Signal enhancement method and system adapted to complex marine environment

Abstract

The invention provides a signal enhancement method and system adapted to a complex marine environment. The method comprises the following steps: receiving an original signal flow from an underwater sensor network, and carrying out preliminary optimization processing to obtain an optimized initial signal; meanwhile, marine environment parameters are obtained based on real-time monitoring equipment; constructing a dynamic adaptive filter bank, and processing the optimized initial signal to obtain a signal subjected to environmental parameter compensation; periodically appearing known reference signals are identified and extracted, fine time delay estimation is carried out on the signals collected by the sensor nodes through the known reference signals, and signals after synchronous correction are generated; analyzing the spectral characteristics of the signals after synchronous correction, selecting a primary function set which can best represent the characteristics of the target signals, constructing an over-complete dictionary, and generating clear signals; and fusing the data outputs of the plurality of heterogeneous sensors to generate a final optimized enhanced signal. According to the technical scheme provided by the invention, the stability and reliability of underwater communication and detection are improved.

Description

Technical Field

[0001] The embodiments of the present application relate to the field of underwater communication and signal processing technology, and in particular, to a signal enhancement method and system that can adapt to complex marine environments. Background Art

[0002] In the field of underwater communication and detection, especially in the complex and changeable marine environment, signal transmission faces many challenges. Underwater sensor networks are widely used in fields such as marine monitoring, environmental research, resource exploration, and military reconnaissance, requiring the ability to accurately transmit and process large amounts of data in real time. However, due to the complexity of the marine environment, such as the influence of factors such as temperature gradients, salinity changes, and water flow movement, serious multipath effects and noise interference occur during signal propagation, which seriously affects the communication quality and detection accuracy. Therefore, there is an urgent need for a signal enhancement method that can adapt to complex marine environments to ensure the stability and reliability of underwater communication and detection systems.

[0003] At present, the main methods for underwater signal processing include preliminary processing of the original signal through traditional filtering and denoising technology to improve multipath effects and noise interference. Use fixed filters with preset parameters to process the signal in an attempt to compensate for the impact of marine environmental factors on the signal. Perform time delay estimation based on known reference signals to achieve time alignment of signals and reduce the impact of phase differences. Use spectrum analysis to select a set of basis functions, and reconstruct the signal through a matching pursuit algorithm to enhance the target information frequency band.

[0004] However, although the existing solutions have improved the quality of underwater signals to a certain extent, they still have the following shortcomings: traditional filtering and denoising technologies cannot be adjusted dynamically, and it is difficult to cope with complex changes in the ocean environment, especially under dynamic conditions such as temperature gradients, salinity changes, and water flow movement. Static filter groups rely on preset parameters and cannot adjust the filter coefficients in real time to adapt to the changing ocean environment, resulting in limited compensation effects. Existing delay estimation methods are usually based on fixed reference signals and fail to fully consider changes in ocean environment parameters, resulting in low time alignment accuracy and affecting the effect of subsequent signal processing. Existing spectrum analysis and reconstruction methods are difficult to effectively separate and enhance target information frequency bands in complex backgrounds, resulting in insufficient clarity of the final generated signal, affecting communication and detection performance.

[0005] In summary, existing solutions have obvious limitations in adapting to complex marine environments, and a more advanced and flexible method is needed to improve the stability and reliability of underwater communication and detection. Summary of the invention

[0006] The embodiments of the present application provide a signal enhancement method and system that are adaptable to complex marine environments, so as to solve the problems of poor stability and reliability of underwater communication and detection in the prior art.

[0007] In a first aspect, an embodiment of the present application provides a signal enhancement method adapted to a complex marine environment, comprising:

[0008] Receive the original signal stream from the underwater sensor network and perform preliminary optimization processing to improve multipath effects and noise interference to obtain an optimized initial signal; at the same time, obtain ocean environmental parameters based on real-time monitoring equipment, which include temperature gradient, salinity changes and water flow movement information;

[0009] According to the optimized initial signal and the real-time acquired ocean environment parameters, a dynamic adaptive filter group is constructed, and the filter coefficients are adjusted to compensate for the influence of temperature gradient, salinity change and water flow movement, and the optimized initial signal is processed to obtain a signal compensated for the environmental parameters;

[0010] Based on the signal compensated for environmental parameters, a periodically appearing known reference signal is identified and extracted, and the known reference signal is used to perform a fine delay estimation on the signal collected by each sensor node to achieve precise time alignment, so as to eliminate the phase difference caused by the propagation speed difference and generate a synchronously corrected signal;

[0011] Analyze the spectral characteristics of the synchronously corrected signal, select the basis function set that best represents the characteristics of the target signal, construct an over-complete dictionary, and reconstruct the synchronously corrected signal through a matching pursuit algorithm to separate and enhance the target information frequency band to generate a clear signal;

[0012] By fusing the data outputs of multiple heterogeneous sensors and combining them with the real-time monitored ocean environment parameters, the clear signals are weighted and combined to generate the final optimized enhanced signal, providing more stable and reliable communication and detection performance.

[0013] Optionally, the method constructs a dynamic adaptive filter group according to the optimized initial signal and the real-time acquired ocean environmental parameters, adjusts the filter coefficients to compensate for the influence of temperature gradient, salinity change and water flow movement, processes the optimized initial signal, and obtains a signal compensated for the environmental parameters, including:

[0014] Using the optimized initial signal and the real-time acquired ocean environment parameters, a filter model suitable for the current ocean conditions is selected to ensure that effective signal compensation can be provided under various complex conditions;

[0015] Based on the selected filter model, a dynamic adaptive filter group consisting of multiple filter units with different characteristics is constructed, each filter unit can independently respond to specific marine environmental parameters and generate a filter characteristic configuration adapted to the current environmental conditions;

[0016] By using the optimized initial signal and the real-time monitored ocean environment parameters, the environmental change trend at future moments is measured based on the machine learning algorithm, and the estimated adjustment direction is provided for the dynamic adaptive filter group to generate a prediction adjustment strategy.

[0017] According to the prediction adjustment strategy, combined with the filter characteristic configuration adapted to the current environmental conditions, the coefficients of each filter unit in the dynamic adaptive filter group are updated in real time, so that the dynamic adaptive filter group can respond to changes in the marine environment in real time, automatically adjust to the best working state, and ensure the optimization of the compensation effect;

[0018] The optimized initial signal is input into a dynamic adaptive filter group that has been configured and adjusted. The signal is processed layer by layer by the dynamic adaptive filter group to remove signal distortion caused by temperature gradient, salinity change and water flow movement, and obtain a signal compensated for environmental parameters.

[0019] Optionally, based on the signal compensated for the environmental parameters, identifying and extracting a periodically appearing known reference signal, using the known reference signal to perform fine delay estimation on the signal collected by each sensor node, achieving precise time alignment to eliminate the phase difference caused by the propagation speed difference, and generating a synchronously corrected signal, including:

[0020] Using the signal compensated for the environmental parameters, through a pattern recognition algorithm, identifying and extracting a known reference signal that periodically appears in the signal compensated for the environmental parameters, to obtain a known reference signal;

[0021] According to the known reference signal, the time delay between the signal collected by each sensor node and the known reference signal is calculated, and a fine delay estimation is performed using a minimum mean square error criterion to ensure that the time alignment accuracy reaches a sub-sample level, thereby obtaining a fine delay estimation result;

[0022] Based on the refined delay estimation result, the time axis of the signal collected by each sensor node is adjusted to accurately align all signals in time, eliminate the phase difference caused by the difference in propagation speed, and generate a time-aligned signal;

[0023] The time-aligned signals are combined to generate synchronously corrected signals, which eliminate the phase difference caused by the difference in propagation speed and provide a basis for subsequent spectrum analysis and signal reconstruction.

[0024] Optionally, the using the signal compensated for the environmental parameters to identify and extract a known reference signal that periodically appears in the signal compensated for the environmental parameters through a pattern recognition algorithm to obtain the known reference signal includes:

[0025] The signal after environmental parameter compensation is input into a pre-trained pattern recognition model to perform pattern analysis and feature extraction on the signal to generate a classification result of a candidate reference signal;

[0026] According to the classification results of the candidate reference signals, high-confidence candidate reference signals are screened out, wherein the candidate reference signals present periodic characteristics in time series, and the candidate reference signals after preliminary screening are obtained;

[0027] Based on the candidate reference signals after the preliminary screening, matching and verification are performed with template signals in a known reference signal library to confirm the validity and accuracy of the candidate reference signals, remove false detection signals, and determine known reference signals;

[0028] The determined known reference signal is extracted from the signal that has been compensated for environmental parameters and used as a basis for subsequent delay estimation and time alignment processing to generate a final known reference signal.

[0029] Optionally, analyzing the spectral characteristics of the synchronously corrected signal, selecting a basis function set that best represents the characteristics of the target signal, constructing an overcomplete dictionary, and reconstructing the synchronously corrected signal through a matching pursuit algorithm, separating and strengthening the target information frequency band, and generating a clear signal, includes:

[0030] Utilizing the synchronously corrected signal, performing spectrum analysis on the synchronously corrected signal to obtain amplitude and phase information at different frequencies, and obtaining detailed spectrum characteristics;

[0031] According to the detailed spectrum characteristics, based on sparse representation theory, a basis function set that best represents the characteristics of the target signal is selected from a predefined basis function library to ensure the accuracy of subsequent processing;

[0032] Based on the basis function set, an overcomplete dictionary containing multiple signal modes is constructed to provide a rich representation basis for signal reconstruction;

[0033] The overcomplete dictionary is combined with the synchronously corrected signal, and a matching pursuit algorithm is used to gradually select a basis function that best matches the signal characteristics, thereby effectively reconstructing the signal and obtaining a reconstructed signal;

[0034] The reconstructed signal is processed to separate and strengthen the target information frequency band, suppress the influence of background noise and other irrelevant frequency bands, and finally generate a clear signal.

[0035] Optionally, combining the overcomplete dictionary with the synchronously corrected signal, using a matching pursuit algorithm, and gradually selecting a basis function that best matches the signal characteristics to achieve effective reconstruction of the signal to obtain a reconstructed signal, includes:

[0036] Initializing the input of a matching pursuit algorithm using the overcomplete dictionary and the synchronously corrected signal to prepare for basis function selection;

[0037] According to the initialized input, based on the matching pursuit algorithm, searching the overcomplete dictionary for the basis function that best represents the current maximum amplitude component of the synchronous correction signal, and recording the basis function and the coefficients corresponding to the basis function to obtain a preliminary selection result;

[0038] Based on the preliminary selection result, calculate and subtract the part reconstructed by the selected basis function from the synchronously corrected signal to obtain a residual signal component;

[0039] Using the residual signal component, repeating the process of selecting the basis function and updating the residual signal until a preset stop condition is met, thereby generating a final selection and update result;

[0040] Based on the final selection and update results, the selected basis functions and the coefficients corresponding to the selected basis functions are recombined, and the original synchronously corrected signal is effectively reconstructed to obtain a reconstructed signal.

[0041] Optionally, the data outputs of multiple heterogeneous sensors are fused, and combined with the real-time monitored ocean environment parameters, the clear signals are weighted and combined to generate a final optimized enhanced signal, providing more stable and reliable communication and detection performance, including:

[0042] Prepare weighted combination processing using the clear signal, data outputs of multiple heterogeneous sensors and real-time monitored ocean environmental parameters;

[0043] Based on the clear signal and the real-time monitored ocean environment parameters, the reliability of each sensor data is evaluated, the weight coefficient of each sensor data is determined, and the sensor data reliability evaluation result is obtained;

[0044] According to the sensor data reliability evaluation result, weighted processing is performed on the data outputs of multiple heterogeneous sensors to obtain a weighted sensor data set;

[0045] Using the weighted sensor data set in combination with the real-time monitored ocean environment parameters, adjusting the parameter settings of the weighted combination algorithm to generate an adjusted weighted combination algorithm;

[0046] Based on the adjusted weighted combination algorithm, the clear signal and the weighted sensor data set are comprehensively processed to generate a final optimized enhanced signal.

[0047] In a second aspect, an embodiment of the present application provides a signal enhancement system adapted to a complex marine environment, including:

[0048] The receiving monitoring module is used to receive the original signal stream from the underwater sensor network and perform preliminary optimization processing to improve the multipath effect and noise interference to obtain the optimized initial signal; at the same time, the marine environmental parameters are obtained based on the real-time monitoring equipment, and the marine environmental parameters include temperature gradient, salinity change and water flow movement information;

[0049] Constructing an adjustment module, which is used to construct a dynamic adaptive filter group according to the optimized initial signal and the real-time acquired ocean environmental parameters, and adjust the filter coefficients to compensate for the influence of temperature gradient, salinity change and water flow movement, and process the optimized initial signal to obtain a signal compensated for the environmental parameters;

[0050] An identification and generation module is used to identify and extract a periodically appearing known reference signal based on the signal compensated for environmental parameters, and use the known reference signal to perform fine delay estimation on the signal collected by each sensor node to achieve precise time alignment, so as to eliminate the phase difference caused by the propagation speed difference and generate a synchronously corrected signal;

[0051] An analysis and reconstruction module is used to analyze the spectral characteristics of the synchronously corrected signal, select a basis function set that best represents the characteristics of the target signal, construct an over-complete dictionary, and reconstruct the synchronously corrected signal through a matching pursuit algorithm to separate and enhance the target information frequency band to generate a clear signal;

[0052] The fusion processing module is used to fuse the data outputs of multiple heterogeneous sensors and combine them with the marine environmental parameters monitored in real time to perform weighted combination processing on the clear signals to generate the final optimized enhanced signals, providing more stable and reliable communication and detection performance.

