基于Bagging集成的球团矿烧结过程混合建模

Modeling for Pellets Induration Process Based on Bagging Method

链篦机回转窑球团矿烧结过程是典型的热工过程,具有非线性、高耦合和大滞后的特点。要建立精确可靠的机理模型十分困难。此外,简化和假定条件与实际过程之间往往存在偏差,因此,单纯利用机理建模方法对球团矿烧结过程进行建模具有一定的局限性。考虑到球团矿烧结过程的复杂性和单纯机理模型的局限性,在机理模型的基础上,利用神经网络集成进行灰箱模型建模,以BP神经网络为集成的个体网络,采用Bagging法来生成样本集,样本用来训练个体网络。结果显示,混合模型具有更高的精度,是一种更优的模型。

The grate-kiln system for iron ore pellet induration has the characteristics of nonlinearity, high coupling and large time delay. Considering the errors between the assumption and realistic process, it is hard to build accurate kinetic models for this typical thermal process. Besides, the pure kinetic model has limitations to describe the induration ,process. In this paper, a hybrid model based on kinetic modeling is built using neural network ensemble. The Bagging method is used for training the sample set, and BP network is used as individual network. The results show that the hybrid model is more accurate and better than the kinetic model.