Abstract
Collection of soil samples is labored, time-consuming and the determination of heavy metal concentrations in the laboratory are expensive. The aim of this study was to fix the functions, algorithms as well as optimization of methods for soft computing system such as ANFIS, SVM, and ANN based on their best performance. In this study, soil samples were collected from eighty five distinct locations in and around of a selected open disposal site at old Rajbandh, Khulna, Bangladesh at a depth 0-30 cm from the existing ground surface. In the laboratory, the concentration of heavy metals such as Pb, Cu, Ni, Zn, Co, Cd, As, Sc, Hg, Mn, Cr, Ti, Sb, Sr, V and Ba in soils were measured. The soft computing systems such as ANFIS, SVM, and ANN were implemented for the analysis of heavy metal concentrations in soil. The result reveals model with SCP, gaussmf, linear and hybrid was the best-fitted model of ANFIS. In addition, in SVM analysis, the model SVM-RBF with 15 folds was selected. In ANN, the model LT (Levenberg-Marqardt and Tansig functions) with neuron structure 2-10-1 was selected. The accuracy of the predicted results was checked based on the acceptable limits of prediction parameters such as R value, RMSE, MAPE, GRI and percentage recovery. The result demonstrates that ANFIS model was a reliable technique than that of other counterparts of SVM and ANN with the acceptable degree of robustness and accuracy. Therefore, the performance of soft computing systems may be expressed by the sequence of ANFIS > SVM > ANN. Here it can be noted that one can easily be computed the concentration of a particular heavy metal in soil by inserting GPS values (latitude and longitude) only in the developed rule viewer of ANFIS.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call