MULTIPLE FUZZY RULES INTERPOLATION WITH WEIGHTED ANTECEDENT VARIABLES IN SPARSE FUZZY RULE-BASED SYSTEMS

作者：Chen Shyi Ming^*; Chang Yu Chuan; Chen Ze Jin; Chen Chia Ling

来源：International Journal of Pattern Recognition and Artificial Intelligence, 2013, 27(5): 1359002.

DOI：10.1142/S0218001413590027

• Fuzzy interpolative reasoning
• polygonal membership functions
• multiple fuzzy rules interpolation
• sparse fuzzy rule-based systems

This paper presents a new method for multiple fuzzy rules interpolation with weighted antecedent variables in sparse fuzzy rule-based systems based on polygonal membership functions. First, the proposed method calculates the normalized weighting vector of each closest fuzzy rule. Then, it calculates the composite weight of each closest fuzzy rule. Then, it calculates the left normal point b(l)* and the right normal point b(r)(*) of the fuzzy interpolative reasoning result B* = (b(0)*, b(1)*, ..., b(l)*, b(r)*, ..., b(t-2)*, b(t-1)*), respectively. Finally, it calculates the characteristic points b(0)*, b(1)*, ..., b(l-1)*, b(r-1)*, ..., b(t-2)* and b(t-1)* of the fuzzy interpolative reasoning result B*, respectively. The experimental results show that the proposed method can generate more reasonable fuzzy interpolative reasoning results than the existing methods for sparse fuzzy rule-based systems. The proposed method can overcome the drawbacks of Chang et al.%26apos;s method (IEEE Trans. Fuzzy Syst. 16(5) (2008) 1285-1301), Chen and Ko%26apos;s method (IEEE Trans. Fuzzy Syst. 16(6) (2008) 1626-1648) and Huang and Shen%26apos;s method (IEEE Trans. Fuzzy Syst. 14(2) (2006) 340-359) for multiple fuzzy rules interpolation. It provides us with a useful way for dealing with multiple fuzzy rules interpolation in sparse fuzzy rule-based systems.

• 出版日期2013-8