Detection And Analysis Of Non-Contact Human Body Feature Information Based On UWB Radar

Human social life is closely related to radar,which constantly promotes the development of radar technology.In the radar family,Ultra Wide Band(UWB)is widely used in intelligent medical treatment,natural disaster detection and military warfare due to its strong anti-interference,high detection accuracy and low power consumption.In the actual detection,the radar signal and the measured target do not need close contact or other intermediate media,and the corresponding work can be easily and efficiently completed within the effective range.However,at present,when UWB radar signal is used to detect human breathing and heartbeat signs,due to the small fluctuation of body surface caused by breathing and heartbeat,radar echo signal is susceptible to the interference of ambient noise,and the sign information will be drowned in the noise.In addition,in the separation process of echo signal,the intensity of heartbeat signal is weaker than that of respiratory signal,and it is often difficult to extract the harmonic component of respiration.In this thesis,the Ensemble Empirical Mode Decomposition(EEMD)algorithm was optimized to establish the vital signs detection model by the back-propagation neural network integrated with genetic algorithm.The Intrinsic Mode Functions(IMF)feature training is strengthened,and human respiratory heartbeat signals with high signal-to-noise ratio can be obtained after reconstruction.This thesis analyzes the requirement background of UWB radar to detect the sign information,and researches and designs the radar system hardware and signal processing algorithm in detail.Firstly,the working principle of UWB radar to detect human signs information is analyzed,and the parameters of radar transmitting receiver,system gain and noise coefficient are formulated.According to the design index,the appropriate chip and electronic components are selected.The whole radar system is divided into several sub-modules and independent circuit schematic design and PCB layout drawing are carried out respectively.Lab View is used to develop the upper computer of radar detection system as the data transmission link between hardware circuit and software simulation.After the basic function verification,the correctness of the radar system hardware index and the effectiveness of the detection of vital signs information are verified.Then the vital signs detection model was built.The static clutter was filtered from the echo signal received by UWB radar by moving target detection method,and the body surface vibration signal was extracted by distance gate selection method.Then,the IMF component was obtained by EEMD decomposition of the signal.Ga-bp neural network with Bayesian regularization constraint was used to conduct feature training on the feature vectors after IMF component transformation,and the reconstructed cardiopulmonary signals were compared with those after original EEMD reconstruction.The simulation results show that the reconstructed signal of GA-BP neural network is in better agreement with the actual physical signs signal.Finally,the function of the designed radar system is verified.When UWB radar is used for target detection,ECG of the tested person is collected synchronously for reference calibration.The vital signs model and the original EEMD model with optimized parameters were used to reconstruct the IMF components,and in-depth analysis was conducted from two dimensions of signal waveform and spectrum.The experimental results show that the signal reconstructed by the new model has a high signal-to-noise ratio and meets the detection standard of physical signs information,which strongly confirms the correctness of the research topic.