A practical design of interval type-2 fuzzy logic systems with application to solar radiation prediction

Abstract

A wide range of real-world applications is associated with chaos and uncertainty, particularly when it comes to forecasting future events. Type-2 fuzzy logic systems (T2FLSs) are renowned for their ability to manage such uncertainties. However, for optimal performance, they require careful design and customization. A practical design capable of handling uncertainties, especially those arising from measurement errors and interdependencies between variables, is necessary due to the higher computational cost of T2FLSs. This article is motivated by the need for a practical model of T2FLSs that appropriately models uncertainty but comes with a higher computational cost compared to type-1 fuzzy logic systems (T1FLSs). Therefore, it presents a novel approach that seeks to build on previous research by constructing a computationally efficient configuration of T2FLSs that is particularly well suited for modeling problems involving various forms of uncertainty. Uncertain rule-based T1FLSs and T2FLSs are used as predictors in a four-stage practical design learning procedure. To achieve optimal results in the trade-off between speed and accuracy, the proposed design employs the most advantageous characteristics of T1FLSs and T2FLSs at various stages. The predictors use a few chosen meteorological and solar radiation variables to forecast the global horizontal irradiance for the following day for various Saudi Arabian locations. The results show that T2FLSs improved prediction over T1FLSs, while the practical design procedure successfully reduced the computations by 93.7%.

Keywords

• Type-2 fuzzy logic systems
• complexity
• solar energy
• prediction
• Fuzzy Systems
• Solar energy