Application of neural network to supervisory control of reheating furance in steel industry

摘要

The efficient and reliabel control of a reheating furnace is a challenging problem due to: a) many different types of billets to process, b) strong intercorrelation among process variables, c) large dimension of the input and output space, d) the strong interaction among process variables, e) a large time delay, and f) highly nonlinear behavior. Thus, the conventional reheatign furnace operation has been heavily dependent upon the look-up table which lists the optimal set points. We have developed a modified modular neural network for the supervisory control of the reheating furnace. Based on the divide-and-conquer concept, a modular network is capable of dividing a complex task into subtasks and modeling each subtask with an expert network. To model such activities, a gating network is used for the classification and allocation of the input data to the corresponding expet network. To overcome the correlation effects among process variables and the dimension problem, principal component analysis (PCA) has been employed to remove the correlation and reduce the problem dimension. From PCA analysis, we were able to decide the optimal dimension for the problem to describe the dynamic behavior of the furnace. The proposed neural network has been trained and tested using operation data from the reheating furnace and has been implemented on the wire rod mill process of POSCO~(TM).