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-driv...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.展开更多
With the introduction of underwater bionic camouflage covert communication,conventional communication signal recognition methods can no longer meet the needs of current underwater military confrontations.However,the r...With the introduction of underwater bionic camouflage covert communication,conventional communication signal recognition methods can no longer meet the needs of current underwater military confrontations.However,the research on bionic communication signal recognition is still not comprehensive.This paper takes underwater communication signals that mimic dolphin whistles through phase-shifting modulation as the research object,and proposes a recognition method based on a convolutional neural network.A time-frequency contour(TFC)masking filtering method is designed,which uses image technology to obtain the TFC mask of whistles and extracts whistles from the obtained mask.Spatial diversity combining is used to suppress the signal fading in multipath channels.The phase derivative spectrum image is obtained by Hilbert transform and continuous wavelet transform,and is then used as the basis for recognition.Finally,the effectiveness of the proposed method is verified by simulations and lake experiments.In the simulations,a recognition accuracy of 90%is achieved at a signal-to-noise ratio(SNR)of 0 dB in multipath channels.In the real underwater communication environment,a recognition accuracy of 81%is achieved at a symbol width of 50 ms and an SNR of 6.36 dB.展开更多
The physical properties of pattern characteristics for typical Acoustic Sea-bed Profiling Records (ASPRs) in the area of Changjiang Estuary and the East China Sea are analyzed in this paper. Nine pattern characterist...The physical properties of pattern characteristics for typical Acoustic Sea-bed Profiling Records (ASPRs) in the area of Changjiang Estuary and the East China Sea are analyzed in this paper. Nine pattern characteristics are summarized and it was shown that 9 geological categories can be determined by 4 pattern characteristics. Based on the above analysis, a Bayes-based pattern characteristics classifier for interpretation of ASPRs is developed.展开更多
An expert system based on the fuzzy set theory has been developed for geological interpretation of Acoustic Seabed Profiling Records(ASPR). After successively extracting each state of several main pattern characterist...An expert system based on the fuzzy set theory has been developed for geological interpretation of Acoustic Seabed Profiling Records(ASPR). After successively extracting each state of several main pattern characteristics shown on the ASPRs, the similarities between this pattern characteristic-state set and the standard ones corresponding to different geological categories of marine sediments are computed respectively By comparillg these values of sidrilarities, the conclusion of geological classification to the ASPR can be derived.展开更多
A computer-based pattern recognition systems has been developed for geological interpretation of Acoustic Sea-bed Profiling Records. Based on practical experience accumu- lated by specialists, the main pattern charact...A computer-based pattern recognition systems has been developed for geological interpretation of Acoustic Sea-bed Profiling Records. Based on practical experience accumu- lated by specialists, the main pattern characteristics of Acoustic Sea-bed Profiling Records (ASPRs) corresponding to typical geological categories of marine sediment layers in the area of the East China Sea have been expressed altogether in 9 aspects, and a dynamic reasoning expert system designed correspondingly. Starting from an initial premise Characteristic and makes the next step reasoning until the final conclusion (i.e. which geological category the sediment layer belongs to.) is derived, in the mean time, for quantitatively estimating the correctness of the final conclusions, the so-called certainty factor is calculated.展开更多
基金the National Natural Science Foundation of China(No.6210011631)in part by the China Postdoctoral Science Foundation(No.2021M692628)。
文摘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.
基金Project supported by the National Natural Science Foundation of China(No.62231011)the Tianjin Outstanding Young Scientists Fund Project(No.24JCJQJC00240)。
文摘With the introduction of underwater bionic camouflage covert communication,conventional communication signal recognition methods can no longer meet the needs of current underwater military confrontations.However,the research on bionic communication signal recognition is still not comprehensive.This paper takes underwater communication signals that mimic dolphin whistles through phase-shifting modulation as the research object,and proposes a recognition method based on a convolutional neural network.A time-frequency contour(TFC)masking filtering method is designed,which uses image technology to obtain the TFC mask of whistles and extracts whistles from the obtained mask.Spatial diversity combining is used to suppress the signal fading in multipath channels.The phase derivative spectrum image is obtained by Hilbert transform and continuous wavelet transform,and is then used as the basis for recognition.Finally,the effectiveness of the proposed method is verified by simulations and lake experiments.In the simulations,a recognition accuracy of 90%is achieved at a signal-to-noise ratio(SNR)of 0 dB in multipath channels.In the real underwater communication environment,a recognition accuracy of 81%is achieved at a symbol width of 50 ms and an SNR of 6.36 dB.
基金The work was supported by the National 863 plan Youth Foundation (820-Q-09).
文摘The physical properties of pattern characteristics for typical Acoustic Sea-bed Profiling Records (ASPRs) in the area of Changjiang Estuary and the East China Sea are analyzed in this paper. Nine pattern characteristics are summarized and it was shown that 9 geological categories can be determined by 4 pattern characteristics. Based on the above analysis, a Bayes-based pattern characteristics classifier for interpretation of ASPRs is developed.
文摘An expert system based on the fuzzy set theory has been developed for geological interpretation of Acoustic Seabed Profiling Records(ASPR). After successively extracting each state of several main pattern characteristics shown on the ASPRs, the similarities between this pattern characteristic-state set and the standard ones corresponding to different geological categories of marine sediments are computed respectively By comparillg these values of sidrilarities, the conclusion of geological classification to the ASPR can be derived.
基金the National 863 Plan Youth Foundation of China !(820-Q-09).
文摘A computer-based pattern recognition systems has been developed for geological interpretation of Acoustic Sea-bed Profiling Records. Based on practical experience accumu- lated by specialists, the main pattern characteristics of Acoustic Sea-bed Profiling Records (ASPRs) corresponding to typical geological categories of marine sediment layers in the area of the East China Sea have been expressed altogether in 9 aspects, and a dynamic reasoning expert system designed correspondingly. Starting from an initial premise Characteristic and makes the next step reasoning until the final conclusion (i.e. which geological category the sediment layer belongs to.) is derived, in the mean time, for quantitatively estimating the correctness of the final conclusions, the so-called certainty factor is calculated.
