Performance analysis of nature-based wastewater treatment (NBT) with native wetland species based on experimental and generalized regression neural network (GRNN) approach

Constructed wetlands as nature-based treatments bring more nature and natural features and processes to the cities and landscapes. These innovative solutions contribute to financial growth, promote employment and upgrade our contentment. Using constructed wetland systems to recycle and reuse wastewater provides an alternative water resource for various purposes. The vegetation in nature-based treatments acts as swampland filters and the microorganisms that thrive on the roots do most of the work, breaking down nitrogen and phosphorus with various toxic chemicals. In this study, the performance of six integrated vegetated and non-vegetated nature-based treatment units was studied at different retention times. Results exhibited that nature-based treatment units planted with Typha Latifolia performed better with a reduction efficiency ranging between 80 and 88% for BOD, 82–86% for COD, 67–72% for TP and 75–80% for TN, respectively, at all retention times. However, an unplanted unit with a hollow plastic tube (GPS) indicated a higher removal percentage in comparison with the GS unit without a hollow tube. Further, statistical analysis was also performed, which showed the reliability of the proportion removal of different parameters over a period of time with different nature-based treatment wetland units. Lastly, the Generalized Regression Neural Network (GRNN) was designed to estimate BOD and COD. In order to predict BOD and COD, GRNN was used, and the obtained RMSE was reduced from 0.3511 to 0.3504. The findings show that the application of GRNN is suitable for modelling wastewater treatment efficiency.