Fuzzy neural network (EFuNN) for modelling dissolved oxygen concentration (DO)

Abstract

The aim of this research is to propose a new fuzzy neural network based model, called evolving fuzzy neural network (EFuNN) that extends existing artificial intelligence methods for modelling hourly dissolved oxygen concentration in river ecosystem. To demonstrate the capability and the usefulness of the EFuNN model, a one year period from 1 January 2014 to 31 December 2014, of hourly dissolved oxygen (DO) and Water quality variables data collected by the United States Geological Survey (USGS), were used for the development of the models. Two stations are chosen: the bottom (USGS station no: 420741121554001) and the top (USGS station no: 11509370), at Klamath River above Keno Dam nr Keno, Oregon, USA. For comparison purposes, a multiple linear regression (MLR) model that was frequently used for predicting water quality variables in previous studies is also built. The inputs variables used for the EFuNN and MLR models are water pH, temperature (TE), specific conductance (SC), and sensor depth (SD). In both models, 60% of the data set was randomly assigned to the training set, 20% to the validation set, and 20% to the test set. The performances of the models are evaluated using root mean square errors (RMSE), mean absolute error (MAE) and correlation coefficient (CC) statistics. The lowest RMSE and highest CC values were obtained with the EFuNN model. The results obtained in the current study demonstrate the potential applicability of the proposed modeling approach in modelling dissolved oxygen concentration in river ecosystem.