化工生产数据具有多变量之间关联、非线性、非正态分布、高噪声等特点。由于数理统计的局限性，为摆脱假设束缚，采用探索性数据分析“问题—数据—分析—模型—结构”的逻辑，基于BMOS工业优化软件，利用聚类分析方法探索乙二醇氧化反应生产监控中采集的数据结构特点；通过主成分分析( PCA)算法提取特征参数，对乙二醇生产数据进行挖掘分析，得出优化方案。该方案为生产运行优化创造了条件。%Chemical product data have characteristics of correlation among multiple variables , non-linear, abnormal distribution, high noise, etc. Due to the limitations of mathematical statistics , and in order to get rid of the hypothesis, exploratory data are used to analyzed the logic of “problem-data-analysis-model-structure”. Based on DMOS industrial optimization software, the clustering analysis method is used to explore the features of structure of data collected in production monitoring of glycol oxidation. Through using principal component analysis (PCA) algorithm, the characteristics of data are extracted. The production data of Glycol oxidation are mining analyzed,and the optimization scheme is obtained. The scheme creates the conditions for optimizing productive operation.
展开▼
