Research On Key Technologies Of Underwater Near-Field Object Localization Based On The Sensing Principle Of Fish Lateral Line

The diversity of missions,the complexity of service environment,etc.pose demands and challenges on near-field object detection and localization for unmanned underwater vehicles(UUVs).Currently,these demands and challenges are difficult to cope with through conventional sonar and optical method.With the lateral line organ,fish is capable of obtaining information about the surrounding flow field precisely and achieves accurate perception of surrounding objects,which provides a new solution for UUV's near-field object detection and localization.Sponsored by the National Natural Science Foundation of China,a novel near-field object localization approach is explored based on the sensing principle of lateral line.Firstly,the near-field object perception mechanism of lateral line is analyzed systematically.Then,the optimization design method for artificial lateral line sensor arrays and the near-field object localization method are studied.The main research contents and conclusions are as follows:(1)Modeling and analysis of the lateral line sensing principleTo solve the problem of how the sensing principle of lateral line is applied to near-field object localization,the near-field object perception mechanism of lateral line is analyzed systematically.Firstly,the basic structure and the flow sensing principle of the fish lateral line are analyzed.The amplitude-frequency response characteristics of the lateral line are analyzed based on the established dynamic model.Then,the equivalent object model is established based on the potential flow theory.The typical characteristic of a general object is represented by a dipole source.Finally,based on the distribution of the flow field generated by a dipole source,the characteristics of the flow field where the lateral line works are analyzed.The effective sensing range of the lateral line is made clear.The analysis results show that(a)the lateral line senses the velocity and pressure variation of the surrounding flow field;(b)the range of sensitive frequency is between several and several tens of hertz;(c)the object that the lateral line perceives has the primary component of a dipole source;(d)the lateral line sensing principle is hopeful of filling the gap left by sonar in near filed.(2)Research on the optimization design method for artificial lateral line sensor arraysTo solve the optimization design problem for artificial lateral line sensor arrays,the Cramer-Rao lower bound based localization performance evaluation method is proposed by employing the excitation pattern mismatch and noise information.The quantitative relationship between the array layout and the localization performance is established.How the length of a uniform linear array and the density of the array elements influence the effective localization area is analyzed.Aiming at designing a simplest array consisting of the least number of sensors,the two-layer model based optimization method for designing a uniform linear array and the multi-objective model based optimization method for designing two typical planar arrays are proposed.Under the condition of meeting the localization performance requirements,the designed artificial lateral line sensor arrays have the minimum number of sensors and the optimal sensor-to-sensor spacing.The simulation results show that the optimized arrays achieve the expected localzation performance.The effectiveness of the proposed optimization design methods is verified.(3)Research on the underwater near-field object localization method based on spatial spectrum estimationTo solve the object localization problem in underwater near pressure field,two dipole source localization methods are proposed based on the MVDR(Minimum Variance Distortionless Response)spatial spectrum estimation and the MUSIC(MUltiple SIgnal Classification)spatial spectrum estimation.The localization accuracy and resolution of these two methods are analyzed from the perspectives of noise and excitation pattern mismatch.Compared with the MVDR method,the MUSIC method has equivalent localization accuracy and better localization resolution.It is proved that the object moving in a general form has the same spatial spectrum as the dipole source,which is verified by the simulated periodic non-harmonic motion and random motion respectively.Namely,no matter how the amplitude and frequency of the moving object change,the previous MVDR method and MUSIC method are applicable.Furthermore,the localization method based on GRNN(general regression neural network)and spatial spectrum estimation is proposed.The localization accuracy degradation caused by the excitation pattern mismatch is suppressed effectively.Based on one-time training,this method is able to localize other dipole sources with different sizes,amplitudes and frequencies.The simulation results show that the proposed method achieves accurate and robust localizations with strong versatility,and is an effective and practical near-field object localization method.(4)Experimental verification of underwater near-field object localization with artificial lateral line sensory arraysTwo typical artificial lateral line sensor arrays,a linear array and a cross array,are developed.An experimental platform is built based on dipole source localization.By taking the real data acquired by the developed arrays as the input of the proposed localization method and comparing the calculated positions with the actual ones,the conclusions drawn in the previous chapters are verified systematically.The experiments of localizing a basic dipole source and dipole sources with different parameters,and the dipole source localization experiments under different array layouts and different array architectures are performed.The effectiveness,robustness and versatility of the proposed spatial spectrum estimation based near-field localization method are validated,as well as the rationality of the optimized artificial lateral line sensor array.Moreover,the implications on the biological study of biological lateral line distribution are discussed based on the experimental results of the linear array and the cross array.Namely,the localization performance differences between these two types of arrays reflect the functional differences between the cephalic lateral line and the trunk lateral line.In conclusion,based on the systematic analysis of the near-field object perception mechanism of lateral line,on the one hand,two optimization methods for designing the simplest artificial lateral line sensory arrays are proposed based on the established quantitative relationship between the array layout and localization performance.Namely,the two-layer model based optimization method for designing a uniform linear array and the multi-objective model based optimization method for designing two typical planar arrays.A uniform linear array,a cross array and a matrix array are then designed.On the other hand,an underwater near-field object localization method is proposed based on spatial spectrum estimation from the perspective of array signal processing.Accurate,robust and versatile near-field localizations are achieved.Finally,the effectivenesses of the proposed methods are verified by experiments.A new technological approach is provided for UUV's near-field object localization in complex environment.This research possesses both theoretical and practical significances in academic study and engineering respectively.