近红外可穿戴设备中脉搏波的呼吸率检测

针对可穿戴设备中光电容积脉搏波检测呼吸速率准确性不高和实时性不够的问题,提出了一种基于时频谱的自适应信号分解算法。该算法采用瞬时中心频率估计方法获得脉搏波时频谱和瞬时心率估计值,对脉搏信号进行相干解调提取呼吸信号成分,进而利用呼吸信号成分检测呼吸速率。实验结果表明,与传统连续小波变换方法相比,本文提出的自适应信号分解算法的呼吸速率计算时间提高了84.68%。通过中位数误差及四分位距误差的方差分析,表明该算法比连续小波分解算法和自回归模型算法具有更好的计算精度,中位数误差均值分别提高了96.001%和97.978%,四分位距误差均值分别提高了75.014%和52.732%。

Detection of respiratory rate using pulse wave on near infrared wearable devices

Aiming at the problems of the inaccuracy and insufficient instantaneity in estimation of respiratory rate from photoplethysmography(PPG) on the wearable devices, an Adaptive Signal Decomposition(ASD) algorithm based on time-frequency spectra was put forward. This algorithm adopted the instantaneous center frequency to obtain the pulse wave time-frequency spectra and instant heart rate estimated values, and the breath signal ingredient was extracted through coherent demodulation on the pulse signals, then the respiration signals can be used to detect the respiratory rate. The result indicates that in comparison with the conventional Continuous Wavelet Transform(CWT), the respiratory rate calculation time of ASD algorithm has been increased by 84.68%. The variance analyses of the median error and the interquartile range error indicate that the ASD algorithm has the better calculation accuracy than the CWT algorithm and the autoregression model algorithm, with the median error means compared to which increased by 96.001% and 97.978% respectively and the interquartile range error means increased by 75.014% and 52.732% respectively.