Rcs Angle Extrapolation Based On Deep Learning And Target Identification Research

With the rapid development of information,communication technology and computer network technology,the distributed radar sensor systems have attracted wide interests from both academic and industrial areas.In wireless positioning or intelligent driving scenarios,it is required a device to detect targets.In the distributed radar sensor network,we transmit signals to the target through sensors at different positions.Meanwhile,considering that radar measurements of targets often fail to cover the full angle domain,we try to predict the radar cross section(RCS)of the target of all angles.In view of the difficulty and expensive cost of obtaining the omnidirectional angle of the target radar scattering cross section in the actual scene,this thesis proposes to use back propagation neural network(BPNN)and long short time memory network(LSTM)to predict the unknown angle RCS sequence based on the known angle RCS sequence.Through the research of this thesis,it is found that these two algorithms can find out the scattering law of RCS and extrapolate the prediction of RCS within a certain range.At the same time,since LSTM can solve the problem,we select LTSM to fit the RCS sequence of two missing fragments.And the simulation results show that the RCS prediction model and extrapolation are correct and effective.Target detection and recognition has been widely used in military and civil applications.It can distinguish targets from the background interferences,determine whether there are targets in the sensing area,and identify the target objects when they exist.A target recognition model based on RCS observation sequence is proposed and 12 effective features are extracted in this thesis.The recognition results obtained by the support vector machine(SVM)method are compared with those obtained by other common methods,indicating that the support vector machine method has a better recognition performance.This thesis takes distributed radar sensor systems as the research background.Aiming at the problem of incomplete RCS sequence and target recognition,this thesis proposes the prediction of RCS sequence and the model of target recognition.The method proposed solves the difficulties in the application of distributed radar sensor systems in real scene.The simulation results show its feasibility and provide theoretical support for the practical application of the system.