Research On Neural Network Algorithm Of Breakout Prediction In Slab Continuous Casting

Breakout is not only the most common, but also the most serious accident in slabcontinuous casting. Once happening, it will bring about serious consequences to theenterprise. Breakout prediction system used in continuous casting is the most effectiveand direct way to solve the problem. In all kinds of breakout prediction systems,neural network breakout prediction system is the most widely used, whose core is thealgorithm of neural network. Therefore, it is of great significance to study thealgorithm of neural network.Firstly, in the paper, the formation mechanism of sticker breakout is discussedand the preventive measures are put forward to prevent the sticker breakout. Then theneural network breakout prediction system model is set up, which includes time seriesnetwork model and the space network model. The time series network model isdesigned to identify temperature changes of single thermocouple on time sequence. Ifthe time series network model alarms, the space network model will identifytemperature changes of surrounding thermocouples on time sequence. Then thebreakout alarm information is generated by logical judgment.Secondly, BP neural network algorithm and RBF neural network algorithm,which are the most commonly used in neural network breakout prediction system, arestudied, According to the problem of local minimum points in BP neural networkalgorithm, genetic algorithm is introduced to the BP neural network, on this basis, BPneural network based on genetic algorithm is established, which can effectively avoidthe problem and have better generalization ability than the RBF neural networkalgorithm.Finally, in the application of the WINDOWS operating system, the continuouscasting breakout prediction simulation system is accomplished with the help ofprogramming software Microsoft Visual C + + 6.0, and BP neural network based ongenetic algorithm is used. Then the temperature data, from a steel plant, is input to thesystem, and offline simulation of continuous casting breakout prediction system in laboratory is realized. The result shows that the continuous casting breakout predictionsimulation system can reduce the false alarm and avoid the missing alarm effectively,and BP neural network based on genetic algorithm can be used in the neural networkbreakout prediction system.