模式识别技术在炼油数据分析中的应用

Pattern recognition technology application in data analysis of oil

为了能够研究模式识别技术在炼油数据分析中的应用,本文以催化裂化反应为例,以324个油样的34个输入变量(其中包括:原料基本物化性质数据、催化剂性质数据、装置操作数据、装置条件数据和加工量数据)和13个输出变量(催化裂化反应产物分布数据)为基础,结合模式识别技术判别分析方法:K-Nearest Neighbor Classification(KNN),建立催化裂化反应产物分布预测模型。同时,利用数学方法对模型的预测结果进行验证,结果表明:利用化学计量学模式识别技术建立的预测模型能够达到较好的预测效果,为模式识别技术在炼油数据分析中的应用提供实验支撑。

Reliable and efficient product yields estimation for unknown oils after fluid catalytic cracking （FCC） reaction is one of the key components in oil processing. This article describes the use of chemometric pattern recognition methods, k-nearest neighbor classification （k-NN）, for building classification models to determine the most similar oil sample to an unknown in a given data set, and to use the FCC yields record of the correspondent oil as the product yields prediction for the unknown sample under the same reaction conditions. Two sided t-test, correlation analysis were performed to assess the quality of the models. The work in this article provides laboratory evidence that k-NN techniques could be employed during heavy oil intelligent processing for FCC product yields estimation.