Method and system for predicting sound propagation loss in internal solitary wave environment by artificial intelligence
By constructing an internal solitary wave environment and training a sound propagation loss prediction network, the problems of insufficient data and long computation time for sound propagation loss in the internal solitary wave environment are solved, achieving efficient and accurate sound propagation loss prediction, and improving computational efficiency and network universality.
Patent Information
- Application Number
- CN202511914148.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-18
- Publication Date
- 2026-01-16
AI Technical Summary
Existing methods for calculating acoustic propagation loss in an internal solitary wave environment suffer from insufficient data, high implementation costs, and long computation times, making it difficult to efficiently predict underwater acoustic propagation loss.
An internal solitary wave environment in the ocean is constructed to generate the data required for training the acoustic propagation loss prediction network. The characteristic parameters of the internal solitary waves are normalized and encoded using artificial intelligence methods to construct a training dataset. The acoustic propagation loss prediction network is then trained using the training dataset to establish a mapping relationship between input and output and obtain the acoustic propagation loss in the target area.
It achieves efficient and accurate prediction of acoustic propagation loss in internal solitary wave environments, improves computational efficiency, reduces costs, and provides a universal acoustic propagation loss prediction network.
Smart Images
Figure CN121350508A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to sound propagation technology in the field of marine environment, specifically to a method and system for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence. Background Technology
[0002] Sound waves, as the only energy carrier capable of long-distance underwater transmission, play an irreplaceable role in many key areas such as ocean exploration, information transmission, environmental monitoring, and resource development. Calculating underwater sound propagation directly affects the accuracy of target location, the effectiveness of detection, the reliability of communication, and the effectiveness of underwater countermeasures. The marine environment has a significant impact on the characteristics of underwater sound propagation loss, thus affecting the effectiveness of sonar.
[0003] Internal solitary waves are massive waves occurring within oceans or lakes. They exist at the interface between two layers of water with different densities, and are usually imperceptible at the sea surface (or lake surface). Internal solitary waves can cause vertical water movement on the order of hundreds of meters. On the one hand, they bring nutrient-rich, cold bottom water to the sunlit surface, promoting phytoplankton growth and impacting the entire marine ecosystem. On the other hand, they exert powerful impacts on underwater robots, underwater structures of oil drilling platforms, submarine cables, and pipelines. Internal solitary waves are a type of nonlinear, large-amplitude wave widely observed in the global oceans. They are characterized by short periods and high propagation speeds, with horizontal extensions reaching tens of kilometers and vertical amplitudes potentially reaching hundreds of meters. Calculating sound propagation loss under internal solitary wave conditions is directly related to underwater vehicle path planning and decision-making, maximizing the effectiveness of underwater acoustic equipment, and is of paramount importance. Currently, methods for studying the impact of internal solitary waves on sound propagation characteristics mainly include measured data analysis, numerical simulation analysis, and artificial intelligence prediction methods. The method of directly observing sound propagation loss data in an internal solitary wave environment using instruments such as underwater acoustic receiver arrays and temperature chains to obtain the sound propagation characteristics of internal solitary waves is highly accurate, but it suffers from problems such as insufficient data and high implementation costs. Numerical simulation methods can simulate the structure of internal solitary waves by setting characteristic parameters, and data acquisition is easy, but the computation time is long. Summary of the Invention
[0004] The technical problem to be solved by the present invention is to provide a method and system for predicting acoustic propagation loss in an internal solitary wave environment using artificial intelligence, in order to address the above-mentioned problems in the prior art. The present invention aims to improve the computational efficiency of predicting acoustic propagation loss in an internal solitary wave environment in the ocean by utilizing artificial intelligence.
