A new shape change quantification method for estimation of power in shape rolling

Abstract A new neural network-based model is introduced to estimate power in shape rolling. Power estimation is important for the design and scheduling of mills. The model to be developed in this work is intended to be capable of estimating power in both flat and shape rolling. For this purpose, a new parameter called the “shape factor” is introduced to quantify shape changes in rolling. This new parameter along with other involved factors are used to develop a power estimator using the neural network techniques. Since this approach is applicable to both flat and shape rolling, experimental and analytical data for flat rolling can be used for training the neural network model. This is shown to be essential in the model development due to the lack of enough accurate experimental or analytical data for shape rolling. The network estimations are compared with data collected from flat and structural rolling mills. The comparisons show a fairly good agreement between network predictions and real data. This approach seems to be a powerful and accurate tool to estimate power in shape rolling by taking advantage of theoretical/empirical flat rolling models.