Study And Application Of Water Quality Prediction For Beilun Estuary Based On SARIMA-LSTM

With the improvement of people’s environmental awareness,environmental protection work has received more and more attention from the government and citizens.Mangrove Wetland in Beilun Estuary,as a key ecological area connecting land and ocean,is very important to ensure its ecological health.As water environment is an important part of mangrove ecosystem,an accurate prediction of water quality is the basis for judging the health status of wetland ecosystem.And it also contributes a lot to the smooth development of environmental protection work.However,the current popular monitoring methods still stay on daily checking and on-the-spot investigation.It is difficult to grasp the future development direction of the environment,and easy to cause serious ecological damage by not taking preventive measures in time.Therefore,it is necessary to change the traditional supervisory maintenance into active preventive maintenance.Based on the data of water quality and weather in Beilun Estuary,Guangxi,this paper chooses permanganate index and the content of ammonia nitrogen,which can reflect the water quality,as prediction targets.And proposing a prediction method based on the combination of traditional time series model and deep learning.In order to improve the accuracy of prediction,the time and interaction between features are analyzed by incorporating multiple related features.The main work of this paper is as follows:(1)It deeply analyses the research background and current status of water quality prediction,and introduces the existing time series models by classification.According to the characteristics and shortcomings of current forecasting methods,the main research contents and innovations of this paper are put forward.And mainly elaborating the theoretical basis of the model proposed in this paper.(2)According to the complexity of the environment in which the research target is located,and in view of the fact that most of the current water quality prediction researches only focus on one feature but ignore other influencing factors,a multi-feature prediction method based on seasonal Autoregressive Integrated Moving Average model and Long Short-Term Memory(SARIMA-LSTM)is proposed in this paper,considering the close correlation between weather conditions and water quality.Firstly,the SARIMA model is used to predict the target sequence and obtain the training residuals.Then,getting residual predictions through LSTM by fusing other features.Finally,the superposition of forecasting results of the two components is taken as the ultimate forecasting value.(3)The prediction effect of SARIMA-LSTM model is analyzed on real data sets of water quality and weather.And the performance of different models is compared according to the prediction results and experimental evaluation indicators.(4)Based on the algorithm proposed in this paper,a prototype system of water quality prediction for Beilun estuary mangrove reserve is designed.Including detailed requirements analysis and system framework description,and design and implementation of the main functional modules of the system.