Effect of Pelleting Parameters on Remediation of Oil Polluted Water Bodies using Kenaf Pellets: A Comparative Analysis using Artificial Neural Network (ANN), Support Vector Machines (SVM) and Response Surface Methodology (RSM)

Abstract

Kadiri O. A1,, Aremu A. K, Raji A. O and Akinoso R

Environmental degradation from oil spilled has adversely affected ecological life. The study researches the utilisation of locally produced durable biodegradable kenaf pellets in remediating oil spills as well as the use of machine learning and response surface methodology in analysing experimental data. Recent advances in data modelling have necessitated the use of machine learning for predicting experimental outcomes. ANN, SVM and RSM were used in the analysis to analyse the effect of parameters on remediated crude oil polluted water samples. Statistical indices such as MSE, RMSE, MAE, MAD, MAPE, RCoV, rMBE, R2 were obtained and compared. Graphical plots were also generated and finally, computational codes and mathematical equations were established. Comparisons were made between the three data modelling methods and it was established that based on the optimal choice would be based on the desired output.

PDF