# Signal Processing Method for Mattress-type Physiological Monitoring

## SHEN Jinpeng, WANG Xin’an

ACTA Scientiarum Naturalium Universitatis Pekinensis - - Contents -

The Key Laboratory of Integrated Microsystems, Peking University Shenzhen Graduate School, Shenzhen 518055; † E-mail: shenjp@pkusz.edu.cn

Abstract An empirical mode decomposition (EMD) based algorithm is proposed for mattress-type physiological monitoring. Fast Fourier Transform (FFT) is executed for the mixed physiological signal to estimate the frequency range of the respiration signal and heartbeat signal. Then the physiological signal is decomposed into several IMF components by EMD method, some of which are used to reconstruct the respiration signal and the heartbeat signal, according to the respiration or heartbeat spectrum energy in proportion of the total spectrum energy. The experiment result shows that the accuracy of measured respiration rate and heartbeat rate are both over 90%, compared with the polysomnography. Key words mattress-type physiological monitoring; EMD; piezoelectric sensor

[3–5]单, 已成为人体生理信号监测的研究热点 。床垫式生理信号监测系统的传感器采集的原始信号包括呼吸、心跳和噪声信号, 所以需要提取呼吸和心跳信号, 才能进一步计算呼吸率和心率。目前, 床垫式生理信号监测系统大多采用常规的滤波方法[6]

[7–8]或小波变换 来提取呼吸和心跳信号。然而, 异

1 床垫式生理信号监测系统介绍

2 信号处理方法的实现

fc 是一个根据原始信号来动态调整的参数, 因此可以根据原始信号的不同, 动态地确定呼吸和心跳的频带范围。

1) 找出原始信号的所有极小值点和极大值点,然后拟合成上包络线e up( t)和下包络线e (t), 计

down算包络平均值:

5)将r ( t) 视为新的数据, 重复步骤1~3, 可以1得到第 2 个 IMF分量c 2( t) , 重复 N 次, 得到 N 个IMF 分量和一个不满足 IMF 条件的残余量rn (t)。原始信号可表示为

Er ( i) , (5) E ( i)

3)将所有组成呼吸信号的有效分量加起来, 得到呼吸信号:

3 实验结果与分析

4 结论