Calibration of Linear Muskingum Model Coefficients and Coefficient Parameters Using Grey Wolf and Particle Swarm Optimization

The anticipation of flooding is crucial for River Engineering. The Muskingum technique is among the most renowned investigations on this topic. There are many versions of this model. This work calibrated the parameters of the existing linear Muskingum technique using the Grey Wolf Optimization (GWO) algorithm and Particle Swarm Optimization (PSO). The calibrating procedure is developed using two methods. Initially, as documented in the literature, the parameters employed in the computation of the coefficients are calibrated, and these parameters are subsequently utilized in the formulas to determine the coefficients. In the second instance, the coefficients are calibrated directly. The particle and iteration counts utilized for calibrations are modified in the study. The impact of GWO and PSO on this issue is also examined through the analysis of these figures. The study utilized the Mollasani flood case documented in the literature. The Percentage Absolute Error serves as the error metric. The results acquired in this phase are compared. This comparison utilizes the Taylor diagram and Percentage Absolute Error. The Standard Deviation of the results pertaining to the model’s reliability is also analyzed. Thus, it is observed that the GWO and PSO algorithms, which are heuristic optimization methods for parameter estimation, have alleviated complex scenarios and are near-accurate outcomes. GWO yields superior results compared to PSO. Consequently, it is noted that the GWO and PSO algorithms, which are heuristic optimization techniques for parameter estimation, have mitigated complicated scenarios and produced near-accurate results. Also, GWO produces more favorable outcomes than PSO. The error margin of each calibration method is comparable; however, the particle and iteration counts employed varies. This difference leads to a change in the duration necessary for coefficient computations. Instead of finding the coefficients directly, calibrating the formula variables used to find the coefficients is more effective in achieving fast and accurate results.