Maneuvering Target Tracking Based On H_? Filter And Multi-model Algorithm

Whether in the military or civilian areas,maneuvering target tracking has a crucial position.However,the model uncertainty of the maneuvering target and the unknown statistical character-istics of the noise,etc,has made a great challenge to the tracking accuracy.Based on the H?filtering and multiple model algorithm,this dissertation focuses on maneuvering target tracking under non-ideal assumptions.The main research work and contributions are as follows:(1)In linear discrete time-invariant system,new steady-state H? filters based on transfer function are proposed.The main differences and the application scenarios between the H2 filter,Kalman filter and H? filter are analyzed and summarized.Some numerical simulations are pre-sented to demonstrate that H? filter performs better than Kalman filter when the system model contains uncertainty.(2)In nonlinear system,a new unscented H? filter is proposed based on the theory of sigma points.A numerical simulation compared with the extended H? filter is presented to prove the efficiency of the new algorithm.(3)The noise statistical characteristics estimation problem is studied when the priori knowl-edge of the noise is unknown.A new noise estimation approach based on H? filter is proposed and applied in maneuvering target tracking.The noise mean and covariance estimate is obtained while system state estimation is completed.(4)Maneuvering target tracking problem is studied in the scenario where the statistical char-acteristics and the priori knowledge of the system noise is unknown.By combination of the H?filter and interacting multiple model algorithm,new approaches are proposed,which are interacting multiple model H?(IMMH)algorithm and H?-interacting multiple model(H?-IMM)algorithm.The noise statistical characteristics estimate is obtained while system state estimation is completed.