Method and device for predicting propagation loss distribution of underwater acoustic channel

Abstract

The invention relates to the technical field of underwater acoustic communication, in particular to an underwater acoustic channel propagation loss distribution prediction method and device, and the method comprises the steps: obtaining a first sound source depth and a target sound source depth which are conjugated with respect to a sound channel axis in a water area based on the sound velocity profile of the water area; based on a resolution demand, determining underwater acoustic channel propagation loss distribution of the first sound source depth, and determining free space propagation loss distribution of the target sound source depth; carrying out amplitude limiting and normalization processing on the two, and combining on a channel dimension; and inputting the merged propagation loss distribution into a target neural network to output a normalized propagation loss distribution prediction result of the target sound source depth so as to determine the underwater acoustic channel propagation loss distribution of the target sound source depth. According to the method, the known propagation loss distribution under any sound source depth can be utilized to predict the propagation loss distribution of the sound source depth which is conjugated with the sound source depth about the sound channel axis, and the practical application capability and accuracy of the target neural network are also extremely high.

Description

Technical Field

[0001] The present application relates to the technical field of underwater acoustic communication, and particularly relates to a method and device for predicting the propagation loss distribution of an underwater acoustic channel. Background Art

[0002] In related technologies, the classical methods for predicting the propagation loss of an underwater acoustic channel mainly include modeling methods such as ray model, normal mode model, fast field model, and parabolic equation. Moreover, these modeling methods mostly obtain the modeling results of the propagation loss of the underwater acoustic channel through simulation by considering physical information in the water area environment, such as factors like sound speed and terrain.

[0003] However, the methods for predicting the propagation loss of an underwater acoustic channel in related technologies have great limitations. On the one hand, these methods only consider using the physical information of the environment as input and cannot utilize other types of input, such as some measured propagation loss data, etc. On the other hand, these methods generally rely on specific mathematical models and assumption conditions, with limited environmental information considered. Since the underwater acoustic channel has characteristics such as a complex scattering environment and strong time-variability, and is greatly affected by geographical environment and random factors, it is likely to lead to low modeling accuracy, and some modeling methods have high complexity, introducing a large computational delay, which urgently needs to be solved. Summary of the Invention

[0004] The present application provides a method and device for predicting the propagation loss distribution of an underwater acoustic channel to solve the problems in related technologies that the methods for predicting the propagation loss of an underwater acoustic channel have great limitations, only consider using the physical information of the environment as input and cannot utilize other types of input, and these methods generally rely on specific mathematical models and assumption conditions, with limited environmental information considered, which is likely to lead to low modeling accuracy, thus affecting the accuracy of the prediction result of the propagation loss of the underwater acoustic channel, and some modeling methods have high complexity, introducing a large computational delay.

[0005] An embodiment of the first aspect of the present application provides a method for predicting the propagation loss distribution of an underwater acoustic channel, including the following steps: Based on the sound speed profile of the water area where the propagation loss distribution of the underwater acoustic channel to be predicted is located, obtain the first sound source depth in the depth pair conjugate to the channel axis in the water area and the target sound source depth corresponding to the propagation loss distribution of the underwater acoustic channel to be predicted; Based on the resolution requirement of the propagation loss distribution of the underwater acoustic channel in the water area, determine the propagation loss distribution of the underwater acoustic channel at the first sound source depth and determine the free space propagation loss distribution at the target sound source depth; Perform clipping processing and normalization processing on the propagation loss distribution of the underwater acoustic channel and the free space propagation loss distribution to obtain the processed propagation loss distribution of the underwater acoustic channel and the free space propagation loss distribution, and merge the processed propagation loss distribution of the underwater acoustic channel and the free space propagation loss distribution in the channel dimension to obtain the merged propagation loss distribution; Input the merged propagation loss distribution into a target neural network to output the predicted result of the normalized propagation loss distribution at the target sound source depth, and determine the predicted result of the propagation loss distribution of the underwater acoustic channel at the target sound source depth according to the predicted result of the normalized propagation loss distribution.

[0006] Optionally, in an embodiment of the present application, the determining the propagation loss distribution of the underwater acoustic channel at the first sound source depth and determining the free space propagation loss distribution at the target sound source depth based on the resolution requirement of the propagation loss distribution of the underwater acoustic channel in the water area includes: determining the resolution in the first target dimension and the resolution in the second target dimension of the propagation loss distribution of the underwater acoustic channel according to the resolution requirement; determining the propagation loss distribution of the underwater acoustic channel at the first sound source depth and the free space propagation loss distribution at the target sound source depth according to the resolution in the first target dimension and the resolution in the second dimension.

[0007] Optionally, in an embodiment of the present application, before determining the resolution in the first target dimension and the resolution in the second target dimension of the propagation loss distribution of the underwater acoustic channel according to the resolution requirement, it further includes: obtaining the application requirement of the propagation loss distribution of the underwater acoustic channel to be predicted and the acquisition method of the propagation loss distribution of the underwater acoustic channel at the first sound source depth; determining the resolution requirement by combining the application requirement and the acquisition method.

[0008] Optionally, in an embodiment of the present application, before inputting the combined propagation loss distribution into the target neural network, it further includes: collecting propagation loss distribution data of depth pairs conjugate to the sound channel axis that meet the resolution requirement in the target water area, and preprocessing the propagation loss distribution data to construct a training dataset based on the preprocessed propagation loss distribution data; training the initialized neural network using the training dataset, the objective function, and the objective optimizer to obtain an initial neural network, where the objective function includes masked mean squared error and masked image gradient; fine-tuning the initial neural network using the propagation loss distribution data of the water area to determine the target neural network.

[0009] Optionally, in an embodiment of the present application, the determining the underwater acoustic channel propagation loss distribution of the target sound source depth according to the normalized propagation loss distribution prediction result includes: performing inverse processing of the clipping process and inverse processing of the normalization process on the normalized propagation loss distribution prediction result to determine the prediction result of the underwater acoustic channel propagation loss distribution of the target sound source depth.

[0010] An embodiment of the second aspect of the present application provides a prediction device for underwater acoustic channel propagation loss distribution, including: an acquisition module, configured to obtain the first sound source depth in the depth pairs conjugate to the sound channel axis in the water area and the target sound source depth corresponding to the underwater acoustic channel propagation loss distribution to be predicted based on the sound speed profile of the water area where the underwater acoustic channel propagation loss distribution to be predicted is located; a determination module, configured to determine the underwater acoustic channel propagation loss distribution of the first sound source depth and determine the free space propagation loss distribution of the target sound source depth based on the resolution requirement of the underwater acoustic channel propagation loss distribution to be predicted; a merging module, configured to perform clipping processing and normalization processing on the underwater acoustic channel propagation loss distribution and the free space propagation loss distribution to obtain the processed underwater acoustic channel propagation loss distribution and free space propagation loss distribution, and merge the processed underwater acoustic channel propagation loss distribution and free space propagation loss distribution in the channel dimension to obtain a combined propagation loss distribution; a prediction module, configured to input the combined propagation loss distribution into a target neural network to output a normalized propagation loss distribution prediction result of the target sound source depth, and determine a prediction result of the underwater acoustic channel propagation loss distribution of the target sound source depth according to the normalized propagation loss distribution prediction result.

