On-line identification of multidimensional parametric vector random variation of pulse system

The problem of motion identification of the non-stationary pulse system in on-line mode with multidimensional parametric vector random variation is considered in the paper. Because of the large motion variety, possible in the system of this type with the wide range of parameter variation, the offered approach considers the particular operating motions. The main idea consists in that by means of the procedures offered the phase trajectory of both, transitional and stable motions, is transformed into a vector form in the special spaces. Then, at the multidimensional variation of the parameters, stable motion dynamics regularities are mapped in the form of the multileveled fractal structure. Hence, by means of vector mapping of time series one can subsequently solve the following problems: identify the moment of the transitional process completion, identify the stable system state, and, as a result, identify a corresponding value of the parametric vector. The identification exactness depends on both, absolute and relative ranges of variations of each parameter. The algorithm of approach realization was considered with an example of 4-D parametric vector identification of the pulse system dynamics within the fundamental motion.