Applications of Neural Networks in Continuous Casting

Abstract:- This paperwork describes and refers to the structure of neuronal networks who make up the system we use for predicting wire breaking, the way they have been used and implemented; and the use and implementation of the entire system. Before testing the networks, we must identify the design of the RNA input curves. We should identify it experimentally, using the same measurements as for the continuous cast process. For using the serial-dynamic and space network, we need a large amount of data, more than the data that a thermo-couple uses during 120 seconds. Thus, specialists have had to design new software in order to stimulate the difference curves we should use for each network input. Because dynamic-serial networks follow the same pattern of data input, we have preferred to use only one serial-dynamic network and clone the others. We have performed the same in case of space networks whose input data are the same output data from two of the dynamic-serial networks