Simulation of Re-Aeration Coefficient Using Anfis and Arima Models

Abstract

AbstractIdentification of the assimilative capacity of the river becomes point of interest now a days, as the rivers are receiving a significant amount of wastewater load from urban agglomerations. Re-aeration represents the process of interaction among the air–water interface to absorb the oxygen from the atmosphere and indicates the capacity of the water to hold the pollutant without affecting the state of the river. The soft computing technique, adaptive neuro-fuzzy inference system (ANFIS) is applied to identify the re-aeration coefficient of Yamuna River, Delhi. The hybrid model is developed using ANFIS and autoregressive integrated moving average (ARIMA). The results obtained from the ANFIS are combined with the time series modeling technique ARIMA model. Combination of various hydraulic and water quality parameters were used to design the input for models. Takagi–Sugeno (TS) technique was used to design the ANFIS models and autocorrelation (ACF) and partial autocorrelation (PACF) functions were used for ARIMA model. Performance of the ANFIS models was measured using coefficient of determination (R2), root mean square error (RMSE), and coefficient of correlation (R). Several ANFIS models were designed and results of best model with least RMSE and highest R2 are used as input to ARIMA model to obtain the re-aeration coefficient for each sampling location. Results indicate that ANFIS model with combination of hydraulic and water quality parameter provides the best fit model with both the spatial and temporal variations.KeywordsRe-aeration coefficientYamuna riverWastewaterANFISARIMA