Research On Small Unmanned Aerial Vehicles Based On Small Sample Multi-Classification

With the rapid development of communication network and sending technology,Unmanned Aerial Vehicles(UAVs)have been increasingly applied in many fields,including power lines detection,logistics and distribution,precision agriculture,traffic management so on.However,due to the low price and flexible characteristics of small UAVs,with the increasing popularity of UAVs may also cause serious public safety and privacy problems.The abuse of small UAVs in privacy,protests,crimes and even terrorist attacks pose a big problem.The existing visual and acoustic detection are limited to the recognition of the UAVs and detection distance,it has become a necessity to classify the multi-UAVs from legal ones when they are flying together.In this paper,the features of wireless communication signals and micro-doppler signals of UAVs are used to realize multi-classification detection,which is an urgent problem in the safety of the small UAVs.The low-altitude wireless signal environment of the small UAV is complex,the dynamic feature attribute offset during flight,and the collection of real data samples are difficult,which brings great challenges to the establishment of the identification system and the generation of the training model in real-world scenarios.In this paper,wu introduce the auxiliary classifier wasserstein generative adversarial networks(AC-WGANs)and dynamic attribute-guided augmentation(DAGA)algorithms into small sample multi-classification.The wireless communication signals of small UAVs are collected and the principal component analysis(PCA)is used to reduce the dimensionality.The processed data from UAVs is input to the UAVs discriminant model of the AC-WGANs for classification.The obtained results show the effectiveness of the proposed system,which can achieve a recognition accuracy of around 95%in the indoor environment and can also be suitable in the outdoor environment.What is more,in the absence of wireless signal,the radar-based micro-doppler characteristics detection is used for the classification of the multi-UAVs,Micro-doppler signals of small UAVs rotors are collected,texture features and time-frequency features are extracted,DAGA is proposed to realize dynamic feature recognition of UAVs in operation.