Research On Interference Suppression And Extraction Of Vital Signs Based On FMCW Radar

Vital sign signal detection is a popular research in recent years.Traditional contact vital detection instruments such as electrocardiogram(ECG)and smart bracelet are not suitable for some special scenarios,such as detecting vital signs of burned patients,disaster trapped people,etc.FMCW radar can measure the distance and velocity information of the target,has high sensitivity and anti-interference capability,which has a wide application prospect in the field of non-contact vital sign detection.However,the vital sign signal detection process can be affected by background environmental noise,human jitter and other factors,and the breathing harmonics can also affect the accurate extraction of heartbeat signals.To solve these problems,this paper implements non-contact vital sign detection using FMCW radar and applies various denoising algorithms,which can accurately separate respiration and heartbeat from vital sign signal and accurately estimate the signal frequency.The main work and innovation points are as follows:(1)To introduce the principle of vital sign detection.First,the theoretical knowledge about the principles of vital sign detection is introduced.Secondly,the AWR1642 radar platform is introduced,and the process of vital sign detection of this platform is sorted out.The platform was used to collect target data and a small database of about 500 segments of measured signals was established.Finally,the principle and process of front-end signal pre-processing are outlined,and the front-end signal pre-processing is performed on the collected and collated raw data.Then the possible clutter interference encountered in the process of vital sign detection is analyzed,and the reason why clutter affects vital sign extraction and frequency estimation is pointed out.(2)To study multiple algorithms of respiratory heartbeat separation and interference suppression.Firstly,the simulation signal is established and simulated according to the characteristics of vital signals.Secondly,two major classes of respiratory heartbeat separation and interference suppression algorithms are studied.One is the empirical modal decomposition class of algorithms,including EMD algorithm,EEMD algorithm,CEEMDAN algorithm and ICEEMDAN algorithm.The second is the variational modal decomposition class of algorithms,including the VMD algorithm and the VME algorithm.For each algorithm,the principle of the algorithm is outlined and the signal is simulated and analyzed.Although there are many papers on these two aspects,there is still a lack of very systematic analysis and comparison.In this paper,the simulated and measured signals are analyzed from four perspectives: algorithm characteristics,signal-to-noise ratio,heart rate estimation accuracy,and running time,and the performance of each algorithm is comprehensively compared.Theoretical analysis and extensive simulation experimental results show that the most suitable algorithm for heart rate estimation is the variable modal decomposition class algorithm,among which the VME algorithm has the best performance.(3)The time-frequency analysis method of vital signal is studied,and the TF-Fusion framework applicable to vital sign signal detection is proposed.First,the time domain estimation of the vital signal is performed with multiple denoising methods,including pulse denoising,respiratory heartbeat separation denoising,peak-valley difference denoising,and peak-peak time interval denoising.Then the separated heartbeat signal is estimated in the frequency domain,and the Rife algorithm is applied to insert new values after the sliding window FFT to enhance the denoising effect.Finally,the time-frequency domain heart rate fusion method is proposed to improve the heart rate estimation accuracy.(4)The TF-Fusion framework is validated,taking into account the accuracy of heart rate estimation and the real-time processing of the signal.We applied the VME algorithm,the best algorithm applicable to vital sign detection,to the proposed framework and performed simulation and performance analysis on the framework.The experimental results show that the mean RMSE of the estimated heart rate of the framework is 1.48 bpm,and the mean MAPE is 1.44%,and the mean variance is 0.72 bpm.The experimental results verify the effectiveness of the proposed framework,which has high accuracy and stability of heart rate estimation.