POLYGAUSSIAN DESCRIPTION OF PROBABILITY DISTRIBUTIONS OF PROCESSES GENERATED BY A NONLINEAR LORENZ SYSTEM IMPLEMENTED IN FIXED-POINT NUMBERS

The article is aimed at analyzing the probability distributions of pseudorandom processes generated on the basis of the Lorentz system solution in fixed-point numbers. Numerical solution of the Lo-renz system by the Euler method in single- and double-precision floating-point numbers with a lim-ited number capacity can lead to a breakdown in signal generation. The use of fixed-point numbers contributes to the reduction of computational complexity in the digital implementation of such sys-tems, which ultimately leads to the simplification of their practical implementation on modern chips of programmable logic. This allows to use resources in more efficient way, and increases productiv-ity in the creation and maintenance of such digital systems. The study of approximation of Lorenz system signals using mixtures of Gaussian distributions is of great importance for predictive analyt-ics and stability of the system. Elimination of signal generation failure also contributes to the for-mation of stable modes of generation of chaotic signals with required statistical characteristics.