Applying GMDH-type neural network for modeling and prediction of turbidity and free residual aluminium in drinking water
Alahyar Daghbandan, Saba Khalatbari, Mohammad Mahdi Abbasi · Desalination and Water Treatment · 2019
ABSTRACT Water has a fundamental role in human life. The proper quality and quantity of water is therefore an important issue. Coagulation and flocculation are essential processes for turbidity removal from drinking water. Metals such as aluminium have been implicated in the pathogenesis of Alzheimer’s disease. In this study, group method of data handling (GMDH)-type neural networks have been used for modeling and prediction of turbidity and free residual aluminium in drinking water. To validate the proposed model, a case study was carried out based on the data sets obtained from Guilan WTP. For modeling, the experimental data were divided into train and test sections (70% for training and 30% for testing). Eventually, the results of modeling were compared with experimental data and demonstrated good data compliance, with the coefficient of determination ( R 2 ) in GMDH-type network was 0.8239 and 0.9138 for residual turbidity and residual aluminium, respectively. Moreover, the results of error analysis showed good performance of the proposed models, in this regard can be referred to the mean square error results which obtained 0.0248 for residual turbidity and 0.00000438 for residual aluminium.