[0011] Optionally, in an embodiment of the present application, the determining module includes: a first determining unit, configured to determine the resolution of the underwater acoustic channel propagation loss distribution in a first target dimension and the resolution in a second target dimension according to the resolution requirement; a second determining unit, configured to determine the underwater acoustic channel propagation loss distribution of the first sound source depth and the free space propagation loss distribution of the target sound source depth according to the resolution of the first target dimension and the resolution of the second dimension.

[0012] Optionally, in an embodiment of the present application, it further includes: an obtaining subunit, configured to obtain the application requirement of the underwater acoustic channel propagation loss distribution to be predicted and the obtaining manner of the underwater acoustic channel propagation loss distribution of the first sound source depth before determining the resolution of the underwater acoustic channel propagation loss distribution in the first target dimension and the resolution in the second target dimension according to the resolution requirement; a determining subunit, configured to determine the resolution requirement by combining the application requirement and the obtaining manner.

[0013] Optionally, in an embodiment of the present application, it further includes: a collecting module, configured to collect the propagation loss distribution data of the depth pairs conjugate to the sound channel axis that meet the resolution requirement in the target water area before inputting the merged propagation loss distribution into the target neural network, and preprocess the propagation loss distribution data, so as to construct a training data set according to the preprocessed propagation loss distribution data; a training module, configured to train the initialized neural network by using the training data set, a target function, and a target optimizer to obtain an initial neural network, where the target function includes a masked mean squared error and a masked image gradient; a constructing module, configured to fine-tune the initial neural network by using the propagation loss distribution data of the water area to determine the target neural network.

[0014] Optionally, in an embodiment of the present application, the prediction module includes: a prediction unit, configured to perform the inverse processing of the clipping process and the inverse processing of the normalization process on the normalized propagation loss distribution prediction result, so as to determine the prediction result of the underwater acoustic channel propagation loss distribution of the target sound source depth.

[0015] An embodiment of the third aspect of the present application provides an electronic device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the program to implement the prediction method of the underwater acoustic channel propagation loss distribution as described in the above embodiment.

[0016] An embodiment of the fourth aspect of the present application provides a computer-readable storage medium, where the computer-readable storage medium stores a computer program, and when the program is executed by a processor, it implements the prediction method of the underwater acoustic channel propagation loss distribution as described above.

[0017] The fifth aspect of the present application provides a computer program product, including a computer program, which, when executed, is used to implement the above method for predicting the distribution of underwater acoustic channel propagation loss.

[0018] The embodiment of the present application can use a target neural network to predict the propagation loss distribution of the underwater acoustic channel at the target sound source depth conjugated with the first sound source depth about the sound channel axis based on the known propagation loss distribution of the underwater acoustic channel at the first sound source depth. Thus, the principle that the propagation loss distribution of the two sound source depths has a certain correlation by utilizing the characteristic that the depth conjugated with the sound channel axis has the same sound speed is realized, and the known propagation loss distribution at any sound source depth that has been measured is effectively used to predict the propagation loss distribution at the sound source depth conjugated with the sound source depth about the sound channel axis, which greatly reduces the calculation complexity and calculation time in the prediction process, and the target neural network in the present application can be fine-tuned according to the propagation loss data of the actual application waters, which effectively improves the application ability and scope of application of the target neural network in the actual application waters, thereby improving the accuracy of the propagation loss distribution at the target sound source depth predicted by the present application. This solves the problem that the underwater acoustic channel propagation loss prediction methods in the relevant technologies have great limitations. They only consider the physical information of the environment as input and cannot use other types of input. In addition, these methods are generally based on specific mathematical models and assumptions, and the environmental information considered is limited, which can easily lead to low modeling accuracy, thereby affecting the accuracy of the underwater acoustic channel propagation loss prediction results. In addition, some modeling methods are highly complex and will introduce problems such as large calculation delays.

[0019] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0021] Figure 1 A flowchart of a method for predicting the distribution of propagation loss in an underwater acoustic channel provided according to an embodiment of the present application;

[0022] Figure 2 A flowchart of a neural network training according to an embodiment of the present application;

[0023] Figure 3 This is a flow chart of a neural network reconstruction method for unknown waters according to one embodiment of the present application;

[0024] Figure 4 A flowchart of a method for predicting the distribution of propagation loss in an underwater acoustic channel according to an embodiment of the present application;

[0025] Figure 5 It is a schematic structural diagram of a prediction device for the propagation loss distribution of an underwater acoustic channel provided according to an embodiment of the present application;

[0026] Figure 6 It is a schematic structural diagram of an electronic device provided according to an embodiment of the present application.

[0027] Reference numerals:

[0028] 10 - Prediction device for the propagation loss distribution of an underwater acoustic channel: 100 - Acquisition module, 200 - Determination module, 300 - Merging module, and 400 - Prediction module; 601 - Memory, 602 - Processor, and 603 - Communication interface. Detailed implementation manners

[0029] The embodiments of the present application will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described by referring to the accompanying drawings are exemplary and are intended to explain the present application, and should not be construed as a limitation to the present application.

[0030] The following describes a method and apparatus for predicting the underwater acoustic channel propagation loss distribution according to an embodiment of the present application. In view of the large limitations of the underwater acoustic channel propagation loss prediction methods in the related art mentioned in the above background art, which only consider using the physical information of the environment as input and cannot utilize other types of input, and these methods generally rely on specific mathematical models and assumptions, with limited environmental information considered, easily leading to low modeling accuracy, thereby affecting the accuracy of the underwater acoustic channel propagation loss prediction results, and some modeling methods have high complexity, introducing a large computational delay. The present application provides a method for predicting the underwater acoustic channel propagation loss distribution. In this method, based on the known underwater acoustic channel propagation loss distribution at the first sound source depth, a target neural network is used to predict the underwater acoustic channel propagation loss distribution at the target sound source depth conjugate to the first sound source depth with respect to the sound channel axis. Thus, by leveraging the property that the sound speeds are consistent at depths conjugate to the sound channel axis, and therefore there is a certain correlation between the propagation loss distributions at the two sound source depths, the known propagation loss distribution at any measured sound source depth is effectively utilized to predict the propagation loss distribution at the sound source depth conjugate to this sound source depth with respect to the sound channel axis, greatly reducing the computational complexity and computational time in the prediction process. Moreover, the target neural network in the present application can be fine-tuned according to the propagation loss data of the actual application water area, effectively improving the application ability and applicable range of the target neural network in the actual application water area, and further enhancing the accuracy of the predicted propagation loss distribution at the target sound source depth. Thus, the problems in the related art of the underwater acoustic channel propagation loss prediction method having large limitations, only considering using the physical information of the environment as input, unable to utilize other types of input, and these methods generally relying on specific mathematical models and assumptions, with limited environmental information considered, easily leading to low modeling accuracy, thereby affecting the accuracy of the underwater acoustic channel propagation loss prediction results, and some modeling methods having high complexity, introducing a large computational delay, etc. are solved.

