A reduced-order adaptive neuro-fuzzy inference system model as a software sensor for rapid estimation of five-day biochemical oxygen demand

作者：Noori Roohollah^*; Safavi Salman; Shahrokni Seyyed Afshin Nateghi

来源：Journal of Hydrology, 2013, 495: 175-185.

DOI：10.1016/j.jhydrol.2013.04.052

• Uncertainty analysis
• Reduced-order model
• Five-day biochemical oxygen demand (BOD5)
• Proper orthogonal decomposition
• Adaptive neuro-fuzzy inference system

The five-day biochemical oxygen demand (BOD5) is one of the key parameters in water quality management. In this study, a novel approach, i.e., reduced-order adaptive neuro-fuzzy inference system (ROANFIS) model was developed for rapid estimation of BOD5. In addition, an uncertainty analysis of adaptive neuro-fuzzy inference system (ANFIS) and ROANFIS models was carried out based on Monte-Carlo simulation. Accuracy analysis of ANFIS and ROANFIS models based on both developed discrepancy ratio and threshold statistics revealed that the selected ROANFIS model was superior. Pearson correlation coefficient (R) and root mean square error for the best fitted ROANFIS model were 0.96 and 7.12, respectively. Furthermore, uncertainty analysis of the developed models indicated that the selected ROANFIS had less uncertainty than the ANFIS model and accurately forecasted BOD5 in the Sefidrood River Basin. Besides, the uncertainty analysis also showed that bracketed predictions by 95% confidence bound and d-factor in the testing steps for the selected ROANFIS model were 94% and 0.83, respectively.

• 出版日期2013-7-12