论文部分内容阅读
以汽油、煤油和柴油的混合溶液为研究对象,将柴油作为干扰物,分析混合溶液中汽油和煤油的含量,提出了一种通过分析三维荧光光谱数据测量油类混合物成分及含量的新方法。该方法结合展开偏最小二乘法(UPLS)与残差四线性分析(RQL)处理三维荧光一阶导数光谱。利用Savitzky-Golay多项式拟合微分法分别对油类混合物光谱数据的x轴和y轴求偏导,并将三维荧光光谱扩展为五维导数光谱,再通过U-PLS/RQL建立该四阶数据的校正模型,对预测样品进行分析,使光谱数据得到合理的分解与识别。预测相对误差减小到5.0%以下,预测精度高于三阶多元校正法。
Taking gasoline, kerosene and diesel mixed solution as the research object, the diesel oil was taken as the interferent and the content of gasoline and kerosene in the mixed solution was analyzed. A new method for measuring the composition and content of the oil mixture by analyzing the three-dimensional fluorescence spectrum data was proposed. This method combines the Unfolded Partial Least Squares (UPLS) with the Residual Quadrupole Analysis (RQL) to process three-dimensional fluorescence first derivative spectra. The Savitzky-Golay polynomial fitting differential method is used to derive the partial derivative of the x-axis and y-axis of the oil mixture spectral data, and the three-dimensional fluorescence spectrum is extended to five-dimensional derivative spectra, and then the fourth-order data is established by U-PLS / RQL The calibration model is used to analyze the predicted samples so that the spectral data can be reasonably decomposed and identified. The predicted relative error is reduced below 5.0%, and the prediction accuracy is higher than the third-order multivariate calibration.