Underwater sonar target detection system and method

Abstract

The invention provides an underwater sonar target detection system and method. The underwater sonar target detection system comprises an optical fiber hydrophone array module which comprises a plurality of optical fiber hydrophone units, each unit comprises an optical fiber Bragg grating sensor and a sound pressure signal demodulation device, and the optical fiber hydrophone array module is used for collecting underwater sound wave signals in real time and converting the underwater sound wave signals into first electric signals; the sonar transmitting-receiving array module comprises a broadband sonar transmitter, a multi-beam receiver and a signal preprocessing circuit, and is used for actively transmitting a frequency modulation continuous wave signal, receiving a target reflection echo and generating a second electric signal; and the multi-mode signal fusion processing module comprises a high-speed data acquisition card, a self-adaptive noise suppression unit and a time-frequency analysis unit. The underwater sonar target detection system and method provided by the invention have the advantages of relatively high reliability, relatively high detection performance, relatively accurate cross-modal signal fusion and capability of performing intelligent target identification and tracking.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater sonar target detection, and in particular to an underwater sonar target detection system and method. Background Art

[0002] Underwater sonar target detection technology is a core supporting technology in the fields of marine resource exploration, underwater security, and military anti-submarine warfare. Traditional detection systems mainly rely on a single type of acoustic sensor:

[0003] Sonar systems: While they offer the advantage of active detection, they suffer from issues such as large near-field blind spots and severe reverberation interference. Furthermore, the linear frequency modulation system is prone to velocity-range ambiguity in multi-target scenarios.

[0004] Fiber optic hydrophone: Although highly sensitive, it is limited by its passive working mode and is difficult to locate targets at long distances.

[0005] Multi-sensor fusion system: Existing technologies mostly use simple data-level fusion, which fails to effectively solve problems such as time-frequency mismatch of cross-modal signals and poor feature complementarity.

[0006] Therefore, it is necessary to provide an underwater sonar target detection system and method to solve the above technical problems. Summary of the Invention

[0007] The technical problem solved by the present invention is to provide an underwater sonar target detection system and method with high reliability, high detection performance, relatively accurate cross-modal signal fusion, and the ability to perform intelligent target recognition and tracking.

[0008] To solve the above technical problems, the present invention provides an underwater sonar target detection system and method, comprising: a fiber optic hydrophone array module, comprising a plurality of fiber optic hydrophone units, each unit comprising a fiber Bragg grating sensor and a sound pressure signal demodulation device, for real-time acquisition of underwater sound wave signals and conversion into a first electrical signal;

[0009] A sonar transmitter-receiver array module, comprising a broadband sonar transmitter, a multi-beam receiver, and a signal pre-processing circuit, is used to actively transmit a frequency-modulated continuous wave signal and receive a target reflected echo to generate a second electrical signal;

[0010] a multimodal signal fusion processing module, comprising a high-speed data acquisition card, an adaptive noise suppression unit, and a time-frequency analysis unit, for synchronously acquiring the first electrical signal and the second electrical signal, performing time-domain-frequency domain joint feature extraction, and performing nonlinear signal fusion;

[0011] The target feature intelligent recognition module integrates a deep learning neural network to achieve underwater target classification and trajectory prediction based on the fused multi-dimensional feature data through a trained multi-scale convolutional network model;

[0012] Dynamic calibration and self-test module, with built-in reference sound source and fiber optic phase compensation unit, for real-time calibration of the signal phase difference between the fiber optic hydrophone array and the sonar array, and for performing system fault diagnosis;

[0013] The spatial arrangement of the fiber optic hydrophone array module and the sonar transmitting-receiving array module satisfies: the spacing between the fiber optic hydrophone units is 1 / 4-1 / 2 of the sonar transmitting wavelength, and the two are three-dimensionally orthogonally distributed;

[0014] The multimodal signal fusion processing module adopts a cross-sensor correlation function optimization algorithm to generate an enhanced target feature map by maximizing the mutual information entropy of the fiber optic hydrophone and sonar signals.

