A single vector hydrophone line spectrum signal detection method based on time domain adaptive gain

Abstract

The application relates to the technical field of underwater acoustic signal processing and measurement, and discloses a single-vector hydrophone line spectrum signal detection method based on a time-domain adaptive gain. The method comprises the following steps: collecting sound pressure and particle vibration velocity by using a single-vector hydrophone, recording automatic gain control circuit gear position to perform amplitude correction on data, and performing direction correction on the vibration velocity by using a compass azimuth deflection angle; selecting environment noise shots without line spectrum to construct a time-domain expansion matrix, and iteratively calculating time-domain gain vectors of each channel; combining the to-be-detected shots with the gain vectors to calculate detection errors, obtaining total errors after weighted summation; setting a detection threshold to determine whether each shot has a line spectrum signal and to determine the occurrence time. The application does not need prior information of a target and a sound field, uses environment data to adaptively calculate a gain, has high detection precision for multi-distance and multi-depth line spectrum signals, has low false alarm rate, and has strong environment adaptability.

Description

Technical Field

[0001] This invention relates to the field of underwater acoustic signal processing and measurement technology, and in particular to a method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain. Background Technology

[0002] The low-frequency line spectrum components of underwater target radiated noise are caused by propeller rotation, main engine vibration, and ship structural vibration, and are characterized by concentrated and stable energy, playing a crucial role in underwater acoustic target detection, tracking, and identification. Vector hydrophones can simultaneously measure sound pressure and particle velocity in the sound field, and offer advantages such as small scale and the ability to perform orientation estimation without array formation, making them suitable for small-scale unmanned platforms. However, with the advancement of ship vibration reduction and noise reduction technologies, the signal-to-noise ratio of the radiated noise line spectrum continues to decrease, posing a greater challenge to target detection.

[0003] Currently, the main shortcomings of single-vector hydrophone line spectrum signal detection methods are as follows: First, most existing methods rely on prior information about the ambient sound field and the target signal. Traditional methods are typically based on binary hypothesis testing theory, which depends on prior knowledge of the sound field probability distribution. However, the marine environment is complex and variable, making it difficult to accurately obtain prior information, thus limiting their effectiveness in practical applications. Some methods also rely on the assumption of horizontal isotropy of the sound field; when the actual environment deviates from this assumption, the detection performance drops significantly.

[0004] Second, existing methods lack environmental adaptability and robustness. Marine environments vary significantly across different sea areas and at different times, and environmental noise and interference lead to high false alarm rates. Current research largely focuses on line spectrum orientation estimation, with relatively limited studies on time-domain detection. Single-vector hydrophones can only achieve limited array gain, making it difficult to meet detection requirements under low signal-to-noise ratio conditions.

[0005] Third, existing methods are poorly adaptable to target detection at multiple distances and depths. Changes in the depth and distance of underwater line spectrum sound sources will lead to changes in sound field propagation conditions, while most existing methods use fixed detection thresholds and lack the ability to adapt to changes in the sound field environment.

[0006] Fourth, some existing methods require a large amount of data containing target information for training, resulting in high computational complexity and difficulty in meeting real-time requirements.

[0007] In summary, existing methods still have significant shortcomings in terms of environmental adaptability, detection robustness, and reliance on prior information. There is an urgent need for a time-domain detection method for line spectrum signals that does not require prior information, can adapt to changes in environmental noise, and has good detection performance for targets at multiple distances and depths. Summary of the Invention

[0008] To address the aforementioned technical problems, this invention provides a single-vector hydrophone line spectrum signal detection method based on time-domain adaptive gain. This method aims to achieve high-precision time-domain detection of line spectrum signals without prior information on the target and ambient sound field spectrum, effectively reduce false alarm rates, and adapt to the detection needs of different sea areas and targets at multiple distances and depths.

