Virtual sensor for measuring the concentration of Cu(II) in river waters

Abstract

The possibility of creating a virtual sensor for measuring the concentration of copper ions (Cu(II)) in water bodies was investigated. The creation of a virtual sensor is based on the possibility of using cheap and reliable sensors for easily measurable variables that are mathematically associated with the target variable. The values of the parameters characterizing the water quality have been collected for a long time and the surface water samples have been taken from the same points of the river water. Based on a correlation analysis, two input variables were selected for the virtual sensor - pH and electrical conductivity. At a certain stage of research, temperature was added as an input variable, but this did not lead to a significant increase in the accuracy of the virtual sensor. To create mathematical models for the purposes of the developed virtual sensor, a large number of artificial neural networks with feedforward propagation were trained and regression models were derived. In the article, the mathematical models that describe the measurements carried out with the highest accuracy are presented