论文部分内容阅读
熔融指数是聚丙烯生产过程中重要的质量指标,但难以实现在线实时测量,从而无法实现有效的聚丙烯质量控制。针对聚丙烯生产过程的非线性特性,提出了一种基于核主元分析和组合神经网络(KPCA-SNNs)的软测量建模方法,并将该建模方法应用于聚丙烯软测量研究中。首先利用具有较强非线性特征提取能力的核主元分析对样本数据进行前期处理,然后将其结果作为软测量模型的输入,最后采用具有预测精度高、泛化能力强的组合神经网络(SNNs)建模方法建立软测量模型。仿真结果表明,与单纯SNNs相比,其中均方误差(MSE)由0.0035下降至0.0029,平均绝对误差(MAE)由0.0182下降至0.0139,平均绝对百分百误差(MAPE)由0.0057下降至0.0044。所以采用KPCA-SNNs所建立的聚丙烯熔融指数软测量模型具有更好的鲁棒性和预测精度,满足了聚丙烯生产工艺要求,较好地解决了聚丙烯生产过程中重要质量指标无法在线估计的难题。
Melt index is an important quality indicator in the production of polypropylene, but it is difficult to achieve online real-time measurement, which can not achieve effective polypropylene quality control. Aiming at the non-linearity of polypropylene production process, a soft-sensing modeling method based on kernel principal component analysis (PCA) and combined neural network (KPCA-SNNs) is proposed. The modeling method is applied to the soft-sensing measurement of polypropylene. First, the sample data are preprocessed by using KPCA with strong nonlinear feature extraction ability. Then, the result is used as the input of the soft-sensing model. Finally, a combined neural network (SNNs) with high prediction accuracy and generalization ability is adopted. Modeling Methods Establish a soft-sensing model. The simulation results showed that the mean square error (MSE) decreased from 0.0035 to 0.0029, the mean absolute error (MAE) decreased from 0.0182 to 0.0139, and the mean absolute percentage error (MAPE) decreased from 0.0057 to 0.0044 compared with SNNs. Therefore, the polypropylene softness index soft-sensing model established by KPCA-SNNs has better robustness and predictive accuracy, meets the requirements of polypropylene production process and better solves the problem that the important quality indexes of polypropylene can not be estimated online The problem