[0009] To achieve the above objectives, the technical solution of the present invention is as follows: A method for detecting line spectrum signals from a single-vector hydrophone based on time-domain adaptive gain includes the following steps: (1) Data acquisition and correction: The sound pressure signal and particle velocity signal are acquired by a single-vector hydrophone, the automatic gain control circuit is recorded and the amplitude of the original data is corrected, and the direction of the particle velocity is corrected by the horizontal azimuth deflection angle acquired by the electronic compass, so as to obtain the corrected sound pressure sequence and the axial velocity sequence. (2) Calculation of temporal gain vector: Set the snapshot period, select environmental noise snapshots without line spectrum targets from the corrected data as preprocessed data, construct the temporal expansion matrix of each preprocessed snapshot for the sound pressure channel and each vibration velocity channel, multiply the temporal expansion matrix with the current temporal gain vector of the corresponding channel to obtain the detection output, calculate the error value between the detection output and the preset reference output, and iteratively update the temporal gain vector of the channel according to the minimum mean square error criterion based on the error value until convergence; perform the above iterative process on all preprocessed snapshots to obtain the converged temporal gain vector corresponding to each snapshot, and take the average value of all converged temporal gain vectors as the final temporal gain vector of the channel; (3) Calculation of time-domain detection error: Construct the time-domain extension matrix of each channel of the snapshot to be tested, multiply the time-domain extension matrix of each channel of the snapshot to be tested with the final time-domain gain vector of the corresponding channel to obtain the detection output of each channel; calculate the error value of the detection output of each channel relative to the reference output of the channel, and perform weighted summation of the error values ​​of each channel to obtain the total time-domain detection error of each snapshot to be tested; (4) Line spectrum signal time determination: Set a detection threshold, compare the total time domain detection error with the detection threshold, determine whether each snapshot to be tested has a line spectrum target signal, and determine the time when the line spectrum target signal appears.

[0010] In the above scheme, step (1) involves amplitude correction of the original data, specifically using the following formula: ; in, This is the amplitude-corrected sound pressure sequence. and These are the amplitude-corrected x-axis and y-axis vibration velocity sequences, respectively. , , The original sound pressure, x-axis, and y-axis particle velocity sequences were collected. , , These are the automatic gain control circuit settings for the sound pressure channel, x-axis vibration velocity channel, and y-axis vibration velocity channel, respectively.

[0011] In the above scheme, in step (1), the direction of the particle velocity is corrected, specifically using the following formula: ; in, and These are the x-axis and y-axis vibration velocity sequences after direction correction, respectively. and These are the amplitude-corrected x-axis and y-axis vibration velocity sequences. This represents the horizontal azimuth deflection angle of the vector hydrophone.

[0012] In the above scheme, step (2), constructing the time-domain extension matrix specifically includes: Let the length of the snapshot be... The dimension of the time-domain gain vector for each channel is... ,and For the first Sound pressure data for a quick snapshot , Its time-domain extension matrix for: ; in, This is the index of the sampling points within the snapshot; Similarly, construct the time-domain extended matrix of the x-axis vibration velocity. and the time-domain extended matrix of y-axis vibration velocity .

[0013] In the above scheme, step (2) of iteratively updating the gain vector of each channel based on error specifically includes: For the sound pressure channel, let the first... The temporal extension matrix of each preprocessed snapshot is: Let the first The sound pressure level snapshot data is at the first The time-domain gain vector after the next iteration is Extend the time domain matrix The OK and Multiply to obtain the detection output And calculate the detection output. With reference output error : ; in, For reference output The k Line data, reference output By Quick Shot After It consists of several elements, specifically: ; Based on the minimum mean square error criterion, the time-domain gain vector is updated using the following formula:

[0014] in, To update the step size, the superscript T denotes matrix transpose; repeat the iteration until the time-domain gain vector converges, thus obtaining the converged time-domain gain vector of the snapshot. , This represents the total number of iterations. For all Each preprocessing snapshot undergoes the above iterative process to obtain the converged time-domain gain vector of each snapshot. , The average value is calculated as the final time-domain gain vector of the sound pressure channel using the following formula: ; The above process is repeated for the preprocessed data of the x-axis and y-axis velocity channels to obtain the final time-domain gain vectors corresponding to the velocity channels. and .

[0015] In the above scheme, in step (3), the detection output of each channel... , , Calculate using the following formula: ; in, , , The first The extended matrices of sound pressure, x-axis vibration velocity, and y-axis vibration velocity of the sample to be tested. , , These are the final time-domain gain vectors of the sound pressure channel, x-axis vibration velocity channel, and y-axis vibration velocity channel obtained in step (2), respectively. The reference output Set to: ; in, To preprocess the data snapshot count, The output result is obtained by multiplying the expanded matrix of the preprocessed data with the corresponding channel time-domain gain vector; The error value of each channel's detection output relative to the reference output Calculate using the following formula: ; in, For the first The first photo to be tested The detection output of the channel, For the first The reference output of the channel, Denotes the Euclidean norm; The total time-domain detection error Calculate using the following formula: ; in, For the first The total temporal detection error of the snapshots under test For the first The first photo to be tested The maximum singular value of the channel extension matrix, For the first The first photo to be tested The channel's error value.

