Geographic Information System Technology Combined with Back Propagation Neural Network in Groundwater Quality Monitoring

摘要: This study was conducted to explore the distribution and changes of groundwater resources in research area, promote application geographic information system (GIS) technology its deep learning methods chemical type water quality prediction groundwater. The Shiyang River Basin Minqin County selected as object for analyzing natural components preliminary forecast partial areas. With priority control pollutants, concentration four indicators (including permanganate index) different spatial distributions were analyzed based on GIS technology, so provide a basis prediction. Taking benchmark, this evaluated effects conventional back propagation (BP) neural network (BPNN) model optimized BPNN golden section (GBPNN) wavelet transform (WBPNN). algorithm proposed is compared with several classic algorithms analysis. Groundwater level rules area are technology. results reveal that can characterize analyze it. In contrast, WBPNN has best result. Its average error whole process 3.66%, errors corresponding six predicated values all below 10%, which dramatically better than other two models. maximal accuracy 97.68%, an 96.12%. consistent actual condition, be shown clearly combined algorithm. Therefore, it great significance regional quality, studywill play critical role determining quality.