Prediction of the Impact of Land Use and Soil Type on Concentrations of Heavy Metals and Phthalates in Soil Based on Model Simulation

Abstract

The main objective of this study is to determine the possibility of predicting the impact of land use and soil type on concentrations of heavy metals (HMs) and phthalates (PAEs) in soil based on an artificial neural network model (ANN). Qualitative analysis of HMs was performed with inductively coupled plasma–optical emission spectrometry (ICP/OES) and Direct Mercury Analyzer. Determination of PAEs was performed with gas chromatography (GC) coupled with a single quadrupole mass spectrometry (MS). An ANN, based on the Broyden–Fletcher–Goldfarb–Shanno (BFGS) iterative algorithm, for the prediction of HM and PAE concentrations, based on land use and soil type parameters, showed good prediction capabilities (the coefficient of determination (r2) values during the training cycle for HM concentration variables were 0.895, 0.927, 0.885, 0.813, 0.883, 0.917, 0.931, and 0.883, respectively, and for PAEs, the concentration variables were 0.950, 0.974, 0.958, 0.974, and 0.943, respectively). The results of this study indicate that HM and PAE concentrations, based on land use and soil type, can be predicted using ANN.

Unlocking the Potential of the ANN Optimization in Sweet Potato Varieties Drying Processes

“…Optimization procedures were applied to update these coefficients, minimizing the error between the network’s predictions and experimental outputs. This optimization process involved the sum of squares (SOS) and utilized the Broyden–Fletcher–Goldfarb–Shanno (BFGS) algorithm to expedite and stabilize convergence, as established by Suszyński and Peta [ 29 ] and Stojić et al [ 30 ]. Coefficients of determination served as parameters to assess the performance of the resulting ANN model.…”

Environmental Monitoring and Analysis of Persistent Organic Pollutants

“…The last article “Prediction of the Impact of Land Use and Soil Type on Concentrations of Heavy Metals and Phthalates in Soil Based on Model Simulation” by a team of Serbian researchers [ 6 ] looks into the possibilities of predicting soil pollution with an artificial neural network (ANN) model, more specifically a multi-layer perceptron (MLP) model of three layers (input, hidden, and output) based on the Broyden–Fletcher–Goldfarb–Shanno (BFGS) iterative algorithm. The results confirm good method predictivity and promising application in future studies of the relationship between soil properties and pollutant mass fractions.…”