[0015] Preferably, in the fiber optic hydrophone array module, each fiber Bragg grating sensor adopts a tilted grating structure, the grating period of which varies gradiently along the fiber axis, the sensitivity adjustment range is 0.1-10pm / μPa, and the operating frequency band covers 10Hz-50kHz.

[0016] Preferably, the broadband sonar transmitter of the sonar transmitting-receiving array module adopts a nonlinear frequency modulation waveform, whose instantaneous frequency changes with time according to a hyperbolic tangent function, has a bandwidth dynamic range of 5-200kHz, and a pulse compression ratio greater than 100:1.

[0017] Preferably, the adaptive noise suppression unit of the multimodal signal fusion processing module includes: an ocean environment noise separation subunit based on empirical mode decomposition, a matching field processing and spatial domain filtering joint noise reduction subunit, and a coherence enhancement subunit for optical fiber acoustic signals and sonar echoes.

[0018] Preferably, the deep learning neural network of the target feature intelligent recognition module adopts a dual-branch heterogeneous architecture, including:

[0019] (a) The first branch is a 3D convolutional network that processes the multi-beam range-azimuth information of the sonar array;

[0020] (b) The second branch is a graph convolutional network that processes the topological characteristics of the distributed sound pressure field of the fiber optic hydrophone array;

[0021] (c) The fusion layer dynamically weights the contribution of the two types of features through the attention mechanism and outputs the target type and motion parameters.

[0022] Preferably, the reference sound source of the dynamic calibration and self-test module is a tunable piezoelectric ceramic transducer array, the frequency of its transmitted signal jumps in a chaotic sequence within the range of 1-100 kHz, and the calibration accuracy reaches 0.05° phase error.

[0023] Preferably, a multi-physics field coupling analysis module is also included, which corrects the influence of underwater temperature gradient and salinity change on the sound wave propagation path in real time by solving the coupled solution of the Helmholtz wave equation and the Navier-Stokes equation.

[0024] Preferably, the collaborative working mode of the sonar transmitting-receiving array module and the fiber optic hydrophone array module includes: in near-field detection at a distance of <100 meters, the high sensitivity characteristics of the fiber optic hydrophone array are preferentially adopted; in far-field detection at a distance of ≥100 meters, the active detection mode of the sonar array is activated; and dual-mode data is seamlessly switched and confidence fused through a Bayesian reasoning framework.

[0025] Preferably, the system integrates a multi-target tracking algorithm, including: a track association submodule based on improved particle filtering, a Doppler shift compensation and velocity estimation submodule, and a three-dimensional acoustic imaging and situation visualization submodule.

[0026] The present invention also provides an underwater sonar target detection method, comprising the following steps:

[0027] S1: Collaborative acquisition of multimodal signals:

[0028] S11: collecting underwater sound pressure signals through the optical fiber hydrophone array module to generate a first electrical signal;

[0029] S12: actively transmitting nonlinear frequency-modulated sound waves through the sonar transmitting-receiving array module, and receiving target reflected echoes to generate a second electrical signal;

[0030] S2: Cross-modal data synchronization and preprocessing:

[0031] S21: The multimodal signal fusion processing module synchronously collects the first electrical signal and the second electrical signal to eliminate time domain deviation;

[0032] S22: The dynamic calibration and self-test module calibrates the phase difference between the optical fiber hydrophone and the sonar array in real time to ensure that the data time alignment error is ≤10ns;

[0033] S3: Multi-source feature fusion and enhancement:

[0034] S31: The multimodal signal fusion processing module performs a time-frequency joint analysis on the two signals to extract the sound pressure field distribution characteristics and sonar scattering characteristics;

[0035] S32: Generates a fusion feature map including target position, distance and acoustic characteristics through a cross-sensor correlation function optimization algorithm;

