间歇过程统计监控的OFA同步化方法及其改进

OFA Synchronize Method and Its Improvement for Batch Process

在实际的间歇生产过程中,由于各种原因会导致各批次运行时间的不一致,而无法直接建立有效的统计监控模型。应用正交函数近似(OFA)法,通过把原始数据在正交基上进行投影,并用相应的投影系数来代替原始数据所具有的特征,可以达到轨迹同步化和压缩数据量的目的。本文对OFA法的可行性进行了仿真验证并提出了一种改进的OFA同步化方法,使得该方法可以更快速而有效地处理运行时间长、采样样本大的间歇过程,而且可以提高监控的实时性以适用于某些具有实时性要求的系统过程监控中。

In practice, the effective statistical model can not be built directly because of the runtime difference among the batches caused many resons. By Orthonormal Function Approximation （OFA） method, the orthogonal data are projected base on the orthonormal base, and using the projection coefficient to express the characteristics of the original data we can synchronize the trajectories of each historical batch and reduce the dimension. In this paper, the 0FA method is validated and some improvement on the OFA is made. Using the improved OFA, it possible to deal with deal with the batch process with long time running, large sampling and to deal with the real time requiring systems more efficiently.