is the total number of sampling points, and N = f s T. Calculate the instantaneous sound intensity components I in the x, y, and z directions. x[n],I y [n] and I y [n]:

[0071] I x [n] = p[n]·v x [n]

[0072] I y [n] = p[n]·v y [n]

[0073] I z [n] = p[n]·v z [n];

[0074] Then calculate the amplitude I of the average sound intensity: According to the target azimuth strength formula Calculate the target azimuth intensity, where the reference acoustic intensity I0 = 10. -12 W / m 2 A spherical coordinate system is established with the three-dimensional vector hydrophone as the center, and the azimuth angle... The range is typically 0 to 2π, while the pitch angle θ ranges from 0 to π. Assuming azimuth estimation is performed on a horizontal plane, first fix the pitch angle θ = π / 2 (i.e., only consider the horizontal direction), and then... The upper part is spaced at a certain angle. (like Perform a search. For each The value is calculated based on the above steps to determine the target azimuth strength in the corresponding direction. search The maximum value, and the corresponding azimuth angle This refers to the typhoon's position on the horizontal plane.

[0075] Step S40: Use the target sound pressure data and the target particle velocity data as sample data, and use the real location corresponding to the target sound pressure data and the target particle velocity data as labels to construct a dataset based on the sample data and the labels.

[0076] Specifically, a dataset is constructed based on the sample data and the corresponding labels, and the dataset is divided into a training set and a test set according to a preset ratio; the training set is used for model training and parameter tuning, and the test set is used to evaluate the performance of the model.

[0077] The preprocessed target sound pressure data and target particle velocity data are divided into a training set and a test set according to a set training set ratio (e.g., 7:3). The training set is mainly used to train the model and adjust its parameters (such as the weights of the neural network). By using the data in the training set, the model learns features and patterns for prediction or classification. The test set is used to evaluate the performance of the final model. The results on the test set represent the model's performance in practical applications. The data in the test set is neither used for training nor for parameter tuning, but rather for the final evaluation of the model's accuracy and performance. The performance on the test set verifies the model's ability to generalize to unknown data, ensuring that the model performs well even on unknown data. A reasonable dataset partitioning can avoid overfitting and underfitting problems and improve the model's generalization ability.

[0078] Understandably, in addition to setting training and testing sets, this application can further set up a validation set for tuning model hyperparameters and evaluating model performance. During training, the validation set is used to evaluate the model's performance on unseen data, preventing overfitting. Based on the performance on the validation set, the optimal model parameters can be selected for further model optimization.

[0079] Step S50: Construct a typhoon orientation estimation model, and use the dataset to train and evaluate the typhoon orientation estimation model to obtain the target model.

[0080] A Multilayer Perceptron (MLP) model was chosen for typhoon azimuth estimation. The MLP model consists of an input layer, hidden layers, and an output layer. The input layer receives preprocessed data, with each input node corresponding to a feature. The hidden layer contains multiple neurons, each processing the input through weighted summation and a non-linear activation function to extract features from the data. The output layer outputs the predicted typhoon azimuth and elevation angles. Using backpropagation, the model can adjust the weights and biases between neurons based on the prediction error to improve prediction accuracy.

[0081] Specifically, a multilayer perceptron model is constructed, which includes an input layer, a hidden layer, and a regression output layer. The hidden layer contains multiple neurons, and each neuron processes the data through weighted summation and a nonlinear activation function.

[0082] In this embodiment, a multilayer perceptron (MLP) model is chosen as the typhoon azimuth estimation model. The MLP model has strong nonlinear fitting capabilities and can capture complex features in the data. The MLP model includes an input layer, multiple hidden layers, and a regression output layer. The input layer receives preprocessed data, the hidden layers perform feature extraction and nonlinear transformation, and the regression output layer outputs the predicted values ​​of the typhoon azimuth and elevation angles.

[0083] Understandably, this application may choose to construct different machine learning models to replace the multilayer perceptron, such as support vector machine, decision tree, and random forest models, depending on the actual situation. Support vector machine models have good generalization ability and are suitable for processing high-dimensional data; decision tree and random forest models can handle nonlinear problems and have strong interpretability.

[0084] Furthermore, by increasing the number of hidden layers and the number of neurons, a typhoon location estimation model is obtained. By increasing the number of hidden layers and the number of neurons, the expressive power of the model can be improved, thereby improving the accuracy of the estimation.

