The invention discloses a four-rotor unmanned aerial vehicle attack detection method based on a neural network. The method comprises the following steps: firstly, acquiring operation data of the quad-rotor unmanned aerial vehicle under normal and hidden attacks; secondly, training a feed-forward neural network model with a specific number of layers by using normal data, and detecting hidden attacks by using the model to obtain unmanned aerial vehicle channel information which is not influenced by the attacks; then, aiming at the unmanned aerial vehicle channels which are not influenced by attacks, training an auto-encoder model by using normal data of the unmanned aerial vehicle, and detecting the channels by using an auto-encoder; and finally, fusing the first-stage feedforward neural network detector and the second-stage auto-encoder detector to obtain a final detection result. According to the method, the detection of the hidden attack of the quad-rotor unmanned aerial vehicle is realized. Compared with a traditional attack detection method based on a model, the hidden attack recognition capability can be improved, and a new attack mode can be detected online.
本发明公开了一种基于神经网络的四旋翼无人机攻击检测方法。首先,获取四旋翼无人机正常和隐蔽攻击下的运行数据。其次,利用正常数据训练特定层数的前馈神经网络模型,利用该模型检测隐蔽攻击,得到不受攻击影响的无人机通道信息。然后,针对上述不受攻击影响的无人机通道,使用无人机正常数据训练自编码器模型,利用自编码器检测这些通道。最后融合一级前馈神经网络检测器、二级自编码器检测器得出最终检测结果。本方法实现了四旋翼无人机隐蔽攻击的检测。本发明相对基于模型的传统攻击检测方法,不仅能提高隐蔽攻击的识别能力,也能在线检测新的攻击模式。
Four-rotor unmanned aerial vehicle attack detection method based on neural network
一种基于神经网络的四旋翼无人机攻击检测方法
02.12.2022
Patent
Elektronische Ressource
Chinesisch
Europäisches Patentamt | 2023
|Four-rotor unmanned aerial vehicle for environment detection
Europäisches Patentamt | 2021
|Cylindrical launched reconnaissance attack rotor unmanned aerial vehicle
Europäisches Patentamt | 2021
|