[0031] Specifically, Figure 1 FIG. is a flowchart of a method for predicting the underwater acoustic channel propagation loss distribution provided by an embodiment of the present application.

[0032] As Figure 1 shown, the method for predicting the underwater acoustic channel propagation loss distribution includes the following steps:

[0033] In step S101, based on the sound speed profile of the water area where the underwater acoustic channel propagation loss distribution to be predicted is located, the first sound source depth in the depth pair conjugate to the sound channel axis in the water area and the target sound source depth corresponding to the underwater acoustic channel propagation loss distribution to be predicted are obtained.

[0034] It can be understood that the underwater acoustic channel propagation loss distribution refers to the distribution of sound intensity attenuation caused by various factors (such as spreading, absorption, scattering, boundary reflection, etc.) during the underwater propagation of sound waves at different distances, different frequencies, and different water area environments, which can be represented by a two-dimensional image in a cylindrical coordinate system.

[0035] In some embodiments, the sound speed in the water area will change with different depths. Based on this, the present application can obtain the first sound source depth in the depth pair conjugate to the sound channel axis in the water area where the underwater acoustic channel propagation loss distribution to be predicted is located, and the target sound source depth corresponding to the underwater acoustic channel propagation loss distribution to be predicted.

[0036] Herein, the underwater acoustic channel propagation loss distribution to be predicted refers to the unknown underwater acoustic channel propagation loss distribution at a certain sound source depth in the water area. The sound speed profile of the water area can be understood here as the variation relationship of the sound speed with the water area depth, and can also be called the sound speed-depth function relationship. According to the sound speed profile of the water area, the embodiments of the present application can understand the situation of the sound speed changing with the depth in the water area.

[0037] Among them, the sound speed profile in the embodiments of the present application is adapted to the specific water area characteristics, and the source can be obtained by simulation or measured, and the type can be the Munk sound speed profile or other sound speed profiles. Only exemplary descriptions are made in the embodiments of the present application without specific limitations.

[0038] The conjugate depth of the sound channel axis is closely related to the sound speed profile. Among them, the sound channel axis, also known as the sound channel axis depth, refers to the depth where the sound speed is the smallest in the sound speed profile here. When the sound wave propagates near the sound channel axis depth, due to the change of the sound speed gradient, the sound wave is easily refracted and propagates along the sound channel axis, forming a deep sea sound channel. The conjugate depth refers to the depth below the sound channel axis where the sound speed value is equal to the sound speed value at a certain depth above the sound channel axis (usually the sound source position). That is, two sound source depths located above and below the sound channel axis respectively, and with almost the same sound speed.

[0039] For example, in the embodiments of the present application, a certain depth above the sound channel axis can be represented by the first sound source depth, and the sound source depth corresponding to the underwater acoustic channel propagation distribution loss to be predicted below the sound channel axis can be represented by the target sound source depth. Among them, the target sound source depth is greater than the first sound source depth.

[0040] The depths conjugate to the sound channel axis have the characteristic of consistent sound speed. Therefore, there is a certain correlation between the propagation loss distributions of the two sound source depths. Thus, the embodiments of the present application can use data information such as the first sound source depth and the target sound source depth conjugate to the sound channel axis and the sound speed profile as data support in the process of predicting the underwater acoustic channel propagation distribution loss to be predicted.

[0041] It should be noted that for waters corresponding to sound speed profiles with different parameters, the position of the sound channel axis may be different, the depth pair of sound channel axis conjugates may also be different, and the characteristics of the propagation loss distribution may also be different. However, in the embodiments of the present application, the propagation loss distribution corresponding to the sound source depth below the sound channel axis can be predicted using the propagation loss distribution of the sound source depth above the sound channel axis.

[0042] Step S102: Based on the resolution requirement of the propagation loss distribution of the underwater acoustic channel to be predicted, determine the propagation loss distribution of the underwater acoustic channel at the first sound source depth, and determine the free space propagation loss distribution at the target sound source depth.

[0043] It can be understood that the resolution here refers to the smallest unit that can distinguish changes in propagation loss in the water area; the resolution requirement can be understood as how much the smallest unit requirement is, for example, ten meters, fifteen meters, etc.

[0044] In some embodiments, the present application will first determine the resolution requirement of the propagation loss distribution of the underwater acoustic channel to be predicted, then determine the propagation loss distribution of the underwater acoustic channel corresponding to the first sound source depth through simulation or actual measurement under the condition of meeting this resolution requirement, and determine the free space propagation loss distribution at the target sound source depth through simulation under the condition of meeting this resolution requirement.

[0045] It should be noted that only exemplary explanations are given in the embodiments of the present application on how to specifically obtain the propagation loss distribution of the underwater acoustic channel corresponding to the first sound source depth, without specific limitations. For example, if the first sound source depth is relatively deep or the actual measurement is difficult, simulation can be used to obtain it. Under the condition that it is allowed, both simulation and actual measurement can be used to obtain it.

[0046] Next, a further explanation of this process will be given.

[0047] Optionally, in an embodiment of the present application, based on the resolution requirement of the propagation loss distribution of the underwater acoustic channel to be predicted, determining the propagation loss distribution of the underwater acoustic channel at the first sound source depth and determining the free space propagation loss distribution at the target sound source depth includes: determining the resolution of the propagation loss distribution of the underwater acoustic channel in the first target dimension and the resolution in the second target dimension according to the resolution requirement; determining the propagation loss distribution of the underwater acoustic channel at the first sound source depth and the free space propagation loss distribution at the target sound source depth according to the resolution in the first target dimension and the resolution in the second dimension.

[0048] Based on the relevant descriptions of other embodiments, it can be understood that the present application can determine the propagation loss distribution of the underwater acoustic channel at the first sound source depth based on the resolution requirement of the propagation loss distribution of the underwater acoustic channel to be predicted.

[0049] In the actual implementation process, the present application first determines the resolution sizes m of the propagation loss distribution in two dimensions, such as the first target dimension and the second target dimension, according to the resolution requirements of the propagation loss distribution required for predicting the underwater acoustic channel propagation loss distribution. d ,m r . Among them, the first target dimension can be understood as depth here, but is not limited thereto, and the second target dimension can be understood as the horizontal distance from the sound source here, but is not limited thereto. It should be noted that the sound source in the horizontal distance from the sound source here does not distinguish between the first sound source or the target sound source, because the values of the first sound source and the target sound source in the dimension of the horizontal distance are the same and there is no difference.

