The application of multi-level recursive prediction in time-varying system

The Multi-level recursive prediction is a new method based on statistics, which is suitable for dynamic system. In this paper multi-level recursive prediction method is studied and applied in various time-varying systems. The simulated results from two time-varying systems are presented. One is the malfunction prediction of chemical Continuous Stirred Tank Reactor (CSTR) system and the other is time series prediction of ship rolling movement, from which we conclude that the multilevel recursion method is able to predict malfunction of nonlinear systems with good prediction precision.