Identification of the self-organizing exponential autoregressive models

摘要

A kind of time series model, namely exponential autoregressive that has properties similar to those of nonlinear random vibrations, is achieved to be identified in self-organization. This paper introduces the genetic algorithm hybridized with the recursive least squares method to select the optimum exponential autoregressive model. The final model identified by this evolutionary approach may be not only a full exponential autoregressive mdoel but also a subset model. The simulations of artificial time series and applications to machien tool chatter analysis are given to show the efficiency of the approach proposed.