A comparison of adaptive and conventional Kalman filters in a missle application

Includes bibliographical references (page 59)Kalman filters are used to obtain an estimate of a signal from a\ud set of noisy measurements. Conventional Kalman filters require\ud linear systems in which all variations in the process parameters are\ud known. Many extensions of this theory have been proposed to deal\ud with cases which do not meet these stringent requirements. Y. M. El-Fattah developed an adaptive Kalman filter to be used when the state space model and noise statistics depend on an unknown parameter. Following a brief formulation of the conventional Kalman filter, this paper explains the filtering method proposed by El-Fattah. The\ud performance of each Kalman filter was shown in a missile application.\ud The filter estimated the range to a target using the available noisy\ud measurements. An unknown measurement parameter was introduced for\ud the adaptive example. Simulations of the conventional and adaptive\ud Kalman filters were developed for a digital computer. The performance\ud of the conventional filter simulation for several different sets of initial conditions was shown. The performance of the adaptive filter simulation for the nominal case was compared to that of the non-adaptive filter. It was shown that the adaptive filter\ud was able to quickly provide an accurate estimate of the unknown\ud parameter as well as acceptable range estimates