[0050] After determining the resolution sizes in the first target dimension and the second target dimension, the embodiments of the present application can obtain the underwater acoustic channel propagation loss distribution TL corresponding to the depth of the first sound source through a simulation software, actual measurement, or other means, etc. 1 , which is represented in the form of a matrix as a matrix of size m d ×m r . Among them, each element in the matrix corresponds to the propagation loss at each grid point position of each resolution. And the grid point positions corresponding to the elements in the matrix should be evenly distributed in the two dimensions of depth and distance. Here, the "each grid point position of each resolution" refers to the grid points divided according to a certain resolution in these two dimensions. The higher the resolution, the more points need to be simulated or collected, and the higher the cost.

[0051] After obtaining the underwater acoustic channel propagation loss distribution TL 1 corresponding to the depth of the first sound source, the embodiments of the present application can also obtain the free space propagation loss distribution TL free corresponding to the depth of the target sound source through simulation based on this resolution requirement, that is, the resolution of the free space propagation loss distribution of the depth of the target sound source should be the same as the resolution of the underwater acoustic channel propagation distribution loss distribution of the depth of the first sound source.

[0052] For example, the present application can determine that the resolution sizes of the two dimensions (depth, horizontal distance from the sound source) of the propagation loss distribution are m d =256, m r =512 according to the resolution requirements, and the corresponding true depth range and horizontal distance range are 5000m and 200km respectively.

[0053] After inputting physical information such as the sound speed profile of the water area through relevant channel simulation software, the underwater acoustic channel propagation loss distribution TL 1 corresponding to the depth d of the first sound source is obtained 1 , that is, a matrix of size 256×512, and its elements correspond to the propagation loss at each grid point position of each of the above resolutions. Then, through simulation, the depth of the target sound source is d2 Free space propagation loss distribution TL at free , whose resolution is the same as the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source.

[0054] It should be noted that the resolution in the embodiments of the present application can be utilized in the underwater acoustic channel propagation loss distribution TL corresponding to the depth of the first sound source 1 and the free space propagation loss distribution TL corresponding to the depth of the target sound source free and the propagation loss distributions corresponding to other sound source depths, and the embodiments of the present application do not make specific limitations on this.

[0055] Optionally, in an embodiment of the present application, before determining the resolution in the first target dimension and the resolution in the second target dimension of the underwater acoustic channel propagation loss distribution according to the resolution requirement, it further includes: obtaining the application requirement of the underwater acoustic channel propagation loss distribution to be predicted and the acquisition method of the underwater acoustic channel propagation loss distribution at the depth of the first sound source; determining the resolution requirement by combining the application requirement and the acquisition method.

[0056] In some embodiments, the present application can determine the resolution of the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source and the resolution of the free space propagation loss distribution corresponding to the depth of the target sound source based on the resolution requirement. However, considering the constraints of actual conditions, the embodiments of the present application can obtain the application requirement of the underwater acoustic channel propagation loss distribution to be predicted and the acquisition channel of the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source, and then determine the resolution in the first target dimension and the resolution in the second target dimension.

[0057] For example, if the propagation loss distribution corresponding to the depth of the first sound source is from simulation data and is relatively easy to obtain, a relatively large resolution m d ×m r can be taken; if the propagation loss distribution corresponding to the depth of the first sound source is from measured data and the measurement is difficult, a relatively small resolution m d ×m r can be taken.

[0058] Step S103, perform clipping processing and normalization processing on the underwater acoustic channel propagation loss distribution and the free space propagation loss distribution to obtain the processed underwater acoustic channel propagation loss distribution and the free space propagation loss distribution, and merge the processed underwater acoustic channel propagation loss distribution and the free space propagation loss distribution in the channel dimension to obtain the merged propagation loss distribution.

[0059] In other embodiments, after obtaining the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source and the free space propagation loss corresponding to the depth of the target sound source, for the convenience of subsequent processing, the embodiments of the present application can also perform clipping processing and normalization processing on the two.

[0060] Among them, the upper and lower bounds of amplitude limiting in the embodiments of the present application can be, but are not limited to, determined by the data characteristics of the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source. For example, since the propagation loss in the near field (which can be obtained by simulation or actual measurement) has a certain magnitude (the simulation or actual measurement position is at a certain distance from the sound source, so there is a certain attenuation in the propagation of energy), m d ×m r the minimum value of the propagation loss within the resolution grid can be used as the lower bound of amplitude limiting; the propagation loss in the propagation shadow area is numerically very large, but its physical meaning is the area where energy cannot propagate to. Therefore, the upper bound of amplitude limiting can be taken slightly higher than the maximum value of the propagation loss in the non-propagation shadow area, as long as the amplitude-limited result can characterize the propagation shadow area.

[0061] For example, the lowest propagation loss obtained by simulation is generally about 40 dB. Except for the propagation shadow area, the highest propagation loss is not higher than 140 dB. Therefore, the upper and lower bounds of amplitude limiting can be taken as 40 dB and 140 dB respectively, and normalization processing is performed. After data processing, the propagation loss in the propagation shadow area is normalized to 1. The underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source after data processing can be expressed as:

[0062]

[0063] Similarly, the free space propagation loss distribution corresponding to the depth of the target sound source after processing can be obtained

[0064] After obtaining the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source and the free space propagation loss distribution corresponding to the depth of the target sound source after amplitude limiting processing and normalization processing they can be merged in the channel dimension to obtain the merged propagation loss distribution. After that, the merged propagation loss distribution can be input into the target neural network to output the prediction result of the normalized propagation loss distribution of the target sound source depth, and the prediction result of the underwater acoustic channel propagation loss distribution of the target sound source depth can be determined according to the prediction result of the normalized propagation loss distribution.

[0065] Step S104: Input the merged propagation loss distribution into the target neural network to output the prediction result of the normalized propagation loss distribution of the target sound source depth, and determine the prediction result of the underwater acoustic channel propagation loss distribution of the target sound source depth according to the prediction result of the normalized propagation loss distribution.

[0066] In some embodiments, after obtaining the merged propagation loss distribution, the present application can input the merged propagation loss distribution into the target neural network for processing to predict the normalized propagation loss distribution of the target sound source depth, and then obtain the underwater acoustic channel propagation loss distribution of the target sound source depth.

[0067] Among them, the target neural network can be understood here as a relevant neural network that can process the combined propagation loss distribution and output the results desired by the embodiments of the present application. For example, the target neural network in the embodiments of the present application can adopt an autoencoder structure, that is, use an encoder to compress the input image into a multi-channel feature space, and then use a decoder to restore it to the physical space. This autoencoder can have various different designs, such as using a typical UNet network, and structures such as residual networks and multi-resolution networks can also be considered.

[0068] In addition, the input of the target neural network in the embodiments of the present application includes but is not limited to two channels. One channel can be used to input the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source in the water area, and the other channel can be used to input the free space propagation loss distribution corresponding to the depth of the target sound source.