[0053] In a third aspect, an embodiment of the present application provides a computing device, comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a signal enhancement method that adapts to complex marine environments as described in any one of the first aspects.

[0054] In a fourth aspect, an embodiment of the present application provides a computer storage medium storing a computer program. When the computer program is executed by a computer, a signal enhancement method adapted to a complex marine environment as described in any one of the first aspects is implemented.

[0055] In an embodiment of the present application, an original signal stream from an underwater sensor network is received and preliminarily optimized to improve multipath effects and noise interference, thereby obtaining an optimized initial signal; at the same time, marine environmental parameters are obtained based on real-time monitoring equipment, and the marine environmental parameters cover temperature gradients, salinity changes, and water flow movement information; a dynamic adaptive filter group is constructed based on the optimized initial signal and the marine environmental parameters obtained in real time, and the filter coefficients are adjusted to compensate for the influence of temperature gradients, salinity changes, and water flow movement, and the optimized initial signal is processed to obtain a signal compensated for environmental parameters; based on the signal compensated for environmental parameters, known reference signals that appear periodically are identified and extracted. The method comprises the following steps: first, analyzing the spectral characteristics of the synchronously corrected signal, selecting the basis function set that best represents the target signal characteristics, constructing an over-complete dictionary, and reconstructing the synchronously corrected signal through a matching pursuit algorithm, separating and strengthening the target information frequency band, and generating a clear signal; and second, fusing the data outputs of multiple heterogeneous sensors and combining them with the marine environmental parameters monitored in real time to perform weighted combination processing on the clear signal to generate a final optimized enhanced signal, thereby providing more stable and reliable communication and detection performance.

[0056] The technical solution of this application has the following beneficial effects:

[0057] This application effectively improves the multipath effect and noise interference through preliminary optimization processing and adaptive filter groups, ensuring that the signal can maintain high quality in complex marine environments. The use of known reference signals for fine delay estimation eliminates the phase difference caused by differences in propagation speed and improves the accuracy of signal synchronization. The matching pursuit algorithm is used to reconstruct the synchronized and corrected signal, separate and enhance the target information frequency band, and significantly improve the clarity of the target signal. By weighted combination processing of multiple heterogeneous sensor data and combining real-time marine environmental parameters, the final optimized enhanced signal is generated, ensuring the stability and reliability of communication and detection, and is suitable for complex and changeable marine environments.

[0058] Furthermore, the embodiment of the present application also selects a filter model suitable for the current ocean conditions by using the optimized initial signal and the real-time acquired marine environmental parameters, and constructs a dynamic adaptive filter group composed of multiple filter units with different characteristics. Each filter unit can independently respond to specific marine environmental parameters and generate a filter characteristic configuration adapted to the current environmental conditions. Based on the machine learning algorithm, the environmental change trend at the future moment is predicted, the prediction adjustment strategy is generated, and the coefficients of the filter unit are updated in real time, so that the filter group can respond to the marine environmental changes immediately and automatically adjust to the best working state to ensure the optimization of the compensation effect. Finally, the optimized initial signal is input into the configured and adjusted dynamic adaptive filter group, and the signal distortion caused by the temperature gradient, salinity change and water flow movement is removed layer by layer to obtain a signal compensated by environmental parameters. Based on the signal compensated by environmental parameters, the known reference signal that appears periodically is identified and extracted by the pattern recognition algorithm, the time delay between the signal collected by each sensor node and the known reference signal is calculated, and the minimum mean square error criterion is used to perform fine delay estimation to ensure that the time alignment accuracy reaches the sub-sample level. Based on the results of fine delay estimation, the time axis of the signal collected by each sensor node is adjusted to eliminate the phase difference caused by the difference in propagation speed, and generate a synchronously corrected signal, which provides a basis for subsequent spectrum analysis and signal reconstruction.

[0059] Through the above method, the stability and reliability of underwater communication and detection are significantly improved. Through the real-time adjustment of the dynamic adaptive filter group, the influence of temperature gradient, salinity change and water flow movement is effectively compensated, the signal distortion is significantly reduced, and the quality of signal transmission in complex marine environments is ensured. In addition, the use of known reference signals for fine delay estimation achieves accurate time alignment, eliminates the phase difference caused by the difference in propagation speed, and improves the signal synchronization accuracy. The final generated synchronized and corrected signal provides a solid foundation for subsequent spectrum analysis and signal reconstruction, greatly enhances the clarity and reliability of the target signal, and is suitable for complex and changeable marine environments.

[0060] These and other aspects of the present application will become more clearly understood in the description of the following embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0062] Figure 1 A flow chart of a signal enhancement method adapted to complex marine environments provided in an embodiment of the present application;

[0063] Figure 2 A schematic diagram of the structure of a signal enhancement system adapted to complex marine environments provided in an embodiment of the present application;

[0064] Figure 3 A schematic diagram of the structure of a computing device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0065] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application.

[0066] In some of the processes described in the specification and claims of this application and the above-mentioned figures, multiple operations that appear in a specific order are included, but it should be clearly understood that these operations may not be executed in the order in which they appear in this article or executed in parallel. The serial numbers of the operations, such as 101, 102, etc., are only used to distinguish between different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes may include more or fewer operations, and these operations may be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in this article are used to distinguish different messages, devices, modules, etc., do not represent the order of precedence, and do not limit the "first" and "second" to be different types.

[0067] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work are within the scope of protection of this application.

[0068] Figure 1 A flow chart of a signal enhancement method adapted to a complex marine environment is provided for an embodiment of the present application. Figure 1 As shown, the method includes:

[0069] 101. Receive the original signal stream from the underwater sensor network, and perform preliminary optimization processing to improve the multipath effect and noise interference, and obtain the optimized initial signal; at the same time, obtain the marine environment parameters based on the real-time monitoring equipment, and the marine environment parameters include temperature gradient, salinity change and water flow movement information;

[0070] The original signal stream from the underwater sensor network is received and preliminarily optimized to improve the multipath effect and noise interference, and obtain the optimized initial signal. At the same time, the ocean environment parameters are obtained based on the real-time monitoring equipment, and the ocean environment parameters cover the temperature gradient, salinity change and water flow movement information. The underwater sensor network is composed of multiple distributed sensor nodes. The original signals collected by these nodes contain a lot of noise and multipath reflection, which affects the quality of the signal. Through preliminary optimization processing, such as filtering and denoising technology, the signal quality can be significantly improved, providing a clearer basic signal for subsequent processing. In addition, the real-time monitoring equipment is used to obtain the current ocean environment parameters, which are crucial for the construction of the subsequent dynamic adaptive filter group.

[0071] First, the original signal stream from the underwater sensor network is received. These signals may be affected by the complex marine environment and show multipath effects and noise interference. Then, preliminary optimization processing techniques such as bandpass filtering and wavelet denoising are applied to remove unnecessary frequency components and random noise, thereby improving signal quality. At the same time, the real-time monitoring equipment continuously collects marine environmental parameters, including temperature gradients, salinity changes, and water flow movement information. These data will be used to construct and adjust the dynamic adaptive filter group in subsequent steps.

[0072] In a typical marine monitoring application scenario, assume that an underwater sensor network is deployed to monitor the water quality of a certain sea area. Each sensor node in the network periodically sends the collected raw signal to the central processing unit. After receiving these signals, the central processing unit first applies a bandpass filter to remove high-frequency noise, and then uses a wavelet denoising algorithm to further optimize the signal quality. At the same time, the temperature-salinity-depth instrument CTD and current meter installed at different depths continuously transmit real-time data such as temperature, salinity and water flow velocity to the central processing unit, which provides the necessary environmental parameter support for the subsequent dynamic adaptive filter group.

[0073] 102. According to the optimized initial signal and the real-time acquired ocean environment parameters, a dynamic adaptive filter group is constructed, and the filter coefficients are adjusted to compensate for the influence of temperature gradient, salinity change and water flow movement, and the optimized initial signal is processed to obtain a signal compensated for the environmental parameters;

[0074] According to the optimized initial signal and the real-time acquired ocean environmental parameters, a dynamic adaptive filter group is constructed, and the filter coefficients are adjusted to compensate for the temperature gradient, salinity changes and water flow movement. The optimized initial signal is processed to obtain a signal compensated for environmental parameters. The dynamic adaptive filter group consists of multiple filter units with different characteristics. Each unit can independently respond to specific ocean environmental parameters and generate a filter characteristic configuration that adapts to the current environmental conditions. The machine learning algorithm predicts the environmental change trend at future moments, generates a prediction adjustment strategy, and updates the coefficients of the filter unit in real time, so that the filter group can respond to changes in the ocean environment immediately and automatically adjust to the best working state to ensure the optimal compensation effect.

[0075] Using the optimized initial signal and the real-time monitored ocean environment parameters, the filter model suitable for the current ocean conditions is selected to ensure effective signal compensation under various complex conditions. Based on the selected filter model, a dynamic adaptive filter group consisting of multiple filter units with different characteristics is constructed. Each filter unit can independently respond to specific ocean environment parameters and generate a filter characteristic configuration that adapts to the current environmental conditions. The machine learning algorithm is used to predict the environmental change trend at future moments, generate a prediction adjustment strategy, and update the coefficients of the filter unit in real time, so that the filter group can respond to changes in the ocean environment immediately and automatically adjust to the best working state to ensure the optimal compensation effect.

[0076] Continuing with the above-mentioned ocean monitoring application scenario, the central processing unit starts the construction of the dynamic adaptive filter group after receiving the optimized initial signal and real-time environmental parameters. The system first analyzes the current temperature gradient, salinity changes, and water flow movement information, and selects the filter model that best suits the current conditions. Subsequently, based on the selected model, a filter group consisting of multiple filter units is constructed, each of which is specifically designed to deal with the influence of a specific environmental parameter. For example, one filter unit focuses on compensating for temperature gradients, while another targets salinity changes. The machine learning algorithm predicts the trend of environmental changes in the next few hours, and adjusts the coefficients of each filter unit in advance to ensure that the filter group is always in the best working state, effectively compensates for the impact of environmental factors on the signal, and generates a signal compensated for environmental parameters.

[0077] 103. Based on the signal compensated for the environmental parameters, identify and extract the known reference signal that appears periodically, use the known reference signal to perform fine delay estimation on the signal collected by each sensor node, achieve precise time alignment, eliminate the phase difference caused by the propagation speed difference, and generate a synchronously corrected signal;

[0078] Based on the signal compensated by environmental parameters, the known reference signals that appear periodically are identified and extracted. These known reference signals are used to perform fine delay estimation on the signals collected by each sensor node, and accurate time alignment is achieved to eliminate the phase difference caused by the difference in propagation speed, and generate synchronized and corrected signals. The known reference signal is a signal that appears periodically and has clear characteristics. It can be accurately extracted from the complex background signal through the pattern recognition algorithm. The minimum mean square error criterion is used to calculate the time delay between the signal collected by each sensor node and the known reference signal, ensuring that the time alignment accuracy reaches the sub-sample level, thereby achieving high-precision time synchronization.

[0079] Using the signal compensated for environmental parameters, the pattern recognition algorithm is used to identify and extract the periodically appearing known reference signal to obtain the known reference signal. According to the known reference signal, the time delay between the signal collected by each sensor node and the known reference signal is calculated, and the minimum mean square error criterion is used to perform fine delay estimation to ensure that the time alignment accuracy reaches the sub-sample level. Based on the fine delay estimation results, the time axis of the signal collected by each sensor node is adjusted to accurately align all signals in time, eliminate the phase difference caused by the difference in propagation speed, and generate a time-aligned signal. Finally, the time-aligned signals are combined to generate a synchronously corrected signal, which provides a basis for subsequent spectrum analysis and signal reconstruction.

[0080] In the aforementioned ocean monitoring application scenario, the central processing unit starts the process of identifying and extracting known reference signals after obtaining the signal compensated for environmental parameters. The system uses a pattern recognition algorithm to accurately extract periodically appearing known reference signals from complex background signals, such as pre-set pulse signals or sound wave signals of a specific frequency. Subsequently, the system uses the minimum mean square error criterion to calculate the time delay between the signal collected by each sensor node and the known reference signal to ensure that the time alignment accuracy reaches the sub-sample level. Based on these fine delay estimation results, the time axis of the signal collected by each sensor node is adjusted so that all signals are accurately aligned in time, eliminating the phase difference caused by the difference in propagation speed, and generating a synchronously corrected signal. This step provides a solid foundation for subsequent spectrum analysis and signal reconstruction.

