摘要
对模拟溢油的石油类样品进行近红外光谱检测并进行了主成分分析,以主成分得分作为聚类的特征变量进行模糊聚类,采用近红外光谱测量结合主成分分析和模糊聚类法快速识别模拟溢油种类,并建立了样品聚类与识别模型。采用此模型对柴油、汽油和润滑油等334个与海水体积分数≥4∶10 000的模拟溢油样品进行了准确快速的分类和识别,为近红外光谱技术判别实际溢油样品提供了科学依据。
In this paper, the near-infrared spectral measurement and principal component analysis （PCA） of simulated spilled petroleum samples are made, the princpal component scores are taken as the cluster characteristic variables for making fuzzy clustering, the near-infrared spectrometry and PCA together with fuzzy clustering method are then used for rapid recognition of simulated spilled oils, and the sample clustering and recognition patterns are established. The patterns are used to make correct and repid classification and recognition of 340 simulated spilled oil samples of diesel fuel, gasoline and lube oil with their volumet- ric fraction of seawater greater than 4 ： 10 000, which provides scientific basis for using near-infrared spectrometry to discriminate the real spilled oil samples.
出处
《海洋科学进展》
CAS
CSCD
北大核心
2007年第1期91-94,共4页
Advances in Marine Science
基金
国家高技术研究发展计划项目--海洋石油污染物现场实时监测的光纤近红外传感仪器系统(2001AA635040)
关键词
溢油鉴别
近红外光谱法
主成分分析
模糊聚类
模式识别
spilled oil identification
near-infrared spectrometry
principal component analysis （PCA）
fuzzy clustering
pattern recognition
