Improved maximum likelihood estimation for nonlinear continuous-discrete system and its application to aircraft flight data compatibility checking

Evans-Goodwin et al. (1985) presented a maximum likelihood algorithm for nonlinear continuous-discrete systems. Accounting for both the process and measurement noises, its filtering process is often numerical unstable and the calculation of sensitivities by using the finite-difference method needs to select the size of perturbation, these limit its validity in practical application. This paper gives a improved maximum likelihood method; an algorithm is given to determine the best perturbation. At the same time, a fast triangular formulation of square-root filter is used to replace the conventional one or the so called stabilized extended Kalman filter to improve the numerical stability of the algorithm given. A practical application to compatibility checking of flight test data indicate that the improved ML algorithm is more efficient.<>