[0081] 104. Analyze the spectrum characteristics of the synchronously corrected signal, select a basis function set that best represents the characteristics of the target signal, construct an over-complete dictionary, and reconstruct the synchronously corrected signal through a matching pursuit algorithm to separate and enhance the target information frequency band to generate a clear signal;

[0082] Analyze the spectral characteristics of the synchronously corrected signal, select the set of basis functions that best represent the characteristics of the target signal, build an overcomplete dictionary, and reconstruct the synchronously corrected signal through the matching pursuit algorithm to separate and enhance the target information frequency band and generate a clear signal. Spectral analysis reveals the frequency components of the signal and its intensity distribution, which helps to identify and select the basis functions that best represent the characteristics of the target signal. The overcomplete dictionary is a set of multiple types of basis functions used to represent different features or components of the signal. The matching pursuit algorithm gradually selects the basis function that best matches the signal characteristics to achieve effective reconstruction of the signal, separate and enhance the target information frequency band, and improve the clarity of the signal.

[0083] Analyze the spectral characteristics of the synchronously corrected signal, select the set of basis functions that best represent the characteristics of the target signal, and construct an over-complete dictionary. Reconstruct the synchronously corrected signal through the matching pursuit algorithm, gradually select the basis functions that best match the signal characteristics, separate and enhance the target information frequency band, and generate a clear signal. Specifically, first perform spectrum analysis to determine the main frequency components of the signal, and then select appropriate basis functions to construct an over-complete dictionary. Next, use the matching pursuit algorithm to search the dictionary for the basis function that best represents the current maximum amplitude component of the signal, and record its index and corresponding coefficients. Repeat this process until the preset stop condition is met, generate the final selection and update results, and complete the reconstruction of the signal.

[0084] In the aforementioned ocean monitoring application scenario, the central processing unit starts the spectrum analysis process after obtaining the synchronously corrected signal. By performing a detailed analysis of the spectral characteristics of the signal, the system determines the main frequency components, selects the set of basis functions that best represent the characteristics of the target signal, and constructs an overcomplete dictionary. Next, the system uses the matching pursuit algorithm to search the dictionary for the basis function that best represents the current maximum amplitude component of the signal, and records its index and corresponding coefficients. Through multiple iterations, the target information frequency band is gradually separated and strengthened, and finally a clear reconstructed signal is generated. This step not only improves the clarity of the signal, but also provides a high-quality basic signal for subsequent data fusion and weighted combination processing.

[0085] 105. The data outputs of multiple heterogeneous sensors are integrated and combined with the real-time monitored ocean environment parameters to perform weighted combination processing on the clear signals to generate the final optimized enhanced signal, providing more stable and reliable communication and detection performance.

[0086] By fusing the data outputs of multiple heterogeneous sensors and combining them with the real-time monitored ocean environment parameters, the clear signals are weighted and combined to generate the final optimized enhanced signal, providing more stable and reliable communication and detection performance. Heterogeneous sensors refer to sensors of different types and functions, and the data they collect reflects different aspects of information. Through weighted combination processing, the data characteristics of each sensor can be comprehensively considered to generate a more comprehensive and accurate enhanced signal. The real-time monitored ocean environment parameters further enhance the effect of data fusion, ensuring that the enhanced signal generated in the end has higher stability and reliability.

[0087] The data outputs of multiple heterogeneous sensors are integrated and combined with the real-time monitored ocean environment parameters to perform weighted combination processing on the clear signals to generate the final optimized enhanced signal. Specifically, data from different types of sensors (such as temperature sensors, pressure sensors, optical sensors, etc.) are first collected, and then different weight values ​​are assigned according to the characteristics and importance of each sensor data. Combined with the real-time monitored ocean environment parameters, such as temperature gradients, salinity changes, and water flow movement information, the weight values ​​are further adjusted to ensure the best effect of data fusion. Finally, through weighted combination processing, the final optimized enhanced signal is generated to provide more stable and reliable communication and detection performance.

[0088] In the aforementioned ocean monitoring application scenario, the central processing unit starts the data fusion and weighted combination processing process after obtaining a clear signal. The system first collects data from different types of sensors (such as temperature sensors, pressure sensors, optical sensors, etc.), and then assigns different weight values ​​according to the characteristics and importance of each sensor data. For example, the data of the temperature sensor may be more critical in some cases, so it will be given a higher weight. Combined with the real-time monitored marine environmental parameters, such as temperature gradients, salinity changes, and water flow movement information, the system further adjusts the weight value to ensure the best effect of data fusion. Finally, through weighted combination processing, the final optimized enhanced signal is generated to provide more stable and reliable communication and detection performance. This process not only improves the overall quality of the signal, but also provides solid data support for subsequent decision-making and analysis.

[0089] Through the implementation of steps 101 to 105, this method significantly improves the stability and reliability of underwater communication and detection. First, through preliminary optimization processing and dynamic adaptive filter groups, the influence of temperature gradient, salinity change and water flow movement on the signal is effectively compensated, the multipath effect and noise interference are reduced, and the signal quality is significantly improved. Secondly, the known reference signal is used for fine delay estimation, accurate time alignment is achieved, the phase difference caused by the propagation speed difference is eliminated, and the signal synchronization accuracy is improved. Furthermore, through spectrum analysis and matching pursuit algorithm, the target information frequency band is separated and enhanced, a clear reconstructed signal is generated, and the recognizability of the target signal is enhanced. Finally, by fusing the data output of multiple heterogeneous sensors and combining the real-time monitored marine environment parameters, the final optimized enhanced signal is generated to ensure the stability and reliability of communication and detection. The whole process not only solves the problem of signal processing in complex marine environments, but also provides high-quality and reliable technical support for underwater communication and detection.

[0090] In order to solve the problem of dynamic adaptability of signal compensation in a complex marine environment, in some embodiments, the step 102 constructs a dynamic adaptive filter group according to the optimized initial signal and the real-time acquired marine environmental parameters, and adjusts the filter coefficients to compensate for the influence of temperature gradient, salinity change and water flow movement, and processes the optimized initial signal to obtain a signal compensated for environmental parameters, including:

[0091] The optimized initial signal and the real-time acquired ocean environment parameters are used to select a filter model suitable for the current ocean conditions, so as to ensure that effective signal compensation can be provided under various complex conditions; based on the selected filter model, a dynamic adaptive filter group consisting of a plurality of filter units with different characteristics is constructed, each filter unit can independently respond to specific ocean environment parameters, and generate a filter characteristic configuration adapted to the current environmental conditions; the obtained optimized initial signal and the real-time monitored ocean environment parameters are used to measure the environmental change trend at future moments based on a machine learning algorithm, provide an estimated adjustment direction for the dynamic adaptive filter group, and generate a prediction adjustment strategy; according to the prediction adjustment strategy, combined with the filter characteristic configuration adapted to the current environmental conditions, the coefficients of each filter unit in the dynamic adaptive filter group are updated in real time, so that the dynamic adaptive filter group can respond to changes in the ocean environment in real time, automatically adjust to the best working state, and ensure the optimization of the compensation effect; the optimized initial signal is input into the configured and adjusted dynamic adaptive filter group, and the signal is processed layer by layer by the dynamic adaptive filter group to remove the signal distortion caused by the temperature gradient, salinity change and water flow movement, so as to obtain a signal compensated by the environmental parameters.

[0092] In this embodiment, the filter model suitable for the current ocean conditions is selected to ensure that effective signal compensation can be provided under various complex conditions. It is necessary to select the filter model most suitable for the current ocean conditions according to the optimized initial signal and the real-time acquired ocean environment parameters. These filter models may include but are not limited to Kalman filtering, adaptive noise cancellation, recursive least squares method, etc. Selecting a suitable filter model can ensure the effectiveness and accuracy of subsequent processing. Constructing a dynamic adaptive filter group composed of multiple filter units with different characteristics is based on the selected filter model to construct a dynamic adaptive filter group composed of multiple filter units with different characteristics. Each filter unit can independently respond to specific ocean environment parameters, such as temperature gradient, salinity change and water flow movement information. Through this multi-unit structure, the filter group can generate a filter characteristic configuration adapted to the current environmental conditions, so as to more flexibly cope with complex ocean environment changes. Predicting the environmental change trend at a future moment is to use a machine learning algorithm to analyze the optimized initial signal and the real-time monitored ocean environment parameters to predict the environmental change trend at a future moment. This step is intended to provide an estimated adjustment direction for the dynamic adaptive filter group and generate a prediction adjustment strategy. Common machine learning algorithms include support vector machines, neural networks, and long short-term memory networks, which can predict future environmental changes based on historical data. Real-time updating of the coefficients of the filter units is to update the coefficients of each filter unit in the dynamic adaptive filter group in real time based on the prediction adjustment strategy and the configuration of the filter characteristics adapted to the current environmental conditions. This process enables the filter group to respond to changes in the marine environment instantly and automatically adjust to the best working state to ensure the optimal compensation effect. The real-time update mechanism ensures that the filter group is always in the optimal state, thereby improving the effect of signal processing. Layer-by-layer processing to remove signal distortion is to input the optimized initial signal into the configured and adjusted dynamic adaptive filter group, and process the signal layer by layer through the filter group to remove the signal distortion caused by temperature gradient, salinity change and water flow movement, and finally obtain a signal compensated for environmental parameters. The layer-by-layer processing method ensures that each step can accurately eliminate specific types of interference and improve the overall signal quality.

[0093] In an embodiment of the present application, first, based on the optimized initial signal and real-time ocean environmental parameters, the filter model that best suits the current conditions is selected, such as Kalman filtering or adaptive noise cancellation. Secondly, based on the selected filter model, a dynamic adaptive filter group consisting of multiple filter units is constructed, and each unit focuses on a specific environmental parameter. Then, a machine learning algorithm is used to predict future environmental change trends and generate a prediction adjustment strategy. Furthermore, according to the prediction adjustment strategy, the coefficients of each filter unit are updated in real time to ensure that the filter group is always in the best working state. Finally, the optimized initial signal is input into the dynamic adaptive filter group, processed layer by layer, signal distortion is removed, and a signal compensated for environmental parameters is generated.

[0094] Here is a specific example:

[0095] In a deep-sea exploration application scenario, assume that an underwater sensor network is deployed to monitor seabed geological activities. The central processing unit first receives the optimized initial signal and real-time ocean environmental parameters, such as temperature gradient, salinity changes, and water flow movement information. Based on these data, the system selects the filter model that best suits the current conditions, such as Kalman filtering, because it performs well in processing time-varying systems. Subsequently, the system constructs a dynamic adaptive filter group consisting of multiple filter units, some of which focus on compensating for temperature gradients, while others focus on salinity changes and water flow movement.

[0096] Next, the system uses the long short-term memory network to predict the environmental change trend in the next few hours and generate a forecast adjustment strategy. Based on these predictions, the system updates the coefficients of each filter unit in real time to ensure that the filter group is always in the best working state. Finally, the optimized initial signal is input into the dynamic adaptive filter group, which is processed layer by layer to remove signal distortion caused by temperature gradient, salinity change and water flow movement, and generate a signal compensated for environmental parameters.

[0097] Through this series of steps, the system not only effectively compensates for the impact of environmental factors on the signal, but also significantly improves the quality and stability of the signal, providing high-quality basic signals for subsequent geological data analysis. This method is particularly suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the reliability and accuracy of the data.

[0098] In order to solve the problem of accuracy of signal time alignment in a complex marine environment, in some embodiments, the step 103, based on the signal compensated for environmental parameters, identifies and extracts a periodically appearing known reference signal, uses the known reference signal to perform fine delay estimation on the signal collected by each sensor node, and achieves accurate time alignment to eliminate the phase difference caused by the propagation speed difference, and generates a synchronously corrected signal, including:

[0099] The signal compensated for environmental parameters is used to identify and extract the known reference signal that appears periodically in the signal compensated for environmental parameters through a pattern recognition algorithm to obtain a known reference signal; based on the known reference signal, the time delay between the signal collected by each sensor node and the known reference signal is calculated, and a fine delay estimation is performed using a minimum mean square error criterion to ensure that the time alignment accuracy reaches the sub-sample level to obtain a fine delay estimation result; based on the fine delay estimation result, the time axis of the signal collected by each sensor node is adjusted to accurately align all signals in time, eliminate the phase difference caused by the difference in propagation speed, and generate a time-aligned signal; the time-aligned signals are combined together to generate a synchronously corrected signal, and the synchronously corrected signal eliminates the phase difference caused by the difference in propagation speed, providing a basis for subsequent spectrum analysis and signal reconstruction.

[0100] In this embodiment, the known reference signal is identified and extracted to ensure the accuracy of time alignment. First, it is necessary to identify and extract the periodically appearing known reference signal from the signal compensated for the environmental parameters. The known reference signal is a pre-set, well-characterized and periodically appearing signal, such as a sound wave or pulse signal of a specific frequency. Through pattern recognition algorithms (such as support vector machines, neural networks, etc.), these known reference signals can be accurately extracted from complex background signals. These reference signals are used for subsequent time delay calculations and fine delay estimation. Fine delay estimation is to calculate the time delay between the signal collected by each sensor node and the known reference signal based on the extracted known reference signal. In order to ensure that the time alignment accuracy reaches the sub-sample level, the minimum mean square error criterion is usually used for fine delay estimation. The minimum mean square error criterion finds the best time delay estimation result by minimizing the square error between the predicted value and the actual value, thereby achieving high-precision time alignment. Adjusting the signal time axis is based on the fine delay estimation result, and adjusting the time axis of the signal collected by each sensor node so that all signals are accurately aligned in time. Specifically, the system adjusts the time axis of the collected signal according to the time delay of each sensor node relative to the known reference signal to eliminate the phase difference caused by the difference in propagation speed. This process ensures the synchronization of all signals in time and provides a solid foundation for subsequent processing. Generating a synchronized corrected signal is to combine the time-aligned signals together to generate a synchronized corrected signal. The synchronized corrected signal eliminates the phase difference caused by the difference in propagation speed and provides a high-quality basic signal for subsequent spectrum analysis and signal reconstruction. This step not only improves the synchronization accuracy of the signal, but also enhances the effectiveness and reliability of subsequent processing.

