生物柴油是典型的“绿色能源”, 具备良好的环保性和燃料特性, 通常与柴油混合使用在柴油发动机上。 但是目前世界各国柴油与生物柴油混合的比例标准参次不齐, 没有一个统一的标准, 并且不同比例的柴油/生物柴油混合物具有不同的燃烧性能, 也会对柴油发动机产生一定程度的影响。 为了能够快速、 准确的测量柴油/生物柴油混合物中的生物柴油浓度, 近红外光谱和拉曼光谱在燃油检测方面已经得到广泛的应用。 利用拉曼及近红外光谱对柴油/生物柴油混合物中的生物柴油浓度进行了量化分析研究。 首先采集了柴油/生物柴油混合燃油的拉曼光谱及近红外吸收光谱, 然后利用平滑、 基线校正、 归一化等方法对采集到的光谱进行预处理。 从光谱图中观察到, 在柴油/生物柴油混合物的拉曼光谱和近红外光谱中都有CO特征光谱区域, 且该光谱区域的光谱峰都随生物柴油的浓度增加而越来越明显。 拉曼光谱中, 随生物柴油浓度变化的主要CO特征光谱区域是在1 743 cm-1位置处的特征峰, 在近红外光谱中, 随生物柴油浓度变化的主要CO特征光谱区域是在4 659 cm-1处的特征峰。 然后分别根据强度比方法和偏最小二乘(PLS)回归方法建立了相应的混合燃油中生物柴油浓度预测模型。 结合强度比方法建立特征峰强度比的生物柴油浓度预测模型, 由混合燃油的拉曼光谱和近红外光谱建立的CO特征峰线性预测模型相关系数分别为0.947 2和0.996 2; 结合偏最小二乘(PLS)回归法建立特征光谱区域的生物柴油浓度预测模型, 由混合燃油的拉曼光谱和近红外光谱特征区域建立的相应预测集相关系数(R2)分别为0.981 5和0.991 2, 相应的预测均方根误差(RMSE)分别为0.093 7和0.012 9。 结果表明, 在混合燃油中, 使用近红外光谱中的CO光谱区域建立的生物柴油浓度预测模型会得到更准确的预测结果。

Biodiesel is a typical “green energy” with good environmental protection and fuel characteristics. It is usually mixed with diesel to use in diesel engines. However, there is no uniform standard for the blend proportion of diesel and biodiesel at present, and different proportions of diesel/biodiesel blend present different combustion performance, which also have a certain impact on diesel engines. In order to measure the concentration of biodiesel in diesel/biodiesel blend quickly and accurately, near infrared spectroscopy (NIR) and Raman spectroscopy have been used in fuel detection. In this paper, the concentration of biodiesel in diesel/biodiesel blends was quantitatively analyzed by using Raman spectroscopy combined with NIR. The Raman spectra and NIRs of diesel/biodiesel blends were measured firstly, and then the spectra were pre-processed by smoothing, baseline correction and normalization. The Raman and NIRs of diesel/biodiesel blend present CO characteristic spectral regions, and show corresponding trends with the increase of biodiesel concentration. The main variation of CO characteristic region with biodiesel concentration in Raman spectra is the characteristic peak at 1 743 cm-1, while the main variation of CO characteristic region with biodiesel concentration in NIRs is the characteristic peak at 4 659 cm-1. Afterwards, the concentration prediction models of biodiesel in mixed fuel based on the strength ratio method and partial least squares (PLS) regression method were established respectively. When using intensity ratio method to establish the biodiesel concentration prediction model in the characteristic peak, the correlation coefficients of CO characteristic peak linear prediction model established by Raman spectroscopy and near infrared spectroscopy were 0.947 2 and 0.996 2, respectively. When using partial least squares (PLS) regression method to establish the biodiesel concentration prediction model in the characteristic spectral region, the correlation coefficients (R2) of the prediction set established from the Raman and near infrared spectral characteristic regions of the blended fuel are 0.981 5 and 0.991 2 respectively, and the corresponding RMSE are 0.093 7 and 0.012 9 respectively. The results show that the biodiesel concentration prediction model based on the CO spectral region in near infrared spectroscopy can obtain more accurate prediction results in mixed fuel.