Study On Early Warning Model Of Sudden Water Pollution In Dongzhuang Reservoir Based On BP Neural Network

In recent years,with the rapid development of economy,sudden water pollution accidents occur frequently all over the world,which may lead to catastrophic environmental problems and serious social consequences.The transportation process of pollutants is simulated and analyzed,and build the early warning model of sudden water pollution are of great practical significance to reduce the losses caused by sudden water pollution accidents.Taking Dongzhuang reservoir as an example,this paper uses HD module and ECO lab module of Mike 11 to build one-dimensional hydrodynamic model of Dongzhuang reservoir-Water quality model.The hydrodynamic changes of four typical hydrological years before and after the construction of the reservoir are compared and analyzed.According to the investigation results of pollution sources,COD,NH3-N,TP and TN were determined as the main pollution factors,and the distribution law of the four pollution factors in four typical hydrological years was simulated.Using scenario analysis-method,the temporal and spatial distribution of pollutants under different pollution conditions is simulated and analyzed,and the database of sudden water pollution accidents is established.Based on the database,BP neural network is used to establish the early warning model of TP and TN sudden water pollution.The conclusions are as follows:1、Before the construction of the reservoir,the water depth was shallow,the flow velocity is large,and the water level had little seasonal fluctuation,After the construction of the reservoir,the water level rises obviously,and the flow velocity in the reservoir keeps about Om/s.From the front of the dam to the tail of the reservoir,the variation range of water depth gradually decreases,and the water level has obvious seasonal fluctuation.Under condition of conventional pollution,COD,NH3-N,TP and TN meet the requirements of three kinds of water standards in the reservoir area,and the four water quality indexes show a downward trend on the whole.In the local area of the sewage outlet,the concentration had obvious explosion,The concentrations of the four pollutants decreased most slowly in autumn,followed by spring,summer and winter;In numerical terms,the concentrations of the four pollutants decreased the least in spring,followed by summer,and the largest in autumn and winter.2、The transport and diffusion processes of TP and TN under different pollution conditions were compared and analyzed,The temporal and spatial distribution of pollutants is clarified by using the accident characteristic parameters such as peak concentration,distance from peak concentration to pollution source,affected distance of river channel,arrival time of pollutants and affected time of river.It is found that the propagation range of pollutants is mainly related to the upstream flow.The larger the upstream flow is,the farther the pollutant propagation distance is.The upstream water flow,pollution discharge flow,pollution discharge concentration and pollution discharge time all have an impact on the peak concentration,among which the upstream water flow,pollution discharge flow and pollution discharge concentration have a greater impact,while the pollutant discharge time has a smaller impact.3、TP and TN model based on BP neural network can quickly predict the peak concentration and peak concentration in different time after the water pollution accident,according to the distance of pollution source,the time of river affected and the distance of exceeding the standard of water quality.To a certain extent,it can represent the characteristics of the accident after the sudden water pollution accident,provide the basis for timely,accurate and reasonable emergency measures,and reduce the losses caused by the accident.