[0101] In the embodiment of the present application, first, the signal compensated for environmental parameters is used to identify and extract the known reference signal that appears periodically therein through a pattern recognition algorithm. Secondly, based on the extracted known reference signal, the time delay between the signal collected by each sensor node and the known reference signal is calculated, and the minimum mean square error criterion is used to perform fine delay estimation to ensure that the time alignment accuracy reaches the sub-sample level. Then, based on the fine delay estimation results, the time axis of the signal collected by each sensor node is adjusted so that all signals are accurately aligned in time and the phase difference caused by the difference in propagation speed is eliminated. Finally, the time-aligned signals are combined together to generate a synchronously corrected signal, which provides a basis for subsequent spectrum analysis and signal reconstruction.

[0102] Here is a specific example:

[0103] In an ocean monitoring application scenario, assume that an underwater sensor network is deployed to monitor the water quality of a certain sea area. After obtaining the signal compensated for environmental parameters, the central processing unit starts the process of identifying and extracting the known reference signal. The system uses the convolutional neural network pattern recognition algorithm to accurately extract the periodically appearing known reference signal from the complex background signal, such as a preset specific frequency sound wave signal.

[0104] Next, the system uses the minimum mean square error criterion to calculate the time delay between the signal collected by each sensor node and the known reference signal to ensure that the time alignment accuracy reaches the sub-sample level. For example, for a system with a sampling rate of 1kHz, the time alignment accuracy can reach the microsecond level. Based on these fine delay estimation results, the system adjusts the time axis of the signal collected by each sensor node so that all signals are accurately aligned in time, eliminating the phase difference caused by the difference in propagation speed.

[0105] Finally, the system combines the time-aligned signals to generate synchronized corrected signals. The synchronized corrected signals eliminate the phase difference caused by the propagation speed difference, providing a high-quality basic signal for subsequent spectrum analysis and signal reconstruction. This step not only significantly improves the synchronization accuracy of the signal, but also provides a reliable guarantee for subsequent data processing and analysis. It is suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the stability and accuracy of the data.

[0106] Through this method, the system not only achieves high-precision time alignment, but also greatly improves the signal quality and synchronization accuracy, providing a solid foundation for subsequent spectrum analysis and signal reconstruction.

[0107] In order to solve the problem of accurately extracting known reference signals in complex marine environments, in some embodiments, the step 103 uses the signal compensated for environmental parameters to identify and extract the known reference signals that appear periodically in the signal compensated for environmental parameters through a pattern recognition algorithm to obtain the known reference signals, including:

[0108] The signal compensated for environmental parameters is input into a pre-trained pattern recognition model, and pattern analysis and feature extraction processing are performed on the signal to generate classification results of candidate reference signals; based on the classification results of the candidate reference signals, high-confidence candidate reference signals are screened out, and the candidate reference signals show periodic characteristics in the time series, so as to obtain candidate reference signals after preliminary screening; based on the candidate reference signals after preliminary screening, matching verification is performed with template signals in a known reference signal library to confirm the validity and accuracy of the candidate reference signals, remove false detection signals, and determine known reference signals; the determined known reference signals are extracted from the signal compensated for environmental parameters as a basis for subsequent delay estimation and time alignment processing to generate final known reference signals.

[0109] In this embodiment, the pre-trained pattern recognition model is to ensure the accurate extraction of the known reference signal. First, the signal compensated by the environmental parameters is input into the pre-trained pattern recognition model. The model is usually trained based on a deep learning framework and can perform complex pattern analysis and feature extraction processing on the signal. The pre-trained pattern recognition model is trained using a large amount of labeled data, has strong generalization ability and robustness, and can effectively identify known reference signals that appear periodically under complex backgrounds. The classification result of generating the candidate reference signal is the classification result of the candidate reference signal generated after the pattern recognition model performs pattern analysis and feature extraction processing on the signal. These classification results not only contain the time series information of the signal, but also include its frequency characteristics, amplitude changes and other multi-dimensional characteristics. In this way, the system can preliminarily screen out candidate reference signals with periodic characteristics, providing a basis for further verification. Screening high-confidence candidate reference signals is based on the classification results of the candidate reference signals, and the system screens out high-confidence candidate reference signals. These candidate reference signals show obvious periodic characteristics in the time series, indicating that they may be known reference signals. The screening process is based on the confidence score of the classification result to ensure that only the most likely candidate signals enter the next verification stage. Matching verification and false detection removal are based on the candidate reference signals after preliminary screening. The system matches and verifies the template signals in the known reference signal library. The known reference signal library contains various pre-set reference signal templates for comparing and confirming the validity and accuracy of the candidate reference signals. By calculating the similarity score or using other matching algorithms, the system can confirm whether the candidate reference signal is a valid known reference signal and remove false detection signals. Extracting the final known reference signal is the system extracting a certain known reference signal from the signal that has been compensated for environmental parameters as the basis for subsequent delay estimation and time alignment processing. These known reference signals provide reliable support for precise time alignment, ensuring the accuracy and reliability of subsequent signal processing.

[0110] In an embodiment of the present application, first, the signal compensated for environmental parameters is input into a pre-trained pattern recognition model, and the signal is subjected to pattern analysis and feature extraction processing. Secondly, a classification result of the candidate reference signal is generated, and the candidate reference signals with periodic characteristics are preliminarily screened out. Then, based on the confidence score of the classification result, the candidate reference signals with high confidence are screened out. Furthermore, based on the candidate reference signals after the preliminary screening, a matching verification is performed with the template signal in the known reference signal library to confirm the validity and accuracy of the candidate reference signals and remove false detection signals. Finally, a determined known reference signal is extracted from the signal compensated for environmental parameters as the basis for subsequent delay estimation and time alignment processing.

[0111] Here is a specific example:

[0112] In a deep-sea exploration application scenario, assume that an underwater sensor network is deployed to monitor seabed geological activities. After obtaining the signal compensated for environmental parameters, the central processing unit starts the recognition and extraction process of the known reference signal. The system first inputs the optimized signal into a pre-trained convolutional neural network pattern recognition model, which has been trained with a large amount of annotated data and can effectively recognize periodically occurring known reference signals.

[0113] The pattern recognition model performs pattern analysis and feature extraction on the signal to generate classification results for candidate reference signals. These classification results not only contain the time series information of the signal, but also include multi-dimensional features such as its frequency characteristics and amplitude changes. Based on the confidence scores of the classification results, the system selects high-confidence candidate reference signals, which show obvious periodic characteristics in the time series.

[0114] Next, the system matches and verifies the candidate reference signals after preliminary screening with the template signals in the known reference signal library. The known reference signal library contains various pre-set reference signal templates, such as sound waves or pulse signals of specific frequencies. By calculating similarity scores or using dynamic time warping algorithms, the system confirms the validity and accuracy of the candidate reference signals and removes false detection signals.

[0115] Finally, the system extracts a known reference signal from the environmental parameter-compensated signal as the basis for subsequent delay estimation and time alignment processing. These known reference signals provide reliable support for precise time alignment, ensuring the accuracy and reliability of subsequent signal processing. For example, in a practical application, the system successfully identified and extracted a pre-set 1kHz pulse signal, achieved sub-sample-level time alignment accuracy, and significantly improved the quality of signal synchronization.

[0116] Through this method, the system not only accurately extracts the known reference signal, but also greatly improves the accuracy of time alignment, providing high-quality basic signals for subsequent spectrum analysis and signal reconstruction. It is suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the stability and accuracy of the data.

[0117] This method not only solves the problem of accurate extraction of known reference signals in complex marine environments, but also provides a solid foundation for subsequent signal processing and improves the performance and reliability of the entire system.

[0118] In order to solve the accuracy and clarity problems of signal reconstruction in complex marine environments, in some embodiments, the step 104 analyzes the spectral characteristics of the synchronously corrected signal, selects a basis function set that best represents the characteristics of the target signal, constructs an over-complete dictionary, and reconstructs the synchronously corrected signal through a matching pursuit algorithm, separates and enhances the target information frequency band, and generates a clear signal, including:

[0119] The synchronously corrected signal is used to perform spectrum analysis on the synchronously corrected signal to obtain amplitude and phase information at different frequencies and obtain detailed spectrum characteristics; according to the detailed spectrum characteristics, based on the sparse representation theory, a basis function set that best represents the characteristics of the target signal is selected from a predefined basis function library to ensure the accuracy of subsequent processing; based on the basis function set, an over-complete dictionary containing a variety of signal patterns is constructed to provide a rich representation basis for signal reconstruction; the over-complete dictionary is combined with the synchronously corrected signal, and a matching pursuit algorithm is used to gradually select the basis function that best meets the signal characteristics to achieve effective reconstruction of the signal and obtain a reconstructed signal; the reconstructed signal is processed to separate and enhance the target information frequency band, suppress the background noise and the influence of other irrelevant frequency bands, and finally generate a clear signal.

[0120] In this embodiment, the spectrum analysis process is to obtain the amplitude and phase information at different frequencies. First, the spectrum analysis process is performed on the synchronously corrected signal. The spectrum analysis reveals the distribution of the signal on different frequency components and provides detailed spectrum characteristics. Commonly used spectrum analysis methods include fast Fourier transform, short-time Fourier transform and wavelet transform. These methods can convert time domain signals into frequency domain representations, helping to identify the main frequency components of the signal and their intensity distribution. Sparse representation theory is based on sparse representation theory, and selects a set of basis functions that best represent the characteristics of the target signal from a predefined basis function library. Sparse representation theory believes that any complex signal can be approximated by a linear combination of a small number of basic signals (i.e., basis functions). By selecting appropriate basis functions, the accuracy of subsequent processing can be ensured. The basis function library usually contains multiple types of basis functions, such as sine waves, square waves, Gaussian waves, etc., and each basis function corresponds to different signal characteristics. Constructing an overcomplete dictionary is based on a selected set of basis functions to construct an overcomplete dictionary containing multiple signal patterns. An overcomplete dictionary is a set containing a large number of basis functions, which is used to represent different features or components of a signal. Compared with the traditional orthogonal basis, the overcomplete dictionary provides a richer representation basis and can better capture the subtle changes of the signal. The construction of the overcomplete dictionary not only improves the flexibility of signal representation, but also enhances the effect of signal reconstruction. The matching pursuit algorithm combines the overcomplete dictionary with the synchronously corrected signal, and uses the matching pursuit algorithm to gradually select the basis function that best matches the signal characteristics to achieve effective reconstruction of the signal. The matching pursuit algorithm selects the basis function corresponding to the maximum amplitude component of the current residual signal each time through iterative search, and updates the residual signal until the preset stop condition is met. This method can gradually approach the original signal and improve the accuracy and efficiency of reconstruction. Separating and strengthening the target information frequency band is to separate and strengthen the target information frequency band by processing the reconstructed signal, suppress the influence of background noise and other irrelevant frequency bands, and finally generate a clear signal. Specifically, the system will identify the target information frequency band according to the spectral characteristics, and highlight the information of these frequency bands through filtering or other enhancement techniques, while suppressing the interference of other irrelevant frequency bands. This step significantly improves the clarity and recognizability of the signal.

[0121] In an embodiment of the present application, first, the signal after synchronous correction is used to perform spectrum analysis and processing, and the amplitude and phase information at different frequencies are obtained to obtain detailed spectrum characteristics. Secondly, according to the detailed spectrum characteristics, based on the sparse representation theory, a set of basis functions that best represent the characteristics of the target signal is selected from a predefined basis function library. Then, based on the selected basis function set, an over-complete dictionary containing a variety of signal patterns is constructed to provide a rich representation basis for signal reconstruction. Furthermore, the over-complete dictionary is combined with the signal after synchronous correction, and a matching pursuit algorithm is used to gradually select the basis functions that best match the signal characteristics to achieve effective reconstruction of the signal. Finally, by processing the reconstructed signal, the target information frequency band is separated and strengthened, the background noise and the influence of other irrelevant frequency bands are suppressed, and finally a clear signal is generated.

[0122] Here is a specific example:

[0123] In an ocean monitoring application scenario, assume that an underwater sensor network is deployed to monitor the water quality of a certain sea area. After obtaining the synchronized and corrected signal, the central processing unit starts the spectrum analysis process. The system first uses fast Fourier transform to perform spectrum analysis on the signal, obtains the amplitude and phase information at different frequencies, and obtains detailed spectrum characteristics. For example, the system may find that certain frequency components show significant changes within a specific time period, indicating that these frequencies may be the key features of the target signal.

[0124] Next, based on the sparse representation theory, the system selects the basis function set that best represents the characteristics of the target signal from the predefined basis function library. The basis function library contains various types of basis functions, such as sine waves, square waves, Gaussian waves, etc. The system selects the basis function set that best suits the current signal based on the spectral characteristics to ensure the accuracy of subsequent processing.