[0069] It should be noted that the role of inputting the free space propagation loss distribution corresponding to the depth of the target sound source is mainly to prompt the sound source depth information of the propagation loss distribution to be predicted and the information of the propagation loss changing with the propagation distance. It is not a necessary input for the target neural network in the embodiments of the present application. That is, it is also possible not to provide the free space propagation loss distribution corresponding to the depth of the target sound source to the target neural network, and the present application does not make specific restrictions. However, the underwater acoustic channel propagation loss distribution corresponding to the depth of the first sound source is a necessary input for the target neural network in the embodiments of the present application.

[0070] Optionally, in an embodiment of the present application, determining the underwater acoustic channel propagation loss distribution of the target sound source depth according to the prediction result of the normalized propagation loss distribution includes: performing inverse processing of the clipping process and inverse processing of the normalization process on the prediction result of the normalized propagation loss distribution to determine the prediction result of the underwater acoustic channel propagation loss distribution of the target sound source depth.

[0071] In the actual execution process, when determining the prediction result of the underwater acoustic channel propagation distribution loss of the target sound source depth according to the prediction result of the normalized propagation loss distribution output by the target neural network, it can be but is not limited to being achieved by performing inverse processing of the clipping process and inverse processing of the normalization process on the prediction result of the normalized propagation loss distribution.

[0072] Among them, the inverse processing of the clipping process and the inverse processing of the normalization process can be but is not limited to being achieved by performing coefficient multiplication processing and bias addition processing on the normalized propagation loss distribution in the embodiments of the present application, and finally obtaining the true propagation loss distribution of the target sound source depth

[0073] It should be noted that the coefficient multiplied in the coefficient multiplication processing and the bias added in the bias addition processing in the embodiments of the present application can both be determined by the upper and lower bounds of the above-mentioned clipping operation.

[0074] Optionally, in an embodiment of the present application, before inputting the combined propagation loss distribution into the target neural network, it further includes: collecting propagation loss distribution data of depth pairs conjugate to the sound channel axis that meet the resolution requirements in the target water area, and preprocessing the propagation loss distribution data to construct a training dataset based on the preprocessed propagation loss distribution data; training the initialized neural network using the training dataset, the target function, and the target optimizer to obtain an initial neural network, where the target function includes masked mean squared error and masked image gradient; fine-tuning the initial neural network using the propagation loss distribution data of the water area to determine the target neural network.

[0075] In some embodiments, before inputting the combined propagation loss distribution into the target neural network, the present application needs to first construct the target neural network. During the construction process, it is necessary to first collect the propagation loss distribution data in the target water area, then preprocess the propagation loss distribution data, then randomly initialize the neural network, and use the propagation loss distribution data of the water area, the target loss function, and the target optimizer to train the initialized neural network to determine the target neural network.

[0076] Among them, the target water area can be understood here as any water area environment that can obtain certain propagation loss distribution data and can perform prediction of the propagation loss distribution of the underwater acoustic channel. The target loss function here refers to the loss function established to reduce the prediction error between the prediction result and the true result of the neural network. The target optimizer can be understood here as the optimization algorithm for training the neural network. The target neural network can be understood here as the neural network that has been fine-tuned using the data of any actual application water area and can be applied to any actual application water area after training.

[0077] For example, for a specific water area environment, the embodiment of the present application can first obtain certain propagation loss data through simulation or actual measurement. If a dataset is obtained through simulation, some classic methods can be used, but are not limited to, such as the ray model. It should be noted that the cost of obtaining a dataset through simulation is relatively low, so it is suitable for scenarios with a high resolution of propagation loss modeling. However, since classic modeling methods are prone to problems such as poor modeling accuracy, the performance of the neural network obtained by training with simulation data may decline during subsequent inference; while the cost of obtaining a dataset through actual measurement is relatively high, so it is more suitable for scenarios with a lower resolution requirement for propagation loss modeling; and for the free space propagation loss distribution at the depth of the sound source to be predicted below the sound channel axis in the target water area, it can only be obtained through simulation.

[0078] After obtaining the data, the propagation loss distribution of the underwater acoustic channel corresponding to the source depth above the sound channel axis and the free space propagation loss distribution corresponding to the target source depth below the sound channel axis in the preprocessed target water area are also used as input data, and the true value of the propagation loss distribution of the underwater acoustic channel corresponding to the target source depth below the sound channel axis in the target water area is used as the data label to form a data set, which is then split into a training set and a test set for training and testing the initial neural network. Here, the initial neural network refers to a neural network that has not been trained and tested with a data set from an actual application water area.

[0079] When training the initial neural network using the training set, the input of the initial neural network is also the two-channel propagation loss distribution after preprocessing operations such as clipping and normalization. The two channels are the propagation loss distribution of the underwater acoustic channel corresponding to the source depth above the sound channel axis in the preprocessed target water area and the free space propagation loss distribution corresponding to the target source depth below the sound channel axis in the target water area.

[0080] Next, the embodiment of the present application can use the output of the network, that is, the predicted normalized propagation loss distribution below the sound channel axis, combined with its true value, that is, the true value of the propagation loss distribution of the underwater acoustic channel corresponding to the target source depth below the sound channel axis in the target water area, to calculate the loss function.

[0081] Among them, the target loss function of the initial neural network in the embodiment of the present application can be but is not limited to being composed of masked mean squared error (MSE) and masked image gradient (IG). Among them, the masked mean squared error (MSE) can reduce the gap between the propagation loss distribution output by the target neural network and the true propagation loss distribution used as the label, and the masked image gradient (IG) can enable the target neural network to better learn the change trend of the propagation loss. The role of the mask is to make the network focus on key areas during learning, such as the boundaries of the propagation shadow area and the convergence area with smaller propagation loss, etc.

[0082] Finally, the embodiment of the present application can perform backpropagation by taking the derivative of the target loss function, that is, automatically obtain the gradient of each layer using the chain rule. After obtaining the gradient, the ADAM optimizer can be used to update the parameters of each layer.

[0083] Figure 2 is a flowchart of neural network training for an embodiment of the present application. As Figure 2 shown, the training process of the neural network in the embodiment of the present application can be but is not limited to being expressed as follows:

[0084] S201: Obtain the propagation loss distribution data set through actual measurement or simulation, and perform preprocessing such as channel merging, clipping, and normalization to obtain the preprocessed data set.

[0085] S202: Perform data pre-forwarding, that is, input the preprocessed dataset into the initialized neural network to predict the underwater acoustic channel propagation loss distribution corresponding to the sound source depth below the sound channel axis with the underwater acoustic channel propagation loss distribution corresponding to the sound source depth above the sound channel axis.

[0086] S203: Input the underwater acoustic channel propagation loss distribution corresponding to the sound source depth below the sound channel axis obtained by prediction and the labels in the dataset into the propagation loss function calculator to calculate the loss function. Among them, the loss function includes but is not limited to two parts: masked MSE and masked IG.

