Application of multiple linear regression, central composite design, and ANFIS models in dye concentration measurement and prediction using plastic optical fiber sensor

Abstract

The measurement and prediction of dye concentration is important in the design, planning and management of wastewater treatment. Soft computing techniques can be used as a support tool for analyzing data and making prediction. In this study, Central Composite Design (CCD) and adaptive neuro-fuzzy inference system (ANFIS) are employed to identify and predict the output intensity ratio of light that passes through a plastic optical fiber (POF) sensor in Remazol Black B (RBB) dye solution of different concentrations. The predictive performances of these models are compared to that of the traditional Multiple Linear Regression (MLR). The accuracies of MLR, CCD and ANFIS models are evaluated in terms of square correlation coefficient (R2), root mean square error (RMSE), value accounted for (VAF), and mean absolute percentage error (MAPE) against the empirical data. It is found that the ANFIS model exhibits higher prediction accuracy than the MLR and CCD models.