Application of Extension of Unscented transformation technique to the nonlinear case of error propagation

Error propagation of a random variable for a case of large uncertainty and nonlinearity is studied, for the first time, using the extended unscented transformation technique (EUT), which is an extension of the unscented transformation (UT)technique, by revisiting an example studied in the literature by Smith, Neudecker, and Capote-Noy (Report INDC NDS–0709, 2016). In this example, the first four moments of a nonlinear transformation of a random variable of single dimensionality are determined, using extended unscented transformation technique and compared with 500000 histories of the Monte Carlo (MC) method. It is observed that the EUT method is in better agreement with results obtained using the MC method and superior to the UT method which has been applied by the authors in their earlier papers.