MULTI-ROTOR UAV CLASSIFIER

This article discusses a classifier of radar signals reflected from unmanned aerial vehicles(UAVs), based on neural networks. In the proposed classifier, for formation training data, modelof scattering UAVs is used. Recently, demand UAV classification hasbeen quite high due to significant increase in number models and sales these devices. Increasingthe computing power processors development theory networks allows youto create new types classifiers. When using models, it possible set data that isacceptable network. The convolutional network istrained images obtained model fromUAVs. resulting are modeled taking into account orientation angles relativeto normal coordinate system, flight speed, various propeller parameters simulatedUAV. To form addition signal structure, white noise certain configuration isadded, which helps diversity samples improve learning ability theconvolutional use isdue need large with movement parameters,such as height, direction, space, well wide variety configurationsof vehicles: tricopter (three propellers), quadcopter (four hexacopter(six or octocopter (eight propellers). complicates experimental tocreate classifiers this type.