Predicting chaotic time series using adaptive wavelet-fuzzy inference system

Predicting traffic flow is of extreme importance in traffic modeling and congestion control. The traffic data usually exhibit chaotic dynamics that can be readily modeled and analyzed using time series. Traditional tools for time series analysis have been focused on exploring the statistical properties of the data. On the other hand, it has been long observed that times series can be considered as the output of nonlinear dynamic system. The development of computational intelligence methodology and its composing methods including fuzzy logic and neural networks has provided a new powerful tool for time series analysis. The paper represents a novel method of using a hybrid networks following the fuzzy logic inference mechanism to predict chaotic times series.