硝酸盐氮是水环境中监测中重要的污染指标之一, 通过紫外吸收光谱可以快速无污染地对该污染物进行检测。 针对紫外吸收光谱容易受到浊度干扰的这一情况, 通过实验方法分析了福尔马肼浊度标准液对硝酸盐氮标准液的紫外吸收光谱的影响, 基于此提出了补偿曲线法的浊度补偿方法对硝酸盐氮的紫外吸收光谱进行补偿校正, 并通过实验对该方法进行了验证, 验证结果良好。 首先, 在实验室通过紫外光谱采集测试系统采集了浓度为0.2~10 mg·L-1的12组硝酸盐氮标准液、 5~50 NTU的10组福尔马肼浊度液、 以及福尔马肼浊度液与硝酸盐氮的混合溶液的紫外吸收光谱。 理论上, 根据朗伯-比尔定律, 混合溶液的吸光度应该等于不同溶质吸光度的叠加, 但是通过实验分析, 混合溶液在硝酸盐氮的主要吸收谱区的吸光度并不等于硝酸盐氮和浊度吸光度之和, 这是因为浊度颗粒打破了硝酸盐氮分子的共面性, 造成空间位阻, 使共轭体系被破坏, 导致硝酸盐氮吸光度降低。 因此引入了在0～1之间的补偿系数kN(λ)用来表征浊度对硝酸盐氮吸收谱的影响, 当kN(λ)越接近0时, 表明浊度在此波长处对硝酸盐氮的吸光度影响越大。 根据实验测量的光谱数据求出不同浊度在硝酸盐氮主要吸收谱区的补偿系数, 即可得到不同浊度的补偿曲线。 通过实验分析, 350~400 nm波段的硝酸盐氮吸光度基本为0, 混合溶液吸光度只与浊度相关, 且两者的吸光度基本相同, 因此可以选择此波段的光谱积分来建立浊度回归模型, 从而解算混合溶液的浊度值。 相比于单个波长的建模回归, 该光谱积分回归模型的稳定性好, 不容易受到其他因素干扰。 浊度解算模型的相关系数r的平方为0.998 5, 解算效果较好, 得到浊度值之后即可进行浊度补偿。 通过实验对该补偿方法进行了验证, 并与单波长的浊度补偿与未进行补偿时进行了对比。 验证结果表明, 补偿曲线法进行浊度补偿后, 建立偏最小二乘(PLS)算法的硝酸盐氮预测模型, 预测均方根误差(RMSEP)为0.124, 预测值与真实值的平均绝对误差(MAE)为5.3%, 补偿效果很好, 其他两种都会发生很大偏差。 相比之下, 该文提出的浊度补偿方法效果明显优于其他两种, 此方法可以为硝酸盐氮紫外吸收光谱的浊度补偿提供有效的技术参考。

Nitrate nitrogen is one of the most important pollution indicators in water environment monitoring and can be detected quicklywithout pollution by absorption spectroscopy. In view of the fact that the ultraviolet absorption spectrum is susceptible to turbidity interference, the influence of formaldehyde turbidity standard solution on the ultraviolet absorption spectrum of nitrate nitrogen standard solution is analyzed by experimental method. Based on this, a turbidity compensation method based on compensation curve method is proposed to compensate and correct the ultraviolet absorption spectrum. Then this method is validated by experiments, and the results are good. Firstly, the ultraviolet absorption spectra of 12 groups of nitrate nitrogen standard solution with concentration of 0.2～10 mg·L-1, 10 groups of formaldehyde turbidity solution with concentration of 5～50 NTU and formaldehyde turbidity solution mixed with nitrate nitrogen solution were collected in the laboratory. Theoretically, according to Lambert’s law, the absorbance of the solution should be the superposition of the absorbance of different solutes, but through the experimental analysis, the absorbance of the mixed solution in the main absorption spectrum region of nitrate nitrogen was not equal to the sum of the absorbance of nitrate nitrogen and turbidity. This was because turbidity particles break the coplanarity of nitrate nitrogen molecules, resulting in steric hindrance, which destroys the conjugate system and leads to the decrease of nitrate nitrogen absorbance. Therefore, the compensation coefficient between 0 and 1 was introduced to characterize the effect of turbidity on the absorption spectrum of nitrate nitrogen. The closer to 0, the greater the influence of turbidity on the absorption spectrum of nitrate nitrogen at this wavelength. Based on the measured spectral data, the compensation coefficients of different turbidities in the main absorption spectral region of nitrate nitrogen can be obtained. According to the experimental analysis, the absorbance of nitrate nitrogen in 350～400 nm band is basically 0. The absorbance of mixed solution is only related to turbidity, and the absorbance of both is basically the same. Therefore, the spectral integral of this band can be selected to establish the turbidity regression model, and the turbidity value of mixed solution can be calculated. Compared with single wavelength regression, the spectral integral regression model has good stability and is not easily disturbed by other factors. The square of correlation coefficient R of turbidity calculation model is 0.998 5, and the calculation effect is good. Turbidity compensation can be carried out after the turbidity value is obtained. The compensation method is validated by experiments and compared with single wavelength turbidity compensation and uncompensated of turbidity. The validation results show that after turbidity compensation by compensation curve method, the nitrate nitrogen prediction model based on partial least squares (PLS) algorithm is established. The RMSEP is 0.124 and the average absolute error (MAE) between the predicted value and the real value is 5.3%. The compensation effect is pretty good, and the other two methods will deviate greatly. In contrast, the turbidity compensation method proposed in this paper is obviously better than the other two. This method can provide an effective technical reference for the turbidity compensation of nitrate nitrogen ultraviolet absorption spectrum.