Comparing Fuzzy SARSA Learning and Ant Colony Optimization Algorithms in Water Delivery Scheduling under Water Shortage Conditions

Abstract

Water delivery scheduling was investigated in this study using fuzzy state, action, reward, state, action (SARSA) learning (FSL) and ant colony optimization (ACO) methods to find the advantages of a new robust model (FSL) over a conventional model (ACO) in both normal and emergency conditions. The mathematical models of these methods were developed. Three water shortages of 10%, 20%, and 30% were considered in the East Aghili canal, Iran, for the simulation process. Water depth and delivery indicators were used for evaluating the performance of the developed models. The results revealed that the FSL and ACO methods offered almost the same performance for the normal operation condition with high and acceptable indicators. However, the FSL method outperformed the ACO method in terms of performance in three considered emergency operations. It can be concluded that the FSL, as a new method, can schedule water delivery efficiently, adequately, equitably, and dependably. Furthermore, the FSL method is likely to lead to less maximum absolute error (MAE) and integral of absolute magnitude of Error (IAE) in comparison to the ACO method and is therefore recommended.