[0036] S4: Intelligent target recognition and classification:

[0037] S41: The target feature intelligent recognition module receives the fused feature map and performs target classification through a deep learning neural network;

[0038] S42: Output target type, size and motion trajectory prediction results based on multi-scale convolutional network;

[0039] S5: Dynamic feedback and system self-optimization:

[0040] S51: The dynamic calibration and self-test module reversely adjusts the sensitivity parameters of the optical fiber hydrophone array according to the recognition result;

[0041] S52: Real-time optimization of the frequency modulation parameters of the sonar transmission waveform to adaptively match the current underwater environment and target characteristics.

[0042] Compared with related technologies, the underwater sonar target detection system and method provided by the present invention have the following advantages:

[0043] Beneficial effects:

[0044] The present invention provides an underwater sonar target detection system and method. Through the coordinated work of fiber optic hydrophones and sonar arrays, the detection sensitivity and coverage of underwater targets are significantly improved, especially in the near-field area, the detection blind spots are effectively reduced. The advantages of active sonar and passive fiber optic sensing are combined to enhance the capture capability of weak targets in complex underwater environments. Based on multimodal signal fusion technology, the system effectively suppresses the interference of ocean environmental noise and reverberation, and improves the signal-to-noise ratio of target signals. Through a dynamic calibration mechanism, the time synchronization and phase mismatch problems of multi-sensor data are solved, and the accuracy of signal fusion is improved. Deep learning is used to drive the detection of underwater targets. The dynamic heterogeneous network architecture significantly improves the classification accuracy and recognition speed of various types of underwater targets; realizes real-time prediction and association of multi-target motion trajectories, reducing the risk of misjudgment in track intersection scenarios; through the closed-loop self-calibration function, it automatically compensates for system parameter drift caused by environmental changes and improves long-term detection stability; the built-in multi-physics field coupling analysis module effectively reduces the interference of environmental factors such as water temperature and salinity on the acoustic propagation path; optimizes the collaborative working mode of sonar emission waveform and fiber optic sensing to reduce the overall power consumption of the system; supports modular expansion design, which facilitates the flexible adjustment of the sensor array size according to detection needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 This is a functional block diagram of the underwater sonar target detection system provided by the present invention;

[0046] Figure 2 for Figure 1 The principle block diagram of the fiber optic hydrophone array module is shown;

[0047] Figure 3 for Figure 1The principle block diagram of the sonar transmitting-receiving array module shown;

[0048] Figure 4 for Figure 1 The principle block diagram of the multimodal signal fusion processing module is shown. DETAILED DESCRIPTION

[0049] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0050] Please refer to Figure 1 、 Figure 2 、 Figure 3 and Figure 4 ,in, Figure 1 This is a functional block diagram of the underwater sonar target detection system provided by the present invention; Figure 2 for Figure 1 The principle block diagram of the fiber optic hydrophone array module is shown; Figure 3 for Figure 1 The principle block diagram of the sonar transmitting-receiving array module shown; Figure 4 for Figure 1 The principle block diagram of the multimodal signal fusion processing module is shown. The underwater sonar target detection system and method include:

[0051] A fiber optic hydrophone array module, comprising a plurality of fiber optic hydrophone units, each unit comprising a fiber Bragg grating sensor and an acoustic pressure signal demodulation device, for collecting underwater acoustic wave signals in real time and converting them into a first electrical signal;

[0052] In the fiber optic hydrophone array module, each fiber Bragg grating sensor adopts a tilted grating structure, the grating period of which changes gradiently along the fiber axis, the sensitivity adjustment range is 0.1-10pm / μPa, and the operating frequency band covers 10Hz-50kHz.

[0053] A sonar transmitter-receiver array module, comprising a broadband sonar transmitter, a multi-beam receiver, and a signal pre-processing circuit, is used to actively transmit a frequency-modulated continuous wave signal and receive a target reflected echo to generate a second electrical signal;

[0054] The broadband sonar transmitter of the sonar transmitting-receiving array module adopts a nonlinear frequency modulation waveform, whose instantaneous frequency changes with time according to a hyperbolic tangent function, has a bandwidth dynamic range of 5-200kHz, and a pulse compression ratio greater than 100:1.

