Hits:

Indexed by:期刊论文

Date of Publication:2017-07-01

Journal:IFAC-PapersOnLine

Included Journals:SCIE、EI、CPCI-S、Scopus

Volume:50

Issue:1

Page Number:12823-12828

ISSN No.:24058963

Key Words:BFG pipeline pressure; Membership functions; Multi-objective clustering; Predicting; Fuzzy model

Abstract:Design of reasonable membership functions (MFs) is a primary problem for the fuzzy modeling method. Considering the complex nonlinear characteristics of blast furnace gas (BFG) system in steel industry, a MFs learning method based on clustering analysis is proposed in this paper, where a multi-objective density clustering method is reported by combing the targets of the model accuracy, complexity and interpretability. In order to simplify the modeling process and fit the distribution characteristics of industrial data, a simple type of function is designed and the optimized clustering results are used for determining the parameters of fuzzy MFs. To verify the performance of the proposed method, the practical data coming from a steel plant are employed. The experiment results demonstrate that the MFs designed by the proposed method could effectively improve the accuracy, complexity and interpretability of the fuzzy model, which provide helpful information for the fuzzy modeling of BFG pipeline pressure. ? 2017