诱导期是评价汽油氧化安定性的重要指标。 国内燃油市场汽油流通量较大, 但目前仍采用传统GB/T 8018—87方法测定汽油诱导期, 工序繁琐, 耗时较长, 难以实现对每批油品及时检测, 无法满足汽油品质监管需求。 为此, 本文提出了一种采用傅里叶变换衰减全反射红外光谱(ATR-FTIR)测定汽油诱导期的快速分析方法。 设计和集成了一种采用傅里叶变换衰减全反射红外光谱的智能型专用燃油品质快速分析系统, 集光谱采集、 数据处理、 显示与存储等于一体。 设计的测样附件由9次反射的衰减全反射(ATR)硒化锌晶体和带密封的不锈钢盖组成, 具有光程恒定, 注样和清洗方便等优点。 研究了光谱扫描次数和重复装样次数对光谱信噪比的影响规律, 确定了最佳扫描次数为15次, 装样次数为4次。 从京津地区收集64个不同牌号、 不同生产时间的汽油样本, 采用GB/T 8018—87方法测定其诱导期数值作为建模参考数据。 使用该系统采集油样红外光谱。 对光谱数据进行均值中心化、 一阶求导预处理, 采用偏最小二乘PLS方法将汽油红外光谱与其诱导期数据进行关联, 建立车用汽油诱导期定量校正模型, 其决定系数(R2)0.897, 校正标准偏差(SEC)68.3 min, 预测标准偏差(SEP)91.9 min。 模型专用燃油品质快速分析系统预测汽油诱导期的相对偏差均小于5%, 满足GB方法重复性允差要求。 与传统GB方法相比, 该方法操作简单, 分析速度快, 单个样本检测仅需3 min, 对油品储运及销售过程中快速准确品质监管有积极意义。

A new method is proposed for the fast determination of the induction period of gasoline using Fourier transform attenuated total reflection infrared spectroscopy (ATR-FTIR). A dedicated analysis system with the function of spectral measurement, data processing, display and storage was designed and integrated using a Fourier transform infrared spectrometer module and chemometric software. The sample presentation accessory designed which has advantages of constant optical path, convenient sample injection and cleaning is composed of a nine times reflection attenuated total reflectance (ATR) crystal of zinc selenide (ZnSe) coated with a diamond film and a stainless steel lid with sealing device. The influence of spectral scanning number and repeated sample loading times on the spectral signal-to-noise ratio was studied. The optimum spectral scanning number is 15 times and the optimum sample loading number is 4 times. Sixty four different gasoline samples were collected from the Beijing-Tianjin area and the induction period values were determined as reference data by standard method GB/T 8018—87. The infrared spectra of these samples were collected in the operating condition mentioned above using the dedicated fast analysis system. Spectra were pretreated using mean centering and 1st derivative to reduce the influence of spectral noise and baseline shift. A PLS calibration model for the induction period was established by correlating the known induction period values of the samples with their spectra. The correlation coefficient (R2), standard error of calibration (SEC) and standard error of prediction (SEP) of the model are 0.897, 68.3 and 91.9 minutes, respectively. The relative deviation of the model for gasoline induction period prediction is less than 5%, which meets the requirements of repeatability tolerance in GB method. The new method is simple and fast. It takes no more than 3 minutes to detect one sample. Therefore, the method is feasible for implementing fast determination of gasoline induction period, and of a positive meaning in the evaluation of fuel quality.