不同展开方法在间歇过程故障检测中的应用

Fault Detection for Batch Process Using Different Unfolding Method

间歇过程数据是一个典型的三维数据形式,数据的展开方法在一定程度上影响了所建立的统计模型的精确度。针对这一问题,提出了基于不同展开方式上的核独立元分析(KernelICA)的在线故障检测方法,并应用于青霉素生产过程的数据分析中。仿真结果表明,与传统的在批次方向展开的建模方法相比,所提出的方法大大降低了故障的漏报率,具有更好的故障检测性能。

Batch process data set is a typical three-way array,and the unfolding method impacts the accuracy of statistical models in some extent.In this paper,an online fault detection strategy for batch process that uses different unfolding way and kernel independent component analysis（Kernel ICA） is proposed,and it was applied to data analysis in the simulation benchmark of fed-batch Penicillin production.Simulation results demonstrate the power and advantage of the proposed method in comparison to traditional batch wise modeling method,and the lower missed detection rate is obtained.