基于ShuffleNet的净空保护区飞行违章行为识别算法OA

In order to improve the indentification accuracy of flight violations behavior in the clearance protec-tion zone,a indentification algorithm of flight violation behavior in the clearance protection zone based on Shuf-fleNetV2 is proposed.The improved deep convolution and channel shuffling mechanism of ShuffleNetV2 are used to reduce data computation,SimAM attention mechanism is introduced to dynamically adjust the weight of each neuron,focusing on key and highly recognizable flight violation behavior features while suppressing irrele-vant features,achieving feature extraction optimization.Random vector functional link(RVFL)classifier is used to classify the key features extracted after enhancement,and the classified features is weighted to eliminate rec-ognition errors of flight violation behavior.The experimental results show that the proposed algorithm has an in-dentification accuracy rate of over 90％in distinguishing different types of flying objects and flight violation be-havior,with a floating-point operation frequency of 2.3FLOPS.It also has the ability to indentify flight violation behavior in real time,which helps staff to quickly take intervention measures based on the indentification re-sults.