No. 67 (00188)

Family name : Karimova
Given name : Lyailya
Affiliation : Institute of Mathematics, Almaty, Kazakhstan
Abbreviation : 
E-mail address : chaos@math.kz
Title : THE GENETIC ALGORITHM FOR A SIGNAL ENHANCEMENT
Authors : L.Karimova,Y.Kuandykov, N.Makarenko 
Abstract : 
The paper is devoted to the problem of time series enhancement, which
based on the analysis of local regularity. The model construction using
this analysis does not require any a priori assumption on the structure
of noise and functional relationship between original signal and noise.
The signal itself may be nowhere differentiable with rapidly varying
local regularity, what is overcame with the help of the new technique of
increasing the local Holder regularity of the signal under research. A
new signal with prescribed regularity is constructed using the genetic
algorithm. This approach is applied to enhancement time series in the
paleoclimatology, solar physics, dendrochronology, meteorology and
hydrology.