Forecast of random oscillation processes based on the method of exponential smoothing (Q1920373)

A forecast problem for an unknown nonrandom function \(x(t)\) via oscillating type approximations \(\widehat{x}(\varphi-\theta)= \sum^m_{i=0} [x_{1i}(\varphi) \cos(i\theta)+x_{2i}(\varphi) \sin(i\theta)]\) is considered with the cost function \(\int^\infty_0|x(\varphi-\theta)- \widehat{x}(\varphi- \theta)|^2 \exp(-b\theta)d\theta\). A heuristic algorithm is presented. Simulation results are given.