An Enhanced Gray Wolf Optimizer and Its Application to Parameter Calibration in Open-Channel Flow

Parameter calibration is a key issue in open-channel flow models. Manning’s roughness coefficient, a comprehensive indicator of flow resistance, is highly variable and location-dependent. An optimization objective function is proposed for the parameter calibration model by minimizing the discrepancies between the simulated and measured water levels, and the parameter calibration problem was transformed into an optimization problem (OP). The gray wolf optimizer (GWO) is used widely to solve realistic OPs due to its simple parameters and easy construction. The GWO needs to improve exploitation ability and convergence speed for solving complex and computationally objective problems. Therefore, a novel improved GWO called the enhanced multistrategy gray wolf optimizer (EMSGWO) is proposed. A novel opposition-based learning strategy and a modified social hierarchy different from GWO are introduced to balance the computational efficiency, optimization capability, and computational cost. Several position update strategies, such as wandering hunting, Cauchy mutation, fast hunting, and modified tent chaos, are proposed to achieve better performance. Two sets of benchmark functions, five existing versions of GWO algorithms, and nine other well-known metaheuristic algorithms were used to prove the effectiveness of the EMSGWO. All experimental results validated the applicability, robustness, accuracy, and stability of the EMSGWO in solving complex problems. The EMSGWO was embedded into an open-channel flow model based on the Saint-Venant equations, and was applied to the hydrodynamic parameter calibration of two laboratory experiments and a section of the South-to-North Water Diversion Middle Routine Project. All the results indicated the good performance of the EMSGWO in automatic calibration of model parameters in open-channel flow.