A three-node passive acoustic array ship positioning method based on a CNN-LSTM model

Abstract

The application discloses a three-node passive acoustic array ship positioning method based on a CNN-LSTM model, and relates to the technical field of underwater sound source positioning. The method comprises the following steps: step S1: collecting acoustic signals radiated by a target ship in real time through a three-node passive acoustic network arranged underwater; and pre-processing acoustic signals of each node to obtain interaural intensity difference features and interaural phase difference features; step S2: constructing a CNN-LSTM model, inputting combined feature maps of the interaural intensity difference features and the interaural phase difference features into a spatial feature extraction module of the CNN-LSTM model to perform spatial feature extraction, and outputting a spatial feature vector; step S3: inputting the spatial feature vector into a time sequence feature learning module of the CNN-LSTM model to perform time sequence feature extraction, and outputting a latent feature vector; and step S4: inputting the latent feature vector into an output layer of the CNN-LSTM model to complete regression prediction, and outputting a bearing angle and a distance value of a sound source.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of underwater sound source positioning, in particular to a three-node passive acoustic array ship positioning method based on a CNN-LSTM model. BACKGROUND

[0002] Underwater sound source positioning technology is the basis for many marine application fields such as ocean exploration, national defense security, resource exploration and marine ecological monitoring. The underwater acoustic positioning technology can be generally divided into active positioning and passive positioning. In military reconnaissance and marine environment monitoring, low detectability, accurate and robust positioning of the target sound source is crucial for real-time tracking, environmental assessment and rapid emergency response. However, the underwater acoustic environment is extremely complex and highly dynamic, making the robust positioning of underwater sound sources a highly challenging task. In particular, when using a passive acoustic network (distributed hydrophone network) to locate the radiation noise of targets such as ships, it is a technical problem to be solved in the current underwater acoustic field to obtain high-precision azimuth and distance information under limited observation nodes and complex propagation conditions. SUMMARY

[0003] In view of the above prior art, the present application provides a three-node passive acoustic array ship positioning method based on a CNN-LSTM model, which realizes ship targets in a complex underwater acoustic environment.

[0004] To achieve the above purpose, the technical scheme of the embodiment of the present application is as follows:

[0005] A three-node passive acoustic array ship positioning method based on a CNN-LSTM model, the method comprising the following steps:

[0006] Step S1: Real-time acquisition of acoustic signals radiated by the target ship by deploying a three-node passive acoustic network under water; and pre-processing the acoustic signals of each node to obtain interaural intensity difference features and interaural phase difference features;

[0007] Step S2: Constructing a CNN-LSTM model, inputting the combined feature map of the interaural intensity difference features and the interaural phase difference features into the spatial feature extraction module of the CNN-LSTM model for spatial feature extraction, and then outputting a spatial feature vector;

[0008] Step S3: Inputting the spatial feature vector into the time series feature learning module of the CNN-LSTM model for time series feature extraction, and then outputting a latent feature vector;

[0009] Step S4: Inputting the latent feature vector into the output layer of the CNN-LSTM model for regression prediction, and outputting the azimuth angle and distance value of the sound source.

[0010] As a preferred embodiment of the present invention, the preprocessing of the acoustic signal of each node in step S1 specifically includes: dividing the raw acoustic signal continuously acquired by each node into analysis frames of fixed duration and downsampling them; performing a short-time Fourier transform on the downsampling acoustic signal to obtain the time spectrum of the signal of each node, and then calculating the inter-ear intensity difference feature and the inter-ear phase difference feature; calculating the inter-ear intensity difference feature The formula is as follows:

[0011]

[0012] Calculate the interauricular phase difference characteristics The formula is as follows:

[0013]

[0014] in, and These are the STFT complex matrices of the signals received by the two sensors. Represents frequency, Represents the STFT time frame index.

