Molten Steel Breakout Prediction Based on Genetic Algorithm and BP Neural Network in Continuous Casting Process

摘要

In this paper, a compound sticking breakout prediction model including two kinds of modules, the time-sequence module of single thermocouple and the space module of multi-thermocouple was presented. The GA-BP neural network method with the genetic algorithm optimizing the original weights and thresholds of BP neural network, was used for building time-sequence module. Compared with traditional BP neural network, GA-BP neural network could avoid the defects that the results of traditional BP neural network are easily fall into local minimum point, and identify temperature patterns of sticking breakout more accurately. The testing results show the quote rate and accuracy rate for sticking breakout prediction have both achieved 100%.