论文部分内容阅读
针对软测量建模的特点以及建模过程中存在的主要问题,提出了基于AdaBoost RT集成学习方法的软测量建模方法,并根据AdaBoost RT算法固有的不足和软测量模型在线更新所面临的困难,提出了自适应修改阈值φ和增添增量学习性能的改进方法,使用该建模方法对宝钢300t LF精炼炉建立钢水温度软测量模型,并使用实际生产数据对模型进行了检验,检验结果表明,该模型具有较好的预测精度,能够很好地实现在线更新。
In view of the characteristics of soft sensor modeling and the main problems in the process of modeling, a soft sensor modeling method based on AdaBoost RT integrated learning method is proposed. According to the inherent deficiencies of AdaBoost RT algorithm and the difficulties of online updating of soft sensor model , An improved method to adaptively modify the threshold φ and increase incremental learning performance is proposed. Using this modeling method, a molten steel temperature soft-sensing model is built for Baosteel 300t LF refining furnace, and the actual production data are used to test the model. The test results show , The model has good prediction accuracy and can be well updated online.