[0055] The collaborative working mode of the sonar transmitting-receiving array module and the fiber optic hydrophone array module includes: in near-field detection (distance <100 meters), the high sensitivity characteristics of the fiber optic hydrophone array are preferentially adopted; in far-field detection (distance ≥100 meters), the active detection mode of the sonar array is activated; the dual-mode data is seamlessly switched and confidence fused through the Bayesian reasoning framework.

[0056] a multimodal signal fusion processing module, comprising a high-speed data acquisition card, an adaptive noise suppression unit, and a time-frequency analysis unit, for synchronously acquiring the first electrical signal and the second electrical signal, performing time-domain-frequency domain joint feature extraction, and performing nonlinear signal fusion;

[0057] The adaptive noise suppression unit of the multimodal signal fusion processing module includes: an ocean environment noise separation subunit based on empirical mode decomposition, a matched field processing and spatial domain filtering joint noise reduction subunit, and a coherence enhancement subunit for optical fiber acoustic signals and sonar echoes; the noise suppression depth can reach more than 30dB.

[0058] The target feature intelligent recognition module integrates a deep learning neural network to achieve underwater target classification and trajectory prediction based on the fused multi-dimensional feature data through a trained multi-scale convolutional network model;

[0059] The deep learning neural network of the target feature intelligent recognition module adopts a dual-branch heterogeneous architecture, including:

[0060] (a) The first branch is a 3D convolutional network that processes the multi-beam range-azimuth information of the sonar array;

[0061] (b) The second branch is a graph convolutional network that processes the topological characteristics of the distributed sound pressure field of the fiber optic hydrophone array;

[0062] (c) The fusion layer dynamically weights the contribution of the two types of features through the attention mechanism and outputs the target type and motion parameters.

[0063] Dynamic calibration and self-test module, with built-in reference sound source and fiber optic phase compensation unit, for real-time calibration of the signal phase difference between the fiber optic hydrophone array and the sonar array, and for performing system fault diagnosis;

[0064] The reference sound source of the dynamic calibration and self-test module is a tunable piezoelectric ceramic transducer array, whose transmission signal frequency jumps in a chaotic sequence within the range of 1-100kHz, and the calibration accuracy reaches 0.05° phase error.

[0065] The spatial arrangement of the fiber optic hydrophone array module and the sonar transmitting-receiving array module satisfies: the spacing between the fiber optic hydrophone units is 1 / 4-1 / 2 of the sonar transmitting wavelength, and the two are three-dimensionally orthogonally distributed;

[0066] The multimodal signal fusion processing module adopts a cross-sensor correlation function optimization algorithm to generate an enhanced target feature map by maximizing the mutual information entropy of the fiber optic hydrophone and sonar signals.

[0067] It also includes a multi-physics field coupling analysis module, which corrects the effects of underwater temperature gradients and salinity changes on the sound wave propagation path in real time by solving the coupled solutions of the Helmholtz wave equation and the Navier-Stokes equation.

[0068] The underwater sonar target detection system provided by this method integrates a multi-target tracking algorithm, including: a track association submodule based on an improved particle filter, a Doppler frequency shift compensation and velocity estimation submodule, and a three-dimensional acoustic imaging and situation visualization submodule; the false alarm rate is less than 10^-4 times / hour, and the minimum detectable target speed is 0.1m / s.

