Dissolved oxygen modeling of an urban stream using grid partitioning and
subtractive clustering fuzzy techniques
Abstract
Dissolved oxygen is one of the prime parameters for assessing the water
quality of any stream and the health status of aquatic life. The
dissolved oxygen present in the water body plays an essential role in
deciding water treatment processes to enhance water quality up to the
design standards for the specified water use. Thus, the accurate
estimation of dissolved oxygen concentration is necessary to evolve
measures for maintaining the riverine ecosystem and designing the
appropriate water treatment plants. Machine learning techniques are
becoming useful tools for the prediction and simulation of water quality
parameters. With these viewpoints, a study was carried out in the Delhi
stretch of Yamuna River, India, and physiochemical parameters were
examined for five years to simulate the dissolved oxygen using different
machine learning techniques. Simulation and prediction competencies of
ANFIS grid partitioning (ANFIS-GP) and ANFIS subtractive clustering
(ANFIS-SC) were tested on various water quality parameters. Variation in
dissolved oxygen was examined on various combinations of parameters.
ANFIS-GP has been designed using the Gaussian function, and ANFIS-SC
works on the likelihood of cluster centers. Results obtained from the
models were evaluated using root mean square error (RMSE) and
coefficient of determination (R2) to identify the optimum solution and
appropriate combination of parameters that simulate the observed
dissolved oxygen. Results of ANFIS-GP and ANFIS-SC indicate that both
the models produce suitable solutions for the prediction; however,
ANFIS-GP outperforms the ANFIS-SC and could act as a useful tool for
defining, planning, and management of water quality parameters.