Chaotic time series prediction and Mackey-Glass simulation with fuzzy logic

Abstract

The present study was performed with fuzzy logic (FL) time series prediction modeling on a twenty years hourly averaged wind data, that is, 1985 to 2004 for Quetta, Pakistan. A free fuzzy logic design was followed and hourly wind data for spring prediction were obtained (February, March and April). It was found that the prediction is reliable and precise. Non-stationarity or random walk in wind data exists but it does not influence prediction. Mackey-Glass (MG) simulation of wind data indicated chaos or non periodicity in time series. Moreover, stable attractors were observed in MG-time series, in which the origin is yet unknown. The attractors seen in MG simulation do not influence FL time series prediction. Key words: Fuzzy logic, artificial neural networks, antecedents, the adaptive neural fuzzy inference system, autoregressive integrated moving average.