[0015] As a preferred embodiment of the present invention, in step S2, before inputting the data into the CNN-LSTM model, the extracted interaural intensity difference features and interaural phase difference features are standardized to have zero mean and unit variance, as shown in the following formula:

[0016]

[0017]

[0018]

[0019] in, The mean of a matrix representing a sample. Represents the standard deviation of a matrix. Represents the first in the matrix Line 1 The value of the column, Indicates the number of rows in the matrix. This represents the number of columns in the matrix. This represents the standardized sample matrix.

[0020] As a preferred scheme of the present application, the spatial feature extraction module in the step S2 is composed of two branches, each branch including a shallow convolutional block, a middle convolutional block and a deep convolutional block, the frequency and intensity / phase difference information of the feature map combined by the interaural intensity difference feature and the interaural phase difference feature are preliminarily extracted through the shallow convolutional block, the deep features are further extracted through the middle convolutional block, a batch normalization layer is added in the middle convolutional block to enhance the model expression ability, then the feature correlation, channel dimension reduction and feature reorganization are performed through the deep convolutional block, and finally the spatial feature vector after extraction is output by using the Tanh activation function.

[0021] In order to avoid the information loss problem caused by the pooling layer, the step length of the convolutional layer is set to control the down-sampling of the feature map in the whole spatial feature extraction module without using the pooling layer, and in order to accelerate the convergence speed of the model and enhance the nonlinear expression ability of the model, the ELU activation function is used to introduce the non-saturation characteristics of the negative value interval before the output feature map of the whole convolutional branch, so that the output data maintains a certain gradient on the negative half axis, and the calculation formula of the ELU activation function is as follows:

[0022]

[0023] wherein, is the output of the activation function, is the input of the activation function, is the saturation coefficient on the negative half axis.

[0024] As a preferred scheme of the present application, the step S2 further includes dynamic mask enhancement on the extracted feature map combined by the interaural intensity difference feature and the interaural phase difference feature, including random masking of part of the continuous area in the time domain and part of the frequency band in the frequency domain, and the time domain noise addition and feature mask are jointly used during model training.

[0025] As a preferred scheme of the present application, the spatial feature vector output by each branch of the spatial feature extraction module in the step S3 is converted into a one-dimensional feature vector through a deformation operation, then the time sequence information of the two branches is considered, the two one-dimensional feature vectors are spliced to form a time sequence feature vector containing two frames, and the time sequence feature vector is input into the subsequent time sequence feature learning module;

[0026] The time sequence feature learning module is composed of two layers of LSTM networks, the hidden layer state of the first layer of LSTM is output to the input layer of the second layer of LSTM, the LSTM network effectively learns and utilizes the long-range time dependence in the feature sequence through its unique gating mechanism, L2 regularization constraint is performed on the two layers of LSTM networks to prevent overfitting, and the final latent feature vector is output.

[0027] As a preferred scheme of the present application, the regression prediction in the two fully connected linear layers of the CNN-LSTM model in the step S4 outputs the azimuth and distance values of the sound source, specifically including: the optimization target of the loss function of the CNN-LSTM model is to minimize the sum of the loss values between the predicted azimuth, the predicted distance and the true values, and the loss function is defined as follows:

[0028]

[0029]

[0030]

[0031] wherein, is the total loss function, is the loss function of the predicted azimuth, is the loss function of the predicted distance, represents the number of samples, is the serial number of the sample, and respectively represent the true value and the predicted value of the azimuth, represents the normalized true distance value, represents the normalized distance value predicted by the model, and the distance normalization adopts the minimum-maximum scaling, and the formula is as follows:

[0032]

[0033] wherein, and respectively represent the minimum value and the maximum value of the distance in the complete data set.

[0034] The present application has the following beneficial effects:

[0035] (1) The present method uses the ILD+IPD combined features, combines the powerful spatial feature extraction capability of the CNN and the modeling advantage of the LSTM on the time sequence information, can learn the complex space-time pattern from the input features, directly learns the deep features related to the sound source position from the preprocessed ILD+IPD features, avoids the complex manual feature design and selection process in the traditional method, reduces the dependence on the accurate sound speed profile and other underwater acoustic environment prior knowledge, and simplifies the complexity of the positioning system. At the same time, the data enhancement strategies of time domain noise superposition and time-frequency domain feature mask are adopted, which significantly improves the generalization ability and robustness of the model in the complex underwater noise and signal distortion environment, thereby realizing the synchronous regression prediction of the azimuth and distance of the ship radiated noise.

