Millimeter Wave Radar Indoor Personnel Intelligent Perception

Intelligent perception of indoor personnel is an extremely important and urgent problem in many fields such as smart home,smart elderly care,and smart office.Millimeter-wave radar is one of the most advantageous implementations because of its high resolution,low power consumption,and no risk of privacy leakage.The application background of the thesis is the intelligent perception of indoor personnel,and then a set of principle prototypes are developed using the IWR6843 ISK millimeter-wave radar.Under the high-bandwidth LFMCW signal model,the problems of indoor personnel presence perception and state perception are studied based on the target four-dimensional point cloud information.Obtaining high-quality 4D point cloud information is the basis technology to realizing indoor personnel presence perception and state perception.Therefore,this thesis first expounds the basic theory of object detection under the LFMCW signal model.Among them,the phasor mean cancellation algorithm is used to achieve static clutter filtering,the TDM-MIMO working mode combined with the Capon-BF algorithm is used to achieve angle superresolution,and the CASO-CFAR algorithm is used twice to obtain the distance-angle spectral matrix.The scatter diagram of the target,and finally the Doppler phase compensation algorithm is used to enhance the accuracy of the angle estimation of the moving target.Indoor personnel presence perception is realized by 4D point cloud data combined with group target tracking algorithm.Human motion model and radar system observation model have nonlinear characteristics.Therefore,this thesis establishes a three-dimensional space target motion model and uses Extended Kalman filter algorithm(Extended Kalman filter).Filter,EKF)to achieve target tracking and positioning.In order to suppress indoor multipath effect,several indoor multipath effect models of electromagnetic waves are analyzed,and the room size estimation algorithm and velocity cross-correlation algorithm are used to suppress multipath interference.According to the simulation results and actual measurement analysis,indoor personnel presence perception has an accuracy of more than 95% in the case of multiple targets.Finally,this thesis uses the Support Vector Machine(SVM)classification algorithm to extract the feature quantities of the 4D point cloud data of indoor people walking and falling normally,as the sample data of the SVM algorithm for training,and realizes the normal walking and falling of people.perception of a state.According to the simulation results and actual measurement analysis,the method has an accuracy of more than 85% for personnel fall detection,which verifies that the millimeter-wave radar indoor personnel intelligent perception system based on the target 4D point cloud information has a good effect.