Inferential Model for Industrial Polypropylene Melt Index Prediction with Embedded Priori Knowledge and Delay Estimation

Abstract

Melt index inferential model plays an important role in the control and optimization of polypropylene production. This study proposed a novel multiple-priori-knowledge based neural network (MPKNN) inferential model for melt index prediction. The prior knowledge from the industrial propylene polymerization process is fully exploited and embedded into the construction of multilayer perceptron neural network in the form of nonlinear constraints. Meanwhile, an adaptive PSO-SQP (particle swarm optimization-sequential quadratics programming) is proposed to optimize the network weights. The proposed MPKNN model has good fitting and prediction ability. Meanwhile, it can avoid unwanted zero value and wrong signal of the model gains. By embedding priori knowledge, the model ensures the safety in the quality control of melt index. In addition, a hybrid model combining the MPKNN model with a simplified mechanism model is proposed to enhance the extrapolation capability. A normalized mutual information method is employed to estimate the delay between independent variables and dependent variables. The proposed hybrid inferential model is validated on recorded data from an industrial double-loop propylene-polymerization reaction process.

Industrial melt index prediction with the ensemble anti‐outlier just‐in‐time Gaussian process regression modeling method

“…In contrast to traditional neural networks and SVR‐based deterministic modeling methods, the GPR‐based methods provided probabilistic information for prediction. Comparisons of the predicted variance values for the test samples ( ) with two online local modeling methods, ESVC–JGPR and SVC–JGPR, are shown in Figure .…”

Ensemble local kernel learning for online prediction of distributed product outputs in chemical processes

“…For example, the melt index is often considered as the product quality in polymerization processes (Mat Noor et al, 2010;Liu and Gao, 2015;Lou et al, 2012;Zhang et al, 2006). However, recent studies have shown the fact that the economic gain can be further improved by shaping the distributed output since it can significantly influence the product quality and process efficiency (Alhamad et al, 2005;Braatz and Hasebe, 2002;Chang and Liao, 1999;Nagy, 1999;Nagy and Braatz, 2012;Pigeon et al, 2011;Wang, 2000).…”

Optimal online soft sensor for product quality monitoring in propylene polymerization process

“…To date, different soft sensors have been developed for the prediction of the MI. Lou et al [5] proposed a novel multiple-priori-knowledge based neural network inferential model for melt index prediction. By embedding priori knowledge, the model ensured the safety in controlling the quality of melt index.…”