[0005] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A method for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence includes the following steps: S101, Construct an internal ocean solitary wave environment and generate the data required for training the acoustic propagation loss prediction network, including the characteristic parameters of internal solitary waves and their corresponding acoustic propagation losses; S102, normalize and encode the characteristic parameters of the internal solitary wave, and then construct a training dataset together with the acoustic propagation loss; S103, use the training dataset to train the acoustic propagation loss prediction network so that it learns and establishes the mapping relationship between the normalized and encoded internal solitary wave characteristic parameters of the input and the acoustic propagation loss of the output. S104: Obtain the internal solitary wave characteristic parameters of the target ocean region. Normalize and encode the internal solitary wave characteristic parameters of the target sampling region and input them into the trained acoustic propagation loss prediction network to obtain the acoustic propagation loss of the target ocean region.
[0006] Optionally, in step S101, the characteristic parameters of the internal solitary wave include the sound velocity of seawater, the sound velocity of the seabed, the density of the seabed, and the absorption coefficient of the seabed in the internal solitary wave environment of the ocean.
[0007] Optionally, the functional expression for the speed of sound in the ocean under the isolated wave environment is: ; ; In the above formula, For the sound velocity field distribution, For radius, For depth, For the speed of water in the mezzanine Shanghai, The speed of seawater flowing through the mezzanine. The depth of the upper seawater. The depth of the lower layer of seawater. For the thickness of the transition layer, The sound velocity gradient of the mesas. This represents the depth boundary of the bottom water body.
[0008] Optionally, constructing the ocean-internal solitary wave environment in step S101 includes: S201, Set sound source conditions, including sound source frequency, sound source location, and sound source depth; set receiver parameters, including receiver location and receiver depth; set ocean acoustic parameters, including seawater sound speed, seabed sound speed, seabed density, and seabed absorption coefficient; set internal solitary wave characteristic parameters, including internal solitary wave amplitude and internal solitary wave location. S202, based on the characteristic parameters of internal solitary waves, various internal solitary wave structures are constructed using the KdV equation shown below to form an ocean internal solitary wave environment: ; In the above formula, For the amplitude of the internal solitary wave, The initial amplitude of the internal solitary wave. It is a hyperbolic secant function. Horizontal distance For the location of the internal solitary wave, This is the characteristic width of the internal solitary wave.
[0009] Optionally, in step S101, the data required for training the acoustic propagation loss prediction network includes collecting the characteristic parameters of the isolated wave in the ocean, calculating the environmental sound pressure of the isolated wave in the ocean using the Kraken model, and calculating the acoustic propagation loss according to the following formula: , In the above formula, For sound propagation loss, For sound pressure, For radius, For depth.
[0010] Optionally, the function expression for normalizing and encoding the characteristic parameters of the internal solitary wave in step S102 is: ; ; ; in, These are the normalized characteristic parameters of the internal solitary wave. For the characteristic parameters of internal solitary waves, This represents the maximum value of the characteristic parameters of the internal solitary wave. This represents the minimum value of the characteristic parameters of the internal solitary wave; and These are the horizontal correlation coefficient and the vertical correlation coefficient. and These are the encoding results in the horizontal and vertical directions, respectively.
[0011] Optionally, the acoustic propagation loss prediction network employs an attention-enhanced U-Net model consisting of an encoder and a decoder. The encoder is composed of cascaded multi-level coding units, with the normalized and encoded internal solitary wave feature parameters serving as the input to the first coding unit. Each coding unit consists of two sequentially connected 3×3 convolutional modules, a convolutional attention module, and a 2×2 max-pooling layer for downsampling, and each convolutional module has a ReLU activation function. The decoder is composed of cascaded multi-level decoding units, with the same number of decoding and coding units. Each coding unit includes a skip connection module and two 2×2 deconvolutional modules to perform upsampling. The skip connection module connects the output features of the corresponding coding unit with the input features of the coding unit and outputs the result to the subsequent 2×2 deconvolutional module. The predicted acoustic propagation loss is obtained by the last decoding unit.
[0012] Furthermore, the present invention also provides a system for predicting acoustic propagation loss in an internal solitary wave environment using artificial intelligence, comprising a microprocessor and a memory interconnected thereto, the microprocessor being programmed or configured to perform the method for predicting acoustic propagation loss in an internal solitary wave environment using artificial intelligence.