[0016] In the above scheme, the specific criterion for determining whether each snapshot contains a line spectrum target signal in step (4) is as follows: ; in, For the first The total temporal detection error of the snapshots under test The signal detection threshold. This indicates that the signal within the snapshot contains only ambient noise. This indicates that the signal within the snapshot contains both environmental noise and the target signal in the line spectrum.

[0017] In the above scheme, after determining the appearance time of the target signal of the line spectrum in step (4), robustness processing is also included: saving the signal time period with a duration longer than the first set number of snapshots, deleting the signal time period with a duration shorter than the first set number of snapshots, and splicing and merging adjacent signal time periods with a disappearance time shorter than the second set number of snapshots.

[0018] Through the above technical solution, the single-vector hydrophone line spectrum signal detection method based on time-domain adaptive gain provided by the present invention has the following beneficial effects: (1) No prior information required, wide range of applications This invention does not rely on prior knowledge of the ambient sound field and the target signal spectrum, nor does it require assumptions such as the horizontal isotropy of the sound field. It can directly construct a time-domain gain vector using the ambient noise data collected by a single-vector hydrophone, and has good detection performance for underwater line spectrum targets at multiple distances and depths, significantly reducing the preconditions for the application of the method.

[0019] (2) It has strong environmental adaptability and high detection robustness. This invention utilizes environmental noise snapshots from the same sea area, excluding line spectrum targets, to calculate the time-domain gain vector for each channel. The time-domain gain calculation process fully incorporates the acoustic field characteristics of the current sea area. When the sea environment changes, simply reselecting the current environmental noise snapshot and recalculating the gain vector can adaptively match the new acoustic field environment, effectively eliminating the influence of environmental noise and interference signals on the detection effect, and maintaining stable detection performance under different sea areas and time conditions.

[0020] (3) Effectively reduces false alarm rate and has high detection probability. This invention utilizes the multi-channel information of a vector hydrophone by performing time-domain adaptive gain processing on the sound pressure, x-axis velocity, and y-axis velocity channels separately, and then weighting and summing the results using the maximum singular values ​​of the extended matrices of each channel to obtain the total detection error. Simultaneously, a robust processing strategy is employed at the back end to continuously screen the judgment results, further eliminating false detections caused by transient interference and effectively reducing the false alarm rate. Sea trial data validates that, under the same false alarm rate conditions, the detection probability of this invention is significantly superior to traditional detection methods based on sound power and sound energy.

[0021] (4) The calculation process is simple and has good real-time performance. This invention employs an adaptive iterative algorithm based on the minimum mean square error criterion to solve the time-domain gain vector. It has high computational efficiency, does not require a large amount of data containing target information for training, and has low requirements for storage and computing resources. It is suitable for the real-time target detection needs of small-scale unmanned platforms such as buoys and underwater moorings. Attached Figure Description

[0022] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below.

[0023] Figure 1 This is a schematic flowchart of a single-vector hydrophone line spectrum signal detection method based on time-domain adaptive gain disclosed in an embodiment of the present invention; Figure 2 The spatial distribution of acoustic energy flux density and the azimuth trajectory of the transducer during the experiment (white curve). Figure 3The normalized curves for transducer operating time and time-domain gain error during the experiment; Figure 4 This is the curve showing the relationship between the false alarm probability and the detection probability of the target detector. Detailed Implementation

[0024] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0025] This invention provides a method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain, the process of which is as follows: Figure 1 As shown below, the present invention will be further described in detail using data collected by a single-vector hydrophone during a sea trial as the analysis object.

