数据驱动框架下的非高斯批次过程最小熵性能评估算法

Entropy-based Performance Assessment of Batch Process with non-Gaussian Disturbances under Data-Driven Framework

针对一类带有非高斯干扰的批次过程控制系统,提出一种基于最小熵基准的性能评估算法。为了抑制非高斯噪声的干扰,本算法首次提出引用最小熵来作为批次过程控制系统性能评估的基准,同时考虑到批次过程控制系统的数学模型往往是复杂并难以确定,所以本算法引用了"黄金批次"的概念,突破传统研究中依赖模型的局限,提出一种新颖的数据驱动的评估不同批次性能的基准。通过批次化学反应堆的例子说明了所提出的算法的有效性。

This paper proposed an entropy-based performance assessment method for batch processes with non-Gaussian disturbances.Real batch processes have some intrinsic non-linear characteristics,which makes system output be non-Gaussian even though disturbances obey Gaussian distribution.In this case,minimum variance control(MVC)based performance assessment method is not applieable to batch processes.Consequently,in order to reject non-Gaussian disturbance,entropy is adopted to provide a unified controller performance index.At the same time,the mathematical model of the batch process is often too complicated to build,so the data-driven method is one of the hottest research topics in the area of modeling batch process.This paper referred to concept of “golden benchmark batch” and proposed a data driven entropy-based index for batch processes,which breaks through the limitation of traditional model-based method.The applicability of the proposed method for controller performance assessment of batch processes was illustrated through an industrial example.