[0013] Furthermore, the present invention also provides a computer-readable storage medium storing a computer program or instructions that are programmed or configured to execute, via a processor, the method for predicting sound propagation loss in an internally isolated wave environment using artificial intelligence.
[0014] Furthermore, the present invention also provides a computer program product, including a computer program or instructions that are programmed or configured to execute, via a processor, the method for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence.
[0015] Compared with existing technologies, the present invention has the following main advantages: The method of the present invention includes constructing an internal solitary wave environment in the ocean, generating data required for training the acoustic propagation loss prediction network, including internal solitary wave feature parameters and their corresponding acoustic propagation losses; normalizing and encoding the internal solitary wave feature parameters, and then constructing a training dataset together with the acoustic propagation loss; using the training dataset to train the acoustic propagation loss prediction network to learn and establish a mapping relationship between the input normalized and encoded internal solitary wave feature parameters and the output acoustic propagation loss; obtaining the internal solitary wave feature parameters of the target ocean area, normalizing and encoding the internal solitary wave feature parameters of the target sampling area, and inputting them into the trained acoustic propagation loss prediction network to obtain the acoustic propagation loss of the target ocean area. The present invention can construct a universal acoustic propagation loss prediction network for an internal solitary wave environment based on artificial intelligence, and can predict the acoustic propagation loss in an internal solitary wave environment based on internal solitary wave feature parameters, thereby providing technical support for improving the computational efficiency of acoustic propagation loss in an internal solitary wave environment. Attached Figure Description
[0016] Figure 1 This is a schematic diagram of the basic process of the method in an embodiment of the present invention.
[0017] Figure 2 This is a schematic diagram of a model of an isolated wave environment in the ocean constructed in an embodiment of the present invention.
[0018] Figure 3 This is a schematic diagram of the network structure of the sound propagation loss prediction network in an embodiment of the present invention.
[0019] Figure 4 This is a comparison of the root mean square error of different network models in the embodiments of the present invention.
[0020] Figure 5 This is a comparison of the structural similarity of different network models in the embodiments of the present invention.
[0021] Figure 6 This diagram illustrates the comparison of prediction results between the Kraken numerical model and the acoustic propagation loss prediction network under different internal solitary wave characteristic parameters in this embodiment of the invention. (a1) to (c1) represent the calculation results of the Kraken numerical model, the prediction results of the acoustic propagation loss prediction network, and the errors of the prediction results of the acoustic propagation loss prediction network when the internal solitary wave amplitude is 3m and the internal solitary wave position is 5km, respectively; (a2) to (c2) represent the calculation results of the Kraken numerical model, the prediction results of the acoustic propagation loss prediction network, and the errors of the prediction results of the acoustic propagation loss prediction network when the internal solitary wave amplitude is 3m and the internal solitary wave position is 1km, respectively; (a3) to (c3) represent the calculation results of the Kraken numerical model, the prediction results of the acoustic propagation loss prediction network, and the errors of the prediction results of the acoustic propagation loss prediction network when the internal solitary wave amplitude is 15 and the internal solitary wave position is 5km, respectively; and (a4) to (c4) represent the calculation results of the Kraken numerical model, the prediction results of the acoustic propagation loss prediction network, and the errors of the prediction results of the acoustic propagation loss prediction network when the internal solitary wave amplitude is 15 and the internal solitary wave position is 1km, respectively.
[0022] Figure 7 These are the root mean square error and the average value and confidence interval of the sound propagation loss prediction network under different sound source conditions in the embodiments of the present invention, where (a) is the performance test result under different sound source frequencies and (b) is the performance test result under different sound source depths.
[0023] Figure 8 This invention presents a comparison of the computational efficiency of the Kraken numerical model and the sound propagation loss prediction network in this embodiment. Detailed Implementation
[0024] This invention provides a method for predicting acoustic propagation loss in an internal solitary wave environment using an artificial intelligence prediction network. The invention aims to construct and train an artificial intelligence network model that, by using the characteristic parameters of internal solitary waves as input, can accurately and efficiently predict the underwater acoustic propagation loss distribution under the influence of internal solitary waves. To enable those skilled in the art to better understand the technical solution of this invention, the following will provide a more detailed description of the technical solution in conjunction with the accompanying drawings of the embodiments of this invention.
