| 摘要 |
PURPOSE: A quality stabilization method in continuous casting process utilizing neural network is provided to learn an operation change threshold value using the neural network so as to minimize production failure cost, stabilize the product quality, and improve customer's reliability. CONSTITUTION: The quality stabilization method in continuous casting process utilizing neural network comprises designing a model for measuring effects on a quality of the product in a PLC level through a self learning on the variation factors of the main operation items occurring when a forced puncturing is performed to a sludge nozzle, tracking positions at which the quality of the product is unstable, comparing an estimated quality of the product to be produced and a quality required for the product, and determining quality level of the product based on the comparison result. By adopting the quality stabilization method, it is possible to reduce the defective proportion of the products and increase the productivity and client's reliability on the product.
|
