Abstract
Rotational canal scheduling in arid irrigation districts must satisfy coupled water-volume, timing, and hydraulic constraints while balancing modeled seepage loss and the dispersion of trunk-canal discharge levels. A constrained model with a weighted composite objective was solved using a sigmoid-adaptive genetic algorithm (SIGGA) integrating adaptive operators, feasibility-first comparison, and diversity restoration. The framework was evaluated in the Zhangye (Xijun) and Shihezi oasis irrigation districts, with two irrigation-quota scenarios in Shihezi. For Zhangye, φ= 0.5 gave the closest empirical compromise among five tested weights. Under an equal function-evaluation budget, SIGGA produced a median final composite-objective value 3.9% lower than that of the hybrid particle swarm optimization-simulated annealing genetic algorithm (GSPSO), the closest benchmark, and showed low run-to-run variability under fixed inputs. Removing feasibility-first comparison caused the largest increase in the final composite-objective value. Sensitivity analyses showed that outcomes varied with model inputs, temporal resolution, and adaptive-operator settings. The representative Zhangye schedule satisfied all modeled constraints and completed the modeled rotation in 10 d, compared with the reported 25-d reference schedule, with a canal water-use efficiency of 0.843. The same formulation and prespecified settings yielded feasible schedules for both Shihezi quota scenarios without quota-specific retuning. Across the original 60 jointly perturbed Zhangye scenarios, all 180 scenario-specific reoptimization runs were feasible, whereas the fixed nominal schedule remained feasible in 52 scenarios. An enlarged replay-only analysis yielded median fixed-schedule feasibility proportions of approximately 86.5% at the larger sample sizes within the prescribed bounded scenario space. The framework can support generation and updating of candidate rotational schedules within the examined oasis open-channel systems, although field validation and testing in more diverse canal networks remain necessary.