While evaluating Ship-radiated noise with operational transfer path analysis (OTPA), a question arises that the input of OTPA cannot meet the linear system requirement, which results in a large estimated error in several frequencies. Therefore, to reduce the estimated error, in this paper a method based on OTPA optimized by operation clustering is proposed. First, a complete clustering feature matrix is constructed and weighted principal component analysis (W-PCA) used for feature extraction. Then, several classes of operating mode(OM) are clustered by K-means based on the feature, and the classified OMs are taken as the input of OTPA, which can meet the linear system requirement of OTPA. Experiments with cabin in Thousand Islet Lake show that the statistical probability of the line spectrum with an estimated error less than 3 dB is more than 95%, which is an improvement of 16.7% compared to the error before optimization, and the maximum estimated error of the line spectrum is reduced by 11.3 dB. The estimated error of 1/3 Octave is less than 2 dB in all frequencies, which can be reduced by approximately 7 dB. The results verify the feasibility of OTPA optimized by operation clustering.
Most of the methods for extracting underwater acoustic beaconing signal are currently suitable for a good hydrological environment. Aiming at the problem of longrange detection of acoustic beaconing signal in harsh environments such as platform movement and serious transient noise interference, a method combined "matched filtering with Doppler filter bank, adaptive line enhancement based on variable step frequency-domain least mean square (LMS) algorithm and Zoom-FFT" is proposed. Verification test of the proposed method is carried out on the lake. The results of data processing show that this method could reduce the Doppler sensitivity and suppress the transient interference. Meanwhile, the output signal-to-noise ratio (SNR) and frequency resolution are improved, which provides an excellent signal environment for the back-end acoustic beacon positioning.
Outputting the acoustic field pressure and particle velocity at the same time and at the same point, a single hydrophone can be used to estimate the direction of arrival (DOA) of the target. It has received extensive attention internationally. The DOA estimation methods of single vector hydrophone can be divided into two categories. One is the maximum likelihood estimation method based on acoustic energy flow, representative algorithms including acoustic intensity averager and histogram method; the other is based on the characteristics of the vector hydrophone array flow, and apply high-resolution algorithms commonly used in array signal processing to single vector hydrophones. These high- resolution algorithms include Minimum Variance Distortionless Response (MVDR), Multiple Signal Classfication (MUSIC), and Estimating Signal Parameter via Rotational Invariance Techniques (ESPRIT). The resolution of the MVDR is not as good as the MUSIC and the ESPRIT, but the quantity of computation is much smaller than the latter. Aiming at the problem of wide output spatial energy spectrum peak and poor spatial resolution of the MVDR, this paper proposes a highresolution DOA estimation method for single-vector hydrophones based on the theory of the optimal preselected filter--ECKART filter. This method can significantly improve the spatial resolution and output signal-to-noise ratio (SNR) of the MVDR algorithm based on single-vector hydrophone.
In order to effectively separate multiple target signals and reduce noise interference to underwater target signals, the frequency-wavenumber(f-k) filtering method commonly used in the seismic data processing was applied to the separation of underwater multi-target signals and the denoising processing of underwater acoustic data. By analyzing the f-k spectrum of multi-target underwater acoustic data, the parameters of the sector-filter were designed, then f-k filtering of multi-target underwater acoustic data were carried out. The filtering processing results show that the f-k filtering method can separate the multi-target signals that are aliased in the time domain and the frequency domain. And the f-k filtering method also has a certain inhibitory effect on noise, which can effectively improve the signal-to-noise ratio.
In the field of underwater acoustic signal processing, broadband array signal processing is an important research issue, which is more complex than that of narrowband. Broadband passive localization of underwater acoustic sources is investigated in this study. In order to improve the ability to locate multiple radiation sources which are emitted from ships, the orthogonal propagator method (OPM) is proposed to estimate direction of arrival (DOA) for the target. Based on the theory of spatial spectrum estimation, the linear decomposition of the covariance matrix is adopted to form the orthogonal propagator, and the spatial spectrum is constructed according to orthogonality of subspaces. Then, the directions of ship radiation noises are determined by searching the extreme points of the spatial spectrum. For broadband underwater acoustic signals, the non-coherent subspace method is used to average the effective frequency points in bandwidth to reduce noise interference, and OPM algorithm operating in broadband underwater acoustic signal, on arbitrary sonar array, is derived. In order to verify this method, numerical simulations are conducted to test its performance in localization accuracy, beam width and spatial resolution. During the simulations, a uniform linear array consisting of 96 elements is designed. Additionally, comparisons between the proposed method and conventional beamforming technique are conducted. Comparison results show that the OPM method performs better in spatial directivity and spatial resolution.
