Research On An Improved Particle Filter For Uav Passive Tracking Based On RSS

In recent years,with the rapid development of the UAV(Unmanned Aerial Vehicle)industry,various technologies have gradually matured and the cost has continued to decrease.The UAVs are developing in the direction of miniaturization,low-altitude and diversification,and more and more consumer-grade UAVs entered people’s lives and played many irreplaceable roles in the civilian field.However,the rapid development of UAV has also brought many social problems,and the accidents that threaten public safety frequently occur.Therefore,it has important research significance for the positioning and tracking of non-cooperative UAVs.Compared with time delay and Doppler-based positioning methods,power-based UAV positioning methods have been widely used due to their low cost;however,the signal power is easily affected by factors such as noise,clutter,and antenna directivity.The accuracy of positioning and tracking is often not high,especially in the complex maneuver scenario of the UAVs,there is a lack of effective solutions.Therefore,this paper studies the passive positioning and tracking technology of UAV in this scenario.First of all,in view of the abnormal value of the UAV position observation value in the actual measurement scene based on the power and the deviation caused by the antenna directivity,a particle filter algorithm is proposed that can handle abnormal data and use offline data to correct the deviation.For problems with abnormal data,the correlation gate is used to preprocess the data and merge the valid data.For the problem of position observation deviation caused by antenna directivity,the relationship between UAV’s actual position and position observation deviation in off-line data is used to improve the likelihood function in particle filter algorithm,and then the deviation is corrected by changing the distribution of particle weight.Finally,the effectiveness of the algorithm is verified through field tests,and the errors of position and velocity are analyzed.Then,in the UAV maneuvering scene,the data generation operator is used to process the missing data,and the gray prediction model is introduced on the basis of the improved algorithm above,and a filtering algorithm that can use historical observation data to predict the UAV movement is proposed..In view of the lack of gray sequence data due to abnormal data,etc.,the mean value generation operator and the ratio generation operator are introduced to improve the robustness of the algorithm;For the problem that the motion state of UAV does not conform to the description of the state equation in the maneuvering scene,this paper introduces the grey prediction model.This model uses the position observations to generate some particles to improve the posterior probability distribution.The problem that the importance distribution function can not completely cover the posterior density is solved.Finally,the influence of different parameters on the algorithm is analyzed through simulation,and compared with the algorithm proposed above,it is verified that the proposed algorithm has better performance in maneuvering scenarios.These results are of great significance for the subsequent research on passive location and tracking technology.