Underwater sound propagation loss prediction method and device suitable for dynamic environmental parameters, and medium

Abstract

The invention discloses an underwater sound propagation loss prediction method and device suitable for dynamic environment parameters and a medium, and the method comprises the steps: collecting marine environment parameters in real time through a multi-source sensor, carrying out the preprocessing of the collected marine environment parameters, employing Kalman filtering to remove the noise in the marine environment parameters, and carrying out the prediction of the underwater sound propagation loss. The method comprises the following steps: acquiring marine environment parameters, standardizing the marine environment parameters, extracting key features from the preprocessed marine environment parameters, establishing a multi-modal input feature set, constructing a dynamic propagation loss prediction model based on a physical model and a data driving model, and performing model training on the dynamic propagation loss prediction model by adopting a random forest. And performing propagation loss prediction based on the multi-modal input feature set and the trained dynamic propagation loss prediction model, and outputting an underwater sound propagation loss prediction result. Based on a multi-source data fusion strategy, the adaptive capacity of the model in the dynamic marine environment is improved, the multi-dimensional feature extraction technology is adopted, physical modeling and a data driving method are combined, the precise modeling capacity of propagation loss is improved, and the prediction stability is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of data acquisition, and in particular relates to a method, device and medium for predicting underwater sound propagation loss applicable to dynamic environmental parameters. Background Art

[0002] Predicting underwater acoustic propagation loss is a key technology in underwater communications, sonar detection, and target recognition applications. Currently, existing methods rely primarily on fixed feature inputs, making them difficult to adapt to the dynamic changes in the real-time ocean environment and exhibiting certain limitations.

[0003] Traditional models typically assume that environmental parameters (such as sound velocity profiles and seafloor topography) are static inputs, making them unable to adapt to the dynamic changes in the actual ocean environment, resulting in gradually increasing prediction errors. Furthermore, empirical formulas and ray theory methods lack the ability to dynamically adjust to complex ocean environmental changes (such as thermoclines, temperature gradients, and salinity changes), and are unable to provide high-precision prediction results. Furthermore, existing methods primarily rely on a single data source (such as sound velocity profiles) for propagation loss prediction, failing to fully incorporate multi-source ocean data (such as temperature gradients, seabed sediments, and ocean currents), reducing the model's adaptability and generalization capabilities. Furthermore, in recent years, some studies have attempted to use machine learning methods (such as support vector regression and random forests) to predict underwater sound propagation loss. However, these methods are typically trained based on static data, lack adaptive mechanisms, and struggle to effectively handle dynamic environmental parameter changes.

[0004] At present, although some studies have attempted to introduce new solutions, such as using local real-time sensor data to correct the prediction of short-range underwater acoustic propagation loss, it is difficult to extend it to long distances or complex environments due to the limited range of data acquisition. Summary of the Invention

[0005] In view of the above-mentioned defects in the prior art, the present invention provides a method for predicting underwater sound propagation loss applicable to dynamic environmental parameters, the method comprising the following steps:

[0006] Step 1: Real-time collection of ocean environmental parameters using multi-source sensors, including but not limited to sound velocity profile, temperature gradient, seabed sediments, and sonar signals;

[0007] Step 2: preprocessing the collected ocean environment parameters, including using Kalman filtering to remove noise from the ocean environment parameters and standardizing the ocean environment parameters;

[0008] Step 3, extracting key features from the pre-processed ocean environment parameters to establish a multimodal input feature set;

[0009] Step 4: construct a dynamic transmission loss prediction model based on the physical model and the data-driven model, and use random forest to train the dynamic transmission loss prediction model;

[0010] Step 5: Perform propagation loss prediction based on the multimodal input feature set and the trained dynamic propagation loss prediction model, and output an underwater acoustic propagation loss prediction result.

[0011] Among them, based on the preset data collection frequency, a sound velocity profiler is used to collect the sound velocity in the water layer, a temperature detector is used to collect the water temperature at different depths, a bottom sediment detector is used to obtain the physical properties of the seabed, and a sonar receiver is used to capture the intensity of the sound signal.

