基于基线校正和主元分析的紫外-可见光光谱在线水质异常检测方法
发布时间：2019-01-11 11:00
【摘要】：近年来,饮用水安全问题引起社会的广泛关注。采用紫外-可见光吸收光谱对水质进行异常检测,具有现场原位、无需试剂、分析快速等优点,适合快速在线监测。然而,紫外-可见光光谱数据量大,且易受仪器和水质正常波动的干扰,从而影响水质异常检测结果。提出一种基于基线校正和主元分析的紫外-可见光光谱法来检测污染物引起的水质异常,该方法利用非对称最小二乘校正基线,采用主元分析法从基线校正后的光谱矩阵中降维并提取特征,然后根据残差子空间的Q统计量评估测试样本的离群点,最后采用累计概率来更新异常报告结果。通过苯酚注入的实验,验证了该算法的有效性,实验结果表明,提出的方法与单波长法相比,有效地提高了污染物的检出下限;与未经基线校正采用主元分析进行的异常检测方法相比,提高了检出率,降低了误报率。
[Abstract]:In recent years, the safety of drinking water has attracted wide attention of the society. Using UV-Vis absorption spectrum to detect abnormal water quality, it has the advantages of in situ, no reagent, fast analysis and so on. It is suitable for rapid on-line monitoring. However, the UV-Vis spectrum has a large amount of data and is susceptible to the interference of instrument and water quality fluctuation, thus affecting the detection results of water quality anomalies. A UV-Vis spectral method based on baseline correction and principal component analysis (PCA) is proposed to detect water quality anomalies caused by pollutants. The method uses asymmetric least squares to correct baselines. The principal component analysis (PCA) is used to reduce the dimension and extract the feature from the baseline corrected spectral matrix. Then the outliers of the test samples are evaluated according to the Q statistics in the residual subspace, and the cumulative probability is used to update the anomaly report results. The experimental results show that the proposed method can effectively improve the detection limit of pollutants compared with the single wavelength method. Compared with the anomaly detection method without baseline correction using principal component analysis, the detection rate is improved and the false positive rate is reduced.
【作者单位】： 浙江大学控制科学与工程学院工业控制技术国家重点实验室;
【基金】：国家自然科学基金项目(61573313,U1509208) 浙江省科技厅公益项目(2014C33025) 浙江省重点研发计划项目(2015C03G2010034)资助
【分类号】：O657.3;R123.1
本文编号：2407063
[Abstract]:In recent years, the safety of drinking water has attracted wide attention of the society. Using UV-Vis absorption spectrum to detect abnormal water quality, it has the advantages of in situ, no reagent, fast analysis and so on. It is suitable for rapid on-line monitoring. However, the UV-Vis spectrum has a large amount of data and is susceptible to the interference of instrument and water quality fluctuation, thus affecting the detection results of water quality anomalies. A UV-Vis spectral method based on baseline correction and principal component analysis (PCA) is proposed to detect water quality anomalies caused by pollutants. The method uses asymmetric least squares to correct baselines. The principal component analysis (PCA) is used to reduce the dimension and extract the feature from the baseline corrected spectral matrix. Then the outliers of the test samples are evaluated according to the Q statistics in the residual subspace, and the cumulative probability is used to update the anomaly report results. The experimental results show that the proposed method can effectively improve the detection limit of pollutants compared with the single wavelength method. Compared with the anomaly detection method without baseline correction using principal component analysis, the detection rate is improved and the false positive rate is reduced.
【作者单位】： 浙江大学控制科学与工程学院工业控制技术国家重点实验室;
【基金】：国家自然科学基金项目(61573313,U1509208) 浙江省科技厅公益项目(2014C33025) 浙江省重点研发计划项目(2015C03G2010034)资助
【分类号】：O657.3;R123.1
【相似文献】
相关硕士学位论文 前1条
1 程吉锋;基于多尺度主元分析的丙烯聚合过程故障诊断研究[D];浙江工业大学;2009年
，本文编号：2407063
本文链接：https://www.wllwen.com/kejilunwen/huaxue/2407063.html
教材专著