[0069] The present invention also provides an underwater sonar target detection method, comprising the following steps:

[0070] S1: Synchronous acquisition of multimodal acoustic signals:

[0071] S11: The fiber optic hydrophone array module collects underwater sound pressure signals in real time at a sampling rate of 10kHz-50kHz through multiple fiber Bragg grating sensors to generate a first digital signal sequence D f (t), where the grating period gradient variation of each sensor unit ΔΛ = 5-50 nm / m, and the sensitivity calibration error ≤ ±0.5 dB;

[0072] S12: Control the broadband sonar transmitter of the sonar transmitting-receiving array module to transmit a nonlinear frequency modulation waveform signal s(t), whose instantaneous frequency changes according to the hyperbolic tangent function f(t) = 5 + 97.5 [1 + tanh (0.1 (t-10))] kHz (t∈[0,20]) ms, and simultaneously receive the target reflection echo through the multi-beam receiver at a sampling rate of 200 kHz to generate a second digital signal sequence D s (t).

[0073] S2: Cross-modal signal preprocessing and time-frequency alignment:

[0074] S21: To D f (t) Perform fiber phase demodulation:

[0075]

[0076] Extract the sound pressure change ΔP(t) = K φ Δφ(t), where the phase-to-sound pressure conversion coefficient K φ =0.15μPa / rad;

[0077] S22: To Df (t) Perform pulse compression processing:

[0078]

[0079] Where the matched filter h(t) = s * (Tt)·ω(t), the time domain window function ω(t) is weighted by the hyperbolic tangent derivative, and the pulse width after compression is ≤2μs;

[0080] S23: The reference sound source of the dynamic calibration and self-test module emits a chaotic frequency-hopping signal (1-100kHz, hopping interval 50μs) to calculate the time delay difference between the optical fiber and the sonar signal. where R xy is the cross-correlation function, f s =200kHz, achieving a time synchronization error of the two modal signals ≤5ns.

[0081] S3: Multi-physics joint feature extraction:

[0082] S31: Perform empirical mode decomposition (EMD) on ΔP(t) to separate the first three order intrinsic mode functions (IMFs) as the ocean ambient noise floor N(t), and the remaining components as the target acoustic signal S f (t);

[0083] S32: To S c (t) Perform Doppler compensation:

[0084]

[0085] Where υ is the estimated radial velocity of the target, c = 1500 m / s, f c =100kHz, the compensated signal is recorded as S c (t);

[0086] S33: Constructing the time-frequency joint feature matrix M∈R N×T :

[0087] Fiber signal characteristics: Extract S f (t) short-time energy E f , Mel cepstral coefficients (MFCC) and Wigner-Ville distribution ridges;

[0088] Sonar signal features: Extract S f (t) the distance-Doppler spectrum R(f d ), scattering cross section fluctuation variance σ 2 .

[0089] S4: Heterogeneous Deep Learning Model Inference:

[0090] S41: Input the fiber acoustic pressure field topology data into the graph convolutional network (GCN) branch:

[0091]

[0092] The adjacency matrix A is constructed by the spatial position of the sensor, and the node features

[0093] S42: Input sonar multibeam data into the 3D-CNN branch:

[0094]

[0095] The input tensor dimension is 16×16×32 (azimuth×pitch×range units);

[0096] S43: Generate final features through attention fusion layer:

[0097] α=softmax(W a [H g ;H c ])

[0098] H final =α g H g +α c H c

[0099] where α g +α c =1, the weight is dynamically adjusted in the range of 0.3-0.7.

[0100] S5: Multi-target trajectory prediction and output:

[0101] S51: To H final Perform target clustering:

[0102]

[0103] An improved K-means algorithm is used, and the distance metric includes Mahalanobis distance and Doppler consistency constraint;

[0104] S52: Track update based on particle filtering:

[0105]

[0106] Number of particles N = 1000, resampling threshold N eff =0.5N;

[0107] S53: Output 3D situation map:

[0108] Target type: ship / torpedo / marine life (confidence ≥ 90%);

[0109] Motion parameters: position error ≤ 0.5m, speed error ≤ 0.2m / s;

[0110] Threat Level: Calculated based on closing speed and trajectory intersection probability.

