Abstractperformance of a constructed wetland wastewater treatment plant (CWWTP). The model assesses the Biochemical Oxygen Demand (BOD) concentration at outlet of a treatment plant. Training of ANN models was based on experimental results of a pilot plant study in India. The data used in this work were obtained under various hydraulic and BOD loading. Regular records of BOD were made at inlet, and outlet levels through various stages of the treatment process for over 18 months. The ANN-based models were found to provide an efficient and a robust tool in predicting CWWTP performance. Keywords: neural networks, constructed wetland, model studies, prediction, optimization, biochemical oxygen demand.
IntroductionThe proper operation and management of constructed wetland wastewater treatment plants (CWWTP) is receiving attention because of the rising concern about environmental issues and growing importance of sustainable and natural wastewater treatment techniques. Improper design and operation of a CWWTP may cause serious environmental and public health implications, as its effluent may contaminate receiving water body, causing severe aquatic pollution and spread various water born diseases. For proper design and assessment of quality of non-conventional wastewater treatment and thereafter to conserve the receiving water bodies, reliable prediction of effluent from Constructed Wetland (CW) is essential. A better control can be achieved by developing a mathematical tool for predicting the plant performance based on past observations of certain key parameters. However, modeling a CWWT is a difficult task due to the complexity of the treatment processes. The complex physical, biological and chemical processes involved in constructed wetland treatment process exhibit non-linear behaviors, which are difficult to describe by linear mathematical models. This paper presents predictive models based on the concept of neural networks (NN).Artificial intelligence concepts have been successfully used in a range of engineering, environmental, and financial problems [1,2]. The NN-based model that was applied to a pilot plant study of CW in India have performed consistently well in the face of varying accuracy and size of input data. Using these models, the planner and decision maker can easily make assessment of the expected plant effluent.
Constructed wetland pilot plant and experimental set upWetlands are considered as low-cost alternatives for treating municipal, industrial, and agricultural effluents. Constructed wetlands (CW) are preferred because of low maintenance, shock loading absorbance capacity and less energy consumption [3]. They may be classified as surface flow marshes, vegetated subsurface flow beds, submerged aquatic beds, and floating leaved aquatics [4] This new developing technology may offer a low cost and low maintenance alternative for treatment of domestic wastewater which is especially suitable for developing countries [5,6].The scheme of the units designed in New Delhi is presented in Fig. 1. Wastew...