Abstract This short communication presents a proper detection of illegal additives in Brazilian S-10/common diesel B7/5 using mid-infrared (MIR) spectroscopy and data driven soft independent modeling of class analogy (DD-SIMCA). The one-class classification chemometric model based on DD-SIMCA with MIR spectroscopic data was applied in the following analytical systems: (a) authentic Brazilian S-10 diesel B7 (BX is the amount of biodiesel blended) used as members of the unique target class and Brazilian S-10 diesel B7 adulterated by residual automotive lubricant oil (RALO) and residual solvent used in a dry wash (RSUDW) as non-members of the target class and (b) authentic Brazilian common diesel B5 used as a unique target class and Brazilian common diesel B5 adulterated by RALO, soybean oil and used frying oil (UFO) and contaminated with gasoline. DD-SIMCA model in data set described in (a) was able to properly classify all samples of Brazilian S-10 diesel B7 adulterated as non-members of the target class in the validation phase with a specificity of 100% and also all samples of authentic Brazilian S-10 diesel B7 were correctly accepted as members of the target class with test sensitivity equal 100%, although in the training phase one of the samples was rejected at 95% tolerance of the acceptance area, achieving training sensitivity equal 95%. In the data set described in (b) the modern DD-SIMCA model also achieved excellent results with 100% training sensitivity, test sensitivity and specificity. Finally, multivariate curve resolution-alternating least-squares (MCR-ALS) with the area correlation constraint and first-order MIR spectroscopic data was able to provide pure profiles such as: (i) concentration profiles allowed to quantify the amount of Jatropha biodiesels added in Brazilian S-10 diesel and (ii) spectral profiles allowed to identify the feedstock used in biodiesel production present in the biodiesel/diesel blend, which can provide further investigation for the European Union during the implementation of EU 2015/1513.

中文翻译：

---

根据 EU 2015/1513，在相关约束下使用 DD-SIMCA 和 MCR-ALS 的 MIR 光谱检测巴西 S-10/普通柴油 B7/5 中的非法添加剂和与柴油混合的麻风树生物柴油的量化

摘要 这篇简短的通信介绍了使用中红外 (MIR) 光谱和数据驱动的类类类推软独立建模 (DD-SIMCA) 正确检测巴西 S-10/普通柴油 B7/5 中的非法添加剂。基于 DD-SIMCA 和 MIR 光谱数据的一类分类化学计量模型应用于以下分析系统： (a) 正宗的巴西 S-10 柴油 B7（BX 是混合生物柴油的量）作为唯一目标的成员类和巴西 S-10 柴油 B7 掺有残留汽车润滑油 (RALO) 和干洗中使用的残留溶剂 (RSUDW) 作为非目标类成员，以及 (b) 用作独特目标的正宗巴西普通柴油 B5类和巴西普通柴油 B5 掺有 RALO、大豆油和使用过的煎炸油 (UFO)，并被汽油污染。(a) 中描述的数据集中的 DD-SIMCA 模型能够正确分类在验证阶段掺假为非目标类别成员的巴西 S-10 柴油 B7 的所有样品，特异性为 100%正宗的巴西 S-10 柴油 B7 被正确地接受为目标类别的成员，测试灵敏度等于 100%，尽管在训练阶段，其中一个样本在接受区域的 95% 容差下被拒绝，实现了等于 95% 的训练灵敏度。在（b）中描述的数据集中，现代 DD-SIMCA 模型也以 100% 的训练灵敏度、测试灵敏度和特异性取得了优异的结果。最后，具有面积相关约束和一阶 MIR 光谱数据的多元曲线分辨率交替最小二乘法 (MCR-ALS) 能够提供纯剖面，例如：