[0125] Then, based on the selected set of basis functions, the system builds an overcomplete dictionary containing multiple signal patterns. This dictionary not only contains the selected basis functions, but also their various deformations and combinations, providing a rich representation basis for signal reconstruction. The construction of an overcomplete dictionary enables the system to capture subtle changes in the signal more flexibly and improve the reconstruction effect.

[0126] Subsequently, the system combines the overcomplete dictionary with the synchronously corrected signal and uses the matching pursuit algorithm to reconstruct the signal. The matching pursuit algorithm iteratively searches, each time selecting the basis function corresponding to the maximum amplitude component of the current residual signal, and updates the residual signal until the preset stop condition is met. This process gradually approaches the original signal, significantly improving the accuracy and efficiency of reconstruction.

[0127] Finally, the system processes the reconstructed signal to separate and enhance the target information frequency band, suppress the influence of background noise and other irrelevant frequency bands, and finally generate a clear signal. For example, the system may identify that the signal in a certain frequency range is an important indicator of water quality changes, and highlight the information of these frequency bands through filtering or other enhancement techniques, while suppressing the interference of other irrelevant frequency bands. This step significantly improves the clarity and recognizability of the signal, providing a high-quality basic signal for subsequent data analysis.

[0128] Through this method, the system not only solves the accuracy and clarity problems of signal reconstruction in complex marine environments, but also provides reliable support for subsequent data analysis and decision-making. It is suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the stability and accuracy of the data.

[0129] This method not only improves the quality of signal reconstruction, but also greatly enhances the identifiability of the target signal, providing a solid foundation for subsequent spectrum analysis and signal processing, thereby improving the performance and reliability of the entire system.

[0130] In order to solve the accuracy and efficiency problems of signal reconstruction in complex marine environments, in some embodiments, the overcomplete dictionary in step 104 is combined with the synchronously corrected signal, and a matching pursuit algorithm is used to gradually select the basis function that best meets the signal characteristics to achieve effective reconstruction of the signal, and obtain the reconstructed signal, including:

[0131] Using the overcomplete dictionary and the synchronously corrected signal, the input of the matching pursuit algorithm is initialized to prepare for basis function selection; according to the initialized input, based on the matching pursuit algorithm, the basis function that best represents the current maximum amplitude component of the synchronously corrected signal is searched in the overcomplete dictionary, and the basis function and the coefficients corresponding to the basis function are recorded to obtain a preliminary selection result; based on the preliminary selection result, the part reconstructed by the selected basis function is calculated and subtracted from the synchronously corrected signal to obtain the residual signal component; using the residual signal component, the process of basis function selection and residual signal update is repeated until a preset stop condition is met to generate a final selection and update result; based on the final selection and update result, the selected basis function and the coefficients corresponding to the selected basis function are recombined to effectively reconstruct the original synchronously corrected signal to obtain a reconstructed signal.

[0132] In this embodiment, the input of the matching pursuit algorithm is initialized in order to prepare for the selection of basis functions. First, the input of the matching pursuit algorithm is initialized using an overcomplete dictionary and a synchronously corrected signal. This process includes setting initial parameters, loading an overcomplete dictionary, and importing a synchronously corrected signal. Initialization ensures the accuracy and consistency of subsequent processing and provides a basis for the selection of basis functions. The basis function for searching the maximum amplitude component is based on the initialized input. The matching pursuit algorithm searches the overcomplete dictionary for the basis function that best represents the current maximum amplitude component of the synchronously corrected signal. In each iteration, the algorithm finds the basis function that best matches the maximum amplitude component of the current residual signal through an inner product operation or other similarity measurement method (such as minimum mean square error), and records the basis function and its corresponding coefficients. This step ensures that the basis function selected each time is the most consistent with the current signal characteristics, thereby improving the accuracy of reconstruction. The calculation and update of the residual signal component is based on the preliminary selection results. The system calculates and subtracts the part reconstructed by the selected basis function from the synchronously corrected signal to obtain the residual signal component. The residual signal component represents the part of the signal that is not captured by the current basis function and needs further processing. By continuously updating the residual signal components, the system can gradually approach the original signal and improve the reconstruction effect. The process of repeating basis function selection and residual signal update is to use the residual signal components, and the system repeats the process of basis function selection and residual signal update until the preset stop condition is met. Common stop conditions include reaching a predetermined number of iterations, the residual signal energy is lower than the threshold, or the reconstruction error is less than the set value. This iterative process ensures the flexibility and adaptability of the system, and can find the optimal basis function combination under different circumstances to achieve effective reconstruction of the signal. The recombining basis functions and coefficients is based on the final selection and update results. The system recombines the selected basis functions and their corresponding coefficients, and effectively reconstructs the original synchronously corrected signal to obtain the reconstructed signal. In this way, the system not only achieves accurate reconstruction of the signal, but also enhances the clarity of the target information frequency band and suppresses the influence of background noise and other irrelevant frequency bands.

[0133] In an embodiment of the present application, first, the input of the matching pursuit algorithm is initialized using an overcomplete dictionary and the synchronously corrected signal to prepare for basis function selection. Secondly, based on the initialized input, the basis function that best represents the current maximum amplitude component of the synchronously corrected signal is searched in the overcomplete dictionary, and the basis function and the corresponding coefficients are recorded. Then, based on the preliminary selection result, the part reconstructed by the selected basis function is calculated and subtracted from the synchronously corrected signal to obtain the remaining signal component. Further, using the remaining signal component, the process of basis function selection and remaining signal update is repeated until the preset stop condition is met. Finally, based on the final selection and update results, the selected basis functions and the corresponding coefficients are recombined, and the original synchronously corrected signal is effectively reconstructed to obtain the reconstructed signal.

[0134] Here is a specific example:

[0135] In an ocean monitoring application scenario, assume that an underwater sensor network is deployed to monitor the water quality of a certain sea area. After obtaining the synchronized and corrected signal, the central processing unit starts the signal reconstruction process. The system first uses the overcomplete dictionary and the synchronized and corrected signal to initialize the input of the matching pursuit algorithm to ensure that all parameters are set correctly, providing a basis for subsequent processing.

[0136] Next, based on the initialized input, the system searches the overcomplete dictionary for the basis function that best represents the current maximum amplitude component of the synchronously corrected signal. For example, the system may find that a certain frequency component exhibits a significant energy peak in the signal, select the corresponding basis function to represent this component, and record the basis function and its corresponding coefficients. This step ensures that the basis function selected each time is the most consistent with the current signal characteristics, improving the accuracy of reconstruction.

[0137] Then, based on the preliminary selection results, the system calculates and subtracts the part reconstructed by the selected basis function from the synchronously corrected signal to obtain the residual signal component. The residual signal component represents the part of the signal that is not captured by the current basis function and needs further processing. By continuously updating the residual signal component, the system can gradually approach the original signal and improve the reconstruction effect.

[0138] Subsequently, the system uses the residual signal components to repeat the process of basis function selection and residual signal update until the preset stop condition is met. For example, the system can set an upper limit on the number of iterations or a residual signal energy threshold, and stop the iteration when these conditions are met. This iterative process ensures the flexibility and adaptability of the system, and can find the optimal basis function combination in different situations to achieve effective reconstruction of the signal.

[0139] Finally, based on the final selection and update results, the system recombines the selected basis functions and their corresponding coefficients, effectively reconstructs the original synchronously corrected signal, and obtains the reconstructed signal. By analyzing the reconstructed signal, the system can separate and enhance the target information frequency band, suppress the influence of background noise and other irrelevant frequency bands, and generate a clear and high-quality basic signal. For example, in a practical application, the system successfully reconstructed the signal within a specific frequency range containing important indicators of water quality changes, significantly improving the clarity and recognizability of the signal.

[0140] Through this method, the system not only solves the accuracy and efficiency problems of signal reconstruction in complex marine environments, but also provides reliable support for subsequent data analysis and decision-making. It is suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the stability and accuracy of the data.

[0141] This method not only improves the quality of signal reconstruction, but also greatly enhances the identifiability of the target signal, providing a solid foundation for subsequent spectrum analysis and signal processing, thereby improving the performance and reliability of the entire system.

[0142] In order to solve the reliability and optimization problems of multi-source data fusion in complex marine environments, in some embodiments, the data outputs of multiple heterogeneous sensors are fused in step 105, and the clear signals are weighted combined in combination with the real-time monitored marine environment parameters to generate a final optimized enhanced signal, providing more stable and reliable communication and detection performance, including:

[0143] Prepare weighted combination processing by using the clear signal, data outputs of multiple heterogeneous sensors and real-time monitored ocean environmental parameters; evaluate the reliability of each sensor data based on the clear signal and real-time monitored ocean environmental parameters, determine the weight coefficient of each sensor data, and obtain sensor data reliability evaluation results; perform weighted processing on the data outputs of multiple heterogeneous sensors according to the sensor data reliability evaluation results to obtain a weighted sensor data set; adjust the parameter settings of the weighted combination algorithm by using the weighted sensor data set in combination with the real-time monitored ocean environmental parameters to generate an adjusted weighted combination algorithm; based on the adjusted weighted combination algorithm, perform comprehensive processing on the clear signal and the weighted sensor data set to generate a final optimized enhanced signal.

[0144] In this embodiment, the weighted combination processing is prepared to ensure the effectiveness of data fusion. First, the clear signal, the data output of multiple heterogeneous sensors and the real-time monitored marine environmental parameters are used to prepare for weighted combination processing. This step includes collecting data from different types of sensors (such as temperature sensors, pressure sensors, optical sensors, etc.), as well as real-time monitored marine environmental parameters (such as temperature gradient, salinity changes and water flow movement information). These data will be used for subsequent reliability evaluation and weighted processing. The reliability of each sensor data is evaluated based on the clear signal and the real-time monitored marine environmental parameters. The system evaluates the reliability of each sensor data, determines the weight coefficient of each sensor data, and obtains the sensor data reliability evaluation result. Reliability evaluation can be achieved through a variety of methods, such as calculating the signal-to-noise ratio, error estimation, or using machine learning algorithms to predict data quality. Through evaluation, the system can identify which sensor data is more reliable in the current environment, thereby assigning a higher weight to it. Weighted processing is based on the sensor data reliability evaluation results. The system performs weighted processing on the data output of multiple heterogeneous sensors to obtain a weighted sensor data set. Weighted processing comprehensively considers the characteristics and importance of each sensor data by assigning different weight values ​​to different sensor data. For example, the data of some sensors may be more critical under certain conditions, so they will be given a higher weight. Adjusting the parameter settings of the weighted combination algorithm is to use the weighted sensor data set, combined with the real-time monitored marine environmental parameters, the system adjusts the parameter settings of the weighted combination algorithm to generate an adjusted weighted combination algorithm. Parameter adjustment is intended to optimize the performance of the algorithm so that it can achieve the best effect in the current marine environment. For example, the system can dynamically adjust the weight coefficient based on the real-time monitored temperature gradient and water flow movement information to ensure the best effect of data fusion. The generation of the final optimized enhanced signal is based on the adjusted weighted combination algorithm. The system comprehensively processes the clear signal and the weighted sensor data set to generate the final optimized enhanced signal. This process not only improves the overall quality of the signal, but also enhances its stability and reliability, providing a high-quality basic signal for subsequent communication and detection.

[0145] In an embodiment of the present application, first, a clear signal, the data output of multiple heterogeneous sensors, and the marine environmental parameters monitored in real time are used to prepare for weighted combination processing. Secondly, based on the clear signal and the marine environmental parameters monitored in real time, the reliability of each sensor data is evaluated, the weight coefficient of each sensor data is determined, and the sensor data reliability evaluation result is obtained. Then, according to the sensor data reliability evaluation result, the data output of multiple heterogeneous sensors is weighted to obtain a weighted sensor data set. Further, using the weighted sensor data set, combined with the marine environmental parameters monitored in real time, the parameter settings of the weighted combination algorithm are adjusted to generate an adjusted weighted combination algorithm. Finally, based on the adjusted weighted combination algorithm, the clear signal and the weighted sensor data set are comprehensively processed to generate a final optimized enhanced signal.

[0146] Here is a specific example:

[0147] In a deep-sea exploration application scenario, assume that an underwater sensor network is deployed to monitor geological activities in a certain sea area. After obtaining a clear signal, the central processing unit starts the multi-source data fusion process. The system first collects data from different types of sensors (such as temperature sensors, pressure sensors, optical sensors, etc.), as well as real-time monitored marine environmental parameters (such as temperature gradients, salinity changes, and water flow information), and prepares for weighted combination processing.

[0148] Next, the system evaluates the reliability of each sensor data based on the clear signal and real-time monitoring of the ocean environment parameters, and determines the weight coefficient of each sensor data. For example, the system may find that the temperature sensor data has a higher signal-to-noise ratio in the current environment, so it assigns a higher weight to it. At the same time, for some sensors that perform poorly under certain conditions, the system will reduce their weight to reduce their impact on the final result.