[0036] (2) The method is suitable for a distributed network, does not need to rely on a large number of carefully designed dense arrays as a sensor collection network, has the characteristics of low cost and easy deployment, and effectively solves the problem of performance decline in a distributed network and the problem that existing deep learning methods generally rely on dense arrays through a three-node passive acoustic network design. Effective positioning can be achieved using a small number of hydrophones, effectively reducing the cost and complexity of system deployment and maintenance, and easy to popularize and apply in practice.

[0037] (3) The CNN-LSTM model used in the method can output two key positioning parameters, the azimuth angle and the distance of the sound source, at the same time, providing more comprehensive target position information, which is superior to the traditional method which can only estimate the azimuth angle. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 A step schematic diagram of the three-node passive acoustic array ship positioning method based on the CNN-LSTM model provided by the present application is provided.

[0039] Figure 2 A CNN-LSTM model structure schematic diagram provided by the present application is provided.

[0040] Figure 3 A passive acoustic node arrangement and target trajectory schematic diagram provided by the present application is provided.

[0041] Figure 4 An azimuth angle error distribution diagram provided by the present application is provided.

[0042] Figure 5 A distance error distribution diagram provided by the present application is provided.

[0043] Figure 6 A CNN-LSTM model training stage loss curve schematic diagram provided by the present application is provided.

[0044] Figure 7 A CNN-LSTM model verification stage loss curve schematic diagram provided by the present application is provided. DETAILED DESCRIPTION

[0045] The technical solutions of the present application are further described in detail below in combination with the drawings and specific embodiments. Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by those skilled in the art to which the present application belongs. The terms used in the specification of the present application are only for the purpose of describing specific embodiments and are not intended to limit the present application. In the following description, the expression "some embodiments" describes a subset of all possible embodiments, but it should be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0046] In the following description, numerous specific details are set forth to provide a more thorough understanding of the present application. However, it will be apparent to one of skill in the art that the present application can be practiced without one or more of these specific details. In other instances, well-known features have not been described in order to avoid obscuring the present application. In this description, the terms "include" and "comprise," and conjugations thereof, are used generically and not by way of limitation. These terms encompass the various ways a device, component, step, etc. can be comprised, include, or comprise components of another device, component, step, etc. These terms cover a wide variety of situations, for example, a component can be an element by itself or can be included in a device or system including more than one component. In addition, the term "comprise" or "comprises" when used in a "comprising" manner in this description is used to mean that other features, elements, steps, operations, components, and / or properties can be added to those that are recited. The singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. The terms "coupled" and "connected" and variations thereof, as used in this description, are used generically and exemplify a relationship with respect to the fluid flow between devices, components, steps, etc. and that can include any intermediate structural elements unless otherwise specified. The term "erect" is used to mean that the device is in an upright position.

[0047] It is to be understood that the application can assume various alternative forms of embodiment, all of which fall within the scope of the present application. Rather, the embodiments are provided so that this disclosure will be thorough, and will fully convey the scope to those skilled in the art. The terminology used in the description presented herein is intended to be interpreted in only its broadest, most liberal sense, in accordance with the principles of patent law. Additionally, the use herein of the terms "including", "comprising", "having" and the like are specifically intended to convey a description of subject matter that can include, but is not limited to, one or more recited elements or steps. Moreover, although the terms "first", "second" and the like can be used herein to describe various elements, these elements should not be limited by these terms since such elements can include, but are not limited to, one or more elements.

[0048] It is also to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting. As used herein, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that the terms "comprises" and / or "comprising," when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and / or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. As used herein, the term "and / or" includes any and all combinations of one or more of the associated listed items.

