基于BP神经网络的钢水温度预定模型

A presetting model of molten steel temperature based on BP neural network

针对目前钢水温度预定方法存在不足,在分析钢水温度预定原理的基础上,在邯钢邯宝炼钢厂建立了基于BP神经网络的精炼终点目标温度和转炉终点目标温度的动态预定模型。利用邯宝炼钢厂的历史生产数据对模型进行了训练和测试,并进行了现场应用试验。结果表明,预定模型对转炉和精炼终点目标温度进行了优化,应用预定模型后,LF开始温度命中率提高到75%,中间包温度命中率提高到96.7%。

Based on the principles of presetting of molten steel temperature,a dynamic presetting model of molten steel temperature was established in Hanbao Steel Plant using BP neural network method.This model is intended to overcome the disadvantages of traditional presetting methods.The model has been used to predict and set the target end point temperature of molten steel in second refining and the target end point temperature of basic oxygen furnace.The model was trained and tested by the history production data of the steel plant and then applied in production.The simulation results showed that the target end point temperature of second refining and the target end point temperature of basic oxygen furnace had been optimized,the hit rate of starting temperature in LF increased by 75 % and the hit rate of predicted temperature in tundish enhanced by 96.7 %.