基于数字孪生和AO-ELM融合驱动的RH炉终点温度预报模型

RH furnace endpoint temperature prediction model driven by digital twin and AO-ELM fusion

RH精炼炉作为钢铁冶炼过程中的关键设备,其终点温度对后期铸造和产品质量影响较大。为了尽可能准确地预报钢水的终点温度,提出基于数字孪生的RH炉终点温度预报方法,通过数字孪生体模型实现终点温度的精准预报。首先通过天鹰优化器优化极限学习机(AO-ELM)构建终点温度预报虚拟模型,根据物理空间中得到的实时炼钢数据,由AO-ELM模型获得初始预测值,同时更新孪生数据库;然后通过相似度搜索,在孪生数据库中找到相似的冶炼炉次,对比相似炉次下的预报值与实际值,对初始预报值进行加权误差修正,得到最终预报值。实际算例结果表明,所建模型相较于传统人工智能终点温度预报模型更加精准和可靠,对后续温度控制有较好的指导意义。

RH refining furnace is the key equipment in the process of iron and steel smelting,and its endpoint tem-perature has a great influence on the later casting and product quality.In order to predict the endpoint temperature of molten steel as accurately as possible,a prediction method of RH furnace endpoint temperature based on digital twin was proposed,and the accurate prediction of endpoint temperature was realized by digital twin model.Firstly,the endpoint temperature prediction virtual model was constructed by the Aquila optimizer-extreme learning machine(AO-ELM).According to the real-time steelmaking data obtained in the physical space,the initial prediction value was obtained by the AO-ELM model,and the twin database was updated at the same time.Then,similar smelting furnaces were found in the twin database through similarity search,and the predicted value and actual value under similar furnaces were compared,then the weighted error correction of the initial predicted value was carried out to obtain the final predicted value.The actual calculation results show that the proposed model is more accurate and re-liable than the traditional artificial intelligence endpoint temperature prediction model,and has good guiding signifi-cance for subsequent temperature control.