[0049] For a more complete understanding of the present application, reference is made to the following description taken in conjunction with the accompanying drawings in which:

[0050] Reference is made to the accompanying drawings throughout the following detailed description. Figure 1 and the drawings that form a part hereof, wherein: Figure 2 A method for ship positioning based on a CNN-LSTM model is provided, and the method comprises the following steps:

[0051] Step S1: Real-time acquisition of acoustic signals radiated by a target ship through a three-node passive acoustic network deployed underwater; and preprocessing of acoustic signals of each node to obtain interaurale intensity difference features and interaurale phase difference features;

[0052] Step S2: constructing a CNN-LSTM model, inputting the feature map combined by the interaural intensity difference feature and the interaural phase difference feature into a spatial feature extraction module of the CNN-LSTM model to perform spatial feature extraction, and then outputting a spatial feature vector;

[0053] Step S3: inputting the spatial feature vector into a time sequence feature learning module of the CNN-LSTM model to perform time sequence feature extraction, and then outputting a latent feature vector;

[0054] Step S4: inputting the latent feature vector into an output layer of the CNN-LSTM model to perform regression prediction, and outputting an azimuth angle and a distance value of the sound source.

[0055] As a preferred scheme of the present application, the pre-processing of the acoustic signal of each node in step S1 specifically comprises: cutting the original acoustic signal continuously collected by each node into fixed-length analysis frames and performing down-sampling; performing short-time Fourier transform on the down-sampled acoustic signal to obtain a time-frequency spectrum of each node signal, and calculating an interaural intensity difference feature and an interaural phase difference feature; calculating the interaural intensity difference feature , which reflects the energy difference of the signal arriving at two sensors, and the formula is as follows:

[0056]

[0057] calculating the interaural phase difference feature , which reflects the phase difference of the signal arriving at two sensors, and the formula is as follows:

[0058]

[0059] wherein, and are STFT complex matrices of the signals received by two sensors representing frequency, representing an STFT time frame index.

[0060] In the present embodiment, each hydrophone node records waveform data of sound pressure changing over time, and the sampling rate needs to be greater than 2 kHz.

[0061] In this embodiment, the acoustic signal of each node is preprocessed. First, the raw acoustic signal continuously acquired by each node is divided into analysis frames of fixed duration (0.5s fixed duration). Then, the audio is downsampled to 2kHz to reduce redundant information and focus on the frequency band where ship noise is mainly located. Then, a Short Time Fourier Transform (STFT) is performed on the downsampled audio data to obtain the time spectrum of the signal of each node. The parameters of STFT are selected with a sliding window of 32 milliseconds and a step size of 16 milliseconds. Based on the signals received by different hydrophone pairs in the three-node network, the interaural intensity difference feature (ILD) and interaural phase difference feature (IPD) are calculated. For the three nodes (a, b, c), when a reference hydrophone is selected, two pairs of hydrophone combinations can be formed (for example, if b is selected as the reference hydrophone, the combination ab cb is formed).

[0062] As a preferred embodiment of the present invention, in step S2, before inputting the data into the CNN-LSTM model, the extracted interaural intensity difference features and interaural phase difference features are standardized to have zero mean and unit variance, as shown in the following formula:

[0063]

[0064]

[0065]

[0066] in, The mean of a matrix representing a sample. Represents the standard deviation of a matrix. Represents the first in the matrix Line 1 The value of the column, Indicates the number of rows in the matrix. This represents the number of columns in the matrix. This represents the standardized sample matrix.

[0067] As a preferred embodiment of the present invention, the spatial feature extraction module in step S2 consists of two branches, each including a shallow convolutional block, a middle convolutional block, and a deep convolutional block. The shallow convolutional block initially extracts the frequency and intensity / phase difference information of the feature map combining the interauricular intensity difference feature and the interauricular phase difference feature. Then, the middle convolutional block further refines the deep features and adds a batch normalization layer to the middle convolutional block to enhance the model's expressive power. Then, the deep convolutional block performs feature association, channel dimensionality reduction, and feature recombination. Finally, the Tanh activation function is used to output the extracted spatial feature vector.

