Underwater Dipole Source Localization And Moving Target Characteristics Recognition Based On Bionic Lateral Line

The national strategy of “building a maritime power” puts forward a major demand for marine equipment represented by submersibles,especially unmanned submersibles.Submersible environmental situational awareness is a necessary condition for submarine survival,mission execution and safe operation.With the improvement of the target stealth performance and the increasingly complex submersible service environment,the common main sensing means(sound,vision)are not effective in certain environments(narrow spaces,turbid waters).In order to ensure the successful execution of missions and safe operation of submersibles,it is imperative to develop non-sound sensing technology.Fish has good underwater sensing ability to sense the underwater pressure and other information through the fish lateral line organ,thereby realizing the surrounding flow field environment and target perception.The bionic fish lateral line sensing technology is expected to develop into a new generation of sensing technology.Therefore,this paper draws on the fish lateral line perception principle to carry out underwater sensing technology research,which will be of great significance for improving the submersible sensing ability.The main work of this paper includes:(1)Summarize the basic structure and functional principle of the fish lateral line.Based on the superficial neuromasts and the canal neuromasts,the sensitivity characteristics of the fish lateral line sensing unit such as the frequency response and the time domain response of the neuromast are analyzed.It provides guidance and inspiration for the study of fish lateral line sensing prototype system design and sensing methods.(2)Based on the fish lateral line perception mechanism,we select the high-precision and high-sensitivity underwater pressure sensors to design the two-dimensional pressure sensing array and the bionic fish lateral line sensing prototype system.Focusing on the sensing scenes such as underwater dipole source positioning,integrated with the motion control system,excitation system,etc.,we develop an experimental platform to provide experimental verification conditions for sensing method research.(3)For the underwater dipole source location sensing scene,the positioning method based on generalized regression neural network is studied,and the experimental verification is carried out on the perceptual prototype system and experimental platform.The influences of the perceptual distance,sampling interval and number of sensors used to the positioning accuracy are also studied,preliminarily realizing the two-body-length positioning.(4)For the underwater moving target direction recognition scene,CNN,LSTM,CNN-LSTM and CNN-INCEPTION are used respectively to propose the recognition method based on deep learning,and the underwater direction target recognition experiment is verified.The effects of different neural network structures,sampling direction angular intervals and sampling data length on recognition accuracy are analyzed and compared.