以Spheripol工艺环管聚丙烯生产装置为应用背景,研究了计算机集成控制系统构架和系统设计问题,分析了聚丙烯生产装置计算机集成控制系统的主要技术内容.通过运用主元分析理论和神经元网络方法,研究了聚丙烯生产过程关键参数的在线软测量模型;采用多元统计监控理论的核密度估计器方法,实现了过程运行的状态检测与故障诊断;基于过程运行动态模型和非线性最优化理论,提出了装置连续运行过程的最优牌号切换策略并运用迭代动态规划算法加以数值求解.在以上研究的基础上给出了系统的总体结构,并对其进行了实时数据库和界面接口的设计,该设计充分考虑了控制系统的可扩展性;同时还给出了实时数据库与DCS的接口、实时数据库与人机界面接口和工艺控制软件与实时数据库接口3个方面的具体实现技术.
The framework design and implementatiog of the computer integrated control system for Spheripol propylene reactor.Based on principal component analysis(PCA) and artificial neural network,the online softsensors of key parameters of the process equipment were presented.The process monitoring and fault detection were investigated by using kernel density estimator(KDE) in multivariate statistical process monitoring(MSPM).According to the dynamic process model and nonlinear optimization theory,the optimal grade transition strategy in continuous polymerization process solved by iterative dynamic programming algorithm was proposed.The total system framework and software structure of the computer control system were designed with full consideration of extension of the developed system.The implementation of real-time database,user interface and data interchange with distributed control system(DCS) were also discussed.