Research And Development Of Watershed Water Environment Monitoring And Early Warning System

In recent years,with the continuous advance of national industrialization,industrial wastewater and domestic sewage are gradually polluting the water environment which our lives depend on.Water pollution has been one of the constraints of social development.According to the statistics of the Ministry of Water Resources in 2017,the length of the IV River and below accounted for 21.5%of the total.Among 544 important provincial boundary sections monitored in China,the IV River and below accounted for 33%.Faced with such a severe pollution situation,it is of great significance to develop an effective water quality monitoring and early warning system for monitoring and analyzing the water resources status in the basin.In this paper,the water environment in the basin is considered as the research object.Through searching the relevant information of real-time online monitoring and early warning system,the development status at home and abroad is analyzed.It is concluded that the current water environmental monitoring and early warning system in the basin needs to solve the problems of data reception,real-time data display,over-standard alarm,water quality parameter prediction and water quality evaluation.Price and so on.Furthermore,according to the need of watershed water environmental monitoring and early warning,a watershed water environmental monitoring and early warning system model based on B/S structure is designed.The system is mainly composed of three parts:water quality data receiving and pretreatment,water quality monitoring display and off-line water quality analysis.Each module is based on cloud server,and data exchange is realized through database,which realizes monitoring,alarming and forecasting of water quality in river basin.In order to make full use of the monitored water quality data,the prediction of water quality data is studied in this paper.In the prediction of water quality time series data,in order to improve the prediction accuracy and enhance the adaptability of the model to different monitoring stations and different parameters,so that the model can play a more effective role in practical engineering applications where there are a large number of monitoring stations,this paper combines the recurrent neural network with evidence theory,and develops an improved water quality prediction model.As part of the off-line water quality analysis service,it provides services for the automated in-depth analysis of water quality data.In order to improve the monitoring efficiency of water quality monitoring network,and make full use of the monitoring data,get the distribution status of water quality in the area of virtual monitoring stations,and realize the prediction of water quality in the area of virtual monitoring stations,local denoising criteria and contractive regularization term are introduced into auto-encoder based on extreme learning machine,and a deep contractive denoising extreme learning machine model is developed.Then,the model is used to extract the abstract feature expression of water quality spatial relationship in monitoring network and the prediction of water quality is realized by weighted extreme learning machine,which improves the prediction accuracy of water quality spatial distribution from the model perspective.