The underwater bearing only target tracking is a typical nonlinear filtering problem where a moving or stationary observer monitors the sonar bearings received from an acoustic target as measurements to estimate its position and velocity. However, the target state might not be fully observable because the observing station does not have exact information about the target range. The unavailability of range parameter makes the observability a crucial problem for the bearing only target tracking system. Therefore, the requirement to identify the observability conditions plays a dynamic role in defining the observer trajectories and to attain unique tracking solutions. We can adopt different approaches such as using two or more stationary observing stations or single maneuvering observer to resolve the range observability problem. In this paper, the target is assumed to be moving with constant speed and heading whereas the observing station is non-maneuvering which is the most suitable condition in typical scenarios. Also, the observability conditions are briefly discussed in the case of single and multiple observing stations to achieve unique tracking solutions for target states. The popular estimators such as Extended kalman filter (EKF) and Unscented kalman filter (UKF) are used to estimate the target states in this research. Simulations are performed to see the results of underwater bearing only tracking for two observing stations. The simulation results show substantial effect on the observability and overall tracking performance by changing the distance and initial conditions for two observing stations. The effect of changing both the position and distance has also been analyzed between the two observers. Simulation results have shown better convergence and stability for two observers than single observing station in different scenarios.
As a serviceable tool of underwater targets classification for sonar operators, deep neural network behaves a good work on underwater targets intelligent classification. Since the line frequencies of radiated noise supplied distinct frequency bands for different ships, a deep neural network is proposed based on an attention mechanism to improve the classification accuracy in this paper. The results show that the equilibrium classification accuracy of ACNN-QJ4 is 9.15% higher than that of non-attention network. Finally, by comparing the output features of these two networks with and not with attention mechanism, the superiority on feature extracting of the attention network proposed by this paper has been shown.
Source depth discrimination is important for anti-submarine warfare and marine biology. Given the difference in the vertical correlation of acoustic fields excited by sources at varying depths, a physics-based quantity characterizing the vertical correlation of acoustic field is proposed for passive surface/submerged source classification in a shallow water range-independent waveguide. The eigenvalue decline index of correlation matrix is the ratio of the maximum eigenvalue to the second largest eigenvalue of the receiving signal correlation matrix, which is calculated to characterize the oscillation properties of the correlation coefficient matrix. The eigenvalue decline index is then analyzed and discussed under the simulation condition of the typical isovelocity sound speed profile in shallow water. This decline index presents obvious separability so long as the receiving vertical line array is designed properly. The condition for the vertical correlation characteristics of the acoustic field is that the spacing of the array elements should be no more than one-third of the wavelength. Using the eigenvalue decline index may provide a robust alternative method for binary source depth classification.
In-band full-duplex Underwater acoustic communication (IBFD-UWAC) can significantly improve the utilization of the communication frequency band. Because the local high power interference signal will affect the reception of the expected signal, the self-interference cancellation (SIC) becomes the core of IBFD-UWAC. This paper proposes a real-time SIC testing method based on Simulink Desktop Real-Time ® . This paper uses least mean squares and recursive least squares adaptive algorithms to achieve self-interference cancellation. Hardware-in-loop simulation results show that both algorithms can achieve 55dB real-time SIC under time-varying UWA channels. Compared with the traditional method, the algorithm proposed in this paper is performed in real-time, which dramatically increases the efficiency of communication between the two sides. This method is suitable for IBFD-UWAC.
