The case study is the management of fire emergencies in a mineral oils deposit. Since the functional knowledge of fire emergencies is inadequate, it should be worked out a new one by applying suitable procedures starting from a feasible plant model in firing conditions. Then genetic algorithms have been applied to obtain the knowledge base for the optimal management. These data enable to deduce the set of rules for training a neuro-fuzzy controller. Since this set is too large, a reduced order one has been neurally deduced to implement an on-line help for the operator.
展开▼
