Research On UWB-based Respiratory Signal Detection Algorithm

Respiration,as one of the important vital signs of human body,often plays an early warning role in many diseases that affect cardiopulmonary function,and timely detection can be helpful for subsequent treatment.The sudden outbreak of a new type of coronavirus pneumonia sounds the alarm of "prevention is greater than treatment",and people are pursuing more advanced and efficient respiratory monitoring technology as medical aid.With the vigorous development of communication technology and continuous exploration of radar technology in all walks of life,UWBR(Ultra-wide Band Radar)technology has become one of the hot spots in the field of electronic science.To achieve the detection of human respiratory rate,Puls ON-440 Ultra-Wide Band Radar is used to carry out experiments.The following contents are mainly studied.1.In the echo signal preprocessing,the direct subtraction,pulse cancellation,mean filter and SVD(Singular Value Decomposition)algorithms are compared through experimental validation.At last,the mean filter is selected as the radar echo preprocessing algorithm for subsequent research,and matched filter is used to improve the signal-to-noise ratio.Combining the amplitude tracking algorithm with EMD(Empirical Mode Decomposition)to extract moving target breathing.2.A filtering method for fast time domain distance gate signals in radar echo signals using depth neural network is presented,which solves the problem that respiratory signals cannot be extracted accurately and automatically in complex scenes.The topology structure of the deep neural network is analyzed and designed.The Puls ON-440 Ultra-Wide Band Radar module is used to collect data,and the experimental comparison analysis is completed in the actual scene.The results show that this method can effectively remove non-respiratory signals from radar echo,and further improve the accuracy of respiratory signals and robustness of detection system.3.An improved Hilbert frequency estimation algorithm using EWA(Exponentially Weighted Averages)is presented,which solves the problem of inaccurate estimation of respiratory rate by traditional algorithms.The improved algorithm is compared with STFT(Short-time Fourier Transform)in simulation and real environment,respectively.The results show that the former has faster calculation speed and smaller error when estimating respiratory rate.4.A real-time detection software of human respiratory rate based on Ultra-Wide Band Radar is designed and implemented in Python environment.At the same time,a convolution neural network is used to recognize human body turning and normal respiratory posture.The software can display the waveforms of respiratory signals,respiratory rate and posture information in real time.In this paper,Puls ON-440 Ultra-Wide Band Radar is used to collect the echo data of human respiration,select a better radar echo preprocessing algorithm suitable for the experimental environment,and combine the neural network technology to improve the accuracy and robustness of the respiratory signal detection system.It achieves the real-time detection of human respiratory rate and respiratory state.It is of great significance to the research of human sign signal detection technology.