计算机应用 ›› 2017, Vol. 37 ›› Issue (3): 889-895.DOI: 10.11772/j.issn.1001-9081.2017.03.889

• 应用前沿、交叉与综合 • 上一篇    下一篇

基于多任务学习的炼钢终点预测方法

程进, 王坚   

  1. 同济大学 CIMS研究中心, 上海 200092
  • 收稿日期:2016-08-24 修回日期:2016-10-24 出版日期:2017-03-10 发布日期:2017-03-22
  • 通讯作者: 程进
  • 作者简介:程进(1989-),男,上海人,博士研究生,主要研究方向:机器学习、计算机集成制造;王坚(1961-),男,山东淄博人,教授,博士,主要研究方向:企业信息化、网络化制造与系统集成。
  • 基金资助:
    国家科技支撑计划资助项目(2015BAF22B00)。

Endpoint prediction method for steelmaking based on multi-task learning

CHENG Jin, WANG Jian   

  1. CIMS Research Center, Tongji University, Shanghai 200092, China
  • Received:2016-08-24 Revised:2016-10-24 Online:2017-03-10 Published:2017-03-22
  • Supported by:
    This work is partially supported by the National Key Technology R&D Program of China (2015BAF22B00).

摘要: 钢水质量通常根据终点命中率来判断,但炼钢过程影响因素众多,机理分析难以准确预测终点温度和含碳量,鉴于此,提出一种由数据驱动的多任务学习(MTL)炼钢终点预测方法。首先,分析并提取炼钢过程的输入和输出要素,结合炼钢两阶段吹炼特点选择多个子学习任务;其次,根据子任务与终点参数的相关性选择合适的子任务,提升终点预测的准确度并构建多任务学习模型,再对模型输出结果进行二次优化;最后,通过近端梯度算法对处理后的生产数据进行模型训练,获取多任务学习模型的过程参数。以某钢厂为案例,该方法相比神经网络在终点温度12℃误差范围内和终点含碳量0.01%误差内的准确度提升了10%,误差范围6℃和0.005%的预测准确度分别提升了11%和7%。实验结果表明,多任务学习在实际中能够提升终点预测的准确性。

关键词: 产品质量预测, 炼钢终点预测, 数据驱动, 多任务学习, 近端梯度算法

Abstract: The quality of the molten steel is usually judged by the hit rate of the endpoint. However, there are many influencing factors in the steelmaking process, and it is difficult to accurately predict the endpoint temperature and carbon content. In view of this, a data-driven Multi-Task Learning (MTL) steelmaking endpoint prediction method was proposed. Firstly, the input and output factors of steelmaking process were analyzed and extracted, and a number of sub-learning tasks were selected to combine the two-stage blowing characteristics of steelmaking. Secondly, according to the relativity between the sub-tasks and the endpoint parameters, the appropriate subtasks were selected to improve the accuracy of the endpoint prediction, and the multi-task learning model was constructed, and the model output was optimized twice. Finally, the process parameters of the multitask learning model were obtained by model training of the processed production data through the proximal gradient algorithm. In the case of a steel plant, compared with neural network, the prediction accuracy of the method proposed increased 10% when endpoint temperature error was less than 12℃ and carbon content error was less than 0.01%. The prediction accuracy increased by 11% and 7% respectively with the error range less than 6℃ and 0.005%. The experimental results show that multi-task learning can improve the accuracy of endpoint prediction in practice.

Key words: product quality prediction, endpoint prediction for steelmaking, data driven, Multi-Task Learning (MTL), proximal gradient algorithm

中图分类号: