Recognition of fault signature patterns using fuzzy logic for prevention of breakdowns in steel continuous casting process

Abstract

In the continuous casting process of steel, a thin solid shell gradually forms around the cooling copper mould inner perimeter as liquid steel flows through it. By the time steel emerges from mould bottom, the perimetric solid shell acquires sufficient thickness and strength to support ferrostatic pressure of liquid steel in its interior. However, the formative shell can tear, stick to mold wall or fail to thicken adequately - in all cases leading to massive leakage of liquid steel out of mold bottom and consequent breakdown of continuous casting. All such phenomena leading to process disruption are preceded by typical patterns in temperature history data recorded at arrays of thermocouples embedded in mould walls. A fuzzy system is developed to accurately recognize these patterns in real time and regulate the speed of casting to prevent breakdowns. The system designed and implemented is fail-safe, has a low false-alarm ratio, and provides online diagnostics of the continuous casting process. © Springer-Verlag Berlin Heidelberg 2005.