Ship Target Recognition Research Based On Multi Physics Field Features

China has a vast ocean area,and maritime disputes between our motherland and its neighbours occurred constantly,so it is urgent to develop the naval weaponry.This thesis comes from the project of the Weapons Advanced Research Foundation of the General Armament Department.Taking the ship target recognition research as studying contents,taking the improvement of ship target recognition ability as research target,and the thorough research is progressed in this article about ship physical field simulation method,feature extraction and fusion algorithm and classifier design method.The detection and identification of ships are usually based on the ship's physical fields.The commonly used ship physical fields include sound field,magnetic field,water pressure field and so on.They contain a lot of discriminant information.However,the existing methods of ship identification are usually based on a single physical field,which limits the performance of the classifier model.In this work,we proposed a ship recognition method based on combination of multi-physical fields,i.e.,sound field,magnetic field and water pressure field.To the best of the authors' knowledge,this is the first work to recognize ships based on multi-physical fields.This research including the following contents: On the MATLAB platform,three field linkage simulation models of ship sound field,magnetic field and water pressure field are constructed,the effective signal of ship physics field is obtained by simulation to overcome the difficult acquisition of actual ship physics field signal.Because the envelope spectrum of ship radiated noise contains line spectrum features,the envelope spectrum model is adopted to simulate the ship sound field.For the ship magnetic field,this paper chooses the three axis coil magnet model,which is the most commonly used in recent years,to realize the simulation of magnetic field.During the modeling of hydrodynamic field,the subcritical velocity model is utilized,which simulation results are closest to the real signal.The actual ship parameters are obtained from the reference data and used as input of the simulation model,which include 119 naval samples and17 merchant ship samples.A random perturbation is added to each sample parameter,and the total sample number is extended to 680;Because the wavelet transform has good time-frequency analysis characteristics,a signal feature extraction algorithm based on wavelet transform for ship physical field is established to acquire the characteristics of ship category information;Owing to the support vector machine(SVM)has good generalization ability and is suitable for small sample classification problems,a classifier based on SVM is constructed to classify the extracted single field features and the fusion features obtained by various fusion algorithms.The results of classification show that the highest recognition rate is obtained when combining the three field features,which is 92.6%,higher than the highest recognition rate based on a single field of 4.1% and the highest recognition rate based on two physical fields of0.4%.The effectiveness of ship recognition based on the multi-physical field is validated.The successful implementation of this project will provide a new solution for the research of ship target recognition technology,and contribute meager forces to the enhancement of our country's maritime military strength.