Coupling Square Wave Anodic Stripping Voltammetry with Support Vector Regression to Detect the Concentration of Lead in Soil under the Interference of Copper Accurately

In this study, an effective method for accurately detecting Pb(II) concentration was developed by coupling square wave anodic stripping voltammetry (SWASV) with support vector regression (SVR) based on a bismuth-film modified electrode. The interference of different Cu(2+) contents on the SWASV signals of Pb(2+) was investigated, and a nonlinear relationship between Pb(2+) concentration and the peak currents of Pb(2+) and Cu(2+) was determined. Thus, an SVR model with two inputs (i.e., peak currents of Pb(2+) and Cu(2+)) and one output (i.e., Pb(2+) concentration) was trained to quantify the above nonlinear relationship. The SWASV measurement conditions and the SVR parameters were optimized. In addition, the SVR mode, multiple linear regression model, and direct calibration mode were compared to verify the detection performance by using the determination coefficient (R(2)) and root-mean-square error (RMSE). Results showed that the SVR model with R(2) and RMSE of the test dataset of 0.9942 and 1.1204 μg/L, respectively, had better detection accuracy than other models. Lastly, real soil samples were applied to validate the practicality and accuracy of the developed method for the detection of Pb(2+) with approximately equal detection results to the atomic absorption spectroscopy method and a satisfactory average recovery rate of 98.70%. This paper provided a new method for accurately detecting the concentration of heavy metals (HMs) under the interference of non-target HMs for environmental monitoring.