基于BP神经网络的钢水温度预定模型

• 作 者: 冯春松 肖步庆 何飞
• 机 构: 河北钢铁集团邯钢公司邯宝炼钢厂,北京科技大学冶金与生态工程学院,
• 出 处: 钢铁研究
• 发表日期: 2012年03期
• 关键词: 钢水温度 BP 神经网络 BP神经网络 模型 钢水温度预定 命中率 目标温度 转炉 中间包

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 analysis of the preset principle of molten steel temperature, a dynamic prediction model based on BP neural network was established for the final target temperature and the target temperature of the converter at Handan Iron and Steel Plant, Handan Iron and Steel. Using the historical production data of Handan Steel Mill, the model was trained and tested, and the field application test was carried out. The results show that the target temperature is optimized for the converter and the finish endpoint. The predicted temperature hit rate of LF is increased to 75% and the hit rate of the tundish is increased to 96.7% after applying the predetermined model.