Water Quality Prediction By UV-VIS-NIR Absorbance And Sensor Development

In recent years,the landscape water of China faced water eutrophication problems which have caused tremendous pressure on human production,life and sustainable development of the ecological environment.In order to monitor the water quality of landscape water fast and effectively,this study took advantages of spectroscopy principle,mainly worked on the UV-VIS-NIR absorbance of unprocessed water and its prediction ability of turbidity and chlorophyll a concentration(Chl-a).At the same time,the study also designed a detection sensor.Seven kinds of water samples were artificially prepared in this study:spirulina water samples(S),chlorella water samples(C),turbidity water samples(T),mixed water samples of S-C,S-T,C-T and S-C-T.The absorbance(200-900 nm)of various water samples was measured by the spectrophotometer,and then systematically analyzed the Chl-a,turbidity and water absorbance.The researches mainly included:(1)The turbidity had a strong correlation with the absorbance in all water samples,and linear models were built to predict turbidity for each type of water samples,the Rv2 of each specific model was higher than 0.867.For Chl-a prediction,whether the linear regression model or SVR method,the prediction accuracy were obviously affected by the type of water samples,and the prediction ability for pure algae water samples is higher,but for mixed algae water it is greatly reduced.(2)In order to improve Chl-a prediction accuracy for turbidity mixed samples,ULR-LSSVR method and SA-LSSVR-BP method were used to estimate Ch1-a concentration and turbidity.Furthermore,the Rv2 of the overall Chl-a concentration decoupled models for among S-T,C-T and S-C-T were higher than 0.884,the Rv2 of the overall turbidity models were also higher than 0.867.ULR-LSSVR and SA-LSSVR-BP model were also applied to the landscape water samples,the results showed that SA-LSSVR-BP method was used to estimated Chl-a and turbidity which were increased by 58.8%and 85.04%comparing to the corresponding ULR-LSSVR method.(3)In addition,this paper also designed an active light source water quality sensor.Our goal is to explore technologies to cheaply,conveniently and rapidly detect the Chl-a and turbidity in landscape water,and further to effectively guide the situation of landscape water quality.