| 초록 |
The general WWTP have the properties for complex, non-stationary, temporal correlation, and nonlinear systems. To satisfy with required quality of WWTP’s effluents, the mathematical modeling and neural networkmethods is widely used. However, they have limitations to incorporate the key process characteristics at the WWTP. In this study, a systematic methodology of NN modeling based on sensitivity analysis is proposed to select the key modeling information of the plant and predict the effluent concentration by the temporal and hydraulics characteristics. Sensitivity analysis reveals that important variables are ranked by calculating sensitivity measure for every variable. The proposed method is applied for modeling wastewater quality of a full-scale plant, which is a DNR process. In the experimental results in a full-scale plant, the proposed method shows that it suggests a systematic methodology to NN model and can also improve its prediction capability. Acknowledgement) This work was supported by Brain Korea21 project, the Korea Research Foundation by Grant funded by the Korean Government (MOEHRD) (KRF-2007-331-D00089) and funded by Seoul R&BD Program (CS070160). |
