Fuzzy Forecast Modeling for Gas Furnace Based on Fuzzy Sets and Rough Sets Theory

摘要

This paper describes a new approach to generate optimal fuzzy forecast model for Box and Jenkins' gas furnace from its Input/ Output data (I/O data) by fuzzy set theory and rough set theory (RST). Generally, the nonlinear mapping relations of I/O data can be expressed by fuzzy set theory and fuzzy logic, which are proven to be a nonlinear universal function approximator. One of the most distinguished features of RST is that it can directly extract knowledge from large amount of data without any transcendental knowledge. The fuzzy forecast model determination mainly includes 3 steps: firstly, express I/O data in fuzzy decision table. Secondly, quantitatively determine the best structure of the fuzzy forecast model by RST. The third step is to get optimal fuzzy rules from fuzzy decision table by RST reduction algorithm. Experimental results have shown the new algorithm is simple and intuitive. It is another successful application of RST in fuzzy identification.