基于时间序列分析的水质自动监测数据插补方法

Study on monitoring and estimating pollutant flux in mountain rivers

因自动监测设备故障或运维需要,不可避免地中断自动监测数据,影响水质分析结果。比较基于统计学的插补方法可知,ARIMA模型插值插补效果最好,均值插补结果较差。为进一步提高插补的精度,提出了基于LSTM的数据插补方法,插补效果优于基于统计学的插补方法。为了试验西江不同类型点位插补效果,选择封开城上断面古劳断面和珠海大桥断面进行比对实验,结果表明,封开城上断面插补效果比古劳、珠海大桥好。

Due to the failure of automatic monitoring equipment or the need of operation,the automatic monitor⁃ing data are inevitably interrupted,affecting the results of water quality analysis.Compared with the interpolation methods based on statistics,Arima model has the best interpolation effect and the mean interpolation is poor.In order to improve the interpolation precision,a data interpolation method based on LSTM is proposed,and the interpolation effect is better than that based on statistics.In order to test the interpolation effect of different types of points in Xi River,the Gulao section of the upper section of Fengkaesong and the cross section of Zhuhai Bridge were selected for comparison experiments.The results show that the interpolation effect of the upper section of Fengkaesong is better than that of Gulao and Zhuhai Bridge.