APPLICATION OF INFORMATIVE ATRIBUTES FOR DETERMINATION OF STATIONARITY BREAKDOWN INSTANT OF A RANDOM PROCESS
St. Petersburg State Polytechnical University, Department of Higher Mathematics;
A. P. Filimonov
Petersburg State University of Means of Communications, Department of Physics; Professor
Abstract. Number of inversions of stationary process is shown to be of high informative significance for determination of breakdown instant of stationary operation mode of complex dynamic system. The two criteria are considered: parametric one, associated with representation of variances of elementary harmonic components in canonical decomposition of random narrow-band process as coordinates of a Hilbert space vector, and nonparametric, allowing for process trend extraction under wide prior uncertainty conditions.
Keywords: random process overshoots, signal evaluation, signal filtration, canonical representation of random process.