Multiple-input multiple-output (MIMO) sonar can provide higher spatial resolution, better reverberation mitigation ability, improved processing gain compared to phased-array sonar by waveform diversity. High degree of freedom makes fully space-time adaptive processing impractical in MIMO sonar for reasons of the high demand of secondary data for filter designing. To overcome the problem, a novel space-time adaptive processing method via iterative generalized sidelobe canceller (IGSC) is proposed. IGSC can achieve a fast and accurate estimation of space-time covariance matrix (STCM) using a few snapshots and realize near-optimal reverberation mitigation performance. Numerical experiments validate the effectiveness of IGSC.
Noise and abnormal sound are common in daily life and industrial production, whose existence usually is accompanied by malfunction, so noise source localization has become a hot issue. Self-power spectrum removal focused beamforming is a beamforming sound source localization method based on microphone array. But its massive calculation restricts its application in practice. In order to solve this problem, this paper implements the focused self-power spectrum removal beamforming algorithm based on the Computing Unified Device Architecture (CUDA), and designs experiments to verify the accuracy and operating efficiency of the algorithm. By comparing the sound map generated by the CPU-Based algorithm and the algorithm implemented in this paper, the accuracy of our algorithm is verified. By comparing the processing time of focused beamforming with different signal samples and the scanning surface resolution, the results show that the computation time of our algorithm is 2 orders of magnitude lower than the CPU-based algorithm spent, and the average GPU/CPU acceleration ratio can reach more than 300.
How to extract effective target features from complex underwater acoustic signals and better classify underwater targets and ships has always been an important problem in the field of underwater acoustic countermeasures. Due to the complexity of underwater acoustic environment and continuous development of underwater acoustic countermeasures, the limitation of expert experience system based on traditional spectrum analysis technology is becoming more and more obvious in passive sonar target recognition. In recent years, deep learning has made remarkable progress in image recognition, speech recognition and other fields. In this paper, a new method is developed in which one-dimensional time series data is used as the input, and some popular networks such as Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) are used as deep learning models to mine intrinsic features, and Wasserstein Generative Adversarial Network-Gradient Penalty (WGAN-GP) is used to enhance the training samples data. This method has shown an excellent performance in passive sonar recognition.
This paper proposes an improved multi-base sonar positioning algorithm with Constrained Total Least Squares (CTLS) method, based on the Total Least Squares (TLS) algorithm applying to multi-base sonar system location, which aims to solve the problem that the positioning accuracy is seriously affected by the measurement error. This method is achieved by adding constraints to the TLS. To ensure that the form of system matrix and data vector of the localization equation remain unchanged after the constraints are added, and the CTLS algorithm is implemented by the method of continuous projection. The simulation results show that, based on the working mode of multi-base sonar with one transmitter and multiple receivers, compared with the TLS algorithm, the CTLS algorithm used in this paper can reduce the localization variance by about 12% in the case of errors in both the base station location and the measured values. The CTLS estimation method has superior estimation performance and robustness, and it has important application value for the efficiency and accuracy of target localization.
The detecting performance of active sonar in shallow sea is restricted by seabed reverberation and clutter. In this paper, the seabed reverberation and target-like clutter caused by inhomogeneous water columns are modelled by coupled mode reverberation theory. Taking nonlinear internal waves (NIWs) as examples of the common inhomogeneous water columns, mode coupling matrices are derived by coupled mode to describe the process of energy exchange and mechanism of reverberation clutter. It is found that the oscillation phenomena of the reverberation intensity which is after the clutter arrival are well explained by the coupled mode theory and the oscillation periods are connected with the speeds of nonlinear internal waves. A theoretical basis is proposed in this research for identifying strong interferences of clutter caused by NIWs. Meanwhile, the oscillation mechanism is expected to realize the monitoring of internal waves.
