摘要：This paper provides an overview of lo cally stationary pro cesses, a helpful methodology for handling nonstationary time series. These techniques allow for the smo oth evolution of the model parameters. This work reviews estimation and predictions techniques, illustrating the application of these methods to real-life data examples. These examples show that the locally stationary methods provide a useful theoretical and practical framework for the statistical analysis of nonstationary time series data.
关键词：Kalman filter ; State space system ; Nonstationarity ; Long-range ; dependence ; Local stationarity ; Time-varying models.
