摘要
偏最小二乘(Partial Least Square or Projection to Latent Structures,简称PLS)是一种广泛应用的多元统计方法,可以处理高维、相关度高的海量数据,是一种多元线性回归方法。本文介绍了PLS的发展历史、算法原理并重点介绍了目前在石油化工生产过程上几个比较有代表性的应用领域,包括在石油化工产品中的化学计量学、石油炼制过程监控和故障诊断以及反应动力学和工艺优化。其中,化学计量学领域的应用已较为成熟,而过程监控和故障诊断领域与反应动力学和工艺优化领域则更多停留在实验室研究阶段,应用较少。最后对PLS在石油化工生产中的应用前景做了展望。
Partial Least Squares (PLS) is a multivariate statistical method which is widely used in Science and Technology. It can form a linear relationship between the input data matrix X and output matrix Y, and both with many, noise, collinear, a high dimension and incomplete variables. This paper provides a brief introduction of its principal and history and an review of its application in petrochemical process. The application contains three parts: the application in chemometrics, process monitoring and fault diagnosis in petroleum processing and optimization petroleum refining process. The former one has many application examples in industry, while the latter two are more at the stage of simulation or Laboratory studies than industrial applications. A prospects of PLS in Petrochemical Production is provided in the end.
出处
《计算机与应用化学》
CAS
2016年第7期814-820,共7页
Computers and Applied Chemistry
关键词
偏最小二乘
石油化工
化学计量学
过程监控和故障诊断
Partial Least Squares
petrochemical process
chemometrics
process monitoring and fault diagnosis