Research On Underwater Target Active Location And Recognition Technology Based On Electric Field Detection

Chinese waters has rich resources and military strategic status.In the field of national defense,underwater surveillance robots can monitor frogmen and underwater military equipment in real time.In the civil field,autonomous underwater robots can realize underwater ecological environment monitoring.Underwater electric field sensing technology is a technique inspired by weakly electric fish.Compared with the currently widely used visual imaging technology and sonar technology,due to its characteristics of omnidirectional and strong anti-interference ability,it has become a good complementary detection technology in dark and turbid waters.The main contents of this paper are as follows:Firstly,the quasi-static electric fields theory is used to analyze the underwater electromagnetic field,and the potential disturbance caused by the ideal sphere in a uniform electric field on the surrounding environment is studied,and the relevant formulas are derived.The problem of underwater electric field perception is divided into forward and inverse problem.Then ANSYS Maxwell software was used to simulate the factors affecting the surrounding electric field of underwater targets,which verified the feasibility of underwater electric field detect targetsSecondly,electrical impedance tomography(EIT)technology based on the finite element method(FEM)was used to model and simulate the underwater electrical sensing problem.An eel-like electric sensor model was designed and simulated,and the feasibility and effectiveness of underwater target position and recognition based on EIT method are discussed.Considering that EIT method requires Massive computing resources,a discrete dipole approximation(DDA)method is introduced to reduce the computational cost of forward problem.And established the relationship between DDA and a sphere disturbance formula.Through simulation and experiment,in the underwater electric field perception problem,the execution speed of DDA is more than100 times that of FEM,and the average relative error is only 9%Finally,based on the DDA method,the unscented Kalman Filter(UKF)algorithm and the back propagation neural network(BPNN)algorithm are respectively used to estimate the target position of underwater spheres by using probe electrode matrix model.The potential disturbance obtained from the probe electrode matrix is transformed into an electronic image,and the electronic image is identified and classified by the convolutional neural network(CNN).Based on the previous research,a prototype of an underwater positioning and identification system was designed and implemented,including three parts: electric field excitation module,signal acquisition module and data processing module.The localization and recognition algorithms are verified by experiments with 5×5 and 16×16 electrode matrix prototypes.The experimental results show that the prototype location is fast and accurate within a certain range by using BPNN and UKF,with an average distance error about 1.1cm.The overall success rate of CNN in identifying specified cubes,spheres and cylinders is91.5%.