[0068] In order to avoid the information loss problem caused by the pooling layer, the step of the convolution layer is set to control the down-sampling of the feature map in the whole spatial feature extraction module, without using the pooling layer, and in order to accelerate the convergence speed of the model and enhance the nonlinear expression ability of the model, the ELU activation function is introduced before the output feature map of the whole convolution branch, the non-saturation characteristics of the negative value interval are introduced, the output data maintains a certain gradient on the negative half axis, and the calculation formula of the ELU activation function is as follows:

[0069]

[0070] wherein, is the output of the activation function, is the input of the activation function, is the saturation coefficient on the negative half axis, and is usually 1.

[0071] In the embodiment, the spatial feature extraction module adopts a multi-layer two-dimensional convolutional neural network (Conv2D) structure, wherein shallow convolutional blocks (1-3 layers): a larger convolution kernel (convolution kernel size 5x5) is used to obtain a larger receptive field and preliminarily extract frequency and intensity / phase difference information. Medium convolutional blocks (4-5 layers): a smaller convolution kernel (convolution kernel size 3x3) is used to further finely extract deep features. A batch normalization layer (Batch Normalization, BN) is added after the convolution layer to enhance the expression ability of the model. Deep convolutional blocks (6-9 layers): a larger convolution kernel (convolution kernel size 9x9) is used for feature association, and then 1x1 convolution is used for channel dimension reduction and feature reorganization. The last layer of the deep convolutional block also uses 1x1 convolution to further reorganize the features and output the extracted spatial feature vector using Tanh as the activation function.

[0072] In the embodiment, the step of the convolution layer is set to control the down-sampling of the feature map in the whole spatial feature extraction module, without using the pooling layer. Because the pooling layer does not have a learnable parameter, simply using the pooling layer for down-sampling may cause the loss of unwanted spatial information, which is not conducive to the task of accurately positioning the fine spatial clues in ILD and IPD. Therefore, by setting the step of the convolution layer for down-sampling, the size of the feature map can be reduced while the network learns how to retain the most important spatial features, thereby avoiding the information loss problem that may be caused by the pooling layer.

[0073] In the embodiment, the ELU activation function is used before the output feature map of the whole convolution branch, which is to introduce the non-saturation characteristics of the negative value interval, so that the output data maintains a certain gradient on the negative half axis, enhances the nonlinear expression ability of the model, and accelerates the convergence speed. In addition, the ELU activation function can also alleviate the problem of overfitting to a certain extent, and improve the generalization ability of the model.

[0074] As a preferred scheme of the present application, the step S2 further comprises dynamic mask enhancement on the extracted inter-aural intensity difference feature and the inter-aural phase difference feature combined feature map, including random masking of part of the continuous area in the time domain and part of the frequency band in the frequency domain. Time domain noise addition and feature mask are used jointly during training.

[0075] As a preferred scheme of the present application, in the step S3, the spatial feature vectors output by each branch of the spatial feature extraction module are converted into one-dimensional feature vectors through deformation operation, and then the time sequence information of the double branches is considered, the two one-dimensional feature vectors are spliced to form a time sequence feature vector containing two frames, and the time sequence feature vector is input into the subsequent time sequence feature learning module.

[0076] The time sequence feature learning module is composed of two layers of LSTM networks, and the hidden layer state output of the first layer of LSTM is output to the input layer of the second layer of LSTM. The LSTM network effectively learns and utilizes the long-range time dependence in the feature sequence through its unique gating mechanism (input gate, forget gate, output gate). In order to prevent overfitting, L2 regularization constraint is performed on the two layers of LSTM networks, and finally the latent feature vector is output.

[0077] As a preferred scheme of the present application, in the step S4, regression prediction is performed through two fully connected linear layers of the CNN-LSTM model, and the output of the azimuth angle and distance value of the sound source specifically comprises: the optimization target of the loss function of the CNN-LSTM model is to minimize the sum of the loss values between the predicted azimuth angle, the predicted distance and the true value, and the loss function is defined as follows:

[0078]

[0079]

[0080]

[0081] wherein, is the total loss function, is the loss function of the predicted azimuth angle, is the loss function of the predicted distance, denotes the number of samples, is the serial number of the sample, and denote the true value and the predicted value of the azimuth angle, respectively, denotes the normalized true distance value, denotes the normalized distance value predicted by the model, and the distance normalization adopts the minimum-maximum scaling, and the formula is as follows:

[0082]

[0083] where, and respectively represent the minimum and maximum values of the distance in the complete dataset.

