基于LPP的某型柴油机润滑油光谱数据分析  

Spectrographic oil analysis of one diesel engine based on locality preserving projections

作　　者：李婧[1] 田洪祥[1] 孙云岭[1] 张帅[1] LI Jing;TIAN Hong-xiang;SUN Yun-ling;ZHANG Shuai(College of Power Engineering,Naval Univ.of Engineering,Wuhan 430033,China)

机构地区：[1]海军工程大学动力工程学院,武汉430033

出　　处：《海军工程大学学报》2018年第4期59-61,70,共4页Journal of Naval University of Engineering

基　　金：湖北省自然科学基金资助项目(2010CDB01505)

摘　　要：为解决通过光谱数据进行工况识别困难的问题,提出了一种基于局部保留投影算法(LPP)对柴油机原子发射光谱数据降维的新方法,并对降维后的数据进行了聚类分析。结合某型柴油机实验台架,通过改变汽缸套和活塞间隙,制定了7种磨合工况,获得69个润滑油样本的原子发射光谱仪数据。利用上述方法,能较为有效地将不同工况下的油样聚类。实验结果证明了该方法在润滑油光谱分析中的有效性。In order to solve the difficult condition identification by spectral data,a method of locality projections(LPP)algorithm is proposed to reduce the dimension of atomic emission spectrum data of diesel engine,and then the above data is clustered.The samples of 69 lubricating oil spectral data are obtained from the experimental bench of a certain diesel engine,and seven running conditions are formulated by changing the clearance between the cylinder sleeve and the piston.Oil samples can be clustered effectively under different working conditions by using the above method.The experimental results show that the method is effective in the spectral analysis of lubricating oil.

关 键 词：柴油机 局部保留投影算法 聚类 光谱分析 润滑油 

分 类 号：O657.3[理学—分析化学]