| 摘要 |
PURPOSE:To improve accuracy of the inference result by using the theoretical value of each rule condition part as a weight and applying the weighted average to the manipulated variable of the conclusion result of each rule conclusion part for output of the final manipulated variable. CONSTITUTION:The input variable of a controlled system 4 detected by a sensor part 2 is supplied to each of rules 1-n of a rule part 11. The part 11 applies the rules in response to the input variable value related to each rule for output of the theoretical value Li and the manipulated variable Yi. A weighted average part 12 performs the calculation where the weighted average is applied to the variable Yi with the value Li used as the weight Wi for the groups of both values Li and Yi outputted from each rule of the part 11. Then the part 12 outputs the result of said calculation as the final operation variable. An operation value setting part 3 sets the output operation variable as the control value to be given to the system 4. Thus it is possible to avoid such a case where the priority is given as a weight depending on dispersion showing the ambiguity degree of the fuzzy set.
|
