Contents I. Introduction
In the past decades, fuzzy systems were widely used and extensively studied for a variety of applications, especially in control applications because of their excellent performance in representing nonlinear functions in an intuitive and informative manner. In many applications of fuzzy theory, fuzzy systems involved can be viewed as rule-based function approximators. Many scholars have proved that fuzzy systems are universal approximators in different aspects [1]–[5]. To our knowledge, the smaller the approximation accuracy, the more fuzzy rules are required in order to approximate a certain continuous function, especially in high dimensions. Large numbers of rules render difficulty in designing fuzzy systems, such as high storage consumption, increased computational complexity, etc.