[0087] S204: Take the derivative of the loss function, perform backpropagation based on the loss function, and automatically obtain the gradients of each layer by automatic differentiation.

[0088] S205: Each layer automatically updates the entire network parameters according to the calculated gradients, using an optimizer (such as ADAM) and an appropriate learning rate. The entire training process is reset and looped.

[0089] To improve the generalization of the target neural network in different waters and reduce data acquisition and training costs, the embodiments of the present application can fine-tune the neural network using a small amount of propagation loss distribution data in unknown waters to determine the target neural network applicable to the actual application waters. First, the target neural network P(·) trained for the known water area A can be prepared, and a dataset in the unknown water area B, including a training set and a test set, can be established according to the dataset acquisition method in the embodiments of the present application. Note that the number of data contained in the dataset can be small.

[0090] Then load the model P(·), and on this basis, fine-tune it using the dataset in the unknown water area. The method remains the same, but the parameters of the mask in the loss function can be adjusted according to the actual situation. The embodiments of the present application do not make specific limitations. The fine-tuned model P B (·) can be used for predicting the loss distribution of the underwater acoustic channel propagation distribution in the unknown water area.

[0091] Figure 3 It is a flowchart of the method for reconstructing the neural network in the unknown water area according to an embodiment of the present application. As Figure 3 shown, the process of reconstructing the neural network in the unknown water area in the embodiments of the present application can be but is not limited to expressed as follows:

[0092] S301: Obtain the propagation loss distribution data of the unknown water area B through actual measurement or simulation, and perform preprocessing such as channel merging, clipping, and normalization to construct a propagation loss distribution dataset in the water area B.

[0093] S302: Load the model trained with the dataset in other water areas such as water area A.

[0094] S303: Use the B water area data set for data forward transmission to complete the process of using the propagation loss distribution corresponding to the sound source depth above the sound channel axis to predict the propagation loss distribution corresponding to the sound source depth conjugate with the depth about the sound channel axis.

[0095] S304: Input the predicted propagation loss distribution and the labels in the data set into the propagation loss function calculator to calculate the loss function. The loss function includes two parts: the masked MSE and the masked IG.

[0096] S305: Deriving the loss function, performing back propagation based on the loss function, and automatically deriving to obtain the gradients of each layer.

[0097] S306: Each layer automatically updates the parameters of the entire network based on the calculated gradients using an optimizer (such as ADAM) and a suitable learning rate. The entire training process is reset and repeated.

[0098] It should be noted that the free space propagation loss distribution of the target sound source depth in the embodiment of the present application can also provide the target neural network with the sound source depth and the change of propagation loss with distance of the propagation loss distribution to be predicted, thereby improving the prediction ability of the target neural network in the embodiment of the present application.

[0099] The present application is described in detail below with reference to a specific embodiment.

[0100] Figure 4 This is a flow chart of a method for predicting the distribution of underwater acoustic channel propagation loss according to an embodiment of the present application. Figure 4 As shown:

[0101] S401: Based on the sound velocity profile of water area A, obtain N according to the requirements of the sound source depth corresponding to the propagation loss distribution to be studied. A A pair of depths conjugate with respect to the vocal tract axis. A pair of depths contains two depths d above and below the vocal tract axis, respectively, and the sound speed is almost the same 1 ,d 2 , where d 2 That is, the sound source depth corresponding to the propagation loss distribution to be predicted below the vocal tract axis, d 2 >d 1 ;

[0102] S402: Determine the resolution m of the two dimensions (depth and horizontal distance from the sound source) of the propagation loss distribution according to the required propagation loss distribution resolution requirement d ,m r In this example, m is selected d =256,m r= 512, and the corresponding true depth range and horizontal distance range are 5000 m and 200 km respectively; then, after inputting physical information such as the sound speed profile of Water Area A into the Bellhop channel simulation software, the underwater acoustic channel propagation loss distribution TL 1 is obtained when the sound source depth is d 1 , that is, a matrix of size 256×512, whose elements correspond to the propagation loss at each grid point position of each of the above resolutions; and then the free space propagation loss distribution TL 2 is obtained when the sound source depth is d free , and its resolution is the same as that of the propagation loss distribution when the sound source depth is d 1 ;

[0103] S403: Perform clipping and normalization processing on the above propagation loss distribution. In the embodiments of the present application, the lowest propagation loss obtained by simulation is generally about 40 dB, and except for the propagation shadow area, the highest propagation loss is not higher than 140 dB. Therefore, the clipping upper and lower bounds are taken as 40 dB and 140 dB respectively, and normalization processing is performed. After the data preprocessing operation, the propagation loss in the propagation shadow area is normalized to 1. The propagation loss distribution when the sound source depth after data preprocessing is d 1 can be but is not limited to being expressed as:

[0104]

[0105] Similarly, the free space propagation loss distribution when the sound source depth after preprocessing is d 2 is obtained

[0106] S404: After merging the two propagation loss distributions after preprocessing in the channel dimension, input them into the neural network. In this example, a multi-resolution UNet network is used to obtain the normalized propagation loss distribution when the sound source depth is d 2 can be but is not limited to being expressed as::

[0107]

[0108] where Θ P are the learnable parameters of the network.

[0109] S405: Perform data processing on the propagation loss distribution output by the network, that is, perform the inverse operation of the above data clipping and normalization operations to obtain the propagation loss distribution which can be expressed as:

[0110]

[0111] If it is necessary to reconstruct the model in water area B, only need to load the model, prepare the water area B data set by the same method in S301, and then fine-tune the model to carry out actual applications.

[0112] According to the prediction method of the underwater acoustic channel propagation loss distribution proposed in the embodiments of the present application, based on the known underwater acoustic channel propagation loss distribution at the first sound source depth, a target neural network can be used to predict the underwater acoustic channel propagation loss distribution at the target sound source depth conjugate to the first sound source depth with respect to the sound channel axis. Thus, by using the characteristic that the sound speeds are the same at the depths conjugate to the sound channel axis, and there is a certain correlation between the propagation loss distributions at the two sound source depths, the known propagation loss distribution at any sound source depth that has been actually measured can be effectively used to predict the propagation loss distribution at the sound source depth conjugate to this sound source depth with respect to the sound channel axis, greatly reducing the computational complexity and computational time in the prediction process. Moreover, the target neural network in the present application can be fine-tuned according to the propagation loss data of the actual application water area, effectively improving the application ability and applicable range of the target neural network in the actual application water area, and further improving the accuracy of the predicted propagation loss distribution at the target sound source depth in the present application. Thereby, it solves the problems that the underwater acoustic channel propagation loss prediction methods in the related technologies have great limitations, only consider using the physical information of the environment as input, cannot use other types of input, and these methods are generally based on specific mathematical models and assumption conditions, with limited environmental information considered, which easily leads to low modeling accuracy, thus affecting the accuracy of the underwater acoustic channel propagation loss prediction results, and some modeling methods have high complexity and will introduce large computational delays.

