Research Of A Hybrid Surface Water Quality Prediction Model And Development Of A Prediction System

With the development of industry and agriculture and the acceleration of urbanization in China,industrial sewage,urban sewage,non-point source sewage and a large number of effluent containing pollutants,waste residue are discharged into the surface water bodies,which causes a sharp decline in water quality,and damages the ecological balance,and has a great impact on daily drinking water,so that it seriously endangers people's health.Therefore,it has of great practical significance to predict surface water quality.However,in addition to human factors,water environment,as a complex organic system,is greatly affected by climate,environment and geographical location as well.It is particularly important to make a stable and accurate prediction for water quality.An accurate prediction can further effectively predict the quality of surface water and provide relevant basis for scientific prevention and management of surface water.In view of the above problems,a new hybrid surface water quality prediction model is designed based on the existing machine learning algorithm,and a surface water quality prediction system is developed based on the model in this paper.Firstly,the collected original water quality data are preprocessed in the model,mainly including replacing missing values and outliers in the data with median.The outliers are determined by Pauta criterion.The processed water quality data are analyzed and described by fuzzy c-means.Secondly,the main prediction module of this model is composed of BP neural network,extreme gradient boosting,support vector regression,Pearson correlation coefficient and variance reciprocal weighting,and the optimal combination module is selected to predict the water quality data of each cluster.To verify the effectiveness and robustness of the model,the surface water quality data set in Huaihe River basin is taken as an example,and the pH,DO,NH3-N,CODMn of this river are tested.The experimental results show that this model can provide moreaccurate water quality prediction results than other three models.Furthermore,in this paper,based on web crawler and Java web development,the surface water quality prediction system is designed and developed based on the hybrid prediction model.The performance of this system is tested by ApacheBench,and the test results show that the developed system has a good robustness.