[0085] The typhoon orientation estimation model is trained using the training set, and the loss of the typhoon orientation estimation model on the training set is calculated. In each trial, the model is trained using the training set, and the loss of the model on the training set is calculated. By comparing the loss values ​​under different parameter combinations, the parameter combination with the smallest loss is selected as the optimal parameter (e.g., ...). Figure 4 (As shown).

[0086] Experimenting with different learning rates and mini-batch sizes allows for the selection of optimal parameters for model training. The learning rate controls the step size of weight updates during training, while the mini-batch size determines the number of data samples used in each training iteration. By appropriately adjusting these parameters, the model can converge faster, improving training efficiency and prediction accuracy. Optimizing the multilayer perceptron model through experimentation with different learning rates and mini-batch sizes allows for the selection of the parameter combination with the least loss during model training. This process fully leverages the model's potential, improving its adaptability to complex marine acoustic data and its prediction accuracy. In addition to experimenting with different learning rates and mini-batch sizes, this application can also employ adaptive learning rate strategies, such as the adaptive learning rate mechanisms built into optimizers like AdaGrad, RMSProp, or Adam. These optimizers automatically adjust the learning rate based on parameter updates, improving model training efficiency and stability.

[0087] Furthermore, the parameter combination with the minimum loss is selected as the optimal parameter, and the typhoon orientation estimation model is trained again using the optimal parameter. During the training process, the model continuously adjusts the weights and biases to minimize the error between the predicted value and the true value until the typhoon orientation estimation model converges or reaches the preset number of training rounds, thus obtaining the trained model. Finally, the trained model is evaluated using the test set to obtain the target model.

[0088] Step S60: Collect the sound pressure data and particle velocity data to be predicted using the single-vector hydrophone, and input the sound pressure data and particle velocity data to be predicted into the target model to obtain the estimated typhoon location.

[0089] Specifically, the sound pressure data and particle velocity data to be predicted are input into the input layer of the target model for transformation to obtain a feature vector; the feature vector is input into the hidden layer of the target model for feature extraction and summation to obtain a fused feature; the fused feature is mapped through a nonlinear activation function to obtain the target feature; the target feature is input into the regression output layer of the target model for prediction to output the estimated typhoon location.

[0090] As can be seen, this invention provides more accurate labels for model training by calculating the sound intensity matrix and finding the azimuth and elevation angles corresponding to the maximum sound intensity as true reference values. Then, in the MLP model, a regression output layer directly outputs the estimated values ​​of the typhoon's azimuth and elevation angles. This design allows the model to be trained and optimized more directly for the task of estimating azimuth and elevation angles. This invention systematically integrates multiple steps such as data inspection, filtering, normalization, sound intensity calculation, dataset partitioning, model construction, and tuning, forming a complete and coherent single-vector hydrophone typhoon azimuth and elevation angle estimation process. The design and connection of the entire process ensure that each step works collaboratively, providing reliable support for accurate estimation.

[0091] Furthermore, such as Figure 5 As shown, based on the above-mentioned typhoon azimuth estimation method based on a single-vector hydrophone, the present invention also provides a typhoon azimuth estimation system based on a single-vector hydrophone, wherein the typhoon azimuth estimation system based on a single-vector hydrophone includes:

[0092] The data acquisition module 51 is used to place the target single-vector hydrophone at a preset depth below the sea surface and acquire sound pressure data and particle velocity data through the target single-vector hydrophone.

[0093] The data preprocessing module 52 is used to perform data inspection, filtering and normalization on the sound pressure data and the particle vibration velocity data to obtain target sound pressure data and target particle vibration velocity data.

[0094] The true location calculation module 53 is used to calculate the true location of the typhoon within a preset location search range, based on the target sound pressure data and the target particle vibration velocity data.

[0095] The dataset construction module 54 is used to construct a dataset by using the target sound pressure data and the target particle vibration velocity data as sample data, and the real location corresponding to the target sound pressure data and the target particle vibration velocity data as a label;

[0096] The model training and evaluation module 55 is used to construct a typhoon orientation estimation model, and to train and evaluate the typhoon orientation estimation model using the dataset to obtain the target model.

[0097] The typhoon azimuth estimation module 56 is used to collect the sound pressure data to be predicted and the particle velocity data to be predicted through the single-vector hydrophone, and input the sound pressure data to be predicted and the particle velocity data to be predicted into the target model to obtain the estimated azimuth of the typhoon.

