Time Series Forecasting Based on Interval Type-2 Fuzzy Logic System with PSO

For some problems with unpredictable and random risk, such as building energy consumption, an interval type-2 fuzzy logic system is proposed for time series forecasting to reduce forecast errors. In the design, the particle swarm optimization algorithm adjusts the parameters of the designed interval type-2 fuzzy logic system, and the membership functions of the antecedent and the subsequent are all selected as Gaussian membership function which has uncertain standard deviation. Statistical results and contrast experiment confirm that the interval type-2 fuzzy model which used particle swarm optimization algorithm can simplify the system rules in a great extent and improve the accuracy of predicted results in contrast with the mathematical model which has not particle swarm optimization which has certain research value.