Modeling and identification of flexible link using neural networks

Abstract

The challenge in the endpoint control of flexible manipulators is of obtaining an exact model of the nonlinear flexible system and then simulating it for further analysis. But being an infinite dimensional, they pose problems in terms of predicting the behaviour and also exact mathematical representation. This problem of obtaining an exact model can be solved if the behaviour of the flexible link system can be represented by a neural network, which would help further simulation, analysis and control thus eliminating the requirement of the exact mathematical representation of the system.