杨晔1 林彦君2 陈丹丹1 李书明1

　　(1.国电南京自动化股份有限公司，江苏 南京 210061;

　　2.厦门水务原水投资运营有限公司，福建 厦门 361000)

　　摘要：近年来，水生态环境遭到巨大破坏，建立和完善系统的、专业的水环境监测与评价系统，对水体污染的控制与防治起重要指导作用。设计并开发水环境监测与评价系统，通过报表、图表等，结合在线和离线分析技术，对水质数据变化趋势进行直观展示。将机器学习运用到水质评价中，提高运算的效率及准确度。针对水质时间序列数据的不确定特性建立水质预测模型，对水质中参数在未来一段时间中的含量进行预测，解决水质时间序列多元性、时变性、非线性、耦合性等特点所导致的预测建模困难和管理难度增加的问题，并得到较高的预测精度。该系统集水质数据监测、管理、预警功能为一体，在长泰枋洋水利枢纽得到应用，取得良好的效果。

　　关键词：水环境监测;机器学习;水质评价;水质预测模型;预警;水利枢纽

　　Design of water environment monitoring and evaluation system and its application in Fangyang hydro-junction

　　YANG Ye1，LIN Yanjun2，CHEN Dandan1，LI Shuming1

　　(1. Guodian Nanjing Automation Co., Ltd., Nanjing 210061, China;

　　2. Xiamen Water Resources Investment and Operation Co., Ltd., Xiamen 361000, China)

　　Abstract：In recent years, water ecological environment has been greatly affected. The development of a professional water environment monitoring and evaluation system plays an important guiding role in the control and prevention of water pollution. The water environment monitoring and evaluation system is designed and developed to display variations of water quality data through reports, charts, etc., combined with on-line and off-line analysis technology. Machine learning is applied to water quality evaluation to improve the efficiency and accuracy of computing. Due to uncertainty of time-series water quality data, a water quality prediction model is set up to predict water quality parameters. It can solve the problem of difficulties in prediction modeling and management caused by uncertainty such as diversity, time variability, nonlinearity and coupling, and thus the model has high prediction accuracy. The system integrates functions including water quality data monitoring, management and early warning, and has been applied in Changtai Fangyang hydro-junction, which shows good performance.

　　Key words：water environment monitoring;machine learning; water quality evaluation;water quality prediction model;early warning; hydro-junction