Dealing with structural dynamcis and control constitutes one important discipline in aeronautical and astronautical industry. For aircraft, undesirable structural vibrations may affect mission effectiveness, induce weight penalty, create objectionable pilot workload, increase maintenance cost, and even lead to catastrophes. On the other hand, a well-designed structural control system may enhance maneuverability, improve ride quality, and alleviate load. For space structure, vibrations and jitterings resulting from unknown disturbances or inappropriate control design through structural dynamics may deteriorate pointing accuracy or lead to instability. How to accurately model the structural dynamics and control structural motion becomes an important design exercise. A context sensitive neural network is devised for structural system modeling and control. This neural networks attempts to blend existing structural modeling formulation into the neural network framework. One innovation of the neural network is to represent each structural vibration mode using a neuron. Thus, when the neural system converges, the modal data are identified. Another neural network paradigm that is identified for structural modeling is the adaptive vector quantization technique. It is shown that this learning scheme is able to resolve closely-spaced modes for large space structures. The context sensitive neural networks also allow the injection of robust control design in the context of neural network.
Structural modeling and control using neural networks
Strukturmodellierung und -steuerung mit Hilfe neuronaler Netze
1994
11 Seiten, 11 Bilder, 4 Quellen
Conference paper
English
Structural modeling and control using neural networks
AIAA | 1994
|Modeling and Control of UAVs using Neural Networks
AIAA | 2012
|Modeling and Control of UAVs using Neural Networks
British Library Conference Proceedings | 2012
|Traffic Flow Modeling and Control Using Artificial Neural Networks
British Library Online Contents | 1996
|Traffic flow modeling and control using artificial neural networks
Tema Archive | 1996
|