[0084] In this embodiment, the latent feature vector output by the timing feature learning module is fed into two fully connected linear layers, which directly regress the azimuth angle and distance value of the sound source. The output of the model azimuth angle is the predicted actual angle, and the output of the distance is the predicted actual distance after normalization, the interval is [0, 1], and the actual distance value predicted by the model needs to be obtained by normalizing and inverse transforming according to the distribution range of the distance in the actual.

[0085] In this embodiment, when the algorithm reference point on the sea level is set (the position of the reference point depends on the needs of data processing), the azimuth angle parameter determines the accurate angle of the target ship relative to the reference point on the plane. The distance parameter gives the straight-line distance between the target ship and the reference point on the plane. When the two parameters are determined at the same time, the two-dimensional plane position of the ship can be uniquely marked in the polar coordinate system with the reference point as the origin, and the accurate positioning of the target is realized.

[0086] For example, to verify the effectiveness of the present application and to illustrate the best embodiment, the experimental data, data processing flow, model training details and performance verification will be described in detail below.

[0087] First, the experimental data used by the present application is derived from a passive sound source positioning measurement experiment conducted in the shallow water area of the East China Sea. The target is a ship sailing at a speed of less than 10 knots along a predetermined S-shaped trajectory. As shown in the following FIG. 3, the blue curve is the actual sailing trajectory of the ship, and the three red pentagonal marks in the figure are the distributed passive acoustic nodes a, b, and c.

[0088] The trajectory covers a horizontal range of about 2.5 km x 2.5 km. During the experiment, the dynamic range of the target ship from the array center was 300 meters to 1600 meters. The accurate trajectory information (longitude and latitude, speed, etc.) of the target ship was recorded in real time through the automatic identification system (AIS) of the ship. A distributed three-node passive acoustic network was used. That is, three omnidirectional hydrophones (with a sensitivity of -168 dB re 1 V / μPa in the range of 20 Hz-20 kHz) were placed in the form of beacons on the seabed. The hydrophone array continuously collected sound pressure signals at a sampling rate of 128 kHz, with a duration of 1 hour, and the original data was stored with 24-bit quantization. Each node device has a high-precision time synchronization function.

[0089] Secondly, the collected raw data is segmented and synchronized, and the raw audio data continuously collected by each hydrophone node is cut into 0.5-second fixed-length analysis frames. Through linear interpolation, the AIS system records the ship's geographic coordinates (latitude and longitude) and each 0.5-second acoustic data frame is time-synchronized at the millisecond level. The geographic coordinates of the ship are converted into the azimuth and distance in the local coordinate system with the selected reference point as the origin, and the maximum and minimum values of the distance are recorded. The maximum and minimum values are used to scale the true distance to the [0, 1] interval for model training. From 1 hour of synchronized data, a total of 7199 audio segments - sample data set containing ship GPS positions aligned in time. These samples are divided into 70% (5039 samples) for training set and 30% (2160 samples) for validation set. To focus on the frequency band where the ship noise is mainly located and reduce the amount of calculation, the original audio data (128 kHz) is downsampled to 2 kHz. The three-channel audio data after downsampling is subjected to short-time Fourier transform (STFT), and the STFT parameters are set as: window length 32 ms, frame shift (step) 16 ms. After obtaining the time-frequency spectrum of each node signal, the signal of a hydrophone node is selected as the reference signal to form two pairs of hydrophone combinations, and the interaural intensity difference (ILD) and interaural phase difference (IPD) features are calculated. Finally, the generated two sets of combined features (i.e. ILD+IPD feature maps) are standardized and input into the model.

