Breakout prediction of continuous slab casting based on neural networks and fuzzy neural networks

摘要

A CNN based breakout prediction system developed by ARIM has been in on-line operation at BISC for years, this system has shown much better performance than that of the previously used logic decision-making type breakout prediction system. ARIM is now developing an FNN based breakout prediction system for Benxi Steel with the additional merit that knowledge is expressed explicity in fuzzy rule form thus excluding the requirement of initial training data while still retaining the learning ability of the neural networks.