Algorithm Research Of UV-visibie Removing Deturbidty Of Water Chorophtll A Content And The Sensor Design

In our country's urban freshwater resources,most urban water bodies are polluted by eutrophication in varying degrees due to poor governance and lax controls.In order to be able to quickly determine the chlorophyll a content in water and provide reliable data for water pollution early warning,in this paper,based on spectroscopy analysis,a prediction model related to water chlorophyll a content and a sensor for measuring water chlorophyll a content were designed.In this dissertation,laboratory cultured spirulina specimens,standard turbidity samples and mixed samples with different concentration gradients in the laboratory were used as research objects.Exploring the application and accuracy of various modeling methods in predicting chlorophyll a content in water absorption spectra.Including one-dimensional linear regression analysis,partial least squares(PLS)analysis and BP neural network analysis.In this study,firstly,he absorption spectrum data of all samples in the UV-visible band were obtained for correlation analysis,and on this basis,a single linear regression model for Spirulina water samples,standard turbidity water samples,and mixed water samples was established(Model 1),the absorbance and chlorophyll-a concentration of the spirulina sample,the absorbance and turbidity of the standard turbidity sample,and the absorbance of the mixed sample and the chlorophyll a concentration were 0.992,0.9856,and 0.8857,respectively.In order to better remove the influence of turbidity in the mixed water sample on the prediction of chlorophyll a concentration,the partial least squares(PLS)prediction model(model 2)and the BP neural network prediction model(model 3)were further explored.The R2 between the predicted and measured values of Model 2 and Model 3 were 0.9994 and 0.9972,respectively.The results show that Model 1 predicts the concentration of chlorophyll a in water,which is influenced by turbidity.The prediction error of chlorophyll a concentration is large,and the error percentage is up to 25%.Model 2 can remove turbidity well and predict the concentration of chlorophyll a.The impact,the percentage of prediction error is about 4%,but the model can't predict the value of turbidity while predicting the concentration of chlorophyll a;model 3 can simultaneously predict the value of chlorophyll a concentration and turbidity in the mixed water sample.And the percentage of prediction error of chlorophyll a concentration is within 5%.In addition,based on the ideal model(Model 3),this study explored a set of low cost de-turbidity inversion water chlorophyll a content sensor design.The sensor adopts the active light source design.The MSP430F149 single-chip microcomputer is used as the controller to process the measurement data.The LCD 12864 display is used as the display module to display the measurement results.To guide the design of a sensor that can realize real-time measurement of chlorophyll a content in water,provide accurate data for early warning of eutrophication of water,and guide people to manage water bodies that are about to be contaminated by eutrophication.