[0111] Compared with related technologies, the underwater sonar target detection system and method provided by the present invention have the following advantages:

[0112] Beneficial effects:

[0113] The present invention provides an underwater sonar target detection system and method. Through the coordinated work of fiber optic hydrophones and sonar arrays, the detection sensitivity and coverage of underwater targets are significantly improved, especially in the near-field area, the detection blind spots are effectively reduced. The advantages of active sonar and passive fiber optic sensing are combined to enhance the capture capability of weak targets in complex underwater environments. Based on multimodal signal fusion technology, the system effectively suppresses the interference of ocean environmental noise and reverberation, and improves the signal-to-noise ratio of target signals. Through a dynamic calibration mechanism, the time synchronization and phase mismatch problems of multi-sensor data are solved, and the accuracy of signal fusion is improved. Deep learning is used to drive the detection of underwater targets. The dynamic heterogeneous network architecture significantly improves the classification accuracy and recognition speed of various types of underwater targets; realizes real-time prediction and association of multi-target motion trajectories, reducing the risk of misjudgment in track intersection scenarios; through the closed-loop self-calibration function, it automatically compensates for system parameter drift caused by environmental changes and improves long-term detection stability; the built-in multi-physics field coupling analysis module effectively reduces the interference of environmental factors such as water temperature and salinity on the acoustic propagation path; optimizes the collaborative working mode of sonar emission waveform and fiber optic sensing to reduce the overall power consumption of the system; supports modular expansion design, which facilitates the flexible adjustment of the sensor array size according to detection needs.

[0114] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention description and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An underwater sonar target detection system, characterized in that: include: A fiber optic hydrophone array module, comprising a plurality of fiber optic hydrophone units, each unit comprising a fiber Bragg grating sensor and a sound pressure signal demodulation device, for collecting underwater sound wave signals in real time and converting them into a first electrical signal; A sonar transmitter-receiver array module, comprising a broadband sonar transmitter, a multi-beam receiver, and a signal pre-processing circuit, is used to actively transmit a frequency-modulated continuous wave signal and receive a target reflected echo to generate a second electrical signal; a multimodal signal fusion processing module, comprising a high-speed data acquisition card, an adaptive noise suppression unit, and a time-frequency analysis unit, for synchronously acquiring the first electrical signal and the second electrical signal, performing time-domain-frequency domain joint feature extraction, and performing nonlinear signal fusion; The target feature intelligent recognition module integrates a deep learning neural network to achieve underwater target classification and trajectory prediction based on the fused multi-dimensional feature data through a trained multi-scale convolutional network model; Dynamic calibration and self-test module, with built-in reference sound source and fiber optic phase compensation unit, for real-time calibration of the signal phase difference between the fiber optic hydrophone array and the sonar array, and for performing system fault diagnosis; The spatial arrangement of the fiber optic hydrophone array module and the sonar transmitting-receiving array module satisfies: the spacing between the fiber optic hydrophone units is 1 / 4-1 / 2 of the sonar transmitting wavelength, and the two are three-dimensionally orthogonally distributed; The multimodal signal fusion processing module adopts a cross-sensor correlation function optimization algorithm to generate an enhanced target feature map by maximizing the mutual information entropy of the fiber optic hydrophone and sonar signals.

2. The underwater sonar target detection system according to claim 1, characterized in that: In the fiber optic hydrophone array module, each fiber Bragg grating sensor adopts a tilted grating structure, whose grating period varies gradiently along the fiber axis, with a sensitivity adjustment range of 0.1-10 pm / μPa and an operating frequency band covering 10 Hz-50 kHz.

3. The underwater sonar target detection system according to claim 1, characterized in that: The broadband sonar transmitter of the sonar transmitting-receiving array module adopts a nonlinear frequency modulation waveform, whose instantaneous frequency changes with time according to a hyperbolic tangent function, has a bandwidth dynamic range of 5-200 kHz, and a pulse compression ratio greater than 100:

1.

