A fuzzy neural network and its gradient descent algorithm for prediction intervals

Abstract

This study aims to propose a solution to handle the uncertainty and imprecise knowledge associated with the collected data using interval type 2 fuzzy inference system, or IT2FIS. IT2FIS has been shown to be capable of generalizing functional relationship between input and output while reducing computational complexity. The proposed IT2FIS is a fuzzy neural network realizing Takagi-Sugeno-Kang inference mechanism. IT2FIS structure consists of multiple layers, which evolves automatically based on the incoming data. The parameters are updated using meta-cognitive learning and gradient descent algorithm. Prediction intervals are considered as the end-result of the system. For performance evaluation studies, collected data of wind speed and direction are utilized. Using historical data, the proposed model provides short term forecasting of wind energy parameters. The performance of IT2FIS is compared with existing state of the art fuzzy inference system approaches and results clearly indicate the advantages of lT2FIS based prediction.