Construction And Application Of Water Quality Prediction Model Based On PLS-GWO-SVR

Water environment is one of the most important parts of natural ecological environment.In recent years,due to the rapid development of social economy and the continuous expansion of human activities,the global water environment is becoming more and more serious pollution situation.It is particularly important to develop and utilize water resources reasonably,protect water environment and maintain ecological balance.As an important part of water environment research,water quality prediction has become one of the basic subjects of modern environmental science theory research.Considering the complexity of water environment system,a water quality prediction model based on partial least squares(PLS)and grey wolf optimization(GWO)is proposed.Firstly,according to the characteristics of nonlinear and uncertainty of water environment system,the support vector regression model is proposed to predict water quality.Support vector regression machine has the advantages of global optimization,simple structure,strong generalization ability and suitable for forecasting small sample data,which is more suitable for nonlinear system prediction.At the same time,the performance of SVR depends on the type of kernel function and the parameter value,and the parameter selection is uncertain.In order to get the best parameter value,the gray wolf algorithm is used to optimize the parameters in this model.Due to the redundancy of water quality information caused by the multiple correlation of water quality,the prediction accuracy is reduced.In this model,partial least square method is used to extract the characteristics of the input variables of water quality prediction model.Through this combination algorithm,a water quality prediction model based on PLS-GWO-SVR is established.Secondly,the proposed water quality prediction model is verified by an example,and the data comes from Shihe reservoir.Using data mining related knowledge to preprocess the complex water quality monitoring data of Shihe reservoir,including missing data processing,data denoising,data normalization.The database of the processed water quality monitoring data has laid the foundation for the next simulation experiment.At the same time,the main water quality factors are introduced,and the multicollinearity among the water quality factors of Shihe reservoir is analyzed,which proves the necessity of using partial least square method in this model.Finally,the processed water quality data of Shihe reservoir is input into PLS-GWO-SVR water quality prediction model,which can predict the dissolved oxygen and total nitrogen content of water,and it is also compared with PLS-SVR prediction model,SVR prediction model,BP neural network prediction model for prediction comparison.The simulation results show that the PLS-GWO-SVR water quality prediction model proposed in this study has better prediction accuracy and generalization ability.