Technology Research Of Machine Learning Method In Through-the-wall Radar Imaging

Through-the-wall radar imaging,also known as electromagnetic inverse scattering problem,is to explore the physical characteristics of the target behind the wall or obstacles,including the location,size,shape,number and even electromagnetic parameters of the target.This technology is of a great practical significance to applied science.In recent years,with the rapid development of computer technology and the arrival of machine learning research upsurge,it is a very new and exploratory research field to apply machine learning to through wall radar.In this paper,the main research contents include:1.This paper studies the basic theories of three machine learning methods: support vector machine(SVM),least squares support vector machine(LS-SVM)and extreme learning machine(ELM),and deduces the related training algorithms in detail.2.Aiming at the lack of application of machine learning in through wall radar imaging,this paper studies at how to use SVM to solve the problems encountered in through wall radar imaging.This paper proposes a method to predict the wall parameters(relative permittivity,thickness and conductivity)by SVM.Two different support vector machines(SVM;LS-SVM)are used to verify the feasibility of this method and its high prediction accuracy.Considering the practical application,the simulation also verifies the prediction results when there is interference scattering signal target behind the wall.The results show that the dielectric constant and thickness of the wall can also be predicted well in this case.3.Using machine learning to locate the target behind the wall can effectively predict the coordinate position of the target in three-dimensional environment.At the same time,a hyperbolic characteristic of scattering target echo signal is proposed to solve the problem of target echo matching when there are multiple targets behind the wall,which facilitates the feature extraction of a scattering target echo signal.The simulation results verify the feasibility of the scheme.4.On the premise that SVM can be used to solve the problem of target location in through wall radar,another target location method by using extreme learning machine was tried,which is compared with the method of SVM in location quality and efficiency.The research results show that the method can ensure the location quality and have a high positioning efficiency,which is conducive to real-time positioning.At the same time,the location results under different target radius and different noise interference are compared,and the feasibility of this method is verified by experiments.