期刊名称：International Journal of Grid and Distributed Computing
印刷版ISSN：2005-4262
出版年度：2016
卷号：9
期号：2
页码：73-84
DOI：10.14257/ijgdc.2016.9.2.08
出版社：SERSC
摘要：According to the power load has strong randomness and difficult to forecast, the introduction of the two types of fuzzy logic in order to improve the prediction accuracy. The interval type two non-single valued two type Mamdani fuzzy model for power load time series forecasting, and reverse the spread of a similarity of a singular value decomposition iterative blending algorithm to simplify the redundant rules in the model of fuzzy sets and redundant fuzzy rules, in order to eliminate the adverse effects. For ordinary type-2 fuzzy sets, uncertainty of the trace and once the membership function is the most important factor, therefore in the calculation formula for construction of two kinds of measure when considering these two factors; analysis of the ordinary type-2 fuzzy inclusion degree properties; discussed two kinds of conversion between the new measure of the relationship, revealing its internal relations; finally through an example to verify the performance of the new measure ordinary type-2, and the fuzzy similarity and Yang Shih clustering method combining cluster analysis used in Gauss plain type-2 fuzzy sets, obtained the reasonable clustering results, verify the rationality of the new measure and effectiveness.
关键词：Type-2 fuzzy sets; prediction; time series forecasting; total quantity of knowledge