[0149] Then, the system performs weighted processing on the data outputs of multiple heterogeneous sensors based on the sensor data reliability assessment results to obtain a weighted sensor data set. Weighted processing takes into account the characteristics and importance of each sensor data by assigning different weight values ​​to different sensor data. For example, the data of some sensors may be more critical under certain conditions, so they will be assigned higher weights.

[0150] Subsequently, the system uses the weighted sensor data set, combined with the real-time monitored ocean environment parameters, to adjust the parameter settings of the weighted combination algorithm to generate an adjusted weighted combination algorithm. Parameter adjustment is intended to optimize the algorithm performance to achieve the best effect in the current ocean environment. For example, the system can dynamically adjust the weight coefficient based on the real-time monitored temperature gradient and water flow movement information to ensure the best data fusion effect.

[0151] Finally, based on the adjusted weighted combination algorithm, the system comprehensively processes the clear signal and the weighted sensor data set to generate the final optimized enhanced signal. By analyzing the enhanced signal, the system can provide more stable and reliable communication and detection performance. For example, in an actual application, the system successfully generated high-quality enhanced signals, significantly improving the stability and reliability of the signal, and providing solid support for subsequent data analysis and decision-making.

[0152] Through this method, the system not only solves the reliability and optimization problems of multi-source data fusion in complex marine environments, but also greatly improves the quality and stability of the signal. It is suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the accuracy and reliability of the data.

[0153] This method not only improves the quality and stability of the signal, but also greatly enhances the adaptability and reliability of the system, providing a solid foundation for subsequent spectrum analysis and signal processing, thereby improving the performance and reliability of the entire system.

[0154] This application takes into account that underwater sensor networks face multiple challenges such as multipath effects, noise interference, and propagation speed differences in complex marine environments. These factors lead to significant time delays and phase differences between the signals collected by each sensor node, which seriously affects the signal synchronization accuracy and communication quality. In order to achieve precise time alignment and eliminate the phase difference caused by propagation speed differences, researchers have developed a method based on fine delay estimation.

[0155] The core of this method is to use periodically occurring known reference signals to estimate time delays and optimize the delay estimation results through the minimum mean square error criterion. In addition, weighting coefficients and attenuation factors are introduced to reduce the impact of large time delays and enhance the alignment effect. The final generated time-aligned signal not only improves the synchronization accuracy, but also provides a high-quality basic signal for subsequent spectrum analysis and signal reconstruction. Therefore, a new optional solution is proposed, which includes:

[0156] Based on the signal compensated by the environmental parameters, a periodically appearing known reference signal is identified and extracted, and the known reference signal is used to perform a fine delay estimation on the signal collected by each sensor node to achieve accurate time alignment to eliminate the phase difference caused by the propagation speed difference, and generate a synchronously corrected signal, including:

[0157] Using the signal compensated for the environmental parameters, a known reference signal that periodically appears in the signal compensated for the environmental parameters is identified and extracted through a pattern recognition algorithm to obtain a known reference signal;

[0158] According to the known reference signal, the time delay between the signal collected by each sensor node and the known reference signal is calculated, and the minimum mean square error criterion is used to perform fine delay estimation to ensure that the time alignment accuracy reaches the sub-sample level and obtain a fine delay estimation result;

[0159] The fine delay estimation result is obtained by the following formula:

[0160]

[0161] Among them, s i (n) represents the signal collected by the i-th sensor node; N is the signal length; τ i represents the time delay between the signal collected by the i-th sensor node and the known reference signal; τ is the time delay variable to be optimized; i represents the index of the sensor node, which is used to identify different sensors; r(n-τ) is the value of the known reference signal r(n) at time point n after a time delay of τ; w n is the weight coefficient at time point n, and the specific expression is:

[0162]

[0163] Where μ is the time center of the signal, which is half the length of the signal. σ is the standard deviation of the signal, which controls the width of the weight function and is set according to the signal characteristics; n represents the time index or sampling point index;

[0164] Based on the precise delay estimation results, the time axis of the collected signals of each sensor node is adjusted to accurately align all signals in time, eliminate the phase difference caused by the difference in transmission speed, and generate time-aligned signals.

[0165] The time-aligned signal is generated using the following formula:

[0166] s′ i (n) = s i (n+τ i )·exp(-α|τ i |)

[0167] Among them, s' i (n) represents the signal of the i-th sensor node after time alignment, the value at time point n; τ i represents the time delay between the signal collected by the i-th sensor node and the known reference signal; α is the attenuation coefficient, which is used to reduce the impact of large time delays and enhance the alignment effect; s i (n+τ i ) represents the signal s collected by the i-th sensor node i (n) Move τ on the time axis i The result after

[0168] The time-aligned signals are combined to generate synchronously corrected signals, wherein the synchronously corrected signals eliminate the phase difference caused by the propagation speed difference, thereby providing a basis for subsequent spectrum analysis and signal reconstruction;

[0169] The synchronously corrected signal is generated by the following formula:

[0170]

[0171] Among them, s sync (n) represents the signal after synchronization correction; M is the number of sensor nodes; w i is the weight coefficient of each sensor node signal, which is set according to the reliability or signal quality of the sensor. The specific expression is:

[0172]

[0173] Among them, error i represents the error of the signal of the i-th sensor node relative to the known reference signal, β is an adjustment factor that controls the sensitivity of the weight distribution; j is the index used for summation, indicating that all sensor nodes are traversed when calculating the weight coefficient.

[0174] The following is a detailed explanation of each parameter:

[0175] r(n): The value of the known reference signal at time point n. The periodic known reference signal is identified and extracted from the environmental parameter compensated signal through a pattern recognition algorithm. Using the known reference signal as a benchmark can ensure the accuracy of subsequent delay estimation and synchronization correction. The selection of the reference signal needs to have good stability and repeatability to ensure consistent time alignment between different sensor nodes.

[0176] s i (n): The value of the signal collected by the i-th sensor node at time point n. It is collected directly by each sensor node.

[0177] N: signal length.

[0178] τ i : The time delay between the signal collected by the i-th sensor node and the known reference signal.

[0179] τ: time delay variable to be optimized.

[0180] i: The index of the sensor node, used to identify different sensors.

[0181] r(n-τ): The value of the known reference signal r(n) at time point n after a time delay of τ.

[0182] w n : The weight coefficient at time point n, the specific expression is:

[0183] μ: The time center of the signal, which is half the signal length N / 2.

[0184] σ: standard deviation of the signal, controls the width of the weight function, and is set according to the signal characteristics.

[0185] n: time index or sampling point index.

[0186] μ and σ: are set according to the signal characteristics and are usually determined by statistical analysis of the signal.

[0187] s' i (n): The value of the signal of the i-th sensor node at time point n after time alignment.

[0188] τ i : The time delay between the signal collected by the i-th sensor node and the known reference signal. Obtained through fine delay estimation.

[0189] α: Attenuation coefficient, used to reduce the impact of larger time delays and enhance alignment. Determined by experience or experiment, usually small to avoid excessive attenuation.

[0190] s i (n+τ i ): The signal s collected by the i-th sensor node i (n) Move τ on the time axis i The result after.

[0191] s sync (n): The value of the signal after synchronization correction at time point n.

[0192] M: the number of sensor nodes.

[0193] w i : The weight coefficient of each sensor node signal is set according to the reliability or signal quality of the sensor. The specific expression is:

[0194] error i : The error of the signal of the ith sensor node relative to the known reference signal. It is obtained by comparing the difference between the signal of each sensor node and the known reference signal.

[0195] β: Adjustment factor, which controls the sensitivity of weight distribution. It is set according to system requirements and is usually determined through experiments or theoretical analysis.

[0196] j: The index used for summation, indicating that all sensor nodes are traversed when calculating the weight coefficient.

[0197] Here is a specific example:

[0198] Assume that in a deep-sea exploration application scenario, an underwater sensor network consisting of 5 sensor nodes is deployed to monitor seabed geological activities. The length of the signal collected by each sensor node is N = 1024 sampling points, and the sampling rate is 1kHz. It is known that the reference signal is a sound wave signal of a specific frequency, with a time center μ = N / 2 = 512 and a standard deviation σ = 50. The attenuation coefficient α = 0.01 and the adjustment factor β = 0.5.

[0199] Assume the following specific values:

[0200] Number of sensor nodes M = 5; signal length N = 1024; time delay τ1 = 3, τ2 = -2, τ3 = 1, τ4 = -4, τ5 = 0; attenuation coefficient α = 0.01; adjustment factor β = 0.5; error values ​​error1 = 0.1, error2 = 0.2, error3 = 0.15, error4 = 0.25, error5 = 0.1

[0201] Assume that the system successfully identifies and extracts a periodically occurring known reference signal r(n), which is a preset 1kHz pulse signal.

[0202] The minimum mean square error criterion is used for optimization, and the weight coefficient w n The calculation formula is:

[0203]

[0204] Assume that the time delays obtained through optimization calculation are:

[0205] τ1=3, τ2=-2, τ3=1, τ4=-4, τ5=0

[0206] Based on the refined delay estimation results, the time axis of the signal acquisition of each sensor node is adjusted and the attenuation factor is applied.

[0207] The specific formula is as follows:

[0208] s′ i (n) = s i (n+τ i )·exp(-α|τ i |)

[0209] Assume that the original signal s i (n) The value of n at the time point is [1, 2, 3, ..., 1024]. Substituting the specific time delay value and the attenuation coefficient α = 0.01:

[0210] s′1(n)=s1(n+3)·exp(-0.01×3)≈s1(n+3)·0.97

[0211] s′2(n)=s2(n-2)·exp(-0.01×2)≈s2(n-2)·0.98

[0212] s′3(n)=s3(n+1)·exp(-0.01×1)≈s3(n+1)·0.99

[0213] s′4(n)=s4(n-4)·exp(-0.01×4)≈s4(n-4)·0.96

[0214] s′5(n)=s5(n+0)·exp(-0.01×0)=s5(n)

[0215] According to the error of each sensor node relative to the known reference signal, the weight coefficient w is calculated i :

[0216]

[0217] Combine the time-aligned signals to generate a synchronized corrected signal s sync (n).

[0218] The specific formula is as follows:

[0219]

[0220] Substituting the above weight coefficients and the time-aligned signal:

[0221] s sync (n)=0.215·s′1(n)+0.185·s′2(n)+0.202·s′3(n)+0.177·s′4(n)+0.215·s′5(n)

[0222] Assume that at a certain time point n, the signal values ​​after each time alignment are:

[0223] s′1(n)=1.97, s′2(n)=2.98, s′3(n)=3.99, s′4(n)=4.96, s′5(n)=5.00

[0224] Then the synchronously corrected signal s sync The value of (n) at this time point is:

[0225] s sync(n)=0.215·1.97+0.185·2.98+0.202·3.99+0.177·4.96+0.215·5.00

[0226] s sync (n)=0.42355+0.5511+0.80598+0.878112+1.075

[0227] s sync (n)≈3.733742

[0228] Through the above steps, the system successfully achieves the time alignment of the collected signals of each sensor node, eliminates the phase difference caused by the propagation speed difference, and generates a synchronously corrected signal s sync (n). Specific numerical calculations show that at a certain time point n, the signal value after synchronization correction is approximately 3.734.

[0229] This method not only solves the problem of signal synchronization in complex marine environments, but also greatly improves the performance and reliability of the system. It is suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the accuracy and reliability of data. The signal after synchronization correction significantly improves the signal quality and synchronization accuracy, providing a solid foundation for subsequent data processing.

