Analysis of Sewage Treatment Plant’s Data to evaluate Quality of Effluent using Experimental Results and Prediction based on Artificial Intelligence Models (Case Study: Tehran Wastewater Treatment Plant)

The use of wastewater outflowing from the municipal wastewater treatment plants for irrigation of agricultural lands is one of the important and fundamental issues for reuse of the plants’ effluent. In recent years, the artificial neural networks (ANN) have received considerable attention for modeling the sewage treatment process. Accordingly, the ANN models, fuzzy logic (FL) and adaptive neuro fuzzy inference system (ANFIS) have been utilized in this paper to predict quality of the effluent running out of the treatment plant. For this purpose, first, three scenarios were chosen and their inputs were reduced using the principal component analysis (PCA) method. Finally, the process of modeling was conducted with and without this method and then, the results were compared. Evaluating the results obtained from the predictions using the statistical indicators revealed that the ANFIS model with mean error reduction value of 13.92% compared to that of the FL model and a reduction value of 8.22% in contrast to the ANN model, benefits from a higher accuracy and this trend has been held true with and without PCA. Moreover, calculating the percentage of pollutant removal efficiency indicated that the maximum removal efficiency is obtained when total suspended solids (TSS) are removed which is equal to 96.68%. It bears to mention that the rest of the pollutants had values approximately equal to that of TSS. Based on the results, as a simple, accurate, efficient and reliable approach, the ANN models can be applied to predict quality of the effluents.