[0026] In this embodiment, the sampling frequency of the vector hydrophone is The sampling time was approximately 18 hours. The experiment used a hydroacoustic transducer towed by the experimental vessel as the line spectrum sound source. The transducer was activated in four specific time periods, releasing line spectrum signals at 23, 33, and 43 Hz. During these four activation periods, the hydroacoustic transducer was placed at different depths and horizontal positions: (1) the transducer was placed near the water surface, and the horizontal distance between the transducer and the hydrophone gradually increased; (2) the transducer was placed near the water surface, and the horizontal distance between the transducer and the hydrophone remained fixed; (3) the transducer sank several tens of meters, and the horizontal distance between the transducer and the hydrophone remained fixed; (4) the transducer maintained its sank depth, and the horizontal distance between the transducer and the hydrophone gradually increased. The spatial distribution of the acoustic flux density and the azimuth trajectory of the transducer during the activation time are shown in the figure. Figure 2 As shown in the figure, the background noise field is relatively uniformly distributed; the white azimuth trajectory clearly reflects the azimuth change of the transducer relative to the vector hydrophone during the four activation phases, which matches the four operating conditions (different depths and different horizontal distances) of the experimental design. This figure verifies the effectiveness of the experimental process and provides experimental scenario basis for the subsequent analysis of the time-domain gain error curve.

[0027] The single-vector hydrophone line spectrum signal detection method based on time-domain adaptive gain proposed in this invention includes the following steps in its specific processing.

[0028] Step (1): Data Acquisition and Correction Sound pressure signals and particle velocity signals were acquired using a single-vector hydrophone, while simultaneously recording the settings of each channel in the automatic gain control (AGC) circuit. Let the acquired raw sound pressure sequence be... The particle velocity sequences along the x-axis and y-axis are respectively and The AGC circuit settings for the sound pressure channel, x-axis vibration velocity channel, and y-axis vibration velocity channel are as follows: , , The original data is amplitude corrected using AGC gain, and the correction formula is as follows: ; in, This is the amplitude-corrected sound pressure sequence. and These are the amplitude-corrected x-axis and y-axis vibration velocity sequences, respectively.

[0029] While acquiring signals, the horizontal azimuth deflection angle of the vector hydrophone is recorded using an electronic compass. Based on this deflection angle, the amplitude-corrected particle velocity is oriented, resulting in the oriented-corrected x-axis velocity. and y-axis vibration velocity The corrected formula is as follows: ; in, and These are the x-axis and y-axis vibration velocity sequences after direction correction, respectively. and These are the amplitude-corrected x-axis and y-axis vibration velocity sequences. This represents the horizontal azimuth deflection angle of the vector hydrophone.

[0030] Through the above corrections, a sound pressure sequence that can be used for subsequent processing is obtained. and vibrational velocity sequence , .

[0031] Step (2): Calculation of time-domain gain vector Set the data snapshot period to (In this embodiment, we take) (corresponding to a duration of 10 seconds), temporal gain vectors of each channel , , The dimension is And satisfy (In this embodiment, we take) ).

[0032] Environmental noise snapshots containing no line spectrum target signals were selected from the corrected data as preprocessed data. Let a total of [number] snapshots be selected. One pre-processed snapshot (taken in this embodiment) For the first... Sound pressure data for a quick snapshot ,in Construct its time-domain extension matrix as follows: ; in, This is the index of the sampling points within the snapshot; Similarly, construct the time-domain extended matrix of the x-axis vibration velocity respectively. and the time-domain extended matrix of y-axis vibration velocity .

[0033] For the sound pressure channel, the first The expansion matrix of a quick snapshot Multiply by the time-domain gain vector. Let the first... The sound pressure level snapshot data is at the first The time-domain gain vector after the next iteration is .Will The OK and Multiplication yields the output ,calculate With reference output error : ; in, For reference output The k Line data, reference output By Quick Shot After It consists of several elements, specifically: ; Based on the minimum mean square error criterion, the time-domain gain vector is iteratively updated using the following formula: ; in, To update the step size, superscript This represents the matrix transpose. Repeat the iteration until convergence to obtain the final time-domain gain vector of the snapshot. ( (Total number of iterations).

[0034] Perform the above iterative process on all preprocessed snapshots to obtain... Time-domain gain vectors ( Averaging these values ​​yields the final time-domain gain vector for the sound pressure channel: ; right and The preprocessed data for each channel undergoes the same process described above to obtain the final time-domain gain vector corresponding to the vibration velocity channel. and .