[0025] like Figure 1 As shown, the method for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence in this embodiment includes the following steps: S101, Construct an internal ocean solitary wave environment and generate the data required for training the acoustic propagation loss prediction network, including the characteristic parameters of internal solitary waves and their corresponding acoustic propagation losses; S102, normalize and encode the characteristic parameters of the internal solitary wave, and then construct a training dataset together with the acoustic propagation loss; S103, use the training dataset to train the acoustic propagation loss prediction network so that it learns and establishes the mapping relationship between the normalized and encoded internal solitary wave characteristic parameters of the input and the acoustic propagation loss of the output. S104: Obtain the internal solitary wave characteristic parameters of the target ocean region. Normalize and encode the internal solitary wave characteristic parameters of the target sampling region and input them into the trained acoustic propagation loss prediction network to obtain the acoustic propagation loss of the target ocean region.
[0026] In step S101 of this embodiment, the characteristic parameters of the internal solitary wave include the seawater sound velocity, seabed sound velocity, seabed density, and seabed absorption coefficient under the internal solitary wave environment in the ocean. The seawater sound velocity includes the sound velocity above the mezzanine, the sound velocity below the mezzanine, and the sound velocity below the mezzanine. In this embodiment, the functional expression for the seawater sound velocity under the internal solitary wave environment in the ocean is: ; ; In the above formula, For the sound velocity field distribution, For radius, For depth, For the speed of water in the mezzanine Shanghai, The speed of seawater flowing through the mezzanine. The depth of the upper seawater. The depth of the lower layer of seawater. For the thickness of the transition layer, The sound velocity gradient of the mesas. This represents the depth boundary of the bottom water body.
[0027] To construct an intra-oceanic solitary wave environment, step S101 includes the following: S201, Set sound source conditions, including sound source frequency, sound source location, and sound source depth; set receiver parameters, including receiver location and receiver depth; set marine acoustic parameters, including seawater sound speed, seabed sound speed, seabed density, and seabed absorption coefficient, wherein the seawater sound speed, seabed sound speed, seabed density, and seabed absorption coefficient can be set according to the acoustic environment parameters set in the acoustic simulation experiment completed in the relevant sea area. In this embodiment, the acoustic environment of the SWARM (Shallow Water Acoustic Random Medium) experiment is selected as a reference; set internal solitary wave characteristic parameters, including internal solitary wave amplitude and internal solitary wave location. S202, based on the characteristic parameters of internal solitary waves, various internal solitary wave structures are constructed using the KdV equation shown below to form an ocean internal solitary wave environment: ; In the above formula, For the amplitude of the internal solitary wave, The initial amplitude of the internal solitary wave. It is a hyperbolic secant function. Horizontal distance For the location of the internal solitary wave, This represents the characteristic width of an internal solitary wave. For example, the model of a certain ocean internal solitary wave environment constructed in this embodiment is as follows: Figure 2 As shown.
[0028] In step S101 of this embodiment, the data required for training the acoustic propagation loss prediction network includes collecting characteristic parameters of the internal solitary wave, calculating the environmental sound pressure of the internal solitary wave in the ocean using the Kraken model, and calculating the acoustic propagation loss according to the following formula: , In the above formula, For sound propagation loss, For sound pressure, For radius, For depth.
[0029] In step S102 of this embodiment, the function expression for normalizing and encoding the characteristic parameters of the internal solitary wave is as follows: ; ; ; in, These are the normalized characteristic parameters of the internal solitary wave. For the characteristic parameters of internal solitary waves, This represents the maximum value of the characteristic parameters of the internal solitary wave. This represents the minimum value of the characteristic parameters of the internal solitary wave; and These are the horizontal correlation coefficient and the vertical correlation coefficient. and These are the encoding results in the horizontal and vertical directions, respectively.
