摘要
分析了介电谱技术测定石油产品质量指标的技术原理。利用主成分分析方法实现了车用润滑油单级油和多级油的分类。采用偏最小二乘法建立了介电谱技术定量快速预测润滑油40℃和100℃粘度的数学模型。其中主成分数达到7时,预测模型最优,40℃粘度预测值与实测值的相关系数为0.9996,预测误差平均值为1.054;100℃粘度预测值与实测值的相关系数为0.9984,预测误差平均值为0.0942。研究结果表明,介电谱技术能够准确、有效地评价油品的粘度指标。
Using dielectric spectroscopy （DS）to analyze petroleum products was introduced. Based on the principle component analysis method the differences between low grade and high grade automobile lube oils were identified. A model was set up using partial least squares（PLS）method to predict the viscosity values of lube oil at 40 ℃ and 100 ℃ rapidly. When the number of principle components reached 7, the predicted values and the actual measured data were in good agreement having correlation coefficient of 0. 999 6 and 0. 998 4 for 40 ℃ and 100 ℃, respectively; the average prediction error for 40 ℃ and 100 ℃ were 1. 054 and 0. 094 2,respectively. Results validated that the viscosity of lube oil could be measured by DS accurately and rapidly.
出处
《石油炼制与化工》
CAS
CSCD
北大核心
2008年第2期64-68,共5页
Petroleum Processing and Petrochemicals
关键词
介电谱
偏最小二乘
粘度
定量
分析方法
dielectric spectroscopy
partial least squares
viscosity
quantitative
analytical method
