三维荧光光谱技术结合线性支持向量算法在水体有机污染监测中的应用

针对当前地表水体有机污染的原位快速监测需求，提出一种基于三维荧光光谱技术的水质指标预测模型和水质等级快速判断方法。以扬州市域内多种地表水体的水质监测数据作为模型训练样本，充分利用水体三维荧光光谱信息，结合线性支持向量回归算法（LIBLINEAR），建立了与化学需氧量（COD_Cr）、高锰酸盐指数（COD_Mn）、氨氮（NH₃-N）、总磷（TP）、总氮（TN）和五日生化需氧量（BOD₅）6项有机污染相关水质指标的预测模型。研究结果表明，6项指标预测模型的训练集和测试集决定系数R2均大于0.73，预测值与国标及行业标准方法分析结果的相关系数r达到0.9以上。利用水质指标预测结果进一步判断有机污染指标相关水质等级，黑臭水体识别率达86%，对Ⅲ类~重度黑臭共6个水质等级的分类准确率为60%。结果说明该方法通过水体三维荧光光谱信息预测水质有机污染指标具有较好的准确性和精度，为广域时空尺度地表水的高效原位监测提供了一种新的解决方案。

Application of Excitation-Emission Matrix (EEM) Fluorescence Combined With Linear SVM in Organic Pollution Monitoring of Water

In view of the increasingly serious organic pollution of urban waterbodies, this paper proposes a water quality indexes prediction model based on excitation-emission matrix (EEM) fluorescence technology and a method for quickly judging the water quality category. In this study, a large number of diversified surface waters around Yangzhoucity were taken as the training sample of the model. Based on the EEM spectrum of water and linear support vector regression (LIBLINEAR), the prediction models of six water quality indexes were established, including chemical oxygen demand (CODCr) and permanganate index (CODMn) , ammonia nitrogen (NH₃-N), total phosphorus (TP), total nitrogen (TN) and five-day biochemical oxygen demand (BOD5). The test results show that the determination coefficient R2 of the training set and the test set of the six index prediction models are both greater than 73%, while the correlation coefficient between the predicted value and analysis results by the national standard and industry-standard methods is greater than 0.9. Base on the prediction results of the water quality index, the water quality category could be the further judge. The recognition rate of black-odor waterbody reached 86%, and the classification accuracy rate of water bodies above category Ⅲ was 60%. The results show that the method has good accuracy and precision in predicting the water quality index through the three-dimensional fluorescence spectrum information of the waterbodies, which provides a solution for the efficient in-situ monitoring and rapid classification of water quality of urban and surrounding surface water.