Processing of signals as the sum of the determined function and realization of stationary casual process

The paper provides the analysis of processing signals in the form of the sum of not stochastic functions and realization of stationary casual process with use of models of auto-regress - a sliding average. It offers classification of models of trends (main tendencies) according to basic algebraic functions used: fractional-rational functions, functions, representable in the form of the sum of final number quasi monomials, etc. Functions have non-stationary character. It is marked that many time series of an expert detect non-stationary character; they can be carried to homogeneous non-stationary numbers. It is possible to realize a stationary accidental process.