[0090] To improve the generalization ability and robustness of the model to complex underwater environments, the following joint data augmentation strategy is used during model training:

[0091] 1. Time domain enhancement: Gaussian white noise is injected into the 0.5-second original sound pressure signal, and the signal-to-noise ratio (SNR) is randomly selected between -5 dB and 10 dB.

[0092] 2. Time-frequency domain feature enhancement: The extracted ILD+IPD feature maps (training set samples) are dynamically masked, and local masking is performed along the time axis and frequency axis: In the training process, time domain noise addition and time-frequency domain feature masking are used in cascade to generate diverse training samples.

[0093] Thirdly, the CNN-LSTM model is trained using the Adam optimizer, with an initial learning rate of 0.0001. The batch size is 32, and the training epoch is 3000.

[0094] In the experiment, the mean absolute error (MAE) is used to measure the prediction ability of the model. The smaller the MAE, the closer the predicted value of the model to the true value. On the validation set, the MAE of the model's predicted angle reached 4.87°, and the MAE of the distance reached 186.60 meters.

[0095] Figure 4 and Figure 5 The figures show the error distributions of the predicted angles and distances compared to the true values. The error distributions approximate a normal distribution and are concentrated near zero, indicating that the model's prediction bias is small and relatively stable. The mean and standard deviation are shown in Table 1.

[0096] Table 1. Mean and standard deviation of the model's predictions

[0097]

[0098] As shown in Table 1, the average error of the distance prediction results is 80.35 meters, with a standard deviation of 289.84 meters, indicating that the model can effectively estimate the distance to the target sound source and has a small systematic bias. The average error of the angle prediction is 0.63 degrees, with a standard deviation of 8.47 degrees. These data demonstrate the effectiveness of the proposed method in simultaneously estimating the azimuth and distance of the sound source.

[0099] Fourth, such as Figure 6 and Figure 7 As shown in the diagram, the loss curves for training and validation illustrate the model's learning progress during training. The horizontal axis represents the number of training algebras, and the vertical axis represents the calculated loss function. It can be seen that as the number of training algebras increases, both the training set loss and the validation set loss steadily decrease. The training set loss eventually reaches a stable state close to 0, while the validation set loss reaches a stable state close to 5 and does not increase as the training set loss decreases. This indicates that within the limited number of algebras specified by the proposed method, the model does not exhibit overfitting and can learn sufficient effective features from the data to achieve parameter convergence, thereby enabling stable prediction of the target sound source location.

[0100] The model's average inference time per sample was evaluated through a full inference test on a 10-round validation set. The calculated average inference time for a feature sample generated from a single 0.5-second audio clip was (0.40 ± 0.033) milliseconds.

[0101] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A three-node passive acoustic array ship positioning method based on a CNN-LSTM model, characterized in that, The method comprises the following steps: Step S1: Real-time acquisition of acoustic signals radiated by a target ship through a three-node passive acoustic network deployed underwater, and preprocessing of the acoustic signals of each node to obtain interaural intensity difference features and interaural phase difference features; The preprocessing of the acoustic signals of each node specifically includes: cutting the original acoustic signals continuously collected by each node into analysis frames with a fixed time length and performing down-sampling; performing short-time Fourier transform on the down-sampled acoustic signals to obtain time-frequency spectrums of the signals of each node, and then calculating interaural intensity difference features and interaural phase difference features; calculating the interaural intensity difference features The formula of the interaural intensity difference features is as follows: Computing interaural phase difference features The formula is as follows: wherein, and are STFT complex matrices of the signals received by the two sensors respectively, represents the frequency, represents the STFT time frame index; Step S2: Constructing a CNN-LSTM model, inputting the combined feature map of the interaural intensity difference features and the interaural phase difference features into a spatial feature extraction module of the CNN-LSTM model for spatial feature extraction, and then outputting a spatial feature vector; The spatial feature extraction module is composed of two branches, each branch including a shallow convolutional block, a middle convolutional block and a deep convolutional block, the shallow convolutional block is used to preliminarily extract frequency and intensity / phase difference information of the combined feature map of the interaural intensity difference features and the interaural phase difference features, the middle convolutional block is used to further extract deep features, a batch normalization layer is added in the middle convolutional block to enhance the expression ability of the model, the deep convolutional block is used for feature association, channel dimension reduction and feature reorganization, and finally a Tanh activation function is used to output the extracted spatial feature vector; In order to avoid information loss caused by the pooling layer, the step length of the convolutional layer is set to control the down-sampling of the feature map in the entire spatial feature extraction module without using the pooling layer, and in order to speed up the convergence of the model and enhance the nonlinear expression ability of the model, an ELU activation function is used to introduce the non-saturation characteristics of the negative value interval before the output feature map of the entire convolutional branch, so that the output data maintains a certain gradient on the negative half axis, and the calculation formula of the ELU activation function is as follows: wherein, is the output of the activation function, is the input of the activation function, is the negative half-axis saturation coefficient; Step S3: Inputting the spatial feature vector into a time sequence feature learning module of the CNN-LSTM model for time sequence feature extraction, and then outputting a latent feature vector; Step S4: Inputting the latent feature vector into an output layer of the CNN-LSTM model for regression prediction, and outputting the azimuth angle and distance value of the sound source.