[0012] The collected ocean environment parameters are dynamically modeled based on the Kalman filter algorithm to remove random noise, and a state model and an observation model are constructed. The ocean environment parameters are corrected through the steps of state prediction, error covariance prediction, state update, and error covariance update.

[0013] Preprocessing the collected ocean environment parameters further includes standardizing the ocean environment parameters, which is expressed as:

[0014]

[0015] Where x is the original data, μ is the mean, and σ is the standard deviation;

[0016] And the missing data points are filled based on linear interpolation.

[0017] According to the preprocessed data, the physical characteristics, frequency domain characteristics and derivative characteristics of the marine environment parameters are extracted, and a feature matrix X is constructed based on the extracted characteristics, and divided into a training set and a test set;

[0018] Select the best regularization parameter λ through cross-validation, use the training set data to train the LASSO regression model, and calculate the coefficient of each feature in the feature matrix X;

[0019] Evaluate the coefficient of each feature in the feature matrix X, extract features with non-zero coefficients, and construct a multimodal input feature set Xs.

[0020] Among them, the physical model is constructed based on the wave equation, including:

[0021] Construct the three-dimensional acoustic wave equation, expressed as:

[0022]

[0023] Where p is the sound pressure and c is the speed of sound;

[0024] Set the boundary conditions and initial conditions, use the numerical method to solve the wave equation, and obtain the output physical model, which is expressed as:

[0025] p physical (x, y, z, t) = ∫ Ω G(x,y,z;ξ,η,ζ,t)f(ξ,η,ζ)dV

[0026] Where G is the Green's function, which describes the influence of the sound source position (ξ,η,ζ) on the observation point (x,y,z), f is the sound source function, which represents the intensity and distribution of the sound source, and Ω is the area where the sound source is located.

[0027] Among them, the data-driven model is constructed based on LASSO regression, including:

[0028] Construct the LASSO regression objective function, expressed as:

[0029]

[0030] where yi is the sound propagation loss, is the predicted value of the model, n is the total number of samples, and βj is the jth model parameter;

[0031] After training, the output of the LASSO model, pML(X), is expressed as:

[0032]

[0033] Among them, xj is the jth feature in the input features.

[0034] Among them, the physical model and the data-driven model are combined to construct a dynamic propagation loss prediction model, which is expressed as:

[0035]

[0036] Where p^(x,y,z,t) is the final output of the dynamic propagation loss prediction model, which represents the sound pressure prediction at position (x,y,z) and time t, pphysical(x,y,z,t) is the output of the physical model, pML(X) is the output of the data-driven model, and α is an adaptive weight between 0 and 1.

[0037] The adaptive weight α is dynamically adjusted based on the prediction error of the physical model and the machine learning model, which is expressed as:

[0038]

[0039] Among them, Ephysical is the prediction error of the physical model, and EML is the prediction error of the data-driven model.

[0040] wherein, calculating the error index between the output of the dynamic propagation loss prediction model and the measured value, the error index including the root mean square error and the absolute error;

[0041] Based on the error analysis results, the collected ocean environment parameters are adjusted and steps 1 to 5 are continuously executed to achieve continuous optimization of the dynamic propagation loss prediction model.

[0042] The present invention uses multi-source sensors to collect ocean environmental parameters in real time, preprocesses the collected ocean environmental parameters, including using Kalman filtering to remove noise from the parameters and standardize them. Key features are extracted from the preprocessed parameters to establish a multimodal input feature set. A dynamic propagation loss prediction model is constructed based on a physical model and a data-driven model. The dynamic propagation loss prediction model is trained using a random forest. Propagation loss prediction is performed based on the multimodal input feature set and the trained dynamic propagation loss prediction model, and the underwater acoustic propagation loss prediction result is output. By combining multi-source data fusion with adaptive hybrid modeling technology, the accuracy, environmental adaptability, and real-time performance of underwater acoustic propagation loss prediction are significantly improved, providing more accurate and efficient support for applications such as underwater communications, sonar detection, and target recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The above and other objects, features and advantages of the exemplary embodiments of the present disclosure will become readily understood by reading the following detailed description with reference to the accompanying drawings. In the accompanying drawings, several embodiments of the present disclosure are shown in an illustrative and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein:

