Biological learning metaphors for adaptive process control: a general strategy
Jean-Michel Renders, Raymond Hanus · 2003
The authors propose a general strategy for applying biological adaptive metaphors to nonlinear process control. The metaphors considered consists of a mixture of neural networks, immune networks, and genetic algorithms. Issues regarding the fundamental limitations of these metaphors in process control are raised. An approach aimed at overcoming these limitations as far as possible is proposed. In particular, it is shown that the requirement that control be exercised by poorly adapted regimes can be circumvented, and a certain quality control guaranteed. The approach allows current controllers, whether conventional or of novel design (e.g., fuzzy or neural), to be integrated naturally into a coherent control scheme.>