Objective: Wastewater treatment is vital for addressing quantitative and qualitative water shortages, which are crucial for urban development. In this context, maintaining and improving the efficiency of wastewater treatment plants is essential for recovering these water resources. Therefore, accurately modeling the performance of wastewater treatment plants at different stages is of great importance.
Methods: This study used an artificial neural network (ANN) to model the effluent quality of the Bushehr wastewater treatment plant. The input and output data included the parameters BOD, COD, EC, TSS, and TDS. The results demonstrated that the ANN algorithm performed well in modeling the wastewater quality parameters, though with varying success at different treatment stages.
Results: The modeling results showed that accuracy was highest for the EC parameter in the first stage, with a coef-ficient of determination (R²) of 0.84 during the test period. The best accuracy for the TDS parameter was achieved in the third stage. For the COD parameter, the network's accuracy was very favorable in the first stage. The highest accuracy for the TSS parameter was in the third stage during training (R² = 0.79), while the second stage yielded the best result during testing (R² = 0.65). Finally, for the BOD parameter, the highest accuracy was in the third stage.
The best performance for EC and COD was observed in the first stage, with an accuracy exceeding 80%, while for TDS, the highest accuracy was in the second stage.