Underwater Noise Target Recognition Based on Sparse Adversarial Co-Training Model with Vertical Line Array

The automatic identification of underwater noncooperative targets without label records remains an arduous task considering the marine noise interference and the shortage of labeled samples.In particular,the data-driven mechanism of deep learning cannot identify false samples,aggravating the difficulty in noncooperative underwater target recognition.A semi-supervised ensemble framework based on vertical line array fusion and the sparse adversarial co-training algorithm is proposed to identify noncooperative targets effectively.The sound field cross-correlation compression(SCC)feature is developed to reduce noise and computational redundancy.Starting from an incomplete dataset,a joint adversarial autoencoder is constructed to extract the sparse features with source depth sensitivity,aiming to discover the unknown underwater targets.The adversarial prediction label is converted to initialize the joint co-forest,whose evaluation function is optimized by introducing adaptive confidence.The experiments prove the strong denoising performance,low mean square error,and high separability of SCC features.Compared with several state-of-the-art approaches,the numerical results illustrate the superiorities of the proposed method due to feature compression,secondary recognition,and decision fusion.

On fast estimation of direction of arrival for underwater acoustic target based on sparse Bayesian learning

The Direction of Arrival （DOA） estimation methods for underwater acoustic target using Temporally Multiple Sparse Bayesian Learning （TMSBL） as the reconstructing algorithm have the disadvantage of slow computing speed. To solve this problem, a fast underwater acoustic target direction of arrival estimation was proposed. Analyzing the model characteristics of block-sparse Bayesian learning framework for DOA estimation, an algorithm was proposed to obtain the value of core hyper-parameter through MacKay＇s fixed-point method to estimate the DOA. By this process, it will spend less time for computation and provide more superior recovery performance than TMSBL algorithm. Simulation results verified the feasibility and effectiveness of the proposed algorithm.

Extraction and application of the low dimensional dynamical component from underwater acoustic target radiating noise

Signal processing in phase space based on nonlinear dynamics theory is a new method for underwater acoustic signal processing. One key problem when analyzing actual acoustic signal in phase space is how to reduce the noise and lower the embedding dimen- sion. In this paper, local-geometric-projection method is applied to obtain fow dimensional element from various target radiating noise and the derived phase portraits show obviously low dimensional attractors. Furthermore, attractor dimension and cross prediction error are used for classification. It concludes that combining these features representing the geometric and dynamical properties respectively shows effects in target classification.

Passive tracking and size estimation of volume target based on acoustic vector intensity

The special sections of volume target are observed with acoustic vector intensity according to the difference among their radiated-noise characteristics, then three sections are tracked with Kalman filtering, and target size is estimated. Simulation results indicate that in ideal condition three sections of a ship can be tracked and ship's size can be estimated even though one of three sections can not be observed.

Local-linear-prediction analysis for underwater acoustic target radiated noise

Local-linear-prediction in phase space is performed for the underwater acoustic target radiated noise. Relation curve of average prediction error versus neighboring points' number is calculated. The result is used in judging the nonlinearity of radiated noise time series, and obtaining the appropriate form and coefficients of predicting model. The line and continuous spectral component are predicted respectively. Choice of some model parameters minimizing the prediction error is also discussed.

The application of threshold empirical mode decomposition de-noising algorithm for battlefield ambient noise

The detection of the low-altitude acoustic target is an important way to compensate for the weakness of radar.Removing the noise mixed in acoustic signal as much as possible to retain the useful information is a challenging task.Inspired by the wavelet threshold,the de-noising method for low-altitude battlefield acoustic signal based on threshold empirical mode decomposition(EMD-T)is proposed in this paper.Firstly,the noisy signal is decomposed by empirical mode decomposition(EMD)to get the intrinsic mode functions(IMFs).Then the IMFs,whose actual energy exceeds its estimated energy,are processed by the EMD threshold.Finally,the processed IMFs are summed to reconstruct the de-noised signal.To evaluate the performance of the proposed method,extensive simulations are performed using helicopter sound corrupted with four types of typical low-altitude ambient noise under different signal-to-noise ratio(SNR)input values.The performance is evaluated in terms of SNR,root mean square error(RMSE)and smoothness index(SI).The simulations results reveal that the proposed method de-noising method has the perspective of the highest SNR,smallest RMSE and SI in de-noising low-altitude ambient noise compared to other methods,including the wavelet transform(WT)and conventional EMD.