论文部分内容阅读
针对大型聚乙烯工业装置质量指标实时估计和牌号切换的复杂性,基于乙烯聚合原理推导了大型聚乙烯工业装置质量指标实时预测模型,利用模块反推方法推导了参数更新律,提出了一种渐近跟踪状态观测器设计方法用于根据实验室分析数据反馈修正质量指标并实时估计模型参数。参数更新律设计采用新型神经动力学方法实时求解,通过引入切换修正保证了参数更新律的鲁棒性,选取足够大的增益矩阵可使观测器渐近跟踪系统状态。所提方法在大型聚乙烯工业装置上的应用结果证实了其有效性和可行性,为实现大型聚乙烯工业装置先进控制奠定了基础。
According to the real-time estimation of the quality index of large-scale polyethylene industrial plant and the complexity of brand switching, a real-time prediction model of quality index of large-scale polyethylene industrial plant was deduced based on the principle of ethylene polymerization. The parameter updating law was deduced by using the module inverse method. The near-tracking state observer design approach is used to correct quality indicators based on laboratory analysis data and estimate model parameters in real time. The parameter renewal law is designed in real time using a new neuro-dynamics method. The robustness of the parameter updating law is guaranteed by introducing the switching modification. Selecting a sufficiently large gain matrix allows the observer to track the system asymptotically. The application of the proposed method to large-scale polyethylene industrial installations has confirmed its effectiveness and feasibility and laid the foundation for the advanced control of large-scale polyethylene industrial plants.