Fuzzy Rule Base Reduction
Dan Simon
Cleveland State University
d.simon@ieee.org

Various attempts have been made over the years to reduce the rule base of a fuzzy logic system. Rule base reduction may be important for computational reasons in those cases where a fuzzy system has to be implemented in real time. Yeung Yam and his colleagues have recently published an algorithm based on singular value decomposition whereby a fuzzy rule base can be reduced. My recently submitted paper has demonstrated the technique on a fuzzy estimator for motor winding current estimation, where the rule base was reduced from 49 rules to 9 rules. I performed my work as an employee of Cleveland State University.

I have made general-purpose MATLAB code available for fuzzy rule base reduction using Yam's algorithm. The code consists of two files that can be downloaded from this page: Reduce.m (the main file) and FuzzFunc.m (an auxiliary file). Both files are necessary for the rule base reduction algorithm. In order to run the rule base reduction algorithm, download the two files, run MATLAB, make sure that the location of the two files on your hard drive is part of your MATLAB path, and type "Reduce" at the MATLAB prompt. Feel free to contact me and d.simon@ieee.org with any comments or questions. For more information about MATLAB, see The MathWorks' web site.

References

·         D. Simon, "Design and Rule Base Reduction of a Fuzzy Filter for the Estimation of Motor Currents," International Journal of Approximate Reasoning, submitted for publication.

·         Y. Yam, P. Baranyi, and C. Yang, "Reduction of Fuzzy Rule Base Via Singular Value Decomposition," IEEE Transactions on Fuzzy Systems, Volume 7, Number 2, pp. 120-132, 1999.

·         Y. Yam, "Fuzzy Approximation Via Grid Point Sampling and Singular Value Decomposition," IEEE Transactions on Systems, Man, and Cybernetics – Part B: Cybernetics, Volume 27, Number 6, pp. 933-951, 1997.

Home         Credentials         Publications       White Papers

Email Address: d.simon@ieee.org
Phone Number: (216)687-5407

Last Revised: May 29, 2007