Time series modeling for acoustic signals by a generalized adaptive function and its application to actual random noise data

As is well-known, it is impossible to model the actual time series data of acoustic signals by a simple equation. From the engineering viewpoint, effective prediction algorithm of the above time series signal is very important in the regulation and control problems for acoustic noise. It is also useful in saving the frequency band width in the communication line for acoustic information. From the above practical points of view, this paper describes a new trial for predicting the fluctuations of time series data by use of adaptive functions. The present prediction method is established based on not only the information of linear correlation but also that of higher order nonlinear correlations in the actual time series data. The validity of the proposed method has been experimentally confirmed by applying it to actual random noise data.