基于紫外可见吸收光谱的水质检测算法研究  

Research on water quality detection algorithm based on UV-vis absorption spectral

作　　者：林春伟 郭永洪[1] 何金龙[1] LIN Chunwei;GUO Yonghong;HE Jinlong(China Jiliang University,Hangzhou 310018,China)

机构地区：[1]中国计量大学

出　　处：《中国测试》2019年第5期79-84,共6页China Measurement & Test

基　　金：浙江省自然科学基金(Y14F010075)

摘　　要：为实时有效地检测地表水中硝酸根离子和亚硝酸根离子的变化过程,提出一种基于紫外可见吸收光谱的水质检测算法。针对水质光谱数据受到干扰易出现波动误差的问题,采用小波变换对其进行分解以滤除高频噪声,并通过主成分分析对数据特征进行降维以防止模型复杂度较高导致过拟合。水质光谱数据经预处理后采用支持向量机对其进行建模,通过非线性自适应调整变异收缩因子对差分进化算法进行改进,并采用改进差分进化算法对水质预测模型进行参数优化。通过与采用其他常用算法所建模型进行对比分析,实验结果表明:基于该算法所建的硝酸根离子和亚硝酸根离子模型具有更高的预测精度,且其能够以更快的收敛速度使模型达到全局最优。In order to detect the change process of nitrate ions and nitrite ions in surface water effectively and in real time, a water quality detection algorithm based on UV-vis absorption spectra was proposed. Aiming at the problem that water quality spectral data are disturbed and easy to appear wave error, wavelet transform is used to decompose the water quality spectral data to filter out the high frequency noise, and the principal component analysis is used to reduce the dimension of data features to prevent the over-fitting caused by high complexity of the model. The water quality spectral data are modeled by support vector machine after pretreatment, and the differential evolution algorithm is improved by nonlinear adaptive adjustment of variation shrinkage factor, and the parameters of the water quality prediction model are optimized by the improved differential evolution algorithm. By comparing with the models built by other common algorithms, the experimental results show that this algorithm can make the nitrate ion and nitrite ion models have higher prediction accuracy, and it can make the model achieve global optimization with faster convergence speed.

关 键 词：水质检测 光谱分析 支持向量机 改进差分进化算法 小波变换 主成分分析 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]