Implicit rule-based fuzzy-neural networks using the identification algorithm of GA hybrid scheme based on information granulation |
| |
Authors: | Sung-Kwun Oh Witold Pedrycz Ho-Sung Park |
| |
Affiliation: | 1. Middle East Technical University, Turkey;2. University of Turkish Aeronautical Association, Turkey;1. State Key Laboratory of Environmental Criteria and Risk Assessment, Chinese Research Academy of Environmental Sciences, Beijing 100012, China;2. State Environmental Protection Key Laboratory For Lake Pollution Control, Research Center of Lake Environment, Chinese Research Academy of Environmental Sciences, Beijing 100012, China;3. Yunnan Key Laboratory of Pollution Process and Management of Plateau Lake-Watershed, Kunming, 650000, China;4. Dongting Lake Ecological Observation and Research Station, Yueyang, 414000, China |
| |
Abstract: | This paper proposes an identification method for nonlinear models realized in the form of implicit rule-based fuzzy-neural networks (FNN). The design of the model dwells on the technologies of computational intelligence (CI), namely fuzzy sets, neural networks, and genetic algorithm. The FNN modeling and identification environment realizes parameter estimation through a synergistic usage of clustering techniques, genetic optimization and a complex search method. An HCM (Hard C-Means) clustering algorithm helps determine an initial location (parameters) of the membership functions of the information granules to be used in this fuzzy model. The parameters such as apexes of membership functions, learning rates, and momentum coefficients are then adjusted using the optimization algorithm of a GA hybrid scheme. The proposed GA hybrid scheme combines GA with the improved complex method to guarantee both global optimization and local convergence. An aggregate objective function (performance index) is used in the model design in order to achieve a sound balance between its approximation and generalization abilities. The proposed type of the model is experimented with several time series data (gas furnace, sewage treatment process, and NOx emission process data of gas turbine power plant). |
| |
Keywords: | Fuzzy-neural networks Computational intelligence HCM clustering GA hybrid scheme Genetic algorithm Improved complex method |
本文献已被 ScienceDirect 等数据库收录! |
|