[0113] Next, a prediction device for the underwater acoustic channel propagation loss distribution proposed in the embodiments of the present application will be described with reference to the accompanying drawings.

[0114] Figure 5 It is a schematic structural diagram of a prediction device for the underwater acoustic channel propagation loss distribution according to an embodiment of the present application.

[0115] As Figure 5 shown, the prediction device 10 for the underwater acoustic channel propagation loss distribution includes: an acquisition module 100, a determination module 200, a merging module 300, and a prediction module 400.

[0116] Among them, the acquisition module 100 is configured to obtain, based on the sound speed profile of the water area where the underwater acoustic channel propagation loss distribution to be predicted is located, the first sound source depth in the depth pair conjugate to the sound channel axis in the water area and the target sound source depth corresponding to the underwater acoustic channel propagation loss distribution to be predicted.

[0117] The determination module 200 is configured to determine the underwater acoustic channel propagation loss distribution at the first sound source depth and determine the free space propagation loss distribution at the target sound source depth based on the resolution requirement of the underwater acoustic channel propagation loss distribution to be predicted.

[0118] The merging module 300 is used to perform clipping processing and normalization processing on the underwater acoustic channel propagation loss distribution and the free space propagation loss distribution, so as to obtain the processed underwater acoustic channel propagation loss distribution and the free space propagation loss distribution, and merge the processed underwater acoustic channel propagation loss distribution and the free space propagation loss distribution in the channel dimension to obtain the merged propagation loss distribution.

[0119] The prediction module 400 is used to input the merged propagation loss distribution into the target neural network, so as to output the prediction result of the normalized propagation loss distribution of the target sound source depth, and determine the prediction result of the underwater acoustic channel propagation loss distribution of the target sound source depth according to the prediction result of the normalized propagation loss distribution.

[0120] Optionally, in an embodiment of the present application, the determination module 200 includes: a first determination unit, a second determination unit, and a third determination unit.

[0121] Among them, the first determination unit is used to determine the resolution of the underwater acoustic channel propagation loss distribution in the first target dimension and the resolution of the second target dimension according to the resolution requirement.

[0122] The second determination unit is used to determine the underwater acoustic channel propagation loss distribution of the first sound source depth and the free space propagation loss distribution of the target sound source depth according to the resolution of the first target dimension and the resolution of the second dimension.

[0123] Optionally, in an embodiment of the present application, the first determination unit includes: an acquisition subunit and a determination subunit.

[0124] Among them, the acquisition subunit is used to acquire the application requirement of the underwater acoustic channel propagation loss distribution to be predicted and the acquisition method of the underwater acoustic channel propagation loss distribution of the first sound source depth before determining the resolution of the underwater acoustic channel propagation loss distribution in the first target dimension and the resolution of the second target dimension according to the resolution requirement.

[0125] The determination subunit is used to determine the resolution requirement by combining the application requirement and the acquisition method.

[0126] Optionally, in an embodiment of the present application, it further includes: a collection module, a preprocessing module, and a construction module.

[0127] Among them, the collection module is used to collect the propagation loss distribution data of the depth pairs conjugate to the sound channel axis that meet the resolution requirement in the target water area before inputting the merged propagation loss distribution into the target neural network, and preprocess the propagation loss distribution data, so as to construct a training data set according to the preprocessed propagation loss distribution data.

[0128] The training module is used to train the initialized neural network using the training data set, the objective function and the objective optimizer to obtain the initial neural network.

[0129] A building module is used to fine-tune the initial neural network using the propagation loss distribution data of the water area to determine the target neural network.

[0130] Optionally, in one embodiment of the present application, the prediction module 400 includes: a prediction unit, which is used to perform inverse processing of limiting processing and inverse processing of normalization processing on the normalized propagation loss distribution prediction result to determine the underwater acoustic channel propagation loss distribution prediction result of the target sound source depth.

[0131] It should be noted that the aforementioned explanation of the embodiment of the method for predicting the distribution of underwater acoustic channel propagation loss is also applicable to the device for predicting the distribution of underwater acoustic channel propagation loss in this embodiment, and will not be repeated here.

[0132] According to the prediction device for the propagation loss distribution of the underwater acoustic channel proposed in the embodiment of the present application, the target neural network can be used to predict the propagation loss distribution of the underwater acoustic channel at the target sound source depth conjugated with the first sound source depth about the sound channel axis based on the known propagation loss distribution of the underwater acoustic channel at the first sound source depth. Thus, the principle that the propagation loss distribution of the two sound source depths has a certain correlation by utilizing the characteristic that the depth conjugated with the sound channel axis has the same sound speed is realized, and the known propagation loss distribution at any sound source depth that has been measured is effectively used to predict the propagation loss distribution at the sound source depth conjugated with the sound source depth about the sound channel axis, which greatly reduces the calculation complexity and calculation time in the prediction process, and the target neural network in the present application can be fine-tuned according to the propagation loss data of the actual application waters, which effectively improves the application ability and scope of application of the target neural network in the actual application waters, thereby improving the accuracy of the propagation loss distribution at the target sound source depth predicted by the present application. This solves the problem that the underwater acoustic channel propagation loss prediction methods in the relevant technologies have great limitations. They only consider the physical information of the environment as input and cannot use other types of input. In addition, these methods are generally based on specific mathematical models and assumptions, and the environmental information considered is limited, which can easily lead to low modeling accuracy, thereby affecting the accuracy of the underwater acoustic channel propagation loss prediction results. In addition, some modeling methods are highly complex and will introduce problems such as large calculation delays.

[0133] Figure 6 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0134] A memory 601 , a processor 602 , and a computer program stored in the memory 601 and executable on the processor 602 .

[0135] When the processor 602 executes a program, it implements the prediction method for the underwater acoustic channel propagation loss distribution provided in the above embodiments.

[0136] Furthermore, the electronic device further includes:

[0137] A communication interface 603, which is used for communication between the memory 601 and the processor 602.

[0138] A memory 601, which is used to store a computer program that can run on the processor 602.

[0139] The memory 601 may include a high-speed RAM memory, and may also include a non-volatile memory, such as at least one disk memory.

[0140] If the memory 601, the processor 602, and the communication interface 603 are implemented independently, the communication interface 603, the memory 601, and the processor 602 can be interconnected through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the sake of representation, Figure 6 only a thick line is used to represent it in the figure, but it does not mean that there is only one bus or one type of bus.

[0141] Optionally, in a specific implementation, if the memory 601, the processor 602, and the communication interface 603 are integrated on a chip, the memory 601, the processor 602, and the communication interface 603 can communicate with each other through an internal interface.

[0142] The processor 602 may be a Central Processing Unit (CPU), or an Application Specific Integrated Circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0143] The embodiments of the present application also provide a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, it implements the above prediction method for the underwater acoustic channel propagation loss distribution.

