Modeling of gain tuning operation for hot strip looper controller by recurrent neural network

In this paper, neural network model representing gain modulating action by human was developed for looper height controller in hot strip mills. The developed neural network model is that of the recurrent type neural network (RNN) which calculates the appropriate PII) gains of looper height controller based on the modification data of human operation as the training data. Further, learning algorithm for RNN model was developed to accelerate convergence of the gain modification process and to stabilize the looper movement. The neural gain tuning model was applied to the inter-stands looper height controller in hot strip mills. The usefulness of the developed model was checked through numerical experiments. From the experimental results, it was verified that the tuning action by human can be realized by the model. The model could also cope with disturbance such as change in roll gap because of its learning mechanism that may lead to the stabilization of threading operation of hot strip mills.