| 摘要 |
PURPOSE: An automatic generation method of fuzzy functions and rules by using input/output data is provided to generate fuzzy functions and rules automatically by using input/output data of a system, to prevent the increase of the number of fuzzy functions and rules, and to improve the speed of the learning. CONSTITUTION: An automatic generation method of fuzzy functions and rules by using input/output data consists of four steps. At the first step, in a system, when a fuzzy input variable is input, the system finds the maximum inference error point. At the second step, the system decides whether the maximum error point belongs to the renewal area or the generation area. At the third step, as the decision of the second step, if the maximum inference error point belongs to the generation area, a new fuzzy function is generated, the function is inserted between left/right fuzzy functions adjacent to the maximum inference error point, and then, the system changes the range of the left/right fuzzy function. At the fourth step, as the decision of the second step, if the maximum inference error point belongs to the renewal area, the system changes the range of the fuzzy function near the maximum inference error point among left/right fuzzy functions adjacent to the maximum inference error point.
|
