Firefly Optimization Algorithm Effect on Adaptive Neuro-Fuzzy Inference Systems Prediction Improvement of Sediment Transport in Sewer Systems

To optimize and economical designing of sewer systems, minimum flow velocity through sewer should be determined as the input sediment to the channel, is not deposited (limiting velocity at the limit of deposition). In this study, a new hybrid method based on a combination of Adaptive Neural Fuzzy Inference System (ANFIS) and Firefly Algorithm (FFA) was presented to predict limiting velocity at the limit of deposition in sewer systems. The effective parameters in limiting velocity prediction are surveyed, and the four dimensionless parameters namely as volumetric sediment concentration (C_V), a relative of the median diameter of particle size to the hydraulic radius (d/R), dimensionless particle number (D_gr) and flow resistance (λ_s) was selected. Coefficient determination (R²), root relative square error (RRSE), mean absolute percentage error (MAPE), and BIAS are used for calculating the models’ performance. Three sets of experimental data which are presented in the literature are employed to evaluate the models. To examine all possible input combinations, 11 different models were introduced. The limiting velocity was presented by all input combination using ANFIS-FFA. The results indicated that the use of d/R and C_V as input variables lead to most accurate predictions among all models (R²=0.97; MAPE=5.78; RRSE=0.08; BIAS=0.11). The results of ANFIS-FFA was compared with ANFIS and regression-based equations.