[0098] Furthermore, such as Figure 6 As shown, based on the above-mentioned typhoon azimuth estimation method and system based on single-vector hydrophone, the present invention also provides a terminal, which includes a processor 10, a memory 20 and a display 30. Figure 6 Only some of the terminal components are shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0099] In some embodiments, the memory 20 may be an internal storage unit of the terminal, such as a hard disk or memory. In other embodiments, the memory 20 may be an external storage device of the terminal, such as a plug-in hard disk, SmartMedia Card (SMC), Secure Digital (SD) card, or Flash Card. Further, the memory 20 may include both internal and external storage devices. The memory 20 is used to store application software and various types of data installed on the terminal, such as the program code installed on the terminal. The memory 20 can also be used to temporarily store data that has been output or will be output. In one embodiment, the memory 20 stores a typhoon azimuth estimation program 40 based on a single-vector hydrophone, which can be executed by the processor 10 to implement the typhoon azimuth estimation method based on a single-vector hydrophone in this application.

[0100] In some embodiments, the processor 10 may be a central processing unit (CPU), a microprocessor, or other data processing chip, used to run program code stored in the memory 20 or process data, such as executing the typhoon azimuth estimation method based on a single-vector hydrophone.

[0101] In some embodiments, the display 30 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. The display 30 is used to display information on the terminal and to display a visual user interface. The components of the terminal communicate with each other via a system bus.

[0102] In one embodiment, when the processor 10 executes the typhoon azimuth estimation program 40 based on a single-vector hydrophone in the memory 20, the following steps are performed:

[0103] The target single-vector hydrophone is placed at a predetermined depth below the sea surface, and sound pressure data and particle velocity data are collected through the target single-vector hydrophone.

[0104] The sound pressure data and the particle velocity data are subjected to data inspection, filtering and normalization to obtain the target sound pressure data and the target particle velocity data.

[0105] Within the search range of the preset direction, the true direction of the typhoon is calculated based on the target sound pressure data and the target particle velocity data;

[0106] The target sound pressure data and the target particle velocity data are used as sample data, and the real location corresponding to the target sound pressure data and the target particle velocity data is used as a label. A dataset is constructed based on the sample data and the label.

[0107] Construct a typhoon orientation estimation model, and use the dataset to train and evaluate the typhoon orientation estimation model to obtain the target model;

[0108] The predicted sound pressure data and predicted particle velocity data are collected by the single-vector hydrophone, and then input into the target model to obtain the estimated typhoon location.

[0109] The step of placing the target single-vector hydrophone at a predetermined depth below the sea surface and collecting sound pressure data and particle velocity data through the target single-vector hydrophone specifically includes:

[0110] Select a single-vector hydrophone whose sensitivity, frequency response range, working depth, and directionality meet the preset standards as the target single-vector hydrophone;

[0111] The target single-vector hydrophone is placed at a predetermined depth below the sea surface, and sound pressure data and particle velocity data in three orthogonal directions are collected through the target single-vector hydrophone.

[0112] Specifically, the step of performing data inspection, filtering, and normalization on the sound pressure data and the particle velocity data to obtain the target sound pressure data and target particle velocity data includes:

[0113] Check whether the sound pressure data and the particle velocity data are vectors and whether the vector lengths are consistent. If not, determine that the sound pressure data and the particle velocity data were acquired incorrectly and discard the data.

[0114] If so, a fourth-order Butterworth filter is used to filter the sound pressure data and the particle velocity data to obtain filtered data. The filtered data is then mapped to a preset range and normalized to obtain the target sound pressure data and target particle velocity data.

[0115] The true orientation includes the true azimuth angle and the true pitch angle;

[0116] Within the preset search range, the calculation of the typhoon's true location based on the target sound pressure data and the target particle velocity data specifically includes:

[0117] Within the search range of the first preset azimuth angle and the second preset pitch angle, multiple sound intensities in each direction are calculated based on the target sound pressure data and the target particle vibration velocity data.

[0118] A sound intensity matrix is ​​constructed based on the multiple sound intensities, and the true azimuth and true pitch angle of the typhoon are determined based on the sound intensity matrix.

[0119] The step of constructing a dataset based on the sample data and the labels specifically includes:

[0120] A dataset is constructed based on the sample data and the corresponding labels of the sample data, and the dataset is divided into a training set and a test set according to a preset ratio;

[0121] The training set is used for training the model and tuning its parameters, while the test set is used to evaluate the model's performance.

[0122] The step of constructing a typhoon orientation estimation model, and using the dataset to train and evaluate the typhoon orientation estimation model to obtain a target model, specifically includes:

[0123] A multilayer perceptron model is constructed, which includes an input layer, a hidden layer, and a regression output layer. The hidden layer contains multiple neurons, and each neuron processes the data through a weighted summation and a nonlinear activation function.

