| 초록 |
This study proposed cost optimal selection and blending ratio of waste sea shells as SOx absorbent in wet flue gas desulfurization (WFGD) system to solve the high grade limestone depletion using deep neural network (DNN)-based surrogate model. Cost optimization is addressed to the following procedure. First process model was developed to generate the dataset of gypsum purity according to blending ratio. In addition, mathematical model is proposed to calculate the total annualized cost (TAC) and the TAC is added to the dataset. Second, the generated datasets is preprocessed using the z-score normalization and base on the datasets, DNN-based surrogate model is developed. Finally, cost optimal selection and blending ratio is derived using the developed DNN-based surrogate model under the two constraints: gypsum purity and total SOx absorbent consumption. As a results, the TAC could be reduced by about $ 788,469 through derived cost optimal blending ratio. |