2. The three-node passive acoustic array ship positioning method based on the CNN-LSTM model according to claim 1, characterized in that, In step S2, the extracted interaural intensity difference features and interaural phase difference features are respectively standardized to have zero mean and unit variance before being input into the CNN-LSTM model, and the specific formula is as follows: wherein, denotes the mean of a matrix of samples, denotes the standard deviation of a matrix, denotes the value in the row and column of a matrix, denotes the number of rows of a matrix, denotes the number of columns of a matrix, denotes the normalized matrix of samples.

3. The three-node passive acoustic array ship positioning method based on the CNN-LSTM model according to claim 2, characterized in that, Step S2 also includes dynamic mask enhancement of the combined feature map of the extracted interaural intensity difference features and interaural phase difference features, including random masking of part of the continuous area in the time domain and part of the frequency band in the frequency domain, and joint use of time domain noise addition and feature mask during model training.

4. The three-node passive acoustic array ship positioning method based on the CNN-LSTM model according to claim 3, characterized in that, In step S3, the spatial feature vector output by each branch of the spatial feature extraction module is converted into a one-dimensional feature vector through a deformation operation, and then considering the time sequence information of the two branches, the two one-dimensional feature vectors are spliced to form a time sequence feature vector containing two frames, and the time sequence feature vector is input into the subsequent time sequence feature learning module; In step S3, the spatial feature vector output by each branch of the spatial feature extraction module is converted into a one-dimensional feature vector through a deformation operation, and then considering the time sequence information of the two branches, the two one-dimensional feature vectors are spliced to form a time sequence feature vector containing two frames, and the time sequence feature vector is input into the subsequent time sequence feature learning module; The timing feature learning module is composed of two layers of LSTM networks, the hidden layer state output of the first layer of LSTM is output to the input layer of the second layer of LSTM, the LSTM network effectively learns and utilizes the long-range time dependence in the feature sequence through its unique gating mechanism, in order to prevent overfitting, L2 regularization constraint is performed on the two layers of LSTM networks, and the final latent feature vector is output.

5. The three-node passive acoustic array ship positioning method based on the CNN-LSTM model according to claim 4, characterized in that, The regression prediction through the output layer of the CNN-LSTM model in the step S4 outputs the azimuth angle and distance value of the sound source, and specifically includes that the optimization target of the loss function of the CNN-LSTM model is to minimize the sum of the loss values between the predicted azimuth angle, the predicted distance and the true value, and the loss function is defined as follows: where, is the total loss function, is the loss function for predicting azimuth, is the loss function for predicting distance, denotes the number of samples, is the sequence number of the sample, and denote the real value and the predicted value of the azimuth, respectively, denotes the normalized real distance value, denotes the normalized distance value predicted by the model, which uses the minimum-maximum scaling for distance normalization, and the formula is as follows: wherein, and respectively represent the minimum and maximum values of the distances in the complete dataset.

Images