[0044] Figure 1 4 is a flow chart illustrating a method for predicting underwater sound propagation loss applicable to dynamic environmental parameters according to an embodiment of the present invention. DETAILED DESCRIPTION

[0045] To make the objectives, technical solutions, and advantages of the present invention more apparent, the present invention will be further described in detail below with reference to the accompanying drawings. It is apparent that the embodiments described are only some, not all, of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are intended to fall within the scope of protection of the present invention.

[0046] The terms used in the embodiments of the present invention are for the purpose of describing specific embodiments only and are not intended to limit the present invention. The singular forms "a," "an," "the," and "the" used in the embodiments of the present invention and the appended claims are also intended to include plural forms, and unless the context clearly indicates otherwise, "a plurality" generally includes at least two.

[0047] It should be understood that although the terms "first," "second," "third," etc. may be used to describe "...," these "..." should not be limited to these terms. These terms are merely used to distinguish "...." For example, "first..." could also be referred to as "second...", and similarly, "second..." could also be referred to as "first..." without departing from the scope of the present invention.

[0048] It should be understood that the term "and / or" as used herein is merely a description of the relationship between associated objects, indicating that three possible relationships exist. For example, "A and / or B" can represent: A exists alone, A and B exist simultaneously, or B exists alone. Furthermore, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0049] As used herein, the words "if" and "if" may be interpreted as "at the time of" or "when" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrases "if it is determined" or "if (stated condition or event) is detected" may be interpreted as "when it is determined" or "in response to the determination" or "when detecting (stated condition or event)" or "in response to detecting (stated condition or event)," depending on the context.

[0050] It should also be noted that the terms "include," "comprises," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a product or device comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such product or device. In the absence of further limitations, an element defined by the phrase "comprises a..." does not exclude the presence of other identical elements in the product or device comprising the element.

[0051] Accurately predicting underwater acoustic propagation loss faces numerous challenges. Existing propagation models (such as ray models and wave models) are not accurate enough under certain conditions, especially in complex environments. Furthermore, the lack of high-quality environmental data and historical records limits model training and validation. Therefore, an effective method is needed that can more effectively cope with dynamic environmental changes, improve prediction accuracy, optimize computational efficiency, and enhance the model's robustness and generalization capabilities for a wider range of underwater communications, sonar detection, and target recognition applications.

[0052] like Figure 1 As shown, the present invention discloses a method for predicting underwater sound propagation loss applicable to dynamic environmental parameters, the method comprising:

[0053] Step 1: Collect ocean environmental parameters in real time through multi-source sensors. The ocean environmental parameters include but are not limited to sound velocity profile, temperature gradient, seabed sediment and sonar signals.

[0054] Real-time collection of ocean environmental parameters through multi-source sensors to support the prediction of underwater sound propagation loss. This mainly includes the use of sound velocity profile sensors (sound velocity profilers (CTDs)) to record sound velocity, temperature, and salinity data at different depths, usually combined with vertical dives of the sensor in the water column. Configure temperature gradient sensors (such as temperature probes or temperature data loggers) to measure water temperature changes at different depths to help construct temperature gradient maps. Use seabed sediment sensors (sediment samplers or acoustic sediment detectors) to obtain the physical properties of the seabed (such as sediment type, hardness, etc.), which affect the propagation characteristics of sound waves. Set up sonar signal receivers (sonar equipment) to monitor the signal strength and propagation characteristics of underwater sound sources and record echo information.

[0055] In one embodiment, based on a preset data acquisition frequency, a sound velocity profiler is used to collect the sound velocity in the water layer, a temperature detector is used to collect the water temperature at different depths, a bottom sediment detector is used to obtain the physical properties of the seabed, and a sonar receiver is used to capture the intensity of the sound signal.

