Simulating and predicting the performance of a horizontal subsurface flow constructed wetland using a fully connected neural network
Pengyu Li,Tianlong Zheng,Lin Li,Xiuyuan Lv,WenJun Wu,Zhining Shi,Xiaoqin Zhou,Guangtao Zhang,Yingqun Ma,Junxin Liu
DOI: https://doi.org/10.1016/j.jclepro.2022.134959
IF: 11.1
2022-11-05
Journal of Cleaner Production
Abstract:Constructed wetland systems, as an engineered ecological system, are being increasingly employed for wastewater treatment. However, owing to the complex incentives for pollutant removal in ecological treatment systems, it is challenging to simulate and optimize the operation of constructed wetlands to advance ecological wastewater treatment systems. In this study, a horizontal subsurface flow constructed wetland (HSCW) system was constructed and applied to a rural wastewater treatment system. Reeds ( Phragmites australis ) were planted in the HSCW to remove pollutants from the wastewater. Further, a fully connected neural network (FCNN) was designed based on the Adam optimization algorithm with weather conditions, quality, and quantity of influent and effluent as input to simulate and predict the performance of the HSCW. The results of the FCNN simulation analysis showed that the relative errors of the simulated concentrations of COD cr , NH 4 + -N, total nitrogen (TN), and total phosphorus (TP) for the FCNN model were 8.07 ± 10.73%, 18.34 ± 17.75%, 9.90 ± 11.91%, and 9.47 ± 10.98%, respectively. The mean absolute errors (MAEs) of COD cr , NH 4 + -N, TN, and TP for the FCNN model were 2.17, 1.06, 1.21, and 0.54, respectively. The root-mean-squared errors (RMSEs) of COD cr , NH 4 + -N, TN, and TP for the FCNN model were 3.91, 2.05, 2.22, and 0.80, respectively. The correlation coefficients (R 2 ) of COD cr , NH 4 + -N, TN, and TP for the model were 0.99, 0.91, 0.92, and 0.82, respectively. These results indicate that the model performed well. Sensitivity analysis results also showed that temperature, solar radiation intensity, and rainfall had a strong impact on the model accuracy. This study verifies that an artificial neural network can effectively reflect the nonlinear function of each factor and is suitable for simulating HSCW treatment for wastewater under various conditions, providing a new optimization method for wastewater ecological treatment systems.
environmental sciences,green & sustainable science & technology,engineering, environmental