[0124] By increasing the number of hidden layers and the number of neurons, a typhoon location estimation model is obtained.

[0125] The typhoon orientation estimation model is trained using the training set, and the loss of the typhoon orientation estimation model on the training set is calculated.

[0126] The parameter combination with the minimum loss is selected as the optimal parameter. The typhoon orientation estimation model is then trained using the optimal parameter until the typhoon orientation estimation model converges or reaches the preset number of training rounds, thus obtaining the trained model.

[0127] The trained model is evaluated using the test set to obtain the target model.

[0128] Specifically, the step of inputting the sound pressure data to be predicted and the particle velocity data to be predicted into the target model to obtain the estimated typhoon location includes:

[0129] The sound pressure data to be predicted and the particle velocity data to be predicted are input into the input layer of the target model for transformation to obtain feature vectors;

[0130] The feature vector is input into the hidden layer of the target model for feature extraction and summation to obtain fused features. The fused features are then mapped using a non-linear activation function to obtain the target features.

[0131] The target features are input into the regression output layer of the target model for prediction, and the estimated typhoon location is output.

[0132] The present invention also provides a computer-readable storage medium, wherein the computer-readable storage medium stores a typhoon azimuth estimation program based on a single-vector hydrophone, and when the typhoon azimuth estimation program based on a single-vector hydrophone is executed by a processor, it implements the steps of the typhoon azimuth estimation method based on a single-vector hydrophone as described above.

[0133] In summary, this invention proposes a typhoon azimuth estimation method, system, and terminal based on a single-vector hydrophone. The method includes: rigorously checking the sound pressure and three-directional vibration velocity data collected by the single-vector hydrophone to ensure data quality; preprocessing the data, including filtering to remove noise interference, and then normalizing the filtered data; calculating the sound intensity in each direction within a certain azimuth and elevation angle search range to construct a sound intensity matrix, and finding the azimuth and elevation angles corresponding to the maximum sound intensity as true reference values; dividing the processed data into training and testing sets proportionally, constructing a multilayer perceptron model, and optimizing the model by experimenting with different learning rates and mini-batch sizes, selecting the parameters with the least loss to train the optimal model; finally, using the trained model to predict the test set to obtain estimated values ​​of the typhoon azimuth and elevation angles. This invention effectively improves the accuracy of typhoon azimuth estimation.

[0134] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal that includes that element.

[0135] Of course, those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided by this invention can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0136] It should be understood that the application of the present invention is not limited to the examples above. Those skilled in the art can make improvements or modifications based on the above description, and all such improvements and modifications should fall within the protection scope of the appended claims.

Claims

1. A single vector hydrophone based typhoon bearing estimation method, characterized in that, The typhoon direction estimation method based on a single vector hydrophone comprises the following steps: arranging a target single vector hydrophone at a preset depth below the sea surface, collecting sound pressure data and particle velocity data through the target single vector hydrophone, avoiding the interference of direct impact of water waves, and capturing the propagation characteristics of underwater acoustic signals caused by the typhoon; performing data checking, filtering processing and normalization processing on the sound pressure data and the particle velocity data to obtain target sound pressure data and target particle velocity data; calculating the real direction of the typhoon according to the target sound pressure data and the target particle velocity data within a preset search range of direction; the real direction comprises a real azimuth and a real elevation; the calculation of the real direction of the typhoon according to the target sound pressure data and the target particle velocity data within a preset search range of direction comprises: calculating a plurality of sound intensities in each direction according to the target sound pressure data and the target particle velocity data within a search range of a first preset azimuth and a second preset elevation; constructing a sound intensity matrix according to the plurality of sound intensities, and determining the real azimuth and the real elevation of the typhoon according to the sound intensity matrix; constructing a data set according to the sample data and the label; constructing a typhoon direction estimation model, training and evaluating the typhoon direction estimation model using the data set, and obtaining a target model; inputting the predicted sound pressure data and the predicted particle velocity data into the target model to obtain the estimated direction of the typhoon.

2. The single vector hydrophone based typhoon bearing estimation method according to claim 1, wherein, The arrangement of the target single vector hydrophone at a preset depth below the sea surface and the collection of sound pressure data and particle velocity data through the target single vector hydrophone comprise the following steps: selecting a single vector hydrophone with sensitivity, frequency response range, working depth and directivity meeting preset standards as a target single vector hydrophone; arranging the target single vector hydrophone at a preset depth below the sea surface, and collecting sound pressure data and particle velocity data in three orthogonal directions through the target single vector hydrophone.

