Home – Home – Jornals – Avtometriya 2005 number 2

The robust filter synthesis algorithm determining the state vector estimator of a continuous-time linear dynamic system with random jump parameters described by the Markovian chain with the finite number of states for a degenerate or ill-conditioned matrix of measurement noise intensities is considered. It is proposed to choose the filter transfer coefficients from the minimum sum of the trace of covariance matrix of filtering errors and the weighted innovation process covariance with simultaneous averaging over state probabilities of the random jump parameter. Conditions of asymptotic robust filter stability for a stationary system are obtained.