Structural Adaptation of Nonlinear Models in Time Series Processing Problems
摘要
The article proposes a solution to the problem of structural adaptation of nonlinear models for processing time series. The approach is aimed at overcoming the shortcomings of the differential spectrum balance method in the scheme of differential non-Taylor transformations for modeling essentially nonlinear processes. It is proposed to modify the structure of models adaptively to their classes in compliance with the requirements for obtaining a certain system of equations in the scheme of differential spectra balance. The given calculation results prove its effectiveness.