摘要
根据高频往复试验机测定的柴油样品的校正磨斑直径数据,建立了介电谱技术快速预测柴油润滑性能的模型。根据数据预处理的不同,分别建立了采用全谱数据筛选的模型1和小波分析数据预处理的模型2,考察了两模型预测结果的差别。结果表明,将介电谱数据进行3层"db1"小波分解,取其第3层高频系数建立的模型2的预测效果最为理想,预测结果与高频往复试验结果吻合较好,符合其再现性要求。
According to the data of the corrected wear diameter obtained by means of high-frequency reciprocation rig （HFRR）, prediction models for the lubricity of diesel fuels were established through dielectric spectroscopy. With the different data preprocessing, data selection of the whole spectrum and wavelet analysis data preprocessing were both tried to get two predictive models, called as Model 1 and Model 2. The predictive corrected wear diameters were calculated and compared to the measured ones. The results showed that Model 2 with the third layer detail coefficients of dielectric spectroscopy data in three layers ＂db1＂ wavelet decomposition worked best, with which the predictive results were consistent with those of the HFRR and met the requirement of the repeat error standard.
出处
《石油学报（石油加工）》
EI
CAS
CSCD
北大核心
2008年第4期420-425,共6页
Acta Petrolei Sinica(Petroleum Processing Section)
关键词
介电谱
高频往复试验机
小波分析
高频系数
dielectric spectroscopy
high-frequency reciprocation rig
wavelet analysis
detail coefficient
