基于多尺度加权主成分分析的SF6红外光谱分析

利用红外光谱法分析SF6气体及其衍生物是判断气体绝缘组合电器(GIS)运行状态和故障的一种重要手段。传统的诊断方法过程繁琐、效率低下,而且受主观因素的影响较大。本文指出可以采用机器学习的方法实现GIS设备的故障诊断,并提出了多尺度加权主成分分析的特征提取方法。多尺度加权主成分分析结合了主成分分析和多尺度分解的特点,保证了尺度特征信息的最大化,并且修正了特征向量在数据分类时的权重。通过对广西电力研究院提供的SF6及其衍生物的红外光谱进行分析,证明了多尺度加权主成分分析算法对训练样本的分类效果要比标准的主成分分析算法好3～4倍。     

Membrane gas diffusion measurements with MRI

Gas transport across polymeric membranes is fundamental to many filtering and separation technologies. To elucidate transport mechanisms, and understand the behaviors of membrane materials, accurate measurement of transport properties is required. We report a new magnetic resonance imaging (MRI) methodology to measure membrane gas phase diffusion coefficients. The MRI challenges of low spin density and short gas phase relaxation times, especially for hydrogen gas, have been successfully overcome with a modified one-dimensional, single-point ramped imaging with T(1) enhancement, measurement. We have measured the diffusion coefficients of both hydrogen gas and sulfur-hexafluoride in a model polymeric membrane of potential interest as a gas separator in metal hydride batteries. The experimental apparatus is a modified one-dimensional diaphragm cell which permits measurement of the diffusion coefficient in experimental times of less than 1 min. The H(2) gas diffusion coefficient in the membrane was 0.54 +/- 0.01 mm(2)/s, while that of sulfur-hexafluoride was 0.14 +/- 0.01 mm(2)/s, at ambient conditions.