针对荧光峰值法浓度测量中最佳激发和发射波长的选择问题, 提出基于三维荧光光谱的多峰线性回归参数比较法。以市售93号和97号汽油为例, 首先利用FS920型稳态荧光光谱仪扫描纯水和纯油品的三维荧光光谱, 确定特征荧光峰峰位和强度; 其次按不同体积比分别配置7种浓度的油水混合液样品, 提取不同样品的荧光峰峰位、强度等特征参量, 发现浓度对荧光峰特征参量有较大干扰; 然后用排除法确定了93号和97号油水混合液样品分别有23个和22个特征峰; 最后采用最小二乘法对每个峰的强度与浓度作线性回归, 通过对校正决定系数、F值、灵敏度和截距等参数的比较得出, 93号和97号油水混合液的最佳激发/发射波长分别为275/302 nm和285/322 nm。实验结果表明, 根据最佳峰校正模型, 93号和97号汽油浓度测量的绝对误差分别为0.006‰、0.007‰, 比传统的最强峰校正模型精度高1个数量级。

For selecting optimal fluorescence excitation and emission wavelength of peak value method of concentration measurement, a multiple peaks linear regression parameters comparison method based on 3D fluorescence spectra was proposed. Taking 93# and 97# motor gasoline in the market for examples, firstly, 3D fluorescence spectra of purified water and motor gasoline from FS920 steady-state fluorescence spectrometers were analyzed and the characteristic fluorescence peaks were found. Secondly, 7 kinds of mixed liquors of 93# and 97# gasoline and water were prepared by volume ratio, respectively. Peak position and intensity etc. parameters of different sample were extracted. Results show that the characteristic fluorescence peak parameters are different for the different concentration. And then, for the mixed liquors of 93# and 97# gasoline and water, 23 peaks and 22 peaks were found by the exclusive method, respectively. Finally, the peak intensity and concentration data of each sample were fit by using the least square method. The comparison of adj. R-Square, F value, sensitivity and intercept, etc. shows that the optimal excitation/emission wavelength of 93# and 97# gasoline sample is 275/302 nm and 285/322 nm, respectively. Experimental results show that the absolute error of 93# and 97# concentration is 0.006‰ and 0.007‰, respectively. The precision of the calibration model founded by the optimal peak is one order higher than the traditional calibration model founded by the strongest peak.发 光 学 报第38卷 第10期肖长江, 等: 三维荧光光谱多峰校正测量水中汽油浓度