Constraint-based granular computing for fuzzy modeling

With fuzzy constraints serving as the framework, granular computing along with its ensuing processing are applicable to fuzzy modeling, particularly for large-scale systems. In this paper, a novel concept of granular computing regarding to fuzzy modeling is proposed. Put differently, the fuzzy modeling method is viewed as a process of accumulation and granulation of fuzzy granules with the process repeated until a desired model is achieved. In the following presentation, formal definition of each key component of granular computing is introduced first; namely, the definition of fuzzy granule and the definitions of size and granularity of information granule. The mathematical manipulations among granules are given afterwards. Then, the detailed process of attaining a desired fuzzy model is provided. Finally, the simulation of a simplified rotary cement kiln is given in demonstrating the strength of our approach