基于LSSVR的内燃动车组磨损状态监测

本资料包含pdf文件1个，下载需要1积分

The running-in wear state of the diesel locomotive could be judged by the direct reading and atomic emission spectral analysis on the lubricating oil, and reliable basis for the maintenance of ferrography analysis locomotive was pro- vided. The working condition of the diesel multiple units was almost the same, but the two running-in wear state of diesel multiple units （DMUs）were different. Since the oil analysis data were non-stationary and non-equal time-interval time se- ries in general, which were difficult to analyze. The least squares support vector regression（LSSVR） was used to model the small particles of oil direct reading ferrography data and Pb concentration of atomic emission spectral data of the DMUs. The oil measurement data of the two ends of the DMUs and their difference were respectively modeled by LSSVR. It is easy to observe the wear trend and the wear condition difference between two ends of the DMUs,which has certain guiding significance for the wear condition monitoring of the DMUs.油液直读铁谱和原子发射光谱分析有助于判断内燃机车柴油机的磨合磨损情况，为进行机车检修提供可靠的依据。内燃动车组在运用中条件几近相同，但两动车的磨合磨损情况不同。由于油液分析数据通常是非平稳且非等时间间隔时序，给分析工作带来困难。本文利用最小二乘支持向量回归机（LSSVR）对动车组直读铁谱小磨粒数数据和光谱分析铅元素质量分数数据进行建模，对内燃动车组两端动车的油液测量数以及其差值分别进行LSSVR拟合，能够容易看出动车磨损的变化趋势以及两端动车磨损状况的差异，对内燃动车组磨损状态的监测具有一定的指导意义。