Traditional methods for the source depth discrimination encounter the problem of performance degradation when the surface or internal wave motions cause variations of the source depth. While the modal scintillation index (SI) can classify surface/submerged source by using the fluctuations in the received pressure field. However, calculating the SI requires normal mode functions in advance, which is difficult to obtain accurately in practice (sound speed profile or long enough aperture desired). To address this problem, we propose a method to utilize modal correlation scintillation index to separate surface and submerged sources on a single receiver requiring only knowledge of the water depth. Specifically, we extract the autocorrelations and the cross-correlations of normal modes from the received signal autocorrelation function via warping transform. The variance in the estimated magnitude of the modal autocorrelations and the cross-correlations normalized by the squares of their expected value over some observation intervals are defined as modal autocorrelation scintillation index (MACSI) and modal cross-correlation scintillation index (MCCSI), respectively. The derivation and the simulation results provide that a source near the surface (all products of mode functions sharing a common zero-crossing) will exhibit scintillation indices with large values, while with small values for a submerged source near at least one product extremum. The method can be used to determine the depth of the source with unknown sound speed profile details or sound source range, and no need for the source movement.
Aiming at the problems of grayscale distortion, low contrast, and unobtrusive targets of side-scan sonar images, this paper uses five image enhancement methods for comparative experimental analysis. Analyzed five image enhancement algorithms: histogram equalization (HE), grayscale stretching (GS), mean filter (MF), Retinex (MSR) and wavelet transform (WT). Experiments were performed on three side scan sonar images. Experiments show that the three methods HE, GS, and MSR can effectively improve the image quality to a certain extent. The image enhancement effect of the MF method is not obvious, and the effect of the WT method is poor.
Based on time domain, frequency-domain and time-frequency domain feature extraction algorithm proposes a composite feature vector generation method, combined with support vector machine (SVM) to classify three kinds of typical underwater target radiated noise identification, the results show that, based on the characteristics of composite target recognition probability is 94.29%, were higher than the recognition probability of independent features, to verify the effectiveness of this method in classification of underwater acoustic target recognition.
The dispersion of normal modes in shallow water leads to interference striations with certain slopes on intensity spectrogram of the range-frequency domain, which also causes the performance of conventional beamforming (CBF) to degrade in both array gain and output fidelity. The striation-based beamforming (SBF) has been proved to be one effective technique to improve these problems for pulse signals but the phase shift related to the waveguide invariant can deflect its main-lobe to a position that is not corresponding to the source azimuth. In this paper, the phase delay introduced by waveform truncation is analyzed. It is shown that the phase shift related to the waveguide invariant can be compensated when the header truncation point (HTP) is exactly set at the arrival point of the dominating modes (APDM) from the source to the reference array element. Simulation and experimental results show that the arrival point of signal peak (APSP) could be one effective estimation of the APDM. SBF is applied successfully on such specially truncated signals for enhanced passive azimuth estimation without the knowledge of waveguide invariant or the source range.
In this paper, we propose an iterative hybrid equalization algorithm, where our main aim is to provide a better solution towards the problem of high computational complexity during the transmission of Orthogonal Signal Division Multiplexing (OSDM) over underwater acoustics (UWA) channel. Our projected hybrid algorithm is a combination of block and least square QR (LSQR) equalization algorithms. Existing block equalization algorithm followed the iterative matrix inversion in doubly-selective channels, without the consideration of severe impairments in UWA channel. Thus, we added LSQR equalization along with this algorithm and offered a hybrid algorithm. LSQR exhibits compensation of UWA channel matrix inversions, by halting the process of iterations and thus, results in low computational complexity. Hence, by the addition of LSQR, our proposed algorithm also exhibits above mentioned advantages. Simulation results of our recommended hybrid method validate its significance over existing block and LSQR equalization techniques.
In the acquisition of vessel radiated noise characteristics based on acoustic array, element failure or acoustic propagation or other reasons will cause the random errors in amplitude and phase of each channel signal, and lead to vessel radiated noise power spectrum acquired by beam-forming. In order to modify this error, based on mathematical model, the difference between mean power spectrum of all channel signals and power spectrum of beam-forming with random errors in amplitude and phase was analyzed, and a correction method for power spectrum amplitude was established. The simulation results show that the effect of errors in phase is much greater than the errors in amplitude, and the correction method in this paper can effectively reduce the beam-forming power spectrum amplitude error. And the analysis results of measured data show that, taking the mean power spectrum of all channel signals as the reference, the error before correction was 3dB~5dB, and the corrected error was 1dB~3dB, meeting the error range requirements.