[0030] like Figure 3 As shown, in this embodiment, the acoustic propagation loss prediction network adopts a U-Net model consisting of an encoder and a decoder. The encoder is composed of cascaded multi-level coding units. The normalized and encoded internal solitary wave feature parameters are used as the input of the first coding unit. Each coding unit consists of two 3×3 convolutional modules, a convolutional attention module, and a 2×2 max pooling layer for downsampling, and each convolutional module has a ReLU activation function. The decoder is composed of cascaded multi-level decoding units. The number of decoding units is the same as that of the coding units. Each coding unit includes a skip connection module and two 2×2 deconvolutional modules to implement upsampling. The skip connection module is used to connect the output features of the corresponding coding unit and the input features of the coding unit and output them to the subsequent 2×2 deconvolutional modules. The predicted acoustic propagation loss is obtained by the last decoding unit. This implementation of an artificial intelligence method for predicting acoustic propagation loss in an internal solitary wave environment mainly includes three steps: generating an input dataset (constructing an internal solitary wave environment in the ocean: inputting underwater acoustic environment parameters, constructing an internal solitary wave structure based on internal solitary wave characteristic parameters, setting sound source conditions; calculating the acoustic propagation loss of the sound field in the internal solitary wave environment using the Kraken model; processing the internal solitary wave characteristic parameters using normalization and encoding methods and constructing a dataset), constructing and training an acoustic propagation loss prediction network (constructing a U-Net model using a convolutional attention module to train and predict acoustic propagation loss in an internal solitary wave environment), and validating and analyzing the performance of the acoustic propagation loss prediction network model (validating and analyzing the performance of the acoustic propagation loss prediction network model under different sound source and sample number conditions).
[0031] To verify the training effect of the attention-enhanced U-Net model used in the acoustic propagation loss prediction network of the artificial intelligence method for predicting acoustic propagation loss in an internal solitary wave environment in this embodiment, this embodiment uses the existing U-Net model and the MultiScale-DUNet model for comparison, and uses root mean square error and structural similarity as performance evaluation indicators. The results are as follows: Figure 4 and Figure 5 As shown, Figure 4 This example compares the root mean square error of the attention-enhanced U-Net model with existing U-Net and MultiScale-DUNet models. Figure 5 This example compares the structural similarity between the attention-enhanced U-Net model and existing U-Net and MultiScale-DUNet models. Figure 4 and Figure 5 The distribution characteristics of root mean square error and structural similarity of the U-Net model, MultiScale-DUNet model and attention-improved U-Net model are given, indicating that the attention-improved U-Net model used in the acoustic propagation loss prediction network of the artificial intelligence prediction method for internal solitary wave environment in this embodiment has the best training effect.
[0032] To verify the prediction effect of the attention-enhanced U-Net model used in the acoustic propagation loss prediction network of the artificial intelligence prediction method for acoustic propagation loss under internal solitary wave environment in this embodiment, this embodiment compares the results calculated by the Kraken numerical model and the prediction results of the acoustic propagation loss prediction network for different internal solitary wave parameters, and calculates the error. In this embodiment, the characteristic parameters of internal solitary waves are calculated when the sound source frequency is 100Hz and the sound source depth is 50m: the internal solitary wave amplitude is 3m and 15m, and the internal solitary wave position is 5km and 15km. The results of the Kraken numerical model and the prediction results of the acoustic propagation loss prediction network are compared. Figure 6 This diagram illustrates the comparison of prediction results between the Kraken numerical model and the acoustic propagation loss prediction network under different internal solitary wave characteristic parameters in this embodiment. (a1) to (c1) represent the Kraken numerical model calculation results, the acoustic propagation loss prediction network prediction results, and the errors of the acoustic propagation loss prediction network prediction results when the internal solitary wave amplitude is 3m and the internal solitary wave position is 5km, respectively. (a2) to (c2) represent the Kraken numerical model calculation results, the acoustic propagation loss prediction network prediction results, and the errors of the acoustic propagation loss prediction network prediction results when the internal solitary wave amplitude is 3m and the internal solitary wave position is 1km, respectively. (a3) to (c3) represent the Kraken numerical model calculation results, the acoustic propagation loss prediction network prediction results, and the errors of the acoustic propagation loss prediction network prediction results when the internal solitary wave amplitude is 15 and the internal solitary wave position is 5km, respectively. (a4) to (c4) represent the Kraken numerical model calculation results, the acoustic propagation loss prediction network prediction results, and the errors of the acoustic propagation loss prediction network prediction results when the internal solitary wave amplitude is 15 and the internal solitary wave position is 1km, respectively. Figure 6 The results show that the sound propagation loss prediction network has good prediction performance under different internal solitary wave parameters. (3) Performance evaluation of the sound propagation loss prediction network under different sound source conditions; in this embodiment, the root mean square error and structural similarity are calculated when the sound source frequency is 40-200Hz (interval 20Hz) and the sound source depth is 5-50m (interval 5m).
