Masri M. Methods and toolkit of building information models of nonlinear dynamic objects for diagnostic purposes

In the thesis the new solution of urgent scientific and technical objectives is proposed - the improvement of methods and means of mathematical and computer modeling of nonlinear continuous dynamic systems based on Volterra models. These models are intended for use for studying the objects of different physical nature, the creation of information technology and intelligent computing systems diagnosis. On the basis of theoretical and experimental studies using computer simulation tools have been further developed the methods and computational algorithms of deterministic identification of nonlinear dynamical systems in time domain. The corresponding software tools are applied for building models as the sequences of multidimensional Volterra kernels and multidimensional transition functions on the basis of the "Input-Output" experiments taking measurement errors into account. The modified approximation method allow increasing the accuracy of identification in 2-4 times. The wavelet filtering application for smoothing the data of experiments allow the increasing of accuracy in 2-3,4 times having the results of identification smooth.