The Transfer Alignment Technique Research Based On The H_∞ Filter

The carrier aircraft often takes off and weapon system launches weapon when the ship is moving. So initial alignment of Strap-down Inertial Navigation System of the carrier aircraft or weapon system need to be fast on moving base. Considering all the modern ship being fixed high precise Master Inertial Navigation System (MINS), transfer alignment is a perfect choose for initial alignment of the carrier aircraft and weapon's Slave Inertial Navigation System (SINS).The research of the thesis is based on Strap-down Inertial Navigation System. We deduce systemically the error model of velocity matching transfer alignment, and we get its state equation and measured equation base on this error model. Considering the velocity plus attitude matching method's dynamic arrangment equations of error angle being showed in carriered reference frames of MINS and SINS, it's form is not same as velocity matching method's and refer outside observable attitudes to transfer alignment, so we also deduce error model for velocity plus attitude matching method, and get its measured model.We apply PWCS theory to transfer alignment of SINS, and analyse its observability and degree of observability using SOM matrix instead of TOM matrix. For velocity matching and velocity plus attitude matching transfer alignment, We use SVD method to analyse the observability and degree of observability on various moving states in the thesis. We delete the states of lower degree of observability to get absolutely observable state variables.We introduce knowledge about H_∞filter, and discuss the similarities and differences between H_∞filter and Kalman filter from theory. Traditional Kalman filter requires that the noise is the white noise and we often could not meet the need during the actual application, so we adopt H_∞filter to estimate misalignment angles for forward two matching methods, and we resolve H_∞filter by Riccati equation. We design H_∞filter and Kalman filter to estimate misalignment angles in Matlab for white noise and colored noise separated, and contrast to analyse estimating results of two filters. From the simulation, we can conclude that the estimating precision of H_∞filter is lower than Kalman filter's a little for white noise and H_∞filter converging speed is more faster, but for colored noise, not only converging speed but also estimating precision of H_∞filter precede Kalman filter evidently.