[0056] Furthermore, sensor data is transmitted to the central processing unit in real time via wireless communication technologies (such as radio, satellite, or optical fiber) to ensure data timeliness. A timed sampling mechanism is also set up to regularly acquire data to capture environmental changes, such as day and night cycles and seasonal changes.

[0057] Real-time acquisition of ocean environmental parameters using multiple sensors can significantly improve the accuracy and timeliness of underwater sound propagation loss predictions. This approach allows researchers to dynamically monitor key factors such as sound velocity profiles, temperature gradients, and seabed sediments, enabling models to promptly reflect environmental changes. This allows for more effective responses to complex ocean conditions and improves the performance and reliability of sonar systems.

[0058] Step 2: preprocessing the collected ocean environment parameters, including using Kalman filtering to remove noise in the ocean environment parameters and standardizing the ocean environment parameters.

[0059] Kalman filtering estimates system states by combining prediction models with measurement data, making it suitable for removing noise from time series data. Based on the sensor's measurement model and the system's dynamic model, the initial state and covariance matrix are set, and the state estimate and covariance are updated each time new data arrives, effectively removing random noise.

[0060] In one embodiment, the collected ocean environment parameters are dynamically modeled based on the Kalman filter algorithm to remove random noise, and a state model and an observation model are constructed. The ocean environment parameters are corrected through state prediction, error covariance prediction, state update, and error covariance update steps;

[0061] Preprocessing the collected ocean environment parameters further includes standardizing the ocean environment parameters, which is expressed as:

[0062]

[0063] Where x is the original data, μ is the mean, and σ is the standard deviation;

[0064] And the missing data points are filled based on linear interpolation.

[0065] Furthermore, by preprocessing the ocean environment parameters, specifically using Kalman filtering to remove noise and standardize them, the accuracy and consistency of the data were significantly improved. This approach ensures data reliability, making subsequent analysis and model building more effective and more accurate in reflecting the dynamic changes in the ocean environment, thereby improving the accuracy and practicality of underwater sound propagation loss predictions.

[0066] Step 3: extract key features from the preprocessed ocean environment parameters and establish a multimodal input feature set.

[0067] Identify features closely related to underwater sound propagation, such as sound velocity, temperature gradient, salinity, and seabed properties, and assess their importance using statistical analysis or machine learning algorithms (e.g., principal component analysis, random forests). Calculate derived features based on existing features, such as the rate of change of sound velocity and temperature, to capture the impact of dynamic changes.

[0068] In one embodiment, the physical characteristics, frequency domain characteristics and derivative characteristics of the ocean environment parameters are extracted based on the preprocessed data, and a feature matrix X is constructed based on the extracted characteristics and divided into a training set and a test set.

[0069] Furthermore, physical features include extracting sound velocity from sound velocity profile data, temperature from temperature gradient sensor data, salinity information from sensor data, and environmental parameters corresponding to each depth. The rate of change of sound velocity, temperature, and salinity, as well as the interactions between these features, are calculated.

[0070] Furthermore, the sonar signal is analyzed using Fast Fourier Transform (FFT) to extract the frequency components and their amplitudes. The main frequency components and energy distribution are identified, and features such as the main frequency and bandwidth are extracted.

[0071] The extracted physical features, frequency domain features, and derived features are integrated into a feature matrix X, where rows represent samples and columns represent features. The feature matrix X is divided into a training set and a test set. Typically, 80% of the data is used as the training set and 20% as the test set, ensuring random sampling to maintain representativeness.

[0072] The optimal regularization parameter λ is selected through cross-validation, the LASSO regression model is trained using the training set data, and the coefficient of each feature in the feature matrix X is calculated.

[0073] Furthermore, the k-fold cross validation method is used to set different λ values (regularization parameter) to evaluate the performance of the LASSO regression model on the training set under each λ. 2 ) and other indicators to evaluate the model performance and select the best performing λ value.

[0074] Evaluate the coefficient of each feature in the feature matrix X, extract features with non-zero coefficients, and construct a multimodal input feature set Xs.