3. The single vector hydrophone based typhoon bearing estimation method according to claim 1, wherein, The data checking, filtering processing and normalization processing of the sound pressure data and the particle velocity data to obtain target sound pressure data and target particle velocity data comprise the following steps: checking whether the sound pressure data and the particle velocity data are vectors and whether the vector lengths are consistent, and if not, determining that the sound pressure data and the particle velocity data are collected incorrectly and throwing out; if yes, filtering the sound pressure data and the particle velocity data using a fourth-order Butterworth filter to obtain filtered data, and performing normalization processing on the filtered data mapped to a preset range to obtain target sound pressure data and target particle velocity data.

4. The single vector hydrophone based typhoon bearing estimation method according to claim 1, wherein, The construction of a data set according to the sample data and the label comprises the following steps: construct a data set according to the sample data and labels corresponding to the sample data, and divide the data set into a training set and a test set according to a preset ratio; the training set is used for training and parameter adjustment of the model, and the test set is used for evaluating the performance of the model.

5. The single vector hydrophone based typhoon bearing estimation method according to claim 4, wherein, The typhoon direction estimation model is constructed, the typhoon direction estimation model is trained and evaluated using the data set, and a target model is obtained, specifically including: constructing a multi-layer perception model, the multi-layer perception model including an input layer, a hidden layer and a regression output layer, the hidden layer containing a plurality of neurons, each neuron processing data through weighted summation and a nonlinear activation function; increase the number of hidden layers and the number of neurons to obtain a typhoon direction estimation model, train the typhoon direction estimation model using the training set, and calculate the loss of the typhoon direction estimation model on the training set; select the parameter combination with the smallest loss as the optimal parameter, continue to train the typhoon direction estimation model using the optimal parameter until the typhoon direction estimation model converges or reaches a preset number of training rounds, and obtain a trained model; evaluate the trained model using the test set to obtain a target model.

6. The single vector hydrophone based typhoon bearing estimation method according to claim 5, wherein, The target model is input into the target model to obtain an estimated direction of the typhoon, specifically including: input the to-be-predicted sound pressure data and the to-be-predicted particle velocity data into the input layer of the target model for conversion to obtain a feature vector; input the feature vector into the hidden layer of the target model for feature extraction and summation to obtain a fusion feature, and map the fusion feature through a nonlinear activation function to obtain a target feature; input the target feature into the regression output layer of the target model for prediction to output an estimated direction of the typhoon.

7. A single vector hydrophone based typhoon bearing estimation system, characterized in that, The typhoon direction estimation system based on a single vector hydrophone is used to implement the typhoon direction estimation method based on a single vector hydrophone according to any one of claims 1-6, and the typhoon direction estimation system based on a single vector hydrophone includes: a data acquisition module configured to arrange a target single vector hydrophone at a preset depth below the sea surface, and acquire sound pressure data and particle velocity data through the target single vector hydrophone; a data preprocessing module configured to perform data checking, filtering processing and normalization processing on the sound pressure data and the particle velocity data to obtain target sound pressure data and target particle velocity data; a real direction calculation module configured to calculate a real direction of a typhoon according to the target sound pressure data and the target particle velocity data within a preset search range of the real direction; a data set construction module configured to construct a data set by taking the target sound pressure data and the target particle velocity data as sample data and taking real directions corresponding to the target sound pressure data and the target particle velocity data as labels; a model training and evaluation module configured to construct a typhoon direction estimation model, train and evaluate the typhoon direction estimation model using the data set, and obtain a target model. The typhoon azimuth estimation module is configured to collect predicted sound pressure data and predicted particle velocity data by the single vector hydrophone, input the predicted sound pressure data and the predicted particle velocity data into the target model, and obtain an estimated azimuth of the typhoon.

8. A terminal, characterized by comprising: The terminal comprises a memory, a processor, and a single vector hydrophone-based typhoon azimuth estimation program stored in the memory and executable on the processor. When the single vector hydrophone-based typhoon azimuth estimation program is executed by the processor, the steps of the single vector hydrophone-based typhoon azimuth estimation method according to any one of claims 1-6 are implemented.

9. A computer-readable storage medium, characterized in that, The computer readable storage medium stores a single vector hydrophone-based typhoon azimuth estimation program. When the single vector hydrophone-based typhoon azimuth estimation program is executed by the processor, the steps of the single vector hydrophone-based typhoon azimuth estimation method according to any one of claims 1-6 are implemented.

Images