高炉炼铁数据缺失处理研究初探

针对高炉炼铁过程中的数据缺失问题,提出以单维结合多维的系统化数据填补模式。总结并阐述数据缺失填补办法的发展现状以及优劣势比较。在此基础上,通过对河北某钢铁厂的实际高炉生产数据进行分类比较,并结合填补办法的优缺点,针对高炉炼铁数据提出一套以简单统计类办法、线性插值法、机器学习法等多种办法相结合的方案,以实现高炉数据的深度整合及处理,满足数据挖掘工作的供数需求。同时选取了炉顶温度、氧气管道温度作为数据样本,利用临近点中间值法、临近点均值法、线性插值法、Adaboost算法等对样本数据进行了填补且填补效果较为理想,充分验证了方案的可行性。

Preliminary study on missing data processing of blast furnace ironmaking

Aiming at the problem of missing data in the process of blast furnace ironmaking, a systematic data filling model with a single dimension and multi-dimension was proposed. The development status and comparison of the advantages and disadvantages of the method of filling the missing data were summarized and expounded. On this basis, through the classification and comparison of the actual production data of a steel plant in Hebei province, combined with the advantages and disadvantages of the filling method, aiming at blast furnace ironmaking data, a set of solutions combined with simple statistical methods, linear interpolation methods, machine learning methods and other methods were proposed to realize the in-depth integration and processing of blast furnace data and meet the demand for data mining. At the same time, the furnace top temperature and oxygen pipeline temperature were selected as data samples, and the sample data are filled with the near point intermediate value method, the near point mean value method, linear interpolation method and Adaboost algorithm. And the filling effect was ideal, which can fully verify the feasibility of the scheme.