Research On Target Recognition And Parameter Estimation Technology Of Rotor-wing UAV

In recent years,small unmanned aerial vehicles have been widely used in production and life due to their outstanding maneuverability,low price,easy control,and undetectable characteristics.Potential safety hazards also pose a huge threat to air safety control.Therefore,the prevention of unmanned aerial vehicles has become an urgent problem to be solved.In recent years,people have conducted extensive research on the detection and identification of UAVs by radar systems.However,due to the low radar cross section(RCS)of small UAVs,they have a lower flying height and slower speed than traditional aircraft The "low,small,and slow target" is difficult to detect from the complex ground clutter background using traditional methods.The UAV micromotion feature contains the unique motion information of the UAV.In this paper,by combining the UAV micromotion feature,time domain and frequency domain features,the research is focused on the target recognition and parameter estimation of the rotor UAV,mainly The research content is as follows:(1)Firstly,on the basis of analyzing the micro-Doppler of rigid body rotation,the mathematical model of the rotor blade echo of the rotor UAV is introduced,and on this basis,the echo signal of the rotor radar of the multi-rotor UAV is derived.Next,the influence of each parameter on the time domain,frequency domain and frequency characteristics of the rotor UAV is analyzed.And study the applicability and performance of different time-frequency algorithms to describe the rotor UAV.(2)In order to solve the problem of classification and parameter estimation of rotor drones,Radon transform algorithm and image edge detection algorithm are introduced.The Radon transform can quickly detect straight lines and edges,and has strong anti-noise ability.It can perform Radon transform on time-frequency signals.At the same time,the PCA algorithm is used to reduce dimensionality and denoise the time-frequency data and Radon transform data,and the number of UAV rotors and blades are classified by the K-nearest neighbor classification algorithm.Then,based on the autocorrelation function,the rotor rotation frequency is estimated.Finally,the edge detection algorithm is used to estimate the micro-Doppler frequency broadening value caused by the rotation of the rotor UAV,and the length of the rotor blade is estimated.Classification and parameter estimation.(3)Aiming at the shortcomings of the feature extraction and parameter estimation ofthe rotor UAV under the condition of low signal-to-noise ratio,a classification and parameter estimation algorithm of the rotor UAV based on the combination of BP and CNN is proposed,and the convolutional neural network is used to extract the time-frequency Signal and Radon transform signal characteristics,use BP neural network to extract time domain signal and frequency domain signal characteristics.Then,through the fully connected layer,the feature extraction results of the BP neural network and the convolutional neural network are fused,and finally the rotor UAV target recognition and parameter estimation are realized.In feature extraction,time domain information,frequency domain information,time frequency information and Radon transform domain information are retained,and an improved neural network is adopted.Therefore,this method improves the classification accuracy of rotor UAV under low signal to noise ratio.Parameter estimation accuracy...