Maximum entropy approach to fuzzy control

摘要

For the same expert knowledge, if one uses different &- and V-operations in a fuzzy control methodology, one ends up with different control strategies. Each choice of these operations restricts the set of possible control strategies. Since a wrong choice can lead to a low quality control, it is reasonable to try to loose as few possibilities as possible. This idea is formalized and it is shown that it leads to the choice of min(a b,1) for V and min(a,b) for &. This choice was tried on NASA Shuttle simulator; it leads to a maximally stable control.