[0033] Figure 7The root mean square error and structural similarity of the sound propagation loss prediction network under different sound source conditions in this embodiment are the average value and confidence interval, where (a) is the performance test result under different sound source frequencies and (b) is the performance test result under different sound source depths. Figure 8 The results show that the error of the sound propagation loss prediction network increases with the frequency of the sound source, and the sound propagation loss prediction network has good generalization ability.
[0034] To verify the computational efficiency of the attention-enhanced U-Net model used in the acoustic propagation loss prediction network of the artificial intelligence method for predicting acoustic propagation loss in an internal solitary wave environment in this embodiment, this embodiment compares the computational efficiency of the numerical model and the acoustic propagation loss prediction network for different sample numbers; in this embodiment, the number of calculation samples is 10, 20, 50, 100, 150, and 300. Figure 8 This example compares the computational efficiency of the Kraken numerical model and the sound propagation loss prediction network. Figure 8 The results show that the computational efficiency of the acoustic propagation loss prediction network is much higher than that of the numerical model, with an average improvement of 91%. Therefore, the artificial intelligence method for predicting acoustic propagation loss under the influence of internal solitary waves in this embodiment can build and train a universally applicable acoustic propagation loss prediction network. This method can predict underwater acoustic propagation loss under the influence of internal solitary waves, outputting the acoustic propagation loss based on the characteristic parameters of the internal solitary waves, thus providing a reference for calculating the acoustic propagation loss under the influence of internal solitary waves.
[0035] Furthermore, this embodiment also provides a system for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence, including a microprocessor and a memory interconnected, wherein the microprocessor is programmed or configured to execute the method for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence. This embodiment also provides a computer-readable storage medium storing a computer program or instructions programmed or configured to execute the method for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence via a processor. This embodiment also provides a computer program product including a computer program or instructions programmed or configured to execute the method for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence via a processor.
[0036] Those skilled in the art will understand that the technical solutions provided by this invention may take the form of a method, system, or computer program product. Therefore, this invention may take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this invention may take the form of a computer program product embodied on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can 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, produce an implementation of the flowchart... Figure 1 One or more processes and / or boxes Figure 1 The computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The functions specified in one or more boxes. 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, thereby providing instructions that execute on the computer or other programmable apparatus for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0037] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.
Claims
1. A method for predicting sound propagation loss in an internal solitary wave environment using artificial intelligence, characterized by, The method comprises the following steps: S101, constructing an internal solitary wave environment in the sea, generating data required for training the sound propagation loss prediction network, including internal solitary wave characteristic parameters and corresponding sound propagation loss; S102, normalizing and encoding the internal solitary wave characteristic parameters, and then constructing a training data set together with the sound propagation loss; S103, training the sound propagation loss prediction network with the training data set to make it learn and establish a mapping relationship between the normalized and encoded internal solitary wave characteristic parameters as input and the sound propagation loss as output; S104, obtaining the internal solitary wave characteristic parameters of a target sea area, inputting the normalized and encoded internal solitary wave characteristic parameters of the target sampling area into the trained sound propagation loss prediction network to obtain the sound propagation loss of the target sea area.
