Research On Non-Contact Heartbeat Detection Technology Algorithm

Respiration and heartbeat are important indicators for measuring human health.However,traditional contact-based detection methods have significant limitations in scenarios such as special patient testing in hospitals and centralized monitoring of elderly people in nursing homes.Therefore,the thesis investigates a non-contact vital feature detection algorithm based on Frequency Modulated Continuous Wave radar.Although there have been some research achievements in this field,there are still some key issues that need to be studied,such as strong disturbance of respiratory harmony to heart rate signals,separation and marking in complex noise environments,and accurate positioning and detection in multi-target situations.In order to solve these problems and improve the accuracy of non-contact vital sign detection based on millimeter-wave radar,the thesis mainly does the following work:Firstly,the intermediate frequency signal of FMCW radar is preprocessed.After distance-FFT processing and constant false alarm rate detection,the instantaneous phase signal of the human target is obtained and the phase unwrapping is performed.In response to the trend item problem in the phase signal,corresponding trend item processing is carried out to prepare for the subsequent separation of vital sign signals.Secondly,in response to the strong interference of respiratory harmonics on heartbeat signals,the thesis compares the reconstruction-based and notch-based respiratory harmonic suppression methods and proposes a reconstruction-least mean square adaptive harmonic suppression algorithm.This method can also perform respiratory harmonic suppression well when the respiratory detection accuracy is not high.Then,in response to the separation and denoising problem in complex noise environments,the thesis proposes a joint denoising and separation algorithm based on ICEEMDAN and improved wavelet thresholding,referred to as ICE-W algorithm.This method decomposes the mixed vital sign signal using ICEEMDAN,uses sample entropy to determine the noisy IMFs,and then performs improved wavelet thresholding denoising on the noisy IMF components.Finally,the respiratory and heartbeat signals are reconstructed based on the correlation coefficient.By comparing simulation and actual data,the ICE-W algorithm effectively improves signal-to-noise ratio and detection accuracy while reducing mean square error.Finally,in response to the multi-target positioning problem,the thesis compares four azimuth estimation algorithms,including digital beamforming,Capon algorithm,MUSIC algorithm,and forward-backward spatial smoothing MUSIC algorithm,and selects the forward-backward spatial smoothing MUSIC algorithm for azimuth estimation to improve accuracy.Combining with virtual antenna array expansion and azimuth estimation,a multi-target vital sign detection scheme based on ICE-W is proposed,and the effectiveness of this scheme is demonstrated through actual data.