基于CNN-BiLSTM的FMCW雷达生命体征信号检测

FMCW radar vital sign signal detection based on CNN-BiLSTM

针对FMCW雷达生命体征检测时存在呼吸谐波和杂波干扰,无法准确地提取生命体征信号,提出了一种基于CNN-BiLSTM神经网络的FMCW雷达生命体征信号检测算法。首先,根据FMCW雷达中频信号定位目标,然后根据目标位置获取目标距离门及其附近距离门的相位信号,并作为神经网络的输入,通过神经网络提取呼吸和心跳信号。实验结果表明,通过CNN-BiLSTM神经网络可以有效提取呼吸和心跳信号,并根据该提取到的呼吸和心跳信号可以准确估计呼吸频率和心率,估计误差分别为7.2%和8.1%。

This paper proposes an FMCW radar vital sign signal detection algorithm based on CNN-BiLSTM neural network to address the respiratory harmonics and clutter interference during vital sign detection,which cannot accurately extract vital sign signals.Firstly,locate the target based on the intermediate frequency signal of the FMCW radar,and then obtain the phase signals of the target range gate and its nearby range gates based on the target position,which are used as inputs to the neural network to extract respiratory and heartbeat signals.The experimental results indicate that the CNN-BiLSTM neural network can effectively extract respiratory and heart rate signals,and accurately estimate respiratory and heart rate based on the extracted respiratory and heart rate signals.