[0075] A LASSO regression model is trained on the training set using the selected lambda value, calculating the regression coefficient for each feature. LASSO regression applies L1 regularization, squeezing the coefficients of certain features to zero, thereby achieving feature selection. The regression coefficients of each feature in the feature matrix X are analyzed, and features with non-zero coefficients are identified, indicating that these features contribute significantly to the model's predictions. Features with non-zero coefficients are combined into a new feature set Xs, which serves as the multimodal input feature set for subsequent model training and prediction.

[0076] By extracting physical, frequency-domain, and derived features, constructing a feature matrix X, and combining it with the LASSO regression model for effective feature selection, the model's predictive power and generalization performance can be significantly improved. This process not only identifies key features that significantly influence underwater sound propagation loss but also reduces the risk of overfitting through regularization, simplifies the feature space, and enhances the model's interpretability, ultimately providing a solid data foundation for accurate underwater sound propagation predictions.

[0077] Step 4: construct a dynamic transmission loss prediction model based on the physical model and the data-driven model, and use random forest to train the dynamic transmission loss prediction model.

[0078] Acoustic theory was applied to establish a preliminary physical model, accounting for the effects of sound velocity, temperature, salinity, and seabed characteristics on sound wave propagation loss. Relevant formulas (such as the propagation loss formula) were used to describe the propagation mechanism of sound waves in water. The parameters of the physical model were adjusted based on actual environmental data to better reflect real-world conditions.

[0079] In one embodiment, constructing a physical model based on a wave equation includes:

[0080] Construct the three-dimensional acoustic wave equation, expressed as:

[0081]

[0082] Where p is the sound pressure and c is the speed of sound;

[0083] Set the boundary conditions and initial conditions, use the numerical method to solve the wave equation, and obtain the output physical model, which is expressed as:

[0084] p physical (x, y, z, t) = ∫ Ω G(x,y,z;ξ,η,ζ,t)f(ξ,η,ζ)dV

[0085] Where G is the Green's function, which describes the influence of the sound source position (ξ,η,ζ) on the observation point (x,y,z), f is the sound source function, which represents the intensity and distribution of the sound source, and Ω is the area where the sound source is located.

[0086] Furthermore, a machine learning model (e.g., LASSO regression, random forest, etc.) is selected to process the sound propagation data.

[0087] In one embodiment, a data-driven model is constructed based on LASSO regression, including:

[0088] Construct the LASSO regression objective function, expressed as:

[0089]

[0090] where yi is the sound propagation loss, is the predicted value of the model, n is the total number of samples, and βj is the jth model parameter.

[0091] After training, the output of the LASSO model, pML(X), is expressed as:

[0092]

[0093] Among them, xj is the jth feature in the input features.

[0094] Combining physical modeling with data-driven models improves the ability to accurately model propagation loss and enhances prediction stability. By combining the output of physical models with data-driven models, a dynamic propagation loss prediction framework is formed to achieve higher-precision predictions.

[0095] In one embodiment, a dynamic propagation loss prediction model is constructed by combining a physical model and a data-driven model, which is expressed as:

[0096]

[0097] Where p^(x,y,z,t) is the final output of the dynamic propagation loss prediction model, which represents the sound pressure prediction at position (x,y,z) and time t, pphysical(x,y,z,t) is the output of the physical model, pML(X) is the output of the data-driven model, and α is an adaptive weight between 0 and 1.

[0098] Based on the random forest algorithm, the model is trained using the partitioned training set data. By constructing multiple decision trees and employing random sampling and feature selection methods, the model's stability and accuracy are improved. Each tree is independently generated during training, and the final prediction results are synthesized through a voting mechanism. K-fold cross-validation is used to evaluate model performance, ensuring consistency and stability across different data subsets. Metrics such as mean squared error (MSE) and mean absolute error (MAE) are used to evaluate model predictions and ensure model reliability.

[0099] Furthermore, the adaptive weight α is dynamically adjusted based on the prediction error of the physical model and the data-driven model, which is expressed as:

[0100]

[0101] Among them, Ephysical is the prediction error of the physical model, and EML is the prediction error of the data-driven model.

