论文部分内容阅读
针对聚丙烯装置熔融指数软测量中的非线性和多工况切换操作问题,提出1种基于卡尔曼滤波-正交最小二乘(Kalman-OLS)的非线性自适应软测量方法。通过对聚丙烯装置反应系统进行机理分析,采用正交最小二乘方法(OLS)来拟和辅助变量和熔融指数之间的非线性关系。OLS方法的优化目标函数中同时考虑基于留一法均方误差(LOO MSE)和模型参数的局部正则化(LR),以提高模型的稀疏性和泛化能力。为适应装置多工况操作的现状,进一步提出使用Kalman滤波器对OLS模型参数进行自适应更新。工业数据应用结果表明,提出的Kal-man-OLS方法能够比偏最小二乘(PLS)、OLS方法更准确的预报聚丙烯熔融指数的变化。
Aiming at the problem of non-linear and multi-working switching in the melt index soft measurement of polypropylene plant, a nonlinear adaptive soft-sensing method based on Kalman filter and orthogonal least squares (Kalman-OLS) is proposed. Through the mechanism analysis of the reaction system of polypropylene plant, orthogonal nonlinear least squares method (OLS) was used to fit the nonlinear relationship between the auxiliary variables and the melt index. The optimization objective function of OLS method considers both local residuals (LOO MSE) and local regularization (LR) based on model parameters to improve the sparsity and generalization ability of the model. In order to adapt to the status quo of multi-condition operation, the Kalman filter is further proposed to adaptively update OLS model parameters. The results of industrial data application show that the proposed Kal-man-OLS method can predict the change of polypropylene melt index more accurately than partial least squares (PLS) and OLS methods.