[0144] An embodiment of the present application further provides a computer program product, including a computer program, which can run computer instructions, and when the computer instructions are executed by a processor, the prediction method for the underwater acoustic channel propagation loss distribution provided by the embodiment of the present application is implemented.

[0145] In the description of this specification, the descriptions referring to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. mean that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic expressions of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or N embodiments or examples. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0146] In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include at least one of the features. In the description of the present application, the meaning of "N" is at least two, such as two, three, etc., unless otherwise clearly and specifically defined.

[0147] Any process or method description shown in the flowchart or described in other ways herein can be understood as representing a module, segment, or part of code including one or N executable instructions for implementing a customized logic function or process, and the scope of the preferred embodiment of the present application includes additional implementations, where the functions can be executed in a manner that is not in the order shown or discussed, including in a substantially simultaneous manner according to the functions involved or in the reverse order, which should be understood by those skilled in the art of the embodiments of the present application.

[0148] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a definable sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or used in combination with these instruction execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transmit a program for use by or in combination with an instruction execution system, apparatus, or device. More specific examples (non-exhaustive list) of computer-readable media include the following: an electrical connection portion (electronic device) having one or N wirings, a portable computer disk cartridge (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory.

[0149] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or combinations thereof. In the above-described embodiments, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. If implemented using hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits having suitable combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), and the like.

[0150] Those of ordinary skill in the art of this technology can understand that all or part of the steps carried by the method of the above-described embodiments can be completed by a program instructing relevant hardware, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0151] In addition, each functional unit in various embodiments of the present application may be integrated into one processing module, may exist separately as individual physical units, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. When the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0152] The above-mentioned storage medium may be a read-only memory, a magnetic disk or an optical disc, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present application.

Claims

1. A method for predicting the distribution of propagation loss in an underwater acoustic channel, characterized in that: The following steps are involved: Based on the sound velocity profile of the water area with the propagation loss distribution of the underwater acoustic channel to be predicted, obtaining a first sound source depth in a depth pair conjugated with respect to the acoustic channel axis in the water area and a target sound source depth corresponding to the propagation loss distribution of the underwater acoustic channel to be predicted; Determine the underwater acoustic channel propagation loss distribution at the first sound source depth based on the resolution requirement of the underwater acoustic channel propagation loss distribution to be predicted, and determine the free space propagation loss distribution at the target sound source depth; Performing a limiting process and a normalization process on the underwater acoustic channel propagation loss distribution and the free space propagation loss distribution to obtain a processed underwater acoustic channel propagation loss distribution and a free space propagation loss distribution, and merging the processed underwater acoustic channel propagation loss distribution and the free space propagation loss distribution in the channel dimension to obtain a merged propagation loss distribution; The combined propagation loss distribution is input into the target neural network to output the normalized propagation loss distribution prediction result of the target sound source depth, and the underwater acoustic channel propagation loss distribution prediction result of the target sound source depth is determined based on the normalized propagation loss distribution prediction result.

2. The method according to claim 1, characterized in that The step of determining the underwater acoustic channel propagation loss distribution at the first sound source depth based on the resolution requirement of the underwater acoustic channel propagation loss distribution to be predicted, and determining the free space propagation loss distribution at the target sound source depth, comprises: Determine the resolution of the underwater acoustic channel propagation loss distribution in the first target dimension and the resolution of the second target dimension according to the resolution requirement; The underwater acoustic channel propagation loss distribution at the first sound source depth and the free space propagation loss distribution at the target sound source depth are determined according to the resolution of the first target dimension and the resolution of the second dimension.

3. The method according to claim 2, characterized in that Before determining the resolution of the underwater acoustic channel propagation loss distribution in the first target dimension and the resolution of the second target dimension according to the resolution requirement, the method further includes: Obtaining application requirements for the underwater acoustic channel propagation loss distribution to be predicted and a method for obtaining the underwater acoustic channel propagation loss distribution at the first sound source depth; The resolution requirement is determined in combination with the application requirement and the acquisition method.

4. The method according to claim 1, characterized in that: Before inputting the combined propagation loss distribution into the target neural network, the method further includes: Collecting propagation loss distribution data of depth pairs conjugated to the sound channel axis in the target water area that meet the resolution requirement, and preprocessing the propagation loss distribution data to construct a training data set based on the preprocessed propagation loss distribution data; Using the training data set, the objective function and the objective optimizer to train the initialized neural network to obtain an initial neural network, wherein the objective function includes a masked mean square error and a masked image gradient; The initial neural network is fine-tuned using the propagation loss distribution data of the water area to determine the target neural network.

5. The method according to claim 1, characterized in that The determining the underwater acoustic channel propagation loss distribution of the target sound source depth according to the normalized propagation loss distribution prediction result includes: The normalized propagation loss distribution prediction result is subjected to the inverse processing of the limiting processing and the inverse processing of the normalization processing to determine the underwater acoustic channel propagation loss distribution prediction result of the target sound source depth.

6. A prediction device for the propagation loss distribution of an underwater acoustic channel, characterized in that: include: An acquisition module, for acquiring, based on a sound velocity profile of a water area in which a propagation loss distribution of an underwater acoustic channel is to be predicted, a first sound source depth in a depth pair conjugated with respect to an acoustic channel axis in the water area and a target sound source depth corresponding to the propagation loss distribution of the underwater acoustic channel to be predicted; A determination module, configured to determine the underwater acoustic channel propagation loss distribution at the first sound source depth and determine the free space propagation loss distribution at the target sound source depth based on the resolution requirement of the underwater acoustic channel propagation loss distribution to be predicted; A merging module, used for performing limiting processing and normalization processing on the underwater acoustic channel propagation loss distribution and the free space propagation loss distribution to obtain processed underwater acoustic channel propagation loss distribution and free space propagation loss distribution, and merging the processed underwater acoustic channel propagation loss distribution and the free space propagation loss distribution in the channel dimension to obtain a merged propagation loss distribution; A prediction module is used to input the combined propagation loss distribution into a target neural network to output a normalized propagation loss distribution prediction result of the target sound source depth, and determine the underwater acoustic channel propagation loss distribution prediction result of the target sound source depth based on the normalized propagation loss distribution prediction result.

7. The device according to claim 6, characterized in that The determining module comprises: A first determining unit, configured to determine a resolution of the underwater acoustic channel propagation loss distribution in a first target dimension and a resolution of a second target dimension according to the resolution requirement; The second determination unit is used to determine the underwater acoustic channel propagation loss distribution of the first sound source depth and the free space propagation loss distribution of the target sound source depth according to the resolution of the first target dimension and the resolution of the second dimension.

8. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the method for predicting the distribution of propagation loss in an underwater acoustic channel as described in any one of claims 1 to 5.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the method for predicting the distribution of propagation loss in an underwater acoustic channel as described in any one of claims 1 to 5.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed, it is used to implement the method for predicting the distribution of underwater acoustic channel propagation loss as described in any one of claims 1-5.

Images