硝酸盐是水中“三氮”（硝酸盐氮、 氨氮、 总氮）之一, 是反映水体受污染程度的一项重要指标。 传统 “现场采样-离线分析” 的硝酸盐化学检测方法操作繁琐、 耗时长, 难以满足现代水环境实时在线检测需求。 由于硝酸根在紫外区具有很强的紫外吸收特性, 并且紫外吸收光谱法具有简便快速、 可实现实时在线监测等特点, 近年来被广泛用于硝酸盐浓度的测量。 但使用紫外吸收光谱法检测水体硝酸盐含量时, 容易受到水体浊度影响, 造成谱线非线性抬升, 导致测量误差。 目前对浊度补偿算法的研究大都用于水中COD含量的检测, 对硝酸盐检测中浊度干扰去除研究较少。 为此提出一种基于一阶导数紫外吸收光谱的硝酸盐浓度测量方法, 该方法可以减小浊度干扰, 从而提高紫外光谱快速检测硝酸盐含量的准确度。 通过测量福尔马肼与硝酸钠标准溶液和它们混合溶液在190~300 nm波段的紫外吸收光谱并做一阶导数光谱处理, 处理后的光谱采用Savitzky-Golay滤波进行去噪平滑处理, 比较浊度与硝酸盐紫外吸收一阶导数光谱特征, 分波段研究浊度对硝酸盐紫外一阶导数光谱影响, 结果表明硝酸盐导数光谱在220~230 nm波段受浊度影响小; 选取220~230 nm波段作为光谱分析区间, 以30种不同浓度混合的福尔马肼与硝酸钠溶液作为训练样本, 利用偏最小二乘算法建立硝酸盐定量分析模型, 使用该建模模型预测剩下的6种不同浓度福尔马肼与硝酸钠混合溶液中硝酸盐的浓度, 结果表明福尔马肼干扰下硝酸盐测量结果的预测决定系数（correlation coefficient, R2）为0994 3, 预测均方根误差（root mean square error of prediction, RMSEP）为0346 9 mg·L-1。 为进一步验证该方法的准确性与稳定性, 使用该建模模型预测高岭土与硝酸钾配制的混合水样中硝酸盐的浓度, 结果表明该方法对高岭土干扰下硝酸盐测量结果的预测决定系数r2为0991 5, 预测均方根误差RMSEP为0362 8 mg·L-1。 综上所述, 提出的硝酸盐浓度紫外导数光谱检测方法, 采用220~230 nm波段的紫外导数光谱数据, 结合PLS建模, 可以快速准确测量在浊度干扰下水体硝酸盐的浓度, 为发展实际水体硝酸盐在线监测技术与设备提供方法基础。

Nitrate is one of the “three nitrogen” (nitrate nitrogen, ammonia nitrogen, total nitrogen) in water, and it is an important indicator to reflect the degree of pollution for water quality. The traditional method for measuring nitrate has many shortcomings, such as operational complexity, time consuming, and it is difficult to meet the real-time online detection requirements of modern water environment. Nitrate has strong UV absorption characteristics in the ultraviolet region. And in recent years, the UV absorption spectroscopy has been widely used for nitrate measurements for its real-time, low cost and easy operation. However, when the ultraviolet absorption spectrum is used to detect the concentration of nitrate, it is easily affected by the turbidity in water, causing the nonlinear lifting of the spectrum and measurement error. At present, the research on turbidity compensation algorithm is mostly used for the detection of the concentration of COD in water, and there are few studies on turbidity interference removal in nitrate detection. In that case, a method for measuring concentration of nitrate based on first derivative ultraviolet absorption spectrum is proposed to reduce turbidity interference and improve the accuracy of rapid detection of the concentration of nitrate. Ultraviolet absorption spectrum of the formalin and sodium nitrate standard solution and their mixed solution are measured in the region from 190 to 300 nm. Its spectrum is processed by first derivative method. In the meantime, Savitzky-Golay filtering is used to smooth the processed spectrum. After comparing the characteristics of turbidity and nitrate ultraviolet absorption gained by first derivative spectrum, studying the effect of turbidity on the first derivative spectrum of nitrate of different region, the results show that the effect of turbidity is small in the region from 220 to 230 nm. So this region is selected as the spectral analysis interval, and 30 kinds of concentrations of mixed solution of formazin and sodium nitrate solution are used as training samples. The partial least squares algorithm is used to establish the nitrate quantitative analysis model. This model is used to predict the concentration of nitrate in the remaining six different concentrations of formalin and sodium nitrate mixed solution. The results show that the predicted coefficient of determination of nitrate measurement is 0994 3 and RMSEP is 0346 9 mg·L-1 under the condition of formazin interference. In order to further verify the accuracy and stability of the method, the model was also used to predict the concentration of nitrate in mixed water samples prepared from kaolin and potassium nitrate. The results showed that the predicted coefficient of determination of nitrate measurement is 0991 5 and RMSEP is 0362 8 mg·L-1 under the condition of kaolin interference. In summary, the method to detect the concentration of nitrate by UV first derivative spectrum is proposed in this paper. The data from the UV derivative spectrum in the region from 220 to 230nm data was adapted, and PLS algorithm was combined. So we can measure the concentration of nitrate in water under turbidity interference quickly and accurately. Moreover, this study laid the foundation for further implementation of online analysis of actual water and further development of equipment.