[0102] By combining physical models with data-driven models and training them using random forests, an efficient and dynamic propagation loss prediction model can be constructed. This approach not only improves the accuracy and stability of predictions but also adapts to changes in the ocean environment in real time, providing more reliable propagation loss estimates. This comprehensive model provides strong support for underwater acoustic propagation research and practical applications, helping to optimize sonar systems and improve the performance of underwater communications.

[0103] Step 5: Perform propagation loss prediction based on the multimodal input feature set and the trained dynamic propagation loss prediction model, and output an underwater acoustic propagation loss prediction result.

[0104] In one embodiment, the processed and feature-extracted multimodal input feature set Xs is fed into a trained dynamic propagation loss prediction model. Based on the input features, the model calculates and outputs a prediction of underwater acoustic propagation loss, providing real-time or near-real-time propagation loss estimation.

[0105] In one embodiment, an error index between an output of a dynamic propagation loss prediction model and a measured value is calculated, wherein the error index includes a root mean square error and an absolute error;

[0106] Based on the error analysis results, the collected ocean environment parameters are adjusted and steps 1 to 5 are continuously executed to achieve continuous optimization of the dynamic propagation loss prediction model.

[0107] By predicting propagation loss based on a multimodal input feature set and a dynamic propagation loss prediction model, and combining it with error analysis for continuous optimization, the model's prediction accuracy and adaptability can be significantly improved. This cyclical process ensures that the model can respond to changes in the ocean environment in real time, continuously improving the reliability of prediction results, providing a solid foundation for underwater acoustic propagation applications and enhancing the performance and efficiency of sonar systems.

[0108] This invention effectively improves the accuracy and real-time nature of propagation loss prediction by collecting and processing marine environmental data in real time, combining physical models with data-driven models. The rich information acquired from multiple sensors, along with Kalman filtering and normalization techniques, ensures data reliability and consistency. Finally, a dynamic prediction model trained using a random forest model can rapidly respond to environmental changes and provide accurate propagation loss estimates, thereby optimizing the performance of underwater acoustic propagation systems and enhancing their application in complex marine environments.

[0109] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination of the two. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or component, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or component. In the present disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries computer-readable program code. Such a propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium that can transmit, propagate, or transport a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium may be transmitted using any suitable medium, including but not limited to wires, optical cables, RF (radio frequency), etc., or any suitable combination thereof.

[0110] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0111] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages, or a combination thereof, including object-oriented programming languages such as Java, Smalltalk, C++, and conventional procedural programming languages such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a stand-alone software package, partially on the user's computer and partially on a remote computer, or entirely on the remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., through the Internet using an Internet service provider).

[0112] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram can represent a module, program segment, or a part of code, and the module, program segment, or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0113] The units involved in the embodiments described in this disclosure may be implemented in software or hardware, wherein the name of a unit does not necessarily limit the unit itself.

[0114] The above introduces the preferred embodiments of the present invention, which is intended to make the spirit of the present invention clearer and easier to understand, and is not intended to limit the present invention. Any modifications, replacements, and improvements made within the spirit and principles of the present invention should be included in the scope of protection outlined by the claims attached to the present invention.

Claims

1. A method for predicting underwater sound propagation loss for dynamic environmental parameters, comprising: Step 1: Real-time collection of ocean environmental parameters using multi-source sensors, including but not limited to sound velocity profile, temperature gradient, seabed sediments, and sonar signals; Step 2: preprocessing the collected ocean environment parameters, including using Kalman filtering to remove noise from the ocean environment parameters and standardizing the ocean environment parameters; Step 3, extracting key features from the pre-processed ocean environment parameters to establish a multimodal input feature set; Step 4: construct a dynamic transmission loss prediction model based on the physical model and the data-driven model, and use random forest to train the dynamic transmission loss prediction model; Step 5: Perform propagation loss prediction based on the multimodal input feature set and the trained dynamic propagation loss prediction model, and output an underwater acoustic propagation loss prediction result.

