论文部分内容阅读
本文分析了调优问题的数学实质,指出关键是建立数值仿真模型,并将建模的主要方法分为理论分析、经验机理分析和统计分析三类.作者考察了若干种多元统计分析方法的特点后,采用了一种将判别和聚类分析相结合的模式分类方法.作者用实际数据说明了复杂过程的时变特性,提出用渐进式调整方法,可达到自适应调优目标.本文指出在线自适应调优的实现需要专家知识的帮助,提出了将专家知识和仿真计算结合而构成专家仿真系统的一种模式.最后还报道了调优软件在工业装置上实际试运行取得的效果示例.
This paper analyzes the mathematical essence of the tuning problem and points out that the key is to establish a numerical simulation model and divide the main modeling methods into three categories: theoretical analysis, empirical mechanistic analysis and statistical analysis.The author examines the characteristics of several multivariate statistical analysis methods After adopting a pattern classification method combining discriminant analysis and cluster analysis, the author explains the time-varying characteristics of complex processes by using the actual data and proposes a gradual adjustment method to achieve adaptive tuning objectives.This paper points out that online The realization of adaptive tuning needs the help of expert knowledge and puts forward a model that combines expert knowledge and simulation calculation to form an expert simulation system.Finally, an example of the effect of tuning software on actual commissioning of industrial devices is also reported.