铅锌烧结过程综合透气性的集成预测模型

Integrated prediction model for synthetic permeability in lead-zinc sintering process

针对铅锌烧结过程透气性的预测具有模型不确定性和输入变量不确定性等特点,建立了综合透气性智能集成预测模型。首先建立了基于满意聚类的T-S综合透气性预测模型,针对聚类后各子模型结论参数的辨识工作计算复杂、容易陷入局部极值的问题,将混合粒子群优化算法用于这些结论参数的辨识;然后利用灰色理论建立了时间序列综合透气性预测模型;最后利用信息熵技术将2个预测模型进行集成,以获得集成预测模型。选取实际生产过程中100组合格的数据,分别用以上3种预测模型来预测相应的综合透气性,其相对误差的平均值分别为2.1%,3.2%,1.8%。实验结果表明,本文提出的集成预测方法能够有效地克服不确定性带来的影响、提高综合透气性的预测精度。

To deal with the uncertainty exist in the prediction model and the input parameters, an integrated prediction model for the synthetic permeability in lead-zinc sintering process is proposed in this paper. Firstly, a T-S prediction model based on satisfactory clustering for synthetic permeability is established. To avoid the problem of computing complexity and liability of falling into local minimum, a hybrid particle swarm optimization algorithm is applied to identify the consequent parameters of each sub-model produced by clustering. Then, a prediction model based on time sequence for synthetic permeability is built by using grey theory. Finally, the two prediction models are integrated by using the information entropy theory. 100 groups of qualified practical data are selected to predict the synthetic permeability. The average value of relative error of the three models mentioned above are 2.1% , 3.2% , 1.8% respectively. Experiment results show that the proposed integrated method can overcome the uncertainty and improve the prediction precision for the synthetic permeability.