Controlling of Processes by Optimized Expertsystems

Abstract

Expertsystems are characterised by storing knowledge, normally in if-then-rules. The human Intelligence implies the ability to comprehend, reason, memorise, learn, adapt and create. The attribute of certainty or precision does not exist in human perception and cognition. Perception and cognition through biological sensors, pain reception and other similar biological events are characterised by many uncertainties. A person can linguistically express perceptions experienced through the senses, but these perceptions cannot be described using conventional statistic theory. The perception and cognition activity of the brain is based on relative grades of information acquired by the human sensory systems. These are the reasons that fuzzy logic has been applied very successfully in many areas where conventional model–based approaches are difficult or not cost-effective to implement. Therefore fuzzy-rule based Expertsystems have many advantages over classical expertsystems. Hybrid neuro-fuzzy-Expertsystems combine the advantages of fuzzy systems, which deal with explicit knowledge which can be explained and understood, and neural networks which deal with implicit knowledge which can be acquired by learning. Different methods are known for combining fuzzy-rule-based-systems with neural networks. But all these methods have some disadvantages and restrictions. We suggest a new model enabling the user to represent a given fuzzy-rule-base by a neural network and to adapt its components as desired.