全光谱水质参数反演方法研究与分析

Research and analysis on the inversion method of water quality parameters with full spectrum

提出了基于随机森林算法和XGBoost算法的水质参数反演方法,实现了对水温、pH值、电导、溶解氧、浊度、高锰酸盐、氨氮、总磷等水质参数的反演预测分析。结果表明,pH值、溶解氧、高锰酸盐、氨氮、总磷反演精度的均方根误差均在0.7以下,电导、浊度反演精度的均方根误差均在18以下,大部份参数的平均绝对百分比均在0.1以下,较好地实现了全光谱水质参数的反演研究,可为水质监测提供参考。

This paper proposes a new inverse method of water quality parameters based on random forest and XGBoost(Extreme Gradient Boosting),which realizes the inversion prediction and analysis of water temperature,PH,conductivity,dissolved oxygen,turbidity,permanganate,ammonia nitrogen,total phosphorus and other water quality parameters.The results show that root-mean-square error of the inversion accuracy of pH value,dissolved oxygen,permanganate,ammonia nitrogen and total phosphorus are all below 0.7,and for conductivity and turbidity are below 18,and the MAPE of most parameters are below 0.1,which achieves good results on the inversion research of full spectrum water quality parameters and can be the important technical support for water quality monitoring.