Rotary-Wing UAV Target Recognition Method Based on Radar Measured Data

Rotary-wing unmanned aerial vehicle (UAV) technology has been applied in many fields with rapid development in recent years. Based on the theoretical and practical analysis of the micro-Doppler characteristic of the UAV, a radar recognition method for UAV targets is proposed in this paper. After range compression and moving target detection (MTD) processing, the Doppler spread feature can be obtained from the range-Doppler (RD) plane of the target. Then the lightweight deep neural network MobileNetV3 which is more suitable for the embedded hardware of radar is utilized to identify the UAV target by using the two-dimensional data segment cut from the range-Doppler plane. Compared with only using one-dimensional Doppler information, using range-Doppler two-dimensional information can better avoid the interference of adjacent targets to UAV target features in the actual radar detection environment. Therefore, the micro-Doppler signal can be better distinguished from non-UAV targets. Experimental results based on actual radar data show that the accuracy of the proposed recognition method can reach 90%.

I. Introduction

With the rapid development of unmanned aerial vehicle (UAV) technology, multi-rotor UA Vs have a wide range of applications in military and civilian fields. To enhance the standardization of the use of rotary-wing UAVs and effectively monitor the rotary-wing UA Vs in the no-fly zone, rotary-wing UAVs should be accurately detected and identified. Radar detection has the characteristics of all-day, all-weather, and long-distance, and is an effective tool to monitor UAV targets. Therefore, how to use radar echo data to accurately identify UAVs is a subject with important research value.