Comparison of Artificial Neural Fuzzy Inference System (ANFIS) and Response Surface Methodology (RSM) Model in Predicting the Outlet Flow Rate of Passive Treatment System Column

Abstract

Acid mine drainage (AMD) is one of the major environmental problems the mining and mineral processing industries face. Treatment of AMD involves active and passive treatment. In the long term, passive treatment is the most effective way to treat acid mine drainage, but it can be expensive. if handled properly. Therefore, the study of flow rate in a passive treatment system is one of the important ways to identify optimum hydraulic retention time to ensure the maximum percentage of heavy metal removal can be achieved while keeping the cost to a minimum level. This study focused on developing and comparing the Response Surface Methodology (RSM) model and Artificial Neural Fuzzy Inference System (ANFIS) model to predict the outlet flow rate of the passive treatment system column based on three parameters inlet flow time, thickness of peat soil bed, and inlet flow rate. The RSM model was created by Design-Expert software whereas MATLAB created the ANFIS model with 80% of data used for the model training and 20% of the data for model testing. The models' performances were compared in terms of statistical errors (AAPE, RMSE, R2, STD, minimum error, and maximum error). It was found the ANFIS model has performed better in predicting the outlet flowrate with R2 value of 0.99 RSM model with the value of 0.97. The inlet flow rate was an insignificant parameter affecting the outlet flow rate of the passive treatment column. From the 3-D surface response plot, the highest outlet flow rate is predicted to be 524 mL/min.