[0035] Step (3): Calculation of time-domain detection error The data collected from the same sea area are photographed and corrected for amplitude and direction according to step (1), and the time-domain extension matrix of each channel is constructed according to formula (3). Let the first... The sound pressure, x-axis velocity, and y-axis velocity extension matrices of the sample photos to be tested are respectively... , , Multiply each of these by the final time-domain gain vector of the corresponding channel to obtain the detection output for each channel: ; Reference output of the signal detection process ( Set as the average value of the output results of each channel of the preprocessed data: ; in, For the first The output result is obtained by multiplying the extended matrix of each preprocessed snapshot with the corresponding channel time-domain gain vector.

[0036] Calculate the error value of each channel's detection output relative to its reference output: ; in, This represents the Euclidean norm.

[0037] Find the maximum singular value of the expansion matrix for each channel. ( Then calculate the first one according to the following formula. Total temporal detection error of the snapshots under test: .

[0038] Step (4): Determining the timing of the line spectrum signal Set signal detection threshold , will the Total temporal detection error of the snapshot under test and The following criteria are used to determine whether a line spectrum target signal exists within the snapshot: ; in, This indicates that the signal within the snapshot contains only ambient noise. This indicates that the signal within the snapshot contains both environmental noise and the target signal in the line spectrum.

[0039] Based on the above determination results, the initial appearance time of the line spectrum target signal is determined. To further improve detection reliability, robust processing is performed on the initial determination results: signal time periods with a duration longer than a first set number of snapshots (10 snapshots in this embodiment) are saved, signal time periods with a duration shorter than the first set number of snapshots are deleted, and adjacent signal time periods with a disappearance time shorter than a second set number of snapshots (5 snapshots in this embodiment) are merged. After robust processing, the accurate appearance time of the line spectrum target signal is finally output.

[0040] Figure 3 The blue curve in the figure represents the normalized error amplitude curve after time-domain gain, and the red area corresponds to the time during which the sound source works (emits line spectrum signals) during the experiment. It can be seen that during the time when the transducer emits line spectrum signals, the time-domain gain error is significantly different from the error amplitude when only environmental noise exists. Figure 4 The graph shows the relationship between the false alarm probability and the detection probability of the detection method proposed in this invention in an embodiment. This embodiment also uses the sound power signal detection method and sound energy signal detection method proposed in the paper "Signal detection method using a single vector hydrophone in ocean acoustics" (J. Acoust. Soc. Am. 152 (2), August 2022) based on the assumption of a horizontally isotropic sound field for comparison. It can be seen that, under the same false alarm rate, the method proposed in this invention has a higher detection probability and can accurately detect the transducer line spectrum signal generation time.

[0041] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain, characterized in that, Includes the following steps: (1) Data acquisition and correction: The sound pressure signal and particle velocity signal are acquired by a single-vector hydrophone, the automatic gain control circuit is recorded and the amplitude of the original data is corrected, and the direction of the particle velocity is corrected by the horizontal azimuth deflection angle acquired by the electronic compass, so as to obtain the corrected sound pressure sequence and the axial velocity sequence. (2) Calculation of temporal gain vector: Set the snapshot period, select environmental noise snapshots without line spectrum targets from the corrected data as preprocessed data, construct the temporal expansion matrix of each preprocessed snapshot for the sound pressure channel and each vibration velocity channel, multiply the temporal expansion matrix with the current temporal gain vector of the corresponding channel to obtain the detection output, calculate the error value between the detection output and the preset reference output, and iteratively update the temporal gain vector of the channel according to the minimum mean square error criterion based on the error value until convergence; The above iterative process is performed on all preprocessed snapshots to obtain the converged time-domain gain vector corresponding to each snapshot. The average value of all converged time-domain gain vectors is taken as the final time-domain gain vector of the channel. (3) Calculation of time-domain detection error: Construct the time-domain extension matrix of each channel of the snapshot to be tested, multiply the time-domain extension matrix of each channel of the snapshot to be tested with the final time-domain gain vector of the corresponding channel to obtain the detection output of each channel; calculate the error value of the detection output of each channel relative to the reference output of the channel, and perform weighted summation of the error values ​​of each channel to obtain the total time-domain detection error of each snapshot to be tested; (4) Line spectrum signal time determination: Set a detection threshold, compare the total time domain detection error with the detection threshold, determine whether there is a line spectrum target signal in each snapshot to be tested, and determine the time when the line spectrum target signal appears; In step (2), updating the gain vector of each channel based on error iteration specifically includes: For the sound pressure channel, let the first... The temporal extension matrix of each preprocessed snapshot is: Let the first The sound pressure level snapshot data is at the first The time-domain gain vector after the next iteration is: Extend the time domain matrix The OK and Multiply to obtain the detection output And calculate the detection output. With reference output error : ； in, For reference output The k Line data, reference output By Quick Shot After It consists of several elements, specifically: ； Based on the minimum mean square error criterion, the time-domain gain vector is updated using the following formula: ； in, To update the step size, the superscript T denotes matrix transpose; repeat the iteration until the time-domain gain vector converges, thus obtaining the converged time-domain gain vector of the snapshot. , This represents the total number of iterations. For all Each preprocessing snapshot undergoes the above iterative process to obtain the converged time-domain gain vector of each snapshot. , The average value is calculated as the final time-domain gain vector of the sound pressure channel using the following formula: ； The above process is repeated for the preprocessed data of the x-axis and y-axis velocity channels to obtain the final time-domain gain vectors corresponding to the velocity channels. and .

