A NEURAL NETWORK-BASED APPROACH FOR THE CONTROL OF AUTOLEVELLER IN DRAW FRAME

摘要

Autoleveller in drawing frame is the key components to the quality of sliver. With the increase of rotational speed, and emergence of uncertainty factors in multi-varieties of products in a small scale of batch process with different raw material, it is almost impossible to establish the exact mathematical model. Moreover autoleveller system is not only strongly disturbed with large time delays but also nonlinear and time-varying. As such the traditional control design methods are unable to cope with the control demands for the autoleveller in drawing frame. This paper studies applications of adaptive neural networks to autoleveller in drawing frame, which can take advantage of the self-organizing and self-learning of the neural network. This approach is effective to nonlinear systems with large time delays and fast varying parameters, and insures stability of the underlying feedback control system through adjusting the synapses of the neural network.