Comparison of EKF and SDDRE Based Estimation of 2D Coordinated Turn Maneuvers

摘要

Estimation of 2D piecewise coordinated turn target maneuvers withrnunknown turning rates is considered. Two models of maneuvering targetrnin 2D are considered: Velocity based Jerk (VJ) and Acceleration basedrnJerk (AJ). The Extended Kalman Filter (EKF) and State DependentrnDifferential-Difference Riccati Equation (SDDRE) state estimationrnapproaches are compared. The comparison is done in 2D and an effort isrnmade to perform it on a common basis. The EKF uses the Jacobian matrixrnas a linear approximation of the nonlinear system. The SDDRErnparameterizes the state equation into the linear structure called StaternDependent Coefficient Form (SDC). The stability of SDDRE and EKF arernconsidered for the representation of the maneuvering target models. Thernsimulations demonstrate that the SDDRE approach has faster transientrnand adaptation capabilities.rnFurther the SDDRE estimates correctly the unknown turning rate of bothrnmodels while the EKF approach does not estimate correctly the unknownrnturning rate with the VJ model. Both filters are sensitive to the powerrnspectral density of the driving noise. In order to optimize the filters'rnperformance one should implement each filter with its optimal powerrnspectral density of the driving noise.