低场核磁结合偏最小二乘回归预测含油污泥水和油含率  

Application of Low-field Nuclear Magnetic Resonance Combined Partial Least Square Regression in Simultaneous Prediction of Water and Oil in Oily Sludge

作　　者：郑晓园[1] 王波[1] 应芝[1] 池涌[2] 金余其[2] 

机构地区：[1]上海理工大学能源与动力工程学院,上海市动力工程多相流动与传热重点实验室,上海200093 [2]浙江大学能源清洁利用国家重点实验室,浙江杭州310027

出　　处：《分析测试学报》2016年第5期600-604,共5页Journal of Instrumental Analysis

基　　金：国家科技支撑计划项目(2012BAB09B01);上海市科委科研计划项目(13DZ2260900);上海高校青年教师培养资助计划(ZZslg15012)

摘　　要：快速准确分析处理过程中含油污泥的含水率和含油率有助于现场评价其原油回收效率和优化处理工艺参数。以Dean-Stark装置测定的含油污泥样品的含水率和含油率作为参考值,利用低场核磁共振结合偏最小二乘回归法建立了样品含水率和含油率校正集模型,考察了回波衰减曲线和横向弛豫时间T2曲线对校正集模型性能的影响。结果表明,采用前者建立的校正集模型性能优于后者;在此基础上,建立了31个样品的含水率和含油率通用校正集模型,其含水率和含油率模型的决定系数(R2)分别为0.965 7和0.978 5,校正标准差(RMSECV)分别为2.73%和2.22%。利用3个不同批次采集的HZ-OS样品对该模型进行验证,对于含水率和含油率模型,其验证集R2分别为0.914 1和0.924 7,预测标准差(RMSEP)分别为1.85%和2.04%,与RMSECV值比较接近,说明该模型的稳定性较好,可用于准确分析样品的含水率和含油率。In the process of oily sludge treatment, ter and oil contents rapidly and accurately, which it is important to simultaneously determine the wa- is helpful to assess the oil recovery efficiency from the oily sludge and to optimize the process parameters on site. In this study, the calibration models of oily sludge samples were developed by the low-field nuclear magnetic resonance （ LF NMR） together with the partial least square regression （PLSR） while the water and oil contents attained by Dean-Stark apparatus as reference values. Both echo decay curves and T2 distribution curves were employed for the model construction to investigate the performance of the calibration model. In the end, the gener- al calibration model with three sample sets was constructed. The results indicated that the model with echo decay curves is superior to that with T2 distribution curves. The general calibration model with echo decay curves was satisfactory with correlation coefficients （R^2） of 0. 965 7 and 0. 978 5 for the water and oil contents, respectively. The corresponding RMSECV were 2.73% and 2.22% . The 3 batches of samples were used to verity the general model. The R^2 of 0. 914 1 and 0. 924 7 was ob- tained for the water and oil contents, respcetively. The corresponding RMSEP of 1.85% and 2.04% was achieved, which was close to the RMSECV of 2.73% and 2.22%. The above results showed that the general model was steady and could meet the standards of accurate and rapid determination with the advantages of time-saving（less than 5 rain） , solvent-free, and non-invasion.

关 键 词：含油污泥 废物处理 核磁共振 含水率 含油率 主元分析 

分 类 号：O657.3[理学—分析化学] TE133.1[理学—化学]