2. The method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain according to claim 1, characterized in that, In step (1), the original data is corrected for magnitude using the following formula: ； in, This is the amplitude-corrected sound pressure sequence. and These are the amplitude-corrected x-axis and y-axis vibration velocity sequences, respectively. , , The original sound pressure, x-axis, and y-axis particle velocity sequences were collected. , , These are the automatic gain control circuit settings for the sound pressure channel, x-axis vibration velocity channel, and y-axis vibration velocity channel, respectively.

3. The method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain according to claim 1, characterized in that, In step (1), the direction of the particle velocity is corrected using the following formula: ； in, and These are the x-axis and y-axis vibration velocity sequences after direction correction, respectively. and These are the amplitude-corrected x-axis and y-axis vibration velocity sequences. This represents the horizontal azimuth deflection angle of the vector hydrophone.

4. The method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain according to claim 1, characterized in that, In step (2), constructing the time-domain extension matrix specifically includes: Let the length of the snapshot be... The dimension of the time-domain gain vector for each channel is... ,and For the first Sound pressure data for a quick snapshot , Its time-domain extension matrix for: ； in, This is the index of the sampling points within the snapshot; Similarly, construct the time-domain extended matrix of the x-axis vibration velocity. and the time-domain extended matrix of y-axis vibration velocity .

5. The method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain according to claim 1, characterized in that, In step (3), each channel detects the output. , , Calculate using the following formula: ； in, , , The first The extended matrices of sound pressure, x-axis vibration velocity, and y-axis vibration velocity of the sample to be tested. , , These are the final time-domain gain vectors of the sound pressure channel, x-axis vibration velocity channel, and y-axis vibration velocity channel obtained in step (2), respectively. The reference output Set to: ； in, To preprocess the data snapshot count, The output result is obtained by multiplying the expanded matrix of the preprocessed data with the corresponding channel time-domain gain vector; The error value of each channel's detection output relative to the reference output Calculate using the following formula: ； in, For the first The first photo to be tested The detection output of the channel, For the first The reference output of the channel, Denotes the Euclidean norm; The total time-domain detection error Calculate using the following formula: ； in, For the first The total temporal domain detection error of the snapshots under test For the first The first photo to be tested The maximum singular value of the channel extension matrix, For the first The first photo to be tested The channel's error value.

6. The method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain according to claim 1, characterized in that, In step (4), the specific criteria for determining whether each snapshot contains a line spectrum target signal are as follows: ； in, For the first The total temporal domain detection error of the snapshots under test The signal detection threshold, This indicates that the signal within the snapshot contains only ambient noise. This indicates that the signal within the snapshot contains both environmental noise and the target signal in the line spectrum.

7. The method for detecting line spectrum signals of a single-vector hydrophone based on time-domain adaptive gain according to claim 1, characterized in that, In step (4), after determining the appearance time of the target signal of the line spectrum, robustness processing is also included: saving signal time periods with a duration longer than the first set number of snapshots, deleting signal time periods with a duration shorter than the first set number of snapshots, and merging adjacent signal time periods with a disappearance time shorter than the second set number of snapshots.

Images