二叉树型分层广义混合模糊系统的结构和规则数分析

模糊系统由一般输入变量、推理规则和实际输出组成,推理规则总数会随着输入变量的增多而呈指数形式急剧上升,容易引起规则爆炸.为此,降低规则总数是必要的.本研究首先对二叉树型分层广义混合模糊系统进行结构分析;其次,依据混合推理规则及计数公式针对分层前与分层后的规则总数进行比较和分析.结果表明,分层后系统内部规则总数的增长速度远远小于不分层的规则总数的增长速度,从而可避免规则爆炸现象.

Analysis of structure and rule numbers for binary tree-type hierarchy generalized hybrid fuzzy systems

Fuzzy systems are usually composed of input variables, the inference rules and output. However, with the increasing numbers of input variables, the total of inference rules of the systems will increase exponentially to cause rules＇ explosion. So it is necessary to adopt the method of reducing the total number of rules. Firstly, the paper analyzes the structure of binary tree-type hierarchy generalized hybrid fuzzy systems. Secondly, the total rules in no-hierarchical system and hierarchical system are compared and analyzed based on generalized inference rules and the counting formula. It shows that the increasing rate of total of inference rules in hierarchical generalized hybrid fuzzy system is far below that of no-hierarchical system, and the rules＇ explosion is effectively avoided.