| Summary: | 碩士 === 國立宜蘭大學 === 電機工程學系碩士班 === 99 === Abstract
In this thesis, compared with the Enhanced Karnik-Mendel Algorithm, a modified EKM (MEKM) algorithm with more intuitive meaning is presented in this paper to further improve the performance for type reduction of interval type-2 fuzzy systems (FST2). The initial estimated outputs in the modified EKM are calculated with the possible left and possible right most membership functions. In the simulation results, the proposed modified Enhanced Karnik-Mendel (MEKM) algorithm is indicated to have less iterations than that required in the EKM in general. Also, the statistical evaluations on the iterations and computation time using EKM and the proposed MEKM are provided to point out that the proposed MEKM behaves better. Moreover, the MEKM is shown to have better performance than the MEKM when the equivalent selection of the variable values exists.
Keywords: Enhanced Karnik-Mendel (EKM)、modified Enhanced Karnik-Mendel (MEKM)
|
|---|
