基于循环神经网络模型的无人机音频信号识别算法研究

Research on Drones Audio Signal Recognition Algorithm Based on RNN Model

近年来,无人机的广泛应用引发诸多隐私和安全问题。针对此问题,研究基于循环神经网络(Recurrent Neural Network,RNN)模型的无人机音频信号识别算法。首先,对于无人机的音频信号进行特征值提取,基于设计的RNN模型,设置3层长短期记忆网络(Long Short-Term Memory,LSTM)和3层Dropout。其次,在输出层使用Softmax作为激活函数输出结果,并在实验过程中不断更新神经元和Dropout的参数。最后,基于实际采集的音频数据进行无人机检测。实验结果表明,所提出算法的准确率达到96.33%,能够满足无人机检测的实际需求。

In recent years,the widespread application of drones has caused many privacy and security issues.To address this issue,a drone audio signal recognition algorithm based on Recurrent Neural Network(RNN)model is studied.Firstly,feature value extraction is performed on the audio signal of the drone.Based on the designed RNN model,a 3-layer Long Short-Term Memory(LSTM)network and a 3-layer Dropout are set up.Secondly,in the output layer,Softmax is used as the activation function to output the results,and the parameters of neurons and Dropout are continuously updated during the experiment.Finally,based on the actual audio data collected,perform drone detection.The experimental results show that the accuracy of the proposed algorithm reaches 96.33%,which can meet the practical needs of drone detection.