Time series prediction using the parallel-structure fuzzy system

This paper presents the parallel-structure fuzzy system (PSFS) for predicting chaotic time series. The PSFS consists of multiple number of fuzzy systems connected in parallel. The fuzzy system contains Sugeno type fuzzy rules modeled using input-output training data. Each fuzzy system in the PSFS predicts the same future value based on input data with different embedding dimension and time delay. The embedding dimension is chosen optimally to have superior performance for each value of time delay. The PSFS determines the final predicted value by averaging the outputs of each fuzzy system excluding the minimum and the maximum values in order to reduce error accumulation effect.