2. The underwater sound propagation loss prediction method for dynamic environmental parameters according to claim 1, characterized in that: Based on the preset data collection frequency, a sound velocity profiler is used to collect the sound velocity in the water layer, a temperature detector is used to collect the water temperature at different depths, a bottom sediment detector is used to obtain the physical properties of the seabed, and a sonar receiver is used to capture the intensity of the sound signal.

3. The underwater sound propagation loss prediction method for dynamic environmental parameters according to claim 1, characterized in that: Based on the Kalman filter algorithm, the collected ocean environment parameters are dynamically modeled to remove random noise, and a state model and an observation model are constructed. The ocean environment parameters are corrected through the steps of state prediction, error covariance prediction, state update, and error covariance update; Preprocessing the collected ocean environment parameters further includes standardizing the ocean environment parameters, which is expressed as: Where x is the original data, μ is the mean, and σ is the standard deviation; And the missing data points are filled based on linear interpolation.

4. The underwater sound propagation loss prediction method for dynamic environmental parameters according to claim 1, characterized in that: Extracting the physical characteristics, frequency domain characteristics, and derivative characteristics of the ocean environment parameters based on the preprocessed data, and constructing a feature matrix X based on the extracted characteristics, and dividing it into a training set and a test set; Select the best regularization parameter λ through cross-validation, use the training set data to train the LASSO regression model, and calculate the coefficient of each feature in the feature matrix X; Evaluate the coefficient of each feature in the feature matrix X, extract features with non-zero coefficients, and construct a multimodal input feature set Xs.

5. The underwater sound propagation loss prediction method for dynamic environmental parameters according to claim 4, characterized in that: Constructing physical models based on wave equations, including: Construct the three-dimensional acoustic wave equation, expressed as: Where p is the sound pressure and c is the speed of sound; Set the boundary conditions and initial conditions, use the numerical method to solve the wave equation, and obtain the output physical model, which is expressed as: p physical (x, y, z, t) = ∫ Ω G(x,y,z;ξ,η,ζ,t)f(ξ,η,ζ)dV Where G is the Green's function, which describes the influence of the sound source position (ξ,η,ζ) on the observation point (x,y,z), f is the sound source function, which represents the intensity and distribution of the sound source, Ω is the area where the sound source is located, and dV represents the volume element.

6. The underwater sound propagation loss prediction method for dynamic environmental parameters according to claim 4, characterized in that: Build a data-driven model based on LASSO regression, including: Construct the LASSO regression objective function, expressed as: where yi is the sound propagation loss, is the predicted value of the model, n is the total number of samples, and βj is the jth model parameter; After training, the output of the LASSO model, pML(X), is expressed as: Among them, xj is the j-th feature in the input features, and β0 is the constant term of the model.

7. The underwater sound propagation loss prediction method for dynamic environmental parameters according to claim 5, characterized in that: Combining the physical model and the data-driven model, a dynamic propagation loss prediction model is constructed, which is expressed as: Where p^(x,y,z,t) is the final output of the dynamic propagation loss prediction model, which represents the sound pressure prediction at position (x,y,z) and time t, pphysical(x,y,z,t) is the output of the physical model, pML(X) is the output of the data-driven model, and α is an adaptive weight between 0 and 1. The adaptive weight α is dynamically adjusted based on the prediction error of the physical model and the data-driven model, which is expressed as: Among them, Ephysical is the prediction error of the physical model, and EML is the prediction error of the data-driven model.

8. The underwater sound propagation loss prediction method for dynamic environmental parameters according to claim 1, characterized in that: Calculating error indicators between the output of the dynamic propagation loss prediction model and the measured value, wherein the error indicators include root mean square error and absolute error; Based on the error analysis results, the collected ocean environment parameters are adjusted and steps 1 to 5 are continuously executed to achieve continuous optimization of the dynamic propagation loss prediction model.

9. A device for predicting underwater sound propagation loss based on dynamic environmental parameters, comprising: at least one processor; as well as at least one memory including computer program code, The at least one memory and the computer program code are configured to, together with the at least one processor, enable the apparatus to perform the method according to any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the method according to any one of claims 1 to 8 when executed by a processor.

Images