Defuzzification process for fuzzy logic control

摘要

In the defuzzifying process for signal-processing fuzzy components, the two process steps filtering of membership functions and defuzzification by the less cumbersome maximum height or maximum mean method replace the prior art and generally costly defuzzification or hyperdefuzzification. This provides functional advantages. They reside first in a much greater flexibility which can be used owing to their transparency for interactive optimisation. Secondly, besides the actual output values, the fuzzy components provide additional information on the reliability of these values. In the third place, with two-way fuzzy components which can process positive and negative rules, the compromise to be made between the recommendations and warnings or prohibitions can be more accurately controlled. The process also provides structural advantages: there are standard commercial components like digital FIR filters or signal processors with the aid of which the membership functions can be economically filtered even for fast real-time applications.