[0230] This application takes into account that in complex environments, the main challenges faced by signal reconstruction include noise interference, multipath effects, and signal distortion. In order to achieve high-precision signal reconstruction, researchers have developed a method based on an overcomplete dictionary and a matching pursuit algorithm. The core of this method is to use a large number of basis functions (i.e., an overcomplete dictionary) to represent the signal, and gradually select the basis function that best matches the signal characteristics through the matching pursuit algorithm, thereby achieving effective reconstruction of the original signal. Therefore, a new optional solution is proposed, which includes:

[0231] The overcomplete dictionary is combined with the synchronously corrected signal, and a matching pursuit algorithm is used to gradually select a basis function that best matches the signal characteristics to achieve effective reconstruction of the signal to obtain a reconstructed signal, including:

[0232] Initializing the input of a matching pursuit algorithm using the overcomplete dictionary and the synchronously corrected signal to prepare for basis function selection;

[0233] Input: D, s sync

[0234] in, represents an overcomplete dictionary, containing K basis functions; s sync (n) represents the value of the signal after synchronization correction at time point n; d k(n) is the value of the kth basis function in the overcomplete dictionary D at time point n; k is the basis function index; K represents the total number of basis functions in the overcomplete dictionary D;

[0235] According to the initialized input, based on the matching pursuit algorithm, searching the overcomplete dictionary for the basis function that best represents the current maximum amplitude component of the synchronous correction signal, and recording the basis function and the coefficients corresponding to the basis function to obtain a preliminary selection result;

[0236] The preliminary selection results are obtained through the following formula:

[0237]

[0238] Where k1 is the selected basis function index; d k is the kth basis function in the overcomplete dictionary D; is the k1th basis function in the overcomplete dictionary D; <·, ·> represents the inner product operation, which is used to measure the correlation between the basis function and the signal; is the selected basis function The corresponding coefficient indicates the contribution of the basis function to the signal; k is the basis function d k The weight coefficient is specifically expressed as:

[0239]

[0240] Among them, μ k is the center of the basis function index k, set to half the number of basis functions in the dictionary; σ k is the standard deviation of the basis function index k, controls the width of the weight function, and is set according to the dictionary characteristics;

[0241] Based on the preliminary selection result, calculate and subtract the part reconstructed by the selected basis function from the synchronously corrected signal to obtain a residual signal component;

[0242] The residual signal component is obtained by the following formula;

[0243]

[0244] Where r1(n) represents the residual signal component after the first iteration; γ is the attenuation coefficient, which is used to reduce the influence of larger coefficients and enhance the alignment effect; is the value of the best matching basis function found in the first iteration at time point n;

[0245] Using the residual signal component, repeating the process of basis function selection and residual signal update until a preset stop condition is met, generating a final selection and update result;

[0246] The final selection and update results are generated by the following formula:

[0247]

[0248] Where m represents the number of iterations; r m The remaining signal component after the mth iteration, the stopping condition is the reconstruction error || r m ||<∈, where ∈ is the set threshold, or the maximum number of iterations M is reached max ; r m (n) represents the residual signal component after the mth iteration, the specific value at time point n; k m is the basis function index selected in the nth iteration; is the coefficient corresponding to the selected basis function in the mth iteration; ||r m || represents the norm of the residual signal component after the mth iteration, which is used to measure the overall energy or amplitude of the residual signal; ∈ is the reconstruction error threshold, and the iteration is stopped when the energy of the residual signal is lower than this threshold; M max is the maximum number of iterations, which prevents infinite loops and ensures that the algorithm converges within a finite number of steps; is the basis function selected in the mth iteration; r m+1 (n) is the value of the residual signal component after the m+1th iteration at time point n; represents the value of the basis function selected in the mth iteration at time point n; γ is the attenuation coefficient, which is used to reduce the influence of larger coefficients and enhance the alignment effect; k is the basis function index, which is used to identify a specific basis function in the overcomplete dictionary D; w k is the basis function d k The weight coefficient of It represents the basis function selected in the mth iteration The corresponding coefficient The absolute value of d k represents the kth basis function in the overcomplete dictionary D;

[0249] Based on the final selection and update results, the selected basis functions and the coefficients corresponding to the selected basis functions are recombined to effectively reconstruct the original synchronously corrected signal to obtain a reconstructed signal;

[0250] The reconstructed signal is obtained through the following formula:

[0251]

[0252] in, represents the reconstructed signal; M is the number of basis functions finally selected; and are the coefficients and basis functions of the i-th selected basis function respectively; η is an adjustment factor that controls the degree of coefficient attenuation and is determined through experiments or theoretical analysis.

[0253] The following is a detailed explanation of each parameter:

[0254] Overcomplete dictionary, containing K basis functions. Pre-built or trained, containing multiple types of basis functions (such as wavelet, Fourier basis, etc.).

[0255] s sync (n): The value of the synchronously corrected signal at time point n. The signal obtained through the synchronous correction process.

[0256] d k (n): The value of the kth basis function in the overcomplete dictionary D at time point n.

[0257] k: basis function index.

[0258] K: The total number of basis functions in the overcomplete dictionary D.

[0259] k1: The selected basis function index.

[0260] d k : The kth basis function in the overcomplete dictionary D.

[0261] The k1th basis function in the overcomplete dictionary D.

[0262] <·,·>: Inner product operation, used to measure the correlation between basis functions and signals.

[0263] Selected basis functions The corresponding coefficient indicates the contribution of the basis function to the signal. The calculated result reflects the correlation between the selected basis function and the synchronously corrected signal.

[0264] w k : Basis function d k The weight coefficient is expressed as follows: According to the basis function index k and the set μ k and σ k calculate.

[0265] μ k : The center of the basis function index k, set to half the number of basis functions in the dictionary.

[0266] σ k : The standard deviation of the basis function index k, which controls the width of the weight function and is set according to the dictionary characteristics.

[0267] r1(n): The residual signal component after the first iteration, at time n. It is obtained by subtracting the part reconstructed by the selected basis function from the synchronously corrected signal.

[0268] γ: Attenuation coefficient, used to reduce the impact of larger coefficients and enhance the alignment effect. Determined based on experience or experiments, usually small to avoid excessive attenuation.

[0269] The attenuation factor is used to adjust the contribution of the selected basis function to make the reconstructed signal smoother. The degree of coefficient attenuation is controlled by exponential function calculation.

[0270] The best matching basis function found in the first iteration, value at time n. The basis function that best represents the current maximum amplitude component is selected from a pre-built overcomplete dictionary.

[0271] m: number of iterations.

[0272] r m : The residual signal component after the mth iteration.

[0273] r m (n): The specific value of the residual signal component after the mth iteration at time point n.

[0274] k m : The basis function indices selected in the mth iteration.

[0275] The coefficients corresponding to the selected basis functions in the mth iteration.

[0276] ||r m ||: The norm of the residual signal component after the mth iteration, used to measure the overall energy or amplitude of the residual signal.

[0277] ∈: reconstruction error threshold. When the energy of the residual signal is lower than this threshold, the iteration stops. It is set according to system requirements and is usually determined through experiments or theoretical analysis.

[0278] M max : Maximum number of iterations, to prevent infinite loops and ensure that the algorithm converges within a finite number of steps. Set according to system requirements to ensure that the algorithm converges within a finite number of steps.

[0279] The following is an introduction to the design reasons of each sub-item:

[0280] This term combines the contribution of the selected basis function and the attenuation factor to generate an adjusted basis function contribution. The multiplication operation combines the shape, coefficients and attenuation factors of the basis function to ensure that each selected basis function can be added to the reconstruction process in an appropriate proportion, thereby gradually approaching the original signal.

[0281] The subtraction operation in the formula is to obtain the synchronously corrected signal s sync (n) Remove the part reconstructed by the selected basis function The residual signal component r1(n) is obtained. In this way, new basis functions can be continuously found in subsequent iterations to represent the residual signal and gradually approach the original signal.

[0282] This formula is one of the key steps of the matching pursuit algorithm, which aims to achieve effective reconstruction of the signal by gradually selecting the basis function that best matches the signal characteristics. In each iteration, a basis function that best represents the current residual signal characteristics is selected, and the part reconstructed by the basis function is subtracted from the original signal to obtain a new residual signal component. In this way, the algorithm can gradually approach the original signal and eventually generate a high-quality reconstructed signal. Each iteration reduces the energy of the residual signal, minimizing the final reconstruction error. At the same time, by introducing the attenuation factor, the influence of the larger coefficient can be effectively suppressed, making the reconstructed signal smoother and more stable. By gradually selecting the basis function and updating the residual signal component, the overfitting problem caused by selecting too many basis functions at one time is avoided, and the robustness and accuracy of the reconstruction result are improved.

[0283] In summary, this formula ensures high accuracy and stability in the signal reconstruction process through carefully designed parameters and operations, and is suitable for signal processing tasks in complex environments.

[0284] Here is a specific example:

[0285] Assume that in an ocean monitoring application scenario, an underwater sensor network is deployed to monitor the water quality of a certain sea area. The central processing unit obtains the synchronized and corrected signal s sync (n), the signal reconstruction process is started. The system first uses an overcomplete dictionary containing 100 basis functions Prepare to perform basis function selection.

[0286] The input is the overcomplete dictionary D and the synchronously corrected signal s sync (n):

[0287] Input: D, s sync

[0288] in, represents an overcomplete dictionary, containing 100 basis functions; s sync(n) represents the value of the signal after synchronization correction at time point n; d k (n) is the value of the kth basis function in the overcomplete dictionary D at time point n; k is the basis function index; K=100 represents the total number of basis functions in the overcomplete dictionary D.

[0289] According to the initialized input, based on the matching pursuit algorithm, the basis function that best represents the current maximum amplitude component of the synchronous correction signal is searched in the overcomplete dictionary, and the basis function and the coefficients corresponding to the basis function are recorded to obtain the preliminary selection result. The specific formula is as follows:

[0290]

[0291] Where k1 is the selected basis function index; d k is the kth basis function in the overcomplete dictionary D; is the k1th basis function in the overcomplete dictionary D; <·, ·> represents the inner product operation, which is used to measure the correlation between the basis function and the signal; is the selected basis function The corresponding coefficient indicates the contribution of the basis function to the signal; k is the basis function d k The weight coefficient is expressed as follows:

[0292]

[0293] Assume μ k =50 (center of basis function index), σ k =10 (standard deviation). Substitute the above parameters and calculate the weight coefficient w k :

[0294]

[0295] Assume that the preliminary selection results obtained through optimization calculation are:

[0296]

[0297] Based on the preliminary selection results, the part reconstructed by the selected basis function is calculated and subtracted from the synchronously corrected signal to obtain the residual signal component. The specific formula is as follows:

[0298]

[0299] Assuming the attenuation coefficient γ = 0.01, substitute the specific value:

[0300] r1(n)=s sync (n)-0.8·d 25 (n) exp(-0.01×0.8)

[0301] Assumptions sync (n) The value of n at the time point is [1, 2, 3, ..., 1024], d 25 The value of (n) at time point n is [0.5, 1.0, 1.5, ..., 512]. Then the residual signal component r1(n) is calculated as follows:

[0302] For n=1:

[0303] r1(1)=1-0.8·0.5·exp(-0.008)≈1-0.8·0.5·0.992≈1-0.397≈0.603

[0304] For n = 2:

[0305] r1(2)=2-0.8·1.0·exp(-0.008)≈2-0.8·1.0·0.992≈2-0.794≈1.206

[0306] In this way, the residual signal components r1(n) of all n points can be calculated.

[0307] Using the residual signal component r1(n), repeat the process of basis function selection and residual signal update until the preset stop condition is met. The specific formula is as follows:

[0308]

[0309] Assume that after multiple iterations, the final selected basis function indices and corresponding coefficients are:

[0310]

[0311] The stopping condition is the reconstruction error ||r m ||<∈, where ∈=0.01 or the maximum number of iterations M is reached max =50.

[0312] Assume that after the fifth iteration, the energy of the remaining signal component is lower than the threshold ∈=0.01, and the iteration is stopped.

[0313] Based on the final selection and update results, the selected basis functions and the coefficients corresponding to the selected basis functions are recombined to effectively reconstruct the original synchronously corrected signal to obtain the reconstructed signal. The specific formula is as follows:

[0314]

[0315] Assume that the number of basis functions finally selected is M = 5 and the adjustment factor η = 0.02. Substitute the specific values:

[0316]

[0317] For example, for n=1:

[0318]

[0319] Simplified calculation:

[0320]

[0321] For n=2:

[0322]

[0323] Simplified calculation:

[0324]

[0325] By analogy, the reconstructed signals of all n points can be calculated

[0326] The matching pursuit algorithm is used to gradually select the basis function that best represents the signal characteristics, thus effectively reconstructing the original signal. Close to the original synchronous corrected signal s sync (n), significantly improving the reconstruction accuracy. The basis function selected in each iteration is the one that best represents the current residual signal characteristics, ensuring that each selection can minimize the energy of the residual signal. The introduction of weight coefficients highlights important basis functions, reduces the influence of edge noise, and enhances the reconstruction effect. The attenuation factor reduces the influence of larger coefficients, enhances the reconstruction effect, and makes the reconstructed signal smoother. Set the reconstruction error threshold or the maximum number of iterations to ensure that the algorithm converges within a finite number of steps and avoid infinite loops.

[0327] This method not only solves the problem of signal reconstruction in complex environments, but also greatly improves the performance and reliability of the system. It is suitable for long-term monitoring and underwater communication and detection tasks in complex environments, ensuring the accuracy and reliability of the data. The signal after synchronization correction significantly improves the signal quality and synchronization accuracy, providing a solid foundation for subsequent data processing.

[0328] Figure 2 The present invention provides a schematic diagram of a signal enhancement system adapted to complex marine environments. Figure 2 As shown, the system includes:

[0329] The receiving monitoring module 21 is used to receive the original signal stream from the underwater sensor network, and perform preliminary optimization processing to improve the multipath effect and noise interference to obtain the optimized initial signal; at the same time, the marine environmental parameters are obtained based on the real-time monitoring equipment, and the marine environmental parameters include temperature gradient, salinity change and water flow movement information;

[0330] Constructing an adjustment module 22, which is used to construct a dynamic adaptive filter group according to the optimized initial signal and the real-time acquired ocean environmental parameters, and adjust the filter coefficients to compensate for the influence of temperature gradient, salinity change and water flow movement, and process the optimized initial signal to obtain a signal compensated for the environmental parameters;

[0331] An identification and generation module 23 is used to identify and extract a periodically appearing known reference signal based on the signal compensated for environmental parameters, and use the known reference signal to perform a fine delay estimation on the signal collected by each sensor node to achieve precise time alignment, so as to eliminate the phase difference caused by the propagation speed difference and generate a synchronously corrected signal;

[0332] The analysis and reconstruction module 24 is used to analyze the spectrum characteristics of the synchronously corrected signal, select the basis function set that best represents the characteristics of the target signal, construct an over-complete dictionary, and reconstruct the synchronously corrected signal through a matching pursuit algorithm to separate and enhance the target information frequency band to generate a clear signal;

[0333] The fusion processing module 25 is used to fuse the data outputs of multiple heterogeneous sensors, and combine them with the marine environmental parameters monitored in real time to perform weighted combination processing on the clear signals to generate the final optimized enhanced signals, thereby providing more stable and reliable communication and detection performance.

