Neural networks for electric furnaces

摘要

Properly designed neural networks can recognize the complex relationships among variables and learn to control and optimize a manufacturing process. Mills can train the systems to minimize costs, maximize throughput, or achieve some balance between different objectives. Neural Applications' intelligent-arc-furnace (IAF) controller is based on neural networks. It has an IBM-compatible 486 PC as a supervisory control computer and a PC with a Pentium processor that controls electrode positioning in the furnace using three neural-network models.