4. The underwater sonar target detection system according to claim 1, characterized in that: The adaptive noise suppression unit of the multimodal signal fusion processing module includes: an ocean environment noise separation subunit based on empirical mode decomposition, a matched field processing and spatial domain filtering joint noise reduction subunit, and a coherence enhancement subunit for optical fiber acoustic signals and sonar echoes.

5. The underwater sonar target detection system according to claim 1, characterized in that: The deep learning neural network of the target feature intelligent recognition module adopts a dual-branch heterogeneous architecture, including: (a) The first branch is a 3D convolutional network that processes the multi-beam range-azimuth information of the sonar array; (b) The second branch is a graph convolutional network that processes the topological characteristics of the distributed sound pressure field of the fiber optic hydrophone array; (c) The fusion layer dynamically weights the contribution of the two types of features through the attention mechanism and outputs the target type and motion parameters.

6. The underwater sonar target detection system according to claim 1, characterized in that: The reference sound source of the dynamic calibration and self-test module is a tunable piezoelectric ceramic transducer array, whose transmitted signal frequency jumps in a chaotic sequence within the range of 1-100 kHz, and the calibration accuracy reaches 0.05° phase error.

7. The underwater sonar target detection system and method according to claim 1, characterized in that: It also includes a multi-physics field coupling analysis module, which corrects the effects of underwater temperature gradients and salinity changes on the sound wave propagation path in real time by solving the coupled solutions of the Helmholtz wave equation and the Navier-Stokes equation.

8. The underwater sonar target detection system according to claim 1, characterized in that: The collaborative working mode of the sonar transmitting-receiving array module and the fiber optic hydrophone array module includes: in near-field detection at a distance of less than 100 meters, the high sensitivity characteristics of the fiber optic hydrophone array are preferentially adopted; in far-field detection at a distance of ≥100 meters, the active detection mode of the sonar array is activated; and dual-mode data is seamlessly switched and confidence fused through a Bayesian inference framework.

9. The underwater sonar target detection system according to claim 1, characterized in that: The system integrates a multi-target tracking algorithm, including: a track association submodule based on improved particle filtering, a Doppler frequency shift compensation and velocity estimation submodule, and a three-dimensional acoustic imaging and situation visualization submodule.

10. An underwater sonar target detection method using the underwater sonar target detection system according to claim 1, characterized in that: The following steps are involved: S1: Collaborative acquisition of multimodal signals: S11: collecting underwater sound pressure signals through the optical fiber hydrophone array module to generate a first electrical signal; S12: actively transmitting nonlinear frequency-modulated sound waves through the sonar transmitting-receiving array module, and receiving target reflected echoes to generate a second electrical signal; S2: Cross-modal data synchronization and preprocessing: S21: The multimodal signal fusion processing module synchronously collects the first electrical signal and the second electrical signal to eliminate time domain deviation; S22: The dynamic calibration and self-test module calibrates the phase difference between the fiber optic hydrophone and the sonar array in real time to ensure that the data time alignment error is ≤10 ns; S3: Multi-source feature fusion and enhancement: S31: The multimodal signal fusion processing module performs a time-frequency joint analysis on the two signals to extract the sound pressure field distribution characteristics and sonar scattering characteristics; S32: Generates a fusion feature map including target position, distance and acoustic characteristics through a cross-sensor correlation function optimization algorithm; S4: Intelligent target recognition and classification: S41: The target feature intelligent recognition module receives the fused feature map and performs target classification through a deep learning neural network; S42: Output target type, size and motion trajectory prediction results based on multi-scale convolutional network; S5: Dynamic feedback and system self-optimization: S51: The dynamic calibration and self-test module reversely adjusts the sensitivity parameters of the optical fiber hydrophone array according to the recognition result; S52: Real-time optimization of the frequency modulation parameters of the sonar transmission waveform to adaptively match the current underwater environment and target characteristics.

Images