Design Of Narrow-band Radar Target Recognition Air Target Classification System Based On Micro-Doppler

With the development of intelligent information technology,the research of air target recognition technology for narrow-band radar has attracted wide attention.It is more and more necessary for narrow-band air defense radar to automatically classify target attributes while discovering targets.In recent years,fretting characteristics have attracted wide attention in radar target detection and recognition.Fretting refers to the vibration or rotation of radar target besides centroid translation.The Doppler frequency generated by fretting is micro-Doppler frequency.Different Micro-motions will produce different micro-Doppler.Micro-Doppler effect can reflect the geometric composition and motion characteristics of target components,which is the essential feature of the target.Based on the difference of target fretting,features with good classification performance can be extracted to achieve target classification and recognition.However,there may be clutter interference in the actual air defense radar,and the strong clutter energy will affect the extraction of fretting characteristics.Therefore,this paper mainly studies the condition of clutter interference,achieves the separation of target micro-Doppler signals,and designs a narrowband radar classification system based on micro-Doppler characteristics.The main contents are as follows: 1.The micro-Doppler modulation model of the rotor is studied,and on this basis,the micro-Doppler modulation model of three common targets in the air is established.The influence of the structural parameters of the rotor on the time-domain and frequency-domain characteristics of the three models is simulated and analyzed.The obvious differences between the three models provide a theoretical basis for the classification of aircraft targets;2.The possible existence of the actual air defense radar.In this paper,the statistical characteristics of clutter are analyzed and modeled,and the clutter suppression pretreatment of radar echo signal is carried out.Aiming at the problems of excessive energy loss,non-linear amplification of signal characteristics and mode aliasing in traditional clutter suppression algorithms,a modified variable mode decomposition algorithm can remove clutter and preserve the effective features of target signal as much as possible.3.There are few conditions to consider clutter interference in the existing narrowband radar air target recognition system.Based on the interference environment that may exist in the actual recognition process,this paper proposes a method of time-frequency image feature recognition based on micro-Doppler feature.After processing,CNN-LeNet-5 network is input,Dropout algorithm is introduced to reduce the over-fitting of the network,and simulation under different influence factors is carried out.Compared with traditional artificial feature extraction methods,it reduces the subjective interference,realizes the autonomous learning and recognition of features,and has better robustness and higher recognition rate.