2. The method of claim 1, wherein, In step S101, the internal solitary wave characteristic parameters include seawater sound speed, seabed sound speed, seabed density and seabed absorption coefficient in the internal solitary wave environment in the sea.
3. The method of claim 2, wherein, The function expression of the seawater sound speed in the internal solitary wave environment in the sea is: ； ； In the above formula, is the sound speed field distribution, is the radius, is the depth, is the upper layer sea water velocity, is the lower layer sea water velocity, is the upper layer sea water depth, is the lower layer sea water depth, is the transition layer thickness, is the sound speed gradient of the jump layer sea water, is the depth boundary of the bottom layer water body.
4. The method of claim 1, wherein, The construction of the internal solitary wave environment in step S101 comprises: S201, setting sound source conditions, including sound source frequency, sound source position and sound source depth, setting receiver parameters, including receiver position and receiver depth, setting marine acoustic parameters, including seawater sound speed, seabed sound speed, seabed density and seabed absorption coefficient, and setting internal solitary wave characteristic parameters, including internal solitary wave amplitude and internal solitary wave position; S202, constructing multiple internal solitary wave structures according to the internal solitary wave characteristic parameters to form the internal solitary wave environment in the sea by using the KdV equation shown in the following formula: ； in the above formula, is the internal solitary wave amplitude, is the initial internal solitary wave amplitude, is the hyperbolic secant function, is the horizontal distance, is the internal solitary wave position, is the characteristic width of the internal solitary wave.
5. The method of claim 1, wherein, In step S101, generating data required for training the sound propagation loss prediction network comprises collecting internal solitary wave characteristic parameters, calculating the sound pressure in the internal solitary wave environment in the sea by using the kraken model, and calculating the sound propagation loss according to the following formula: ， In the above formulae, is the sound propagation loss, is the sound pressure, is the radius, is the depth.
6. The method of claim 1, wherein, In step S102, the function expression for normalizing and encoding the internal solitary wave characteristic parameters is: ； ； ； wherein, is a normalized internal solitary wave characteristic parameter, is an internal solitary wave characteristic parameter, is a maximum value of the internal solitary wave characteristic parameter, is a minimum value of the internal solitary wave characteristic parameter; and are a horizontal correlation coefficient and a vertical correlation coefficient, and are encoding results in horizontal and vertical directions, respectively.
7. The method of claim 1, wherein, The sound propagation loss prediction network adopts an attention improved U-Net model composed of an encoder and a decoder, the encoder is composed of multiple levels of encoding units in cascade, the normalized and encoded internal solitary wave characteristic parameters are input into the first encoding unit, each level of encoding unit is composed of two 3×3 convolution modules connected in turn, one convolution attention module and one 2×2 maximum pooling layer for downsampling, and each convolution module has a Relu activation function; the decoder is composed of multiple levels of decoding units in cascade, the number of decoding units is the same as that of encoding units, each encoding unit includes a skip connection module and two 2×2 deconvolution modules to realize upsampling operation, the skip connection module is used to connect the output features of the corresponding encoding unit and the input features of the encoding unit and then output to the subsequent 2×2 deconvolution module, and the predicted sound propagation loss is obtained from the last decoding unit.
8. A system for artificial intelligence prediction of sound propagation loss in an internal solitary wave environment, comprising a microprocessor and a memory interconnected, characterized in that, The microprocessor is programmed or configured to perform the method of any one of claims 1-7 for predicting the sound propagation loss in the internal solitary wave environment by artificial intelligence.
9. A computer-readable storage medium having stored therein a computer program or instructions, characterized in that, The computer program or instruction is programmed or configured to execute the method for predicting sound propagation loss in internal solitary wave environment by artificial intelligence according to any one of claims 1-7 by the processor.
10. A computer program product comprising computer programs or instructions, characterized in that, The computer program or instruction is programmed or configured to execute the method for predicting sound propagation loss in internal solitary wave environment by artificial intelligence according to any one of claims 1-7 by the processor.