Research On Detecting Cardiopulmonary Signs In Natural Posture Based On Millimeter Wave Radar

The parameters of cardiopulmonary signs are one of the important characteristics of human health,and real-time monitoring of their dynamic changes has important application value in health evaluation and medical monitoring.Cardiopulmonary signs detection based on millimeter-wave radar has become a research hotspot because of its non-contact,non-invasive and all-weather advantages.And its high-frequency characteristics can sensitively detect the millimeter fluctuation of the chest cavity.But the body movement will inevitably occur when the human body is in the natural posture,and its amplitude can reach several centimeters or even tens of centimeters,corresponding to the broadband noise on the spectrum,and easily covering the cardiopulmonary signals completely,which is a big challenge for the extraction of cardiopulmonary signs.Aiming at this problem,the paper aims to improve the signal-to-noise ratio and accuracy of cardiopulmonary signals from two aspects of improving motion recognition performance and motion noise filtering performance.The main contents are as follows:1)In view of the problem that the clutter generated by random motion is likely to have similar frequency components with breath and heartbeat,and even completely cover cardiopulmonary signal,a body motion elimination algorithm based on amplitude threshold and acceleration information fusion was proposed by comparing the recognition effects of common body motion recognition methods on three different types of body motion: back-and-forth movement,cough and speech.The acceleration information method is used as an auxiliary judgment method to make up for the missing detection problem of the amplitude threshold method for periodic body motion,realizing the rapid and effective recognition of body motion.The improved adaptive noise complete empirical mode decomposition algorithm and variational mode decomposition algorithm are introduced to apply to the body motion data.Experimental results show that the proposed algorithm can effectively detect and filter the body motion.2)For BFM body motion interference,an adaptive noise elimination algorithm based on polynomial fitting is introduced to eliminate body motion.Due to the uncertainty of the time and amplitude of motion,in order to improve the convergence speed and tracking ability of time-varying system without losing steady-state error,a variable step size algorithm was introduced,and a variable step size adaptive noise elimination algorithm based on polynomial fitting is proposed.Experimental results show that the signal-to-noise ratio increases by 0.4069 dB,the mean square error decreases by 0.0119,and the waveform correlation coefficient increases by 0.1192.Furthermore,a highly reliable cardiopulmonary signal extraction algorithm based on dual-channel cross-correlation was proposed to further filter the residual body motion noise.The experimental results show that compared with the single channel treatment results,the SNR of respiratory signal and heartbeat signal increased by 1.114 dB and2.257 dB on average respectively,and the mean error of respiratory signal and heartbeat signal decreased by 0.0076 Hz and 0.0268 Hz respectively after two-channel cross-correlation.