基于机理 AI 混合建模技术的 Novolen 工艺聚丙烯质量预测模型开发与应用

打开文本图片集

DOI：10.19981/j.CN23-1581/G3.2026.12.001

中图分类号：F426

文献标志码：A

文章编号：2095-2945（2026）12-0001-04

Abstract： Novolen gas-phase process is widely used in the production of polypropylene products， which can achieve full range of polypropylene product coverage. During the polypropylene production process， the ratio of various materials， reactor temperature， catalyst flow rate， etc. have a great impact on the quality of the product. Due to the lack of a quality prediction model with strong generalization and interpretation， the quality of polypropylene products is difficult to control. A polypropylene quality prediction model is built by designing a mechanism-based AI hybrid technology， taking material ratio， reactor temperature， catalyst flow rate， etc. as mechanism inputs， and screening those with high correlation with polypropylene quality as noise inputs outside the mechanism. Based on historical data fitting， a polypropylene quality prediction model with strong generalization and good interpretation is achieved. The polypropylene quality prediction model combines advanced process control theory to build an automated， safe and stable advanced process system.

Keywords： Novolen gas-phase method; process mechanism; artificial neural network; loss function; model development

1967年，Basf公司首次实现Novolen气相法工艺工业化，该工艺采用串联或并联立式搅拌床反应器，可覆盖全范围聚丙烯产品，在国内外广泛应用，国内已形成较大规模[1]。（剩余5879字）