Rule based fuzzy logic (FL) time series prediction

Abstract

The singleton and non-singleton type-1 back propagation (BP) designed sixteen rule fuzzy logic system (FLS) on hourly averaged wind data for the years 1985 to 2004 are studied. The BP designed 16 rule non-singleton-type-1 FLS was found relatively a better forecaster than singleton-type-1. There are too many hidden or unraveled uncertainties, such as non-stationarity and stable attractors. These uncertainties make the data chaotic. Non-stationarity in the data can be properly handled with non-singleton type-1 FLS, therefore, there appears no reason to use a type-2 FLS. The stable attractors and non-stationarity in our data do not affect the predicted values as confirmed by Mackey Glass simulation. Parallel structure fuzzy systems and genetic logic may be one of the options to resolve sub crisps and chaos in time series data. Key words: Back propagation, fuzzy logic system, singleton and non-singleton type1-FLS, cascade correlation algorithm, hybridization of intelligent systems with fuzzy logic, stable attractors.