Research On Enhancement And Feature Recognition Technology Of Radiated Noise Signal Of Underwater Acoustic Target

The radiated noise signal of underwater acoustic targets contains various mechanical and navi-gational characteristics of ship targets.The low-frequency narrow-band spectrum signal of the un-derwater acoustic target radiated noise signal is easy to capture because of its stable characteristics,high power,and long propagation distance.It is an important feature for identifying and detecting ship targets.With the development of ship noise reduction technology,the noise power radiated by underwater acoustic targets has been greatly suppressed.This makes the underwater acoustic target radiated noise signal detected by the passive sonar array weak.The narrow-band spectrum character-istic is unstable and is mixed with a lot of non-target interference,which brings great difficulties to the identification and detection of the underwater acoustic target radiated noise.At present,large-aperture towed linear arrays are widely used,which greatly improves the ability of passive sonar arrays to de-tect and identify long-distance weak targets.However,the large-aperture towed linear array sonar system faces two outstanding problems in the actual application process: array distortion and strong multi-target interference.Therefore,based on the enhancement of the narrow-band spectrum feature of the underwater acoustic target radiated noise signal under the distortion array and the research of the narrow-band spectrum feature identification technology in the multi-target strong interference en-vironment has important theoretical and practical significance.Aiming at two outstanding problems in the application of large-aperture towed arrays,this paper has carried out a study on the enhance-ment and identification of the narrow-band spectrum characteristics of the underwater acoustic target radiated noise of the towed sonar array.The main contents and innovations are as follows:1.The enhancement method of underwater acoustic target radiated noise signal under the condition of distortion array is studied.Aiming at the problem of the distortion of the large-aperture towed linear array,this paper proposes a data-driven underwater acoustic target radiated noise signal enhancement method under the condition of the distortion array.This method does not need to set a reference source,and only uses the time delay difference information of multiple narrow-band spectrum signals in the underwater acoustic target radiated noise signal to achieve the enhancement of the underwater acoustic target radiated noise signal.In addition,this method uses the weighted Kalman smoothing algorithm proposed in this paper to estimate the array time delay in the target direction,which can suppress the outlier interference in the observation data to a large extent,and effectively improve the signal enhancement performance.In this paper,the method is verified by simulation experiment and performance analysis.In the lake experiment,the narrow-band spectrum characteristic of the radiated noise signal of the underwater acoustic target obtains a gain of more than 15 dB,which fully verifies the effectiveness of the method.2.The enhancement method of narrow-band spectrum feature in underwater acoustic target ra-diated noise signal based on deep learning is studied.In order to further enhance the narrow-band spectrum characteristics,this paper proposes a fully convolutional deep network for the enhancement of the narrow-band spectrum characteristics in the underwater acoustic target radiated noise signal.This method uses the gradually changing characteristics of the radiated noise narrow-band spectrum signal of the underwater acoustic target and the feature learning ability of the fully convolutional deep network to build a narrow-band spectrum feature enhancement system of the radiated noise signal of the underwater acoustic target.The enhancement of the narrow-band spectrum characteristics of the radiated noise signal of the underwater acoustic target is achieved.The simulation data set verifies the effectiveness of the method in enhancing the narrow-band spectrum characteristics of the radiated noise of underwater acoustic targets.3.The identification method of underwater target acoustic radiation noise narrow-band spectrum in the presence of strong narrow-band spectrum interference is studied.Aiming at the problem of non-target narrow-band spectrum interference in the narrow-band spectrum characteristics of the underwa-ter acoustic target radiated noise signal,this paper firstly proposes a narrow-band spectrum clustering identification method based on principal component analysis.This method uses the time delay char-acteristics of the narrow-band spectrum frequency and passes the principal component analysis and DBSCAN density clustering algorithm effectively identify the narrowband line spectrum.This paper further proposes a narrow-band spectrum identification method based on the Capon spectrum of the narrow-band spectrum.This method makes full use of time and space information of the narrow-band spectrum to effectively identify the narrow-band spectrum.Simulation and experimental analysis verify the effectiveness of these two methods.When SNR is 0dB,the narrow-band spectrum identifi-cation accuracy of these two methods can reach 100%.In the measured data with strong co-frequency interference,the stable narrow-band spectrum can also be identified more accurately.The narrow-band spectrum identification method based on the Capon spectrum has an accuracy rate of 90% for narrow-band spectrum identification.