Vital Signs Extraction Technology Based On UWB Radar

Contact monitoring methods for vital signs are mostly used for patient monitoring in hospitals,and cannot meet the needs of more scenarios,such as sleep monitoring,infectious disease patient monitoring,and burn patient monitoring.Non-contact sign monitoring has gradually attracted people’s attention.Ultra-wide Band（UWB）signal with high bandwidth and narrow pulse width has the characteristics of high distance resolution,strong penetration,strong anti-interference and low power consumption,and can realize non-contact vital sign extraction.This thesis focuses on the extraction of two vital signs of breathing and heartbeat by UWB radar.The research work of this thesis is summarized as follows:1.An ultra-wideband radar sign extraction system is built.An impulse ultra-wideband radar and a directional plane Yagi antenna are used.ECG monitoring is applied as a reference for accuracy.The UWB sign echo signal is modeled,and the harmonic and intermodulation products models are mainly studied.Based on the model above,the technical difficulties that the heartbeat fundamental product is easily submerged in respiratory harmonics and intermodulation products are analyzed.The preprocessing methods of UWB echoes are studied,including static clutter filtering,bandpass filtering in fast sampling dimension,mean filtering,range gate signal selecting and Hanning window processing.2.Aiming at the problem that the heartbeat fundamental wave is easily submerged in the respiratory harmonics and intermodulation products,the harmonics and intermodulation products based fuzzy logic（HIPBFL）algorithm for vital sign extraction is proposed.The proposed algorithm introduces an idea of using the information of harmonics and intermodulation products to reverse heartbeat frequency.Using the frequency relationship between fundamental wave,harmonics and intermodulation products,the possible heartbeat frequencies are extracted in the heartbeat fundamental frequency band,the two intermodulation bands and the heartbeat secondary frequency band respectively.Those possible heartbeat frequencies are scored by the fuzzy logic system.In the continuous monitoring experiment,compared with other commonly used algorithms,HIPBFL has high accuracy,strong robustness and good real-time performance.Aiming at the problem that the variational mode decomposition algorithm needs to set parameters in advance and it is difficult to realize continuous monitoring,the multiple variational mode decomposition（M-VMD）algorithm is proposed.The algorithm implements multiple variational modal decomposition without changing parameter settings for each input time series,and can be applied to different individuals.In the continuous monitoring experiment,M-VMD has high accuracy and can realize continuous monitoring.3.Aiming at the problem that the heartbeat signal with small amplitude is easily overwhelmed by noise,the joint multi-range gate signal by cross-correlation algorithm for vital sign extraction is proposed.Multiple distance gate signals accumulate the same frequency components and eliminate the orthogonal noise.The fundamental respiration and the second harmonic of the heartbeat are respectively extracted in the specific frequency bands.In the continuous monitoring experiment,the proposed algorithm further improves the accuracy of the extraction of breathing frequency and heartbeat frequency.The proposed algorithm has good real-time performance.Experiments are carried out on the eight orientations of the human body to the antenna emission direction.The experiments confirm that the frequency of the breathing signal and the heartbeat signal can be extracted from the human body in any orientation.Breath signals can be extracted at any distance gate within the range of the human body.However,the accuracy of heartbeat signal extraction is related to the choice of distance gate.The selection of the distance gate is also related to the orientation and physiological characteristics of the human body.In the experiment,it is found that when the human body is completely facing the left or the right side,it is difficult to choose the distance gate of the heartbeat signal.The appropriate distance gate may not be at the maximum distance gate,which is related to the physiological structure and its physiological movement.The algorithms proposed in this thesis further improve the accuracy of breathing and heartbeat frequency feature extraction,and ensures real-time and robustness.The human body multi-orientation experiment broadens the application breadth of non-contact vital sign monitoring.The research in this thesis has great application value and prospects in non-contact vital sign monitoring under COVID-19,smart home and even clinical medicine.