決定論的非線形短期予測手法の上水道需要量データへの応用

抄録

One of the possible applications of the chaos theory for engineering is nonlinear short-term prediction. Hetherto, many prediction methods have been proposed. But most of them are based upon linear theories. Therefore, it is difficult to obtain high performance when timeseries is produced by nonlinear dynamics. In this paper, we apply deterministic nonlinear short-term prediction to timeseries data of water demand. As a result, this paper shows timeseries data of water demand have structure of a possible attractor, and prediction accuracy by deterministic nonlinear short-term prediction is higher than that by a typical conventional prediction method with an autoregression method combined with Kalman filter.