A neural network controller designed solely to realize maximum acceptable steady state error may have unsatisfactory transient response such as excessive overshoot and large rise time. Conversely, a controller designed to realize very good transient response may have unacceptable steady state characteristic. In this paper the authors investigate the effectiveness of adding velocity reference feedback for simultaneously improving the transient and steady state response of a neural network controller. Using the direct control scheme, on a nonlinear underwater vehicle dynamics, the authors show that specific improvements in overshoot, rise time and settling time could be achieved using this method. The control scheme and results are discussed in detail.
Improving the dynamic response of neuro-controllers: a novel feedback scheme
Die Verbesserung des dynamischen Verhaltens von Reglern mit neuronalem Verhalten durch eine erweiterte Regelschaltung, aufgezeigt am Beispiel eines Unterwasserfahrzeuges
1993
4 Seiten, 5 Bilder, 6 Quellen
Conference paper
English
künstliche Intelligenz , Expertensystem , lernender Automat , mathematisches Modell , kybernetisches System , selbsteinstellendes System , Regelkreis , Regelgüte , Regelungstheorie , nichtlineare Regelung , dynamisches Verhalten , stationärer Zustand , nichtlineares System , neuronales Netz , Geschwindigkeitsmessung , Unterseeboot , neuronaler Regler
Neuro-Controllers for Adaptive Helicopter Training
SAE Technical Papers | 1993
|Neuro-Controllers for Adaptive Helicopter Training
British Library Conference Proceedings | 1993
|Control Surface Failure Detection and Accomodation Using Neuro-controllers
British Library Conference Proceedings | 1993
|