基于BP神经网络的北京昌平山前平原地下水水质评价

该文采用单因子评价方法对昌平区浅层地下水的超标因子进行筛选,结合水文地球化学理论探讨各因子超标原因,分析浅层地下水水质的空间分布特征,并采用BP神经网络法对水质进行综合评级。从综合评级结果来看,12眼监测井中1眼为Ⅴ类水质,5眼为Ⅳ类水质,6眼为Ⅲ类水质。单因子筛选结果表明,总硬度、总溶解性固体、氮素、氟化物等为该区最主要的超标因子。经分析可知,山前平原地带浅层地下水中氟化物为原生污染,氮素污染物主要来源于地表污染物下渗,总硬度和总溶解性固体的升高主要受地表污染物下渗、氮素的迁移转化等因素的影响。研究可为研究区地下水管理工作提供可靠数据。

Evaluation of groundwater quality in Changping piedmont plain of Beijing based on BP neural network

Abstract: Groundwater quality is closely related with human health and environmental safety. In the suburb of Beijing, the groundwater quality is heavily concerned. In this study, the groundwater quality in Changping piedmont plain was evaluated based on the single factor evaluation method and comprehensively evaluated based on BP neural network. The Changping district is located was in the northwestern area of Beijing. Considering that the main area affected by human activity was the shallow groundwater, we arranged a total of 12 monitoring wells around the plain area. The depths of wells 1#, 2# and 11# were 130 m, the depth of wells 3#, 4# and 7# were about 125 m, and the depths of wells 5#, 6#, 8#, 9#, 10#, 12# were about 120 m. The groundwater samples were collected on April 16, 2015. A total of 27 indexes were determined including pH value, chloride, sulfide, nitrate nitrogen, ammonia nitrogen, heavy metals, fluoride, and so on. In the single factor evaluation, the groundwater quality was evaluated according to the National Groundwater Quality Standards (GB/T14848-93). Based on the single factor evaluation method, water quality in 5 wells of 1#, 4#, 7#, 9# and 11# exceeded the standards. In the 1# well, the turbidity, total hardness, total dissolved solids and nitrate nitrogen exceeded the standards by 1.67, 1.81, 1.48, and 3.08 times, respectively. In the 4# well, the turbidity was 1.33 times exceeding the standard. In the 7# well, the total hardness and total dissolved solids exceeded the standards by 1.2 and 1.15 times, respectively. In the 9# well, the fluoride exceeded the standard by 1.58 times. The ammonia nitrogen in the 11# well exceeded the standard by 3.25 times. Compared the values in 1999, the area with total hardness exceeding the standard was expanded largely. Among the 27 indexes, we chose 17 indexes for the comprehensive evaluation based on BP neural network, including the turbidity, iron, chloride, sulfate, TDS, total hardness, manganese, zinc, potassium permanganate index, nitrate nitrogen, nitrite nitrogen, ammonia nitrogen, fluoride, and so on. The BP evaluation showed that all the 12 wells had the water quality above III. Among the wells, 1 was V grade, 5 were IV grade and 6 were III grade. Compared with the single evaluation results, the comprehensive evaluation based on BP neural network was reasonable. The 1# well had 4 indexes exceeding the standards and thus the water quality was V grade. Compared with previous study, the pollution of nitrate and ammonia nitrogen might be due to surface pollutants infiltration. In the future, the continuous monitoring of shallow groundwater should be conducted, and the surface pollutants infiltration prevention control should be strengthened. The study may provide valuable information for the management of the groundwater in Beijing.