Research On The Detection Method Of Lake Water Quality Based On Artificial Neural Network

In recent years,due to the increasing coverage rate of global industrialization,water environment is being seriously polluted as the root of human survival.Therefore,it is urgent to detect,analyze and treat water environment pollution.In the water environment,the treatment of lake water pollution is more important.Because the content of pollutants in the lake water is large and complex,in order to effectively control the lake water pollution,it is very important to detect and analyze the pollutants affecting the lake water quality.This paper proposes a detection method of lake water quality for the content of pollutants in lake mixed solution.Using fluorescence spectroscopy and spectrophotometry to detect the known pollutants and unknown pollutants in the mixed solution.Establish RBF neural network model based on the detection results.Predict the content of pollutants in the lake,and optimize the model with the particle group algorithm to improve the operation efficiency and prediction accuracy of the model.In view of the detection of known pesticide pollutants in lakes,the lake aqueous solution containing three multicomponent pesticides is taken as experimental samples,which shows that the selected pesticide concentration and the peak intensity have a good fit relationship with the fluorescence peak.The fluorescence spectrum method is used to detect and analyze the solution.Establish RBF neural network model,measure the fluorescence spectrum of the three pesticide mixture solutions.The experimental data are randomly assigned into the training set and the test set.The fluorescence intensity is the input and the concentration is the output.Predict the pollutant concentration of the test set through the corresponding training,optimize the model,and further improve the model operation efficiency and prediction accuracy.The experimental results show that the RBF neural network after particle swarm optimization has less prediction error and higher correlation coefficient,which can improve the prediction accuracy of known pollutants in lakes.For the detection of unknown pollutants in lakes,the degree of lake pollution is indicated by predicting the corresponding water quality parameter COD.Using spectrophotometry to measure seven parameters affecting water quality in lake water.Preprocess the measurement results as the training set and test set of RBF neural network model.Establish RBF neural network model,select five parameters with high correlation with COD,and COD concentration as network input,further improving the operation efficiency and prediction accuracy.The experimental results show that the particle swarm optimized RBF neural network predicts COD parameters with less relative error and higher correlation coefficient,which can better improve the prediction accuracy of water quality parameters in lakes with unknown pollutants.