[0334] Figure 2 The signal enhancement system adapted to complex marine environments can be implemented Figure 1 The implementation principle and technical effect of the signal enhancement method adapted to complex marine environments described in the illustrated embodiment will not be described in detail. The specific manner in which each module and unit performs operations in the signal enhancement system adapted to complex marine environments in the above embodiment has been described in detail in the embodiment of the method, and will not be described in detail here.

[0335] In one possible design, Figure 2 The signal enhancement system adapted to complex marine environments of the embodiment shown can be implemented as a computing device, such as Figure 3 As shown, the computing device may include a storage component 31 and a processing component 32;

[0336] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32 .

[0337] The processing component 32 is used to: receive the original signal stream from the underwater sensor network, and perform preliminary optimization processing to improve the multipath effect and noise interference to obtain an optimized initial signal; at the same time, obtain the ocean environment parameters based on the real-time monitoring equipment, and the ocean environment parameters include temperature gradient, salinity change and water flow movement information; construct a dynamic adaptive filter group according to the optimized initial signal and the real-time acquired ocean environment parameters, and adjust the filter coefficients to compensate for the influence of temperature gradient, salinity change and water flow movement, process the optimized initial signal to obtain a signal compensated for the environmental parameters; based on the signal compensated for the environmental parameters, identify and extract the periodically occurring known Reference signal, use the known reference signal to perform fine delay estimation on the signals collected by each sensor node, achieve precise time alignment, eliminate the phase difference caused by the difference in propagation speed, and generate a synchronously corrected signal; analyze the spectral characteristics of the synchronously corrected signal, select the basis function set that best represents the target signal characteristics, build an over-complete dictionary, and reconstruct the synchronously corrected signal through a matching pursuit algorithm, separate and enhance the target information frequency band, and generate a clear signal; fuse the data outputs of multiple heterogeneous sensors, and combine them with the marine environmental parameters monitored in real time, perform weighted combination processing on the clear signal, generate the final optimized enhanced signal, and provide more stable and reliable communication and detection performance.

[0338] The processing component 32 may include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component may also be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components to perform the above method.

[0339] The storage component 31 is configured to store various types of data to support operations at the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0340] Of course, the computing device may also include other components, such as input / output interfaces, display components, communication components, etc.

[0341] The input / output interface provides an interface between the processing component and the peripheral interface module, which may be an output device, an input device, etc.

[0342] The communication component is configured to facilitate, among other things, wired or wireless communications between the computing device and other devices.

[0343] Among them, the computing device can be a physical device or an elastic computing host provided by a cloud computing platform, etc. In this case, the computing device can refer to a cloud server, and the above-mentioned processing components, storage components, etc. can be basic server resources rented or purchased from the cloud computing platform.

[0344] The present application also provides a computer storage medium storing a computer program, wherein the computer program can achieve the above-mentioned Figure 1 The signal enhancement method of the illustrated embodiment is adapted to complex marine environments.

[0345] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0346] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0347] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0348] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit it. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A signal enhancement method adapted to complex marine environments, characterized in that: include: Receive the original signal stream from the underwater sensor network and perform preliminary optimization processing to improve multipath effects and noise interference to obtain an optimized initial signal; At the same time, the ocean environment parameters are obtained based on real-time monitoring equipment, and the ocean environment parameters include temperature gradient, salinity change and water flow movement information; According to the optimized initial signal and the real-time acquired ocean environment parameters, a dynamic adaptive filter group is constructed, and the filter coefficients are adjusted to compensate for the influence of temperature gradient, salinity change and water flow movement, and the optimized initial signal is processed to obtain a signal compensated for the environmental parameters; Based on the signal compensated for environmental parameters, a periodically appearing known reference signal is identified and extracted, and the known reference signal is used to perform a fine delay estimation on the signal collected by each sensor node to achieve precise time alignment, so as to eliminate the phase difference caused by the propagation speed difference and generate a synchronously corrected signal; Analyze the spectral characteristics of the synchronously corrected signal, select the basis function set that best represents the characteristics of the target signal, construct an over-complete dictionary, and reconstruct the synchronously corrected signal through a matching pursuit algorithm to separate and enhance the target information frequency band to generate a clear signal; By fusing the data outputs of multiple heterogeneous sensors and combining them with the real-time monitored ocean environment parameters, the clear signals are weighted and combined to generate the final optimized enhanced signal, providing more stable and reliable communication and detection performance.

2. The method according to claim 1, characterized in that The method comprises: constructing a dynamic adaptive filter group according to the optimized initial signal and the real-time acquired ocean environmental parameters, adjusting the filter coefficients to compensate for the influence of temperature gradient, salinity change and water flow movement, processing the optimized initial signal to obtain a signal compensated for the environmental parameters, including: Using the optimized initial signal and the real-time acquired ocean environment parameters, a filter model suitable for the current ocean conditions is selected to ensure that effective signal compensation can be provided under various complex conditions; Based on the selected filter model, a dynamic adaptive filter group consisting of multiple filter units with different characteristics is constructed, each filter unit can independently respond to specific marine environmental parameters and generate a filter characteristic configuration adapted to the current environmental conditions; By using the optimized initial signal and the real-time monitored ocean environment parameters, the environmental change trend at future moments is measured based on the machine learning algorithm, and the estimated adjustment direction is provided for the dynamic adaptive filter group to generate a prediction adjustment strategy. According to the prediction adjustment strategy, combined with the filter characteristic configuration adapted to the current environmental conditions, the coefficients of each filter unit in the dynamic adaptive filter group are updated in real time, so that the dynamic adaptive filter group can respond to changes in the marine environment in real time, automatically adjust to the best working state, and ensure the optimization of the compensation effect; The optimized initial signal is input into a dynamic adaptive filter group that has been configured and adjusted. The signal is processed layer by layer by the dynamic adaptive filter group to remove signal distortion caused by temperature gradient, salinity change and water flow movement, and obtain a signal compensated for environmental parameters.

3. The method according to claim 1, characterized in that Based on the signal compensated for the environmental parameters, identifying and extracting a periodically appearing known reference signal, using the known reference signal to perform fine delay estimation on the signal collected by each sensor node, achieving precise time alignment, eliminating the phase difference caused by the propagation speed difference, and generating a synchronously corrected signal, including: Using the signal compensated for the environmental parameters, through a pattern recognition algorithm, identifying and extracting a known reference signal that periodically appears in the signal compensated for the environmental parameters, to obtain a known reference signal; According to the known reference signal, the time delay between the signal collected by each sensor node and the known reference signal is calculated, and a fine delay estimation is performed using a minimum mean square error criterion to ensure that the time alignment accuracy reaches a sub-sample level, thereby obtaining a fine delay estimation result; Based on the refined delay estimation result, the time axis of the signal collected by each sensor node is adjusted to accurately align all signals in time, eliminate the phase difference caused by the difference in propagation speed, and generate a time-aligned signal; The time-aligned signals are combined to generate synchronously corrected signals, which eliminate the phase difference caused by the difference in propagation speed and provide a basis for subsequent spectrum analysis and signal reconstruction.

4. The method according to claim 3, characterized in that The method of using the signal compensated for the environmental parameters to identify and extract a known reference signal that periodically appears in the signal compensated for the environmental parameters through a pattern recognition algorithm to obtain the known reference signal includes: The signal after environmental parameter compensation is input into a pre-trained pattern recognition model to perform pattern analysis and feature extraction on the signal to generate a classification result of a candidate reference signal; According to the classification results of the candidate reference signals, high-confidence candidate reference signals are screened out, wherein the candidate reference signals present periodic characteristics in time series, and the candidate reference signals after preliminary screening are obtained; Based on the candidate reference signals after the preliminary screening, matching and verification are performed with template signals in a known reference signal library to confirm the validity and accuracy of the candidate reference signals, remove false detection signals, and determine known reference signals; The determined known reference signal is extracted from the signal that has been compensated for environmental parameters and used as a basis for subsequent delay estimation and time alignment processing to generate a final known reference signal.

5. The method according to claim 1, characterized in that The method of analyzing the spectral characteristics of the synchronously corrected signal, selecting a set of basis functions that best represent the characteristics of the target signal, constructing an overcomplete dictionary, and reconstructing the synchronously corrected signal through a matching pursuit algorithm, separating and strengthening the target information frequency band, and generating a clear signal includes: Utilizing the synchronously corrected signal, performing spectrum analysis on the synchronously corrected signal to obtain amplitude and phase information at different frequencies, and obtaining detailed spectrum characteristics; According to the detailed spectrum characteristics, based on sparse representation theory, a basis function set that best represents the characteristics of the target signal is selected from a predefined basis function library to ensure the accuracy of subsequent processing; Based on the basis function set, an overcomplete dictionary containing multiple signal modes is constructed to provide a rich representation basis for signal reconstruction; The overcomplete dictionary is combined with the synchronously corrected signal, and a matching pursuit algorithm is used to gradually select a basis function that best matches the signal characteristics, thereby effectively reconstructing the signal and obtaining a reconstructed signal; The reconstructed signal is processed to separate and strengthen the target information frequency band, suppress the influence of background noise and other irrelevant frequency bands, and finally generate a clear signal.

6. The method according to claim 5, characterized in that The overcomplete dictionary is combined with the synchronously corrected signal, and a matching pursuit algorithm is used to gradually select a basis function that best matches the signal characteristics to achieve effective reconstruction of the signal to obtain a reconstructed signal, including: Initializing the input of a matching pursuit algorithm using the overcomplete dictionary and the synchronously corrected signal to prepare for basis function selection; According to the initialized input, based on the matching pursuit algorithm, searching the overcomplete dictionary for the basis function that best represents the current maximum amplitude component of the synchronous correction signal, and recording the basis function and the coefficients corresponding to the basis function to obtain a preliminary selection result; Based on the preliminary selection result, calculate and subtract the part reconstructed by the selected basis function from the synchronously corrected signal to obtain a residual signal component; Using the residual signal component, repeating the process of selecting the basis function and updating the residual signal until a preset stop condition is met, thereby generating a final selection and update result; Based on the final selection and update results, the selected basis functions and the coefficients corresponding to the selected basis functions are recombined, and the original synchronously corrected signal is effectively reconstructed to obtain a reconstructed signal.

7. The method according to claim 1, characterized in that The data output of multiple heterogeneous sensors is integrated and combined with the real-time monitored ocean environment parameters to perform weighted combination processing on the clear signals to generate the final optimized enhanced signal, providing more stable and reliable communication and detection performance, including: Prepare weighted combination processing using the clear signal, data outputs of multiple heterogeneous sensors and real-time monitored ocean environmental parameters; Based on the clear signal and the real-time monitored ocean environment parameters, the reliability of each sensor data is evaluated, the weight coefficient of each sensor data is determined, and the sensor data reliability evaluation result is obtained; According to the sensor data reliability evaluation result, weighted processing is performed on the data outputs of multiple heterogeneous sensors to obtain a weighted sensor data set; Using the weighted sensor data set in combination with the real-time monitored ocean environment parameters, adjusting the parameter settings of the weighted combination algorithm to generate an adjusted weighted combination algorithm; Based on the adjusted weighted combination algorithm, the clear signal and the weighted sensor data set are comprehensively processed to generate a final optimized enhanced signal.

8. A signal enhancement system adapted to complex marine environments, characterized in that: include: The receiving monitoring module is used to receive the original signal stream from the underwater sensor network and perform preliminary optimization processing to improve the multipath effect and noise interference to obtain the optimized initial signal; at the same time, the marine environmental parameters are obtained based on the real-time monitoring equipment, and the marine environmental parameters include temperature gradient, salinity change and water flow movement information; Constructing an adjustment module, which is used to construct a dynamic adaptive filter group according to the optimized initial signal and the real-time acquired ocean environmental parameters, and adjust the filter coefficients to compensate for the influence of temperature gradient, salinity change and water flow movement, and process the optimized initial signal to obtain a signal compensated for the environmental parameters; An identification and generation module is used to identify and extract a periodically appearing known reference signal based on the signal compensated for environmental parameters, and use the known reference signal to perform fine delay estimation on the signal collected by each sensor node to achieve precise time alignment, so as to eliminate the phase difference caused by the propagation speed difference and generate a synchronously corrected signal; An analysis and reconstruction module is used to analyze the spectral characteristics of the synchronously corrected signal, select a basis function set that best represents the characteristics of the target signal, construct an over-complete dictionary, and reconstruct the synchronously corrected signal through a matching pursuit algorithm to separate and enhance the target information frequency band to generate a clear signal; The fusion processing module is used to fuse the data outputs of multiple heterogeneous sensors and combine them with the marine environmental parameters monitored in real time to perform weighted combination processing on the clear signals to generate the final optimized enhanced signals, providing more stable and reliable communication and detection performance.

9. A computing device, characterized in that It comprises a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a signal enhancement method adapted to complex marine environments as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that: A computer program is stored, and when the computer program is executed by a computer, a signal enhancement method adapted to a complex marine environment as claimed in any one of claims 1 to 7 is implemented.

Images