Assessing Storm Water Detention Systems Treating Road Runoff with an Artificial Neural Network Predicting Fecal Indicator Organisms
2010
Kazemi Yazdi, S. | Scholz, M.
This paper examines whether multiple regression analysis and neural network models can be applied successfully for the indirect prediction of the runoff treatment performance with water quality indicator variables in an experimental storm water detention system rig. Five biologically mature experimental storm water detention systems with different designs treating concentrated gully pot liquor (spiked with dog droppings) were assessed. The systems were located on The King's Buildings campus at The University of Edinburgh and were monitored for a period of 18 months. Multiple regression analyses indicated a relatively successful prediction of the biochemical oxygen demand and total suspended solids for most systems, but due to a relatively weak correlation between the predictors and both microbial indicators, multiple regression analyses were not applied for the prediction of intestinal enterococci and total coliform colony-forming units. However, artificial neural network models predicted microbial counts relatively well for most detention systems.
显示更多 [+] 显示较少 [-]