|
Abstract
|
The management improvement of irrigation networks is inevitable because of limited water resources and competition in various sectors of industry, agriculture, environment, and urban water consumption. Application of appropriate water delivery scheduling is one of the methods to improve irrigation management. Several delivery methods are practiced in irrigation networks according to water availability, type of network control system, and exiting capacities. One promising delivery method is the on-request system, which could be applied manually in irrigation networks. High performance of delivery methods arise from the determination of appropriate operation instructions. In previous studies, the Fuzzy SARSA learning (FSL) method was applied to determine operation instruction for on-request delivery in a canal of the Aghili irrigation network located in southwest Iran (SARSA is the definition of state, action, reward, state, action). In this work, the conventional manual operation on the same canal was evaluated, and the results were compared with FSL. The results for both methods in the normal condition of the network were acceptable because the adequacy and efficiency indicators were more than 90%, and the maximum deviation of water depth was less than 10%. However, in the circumstance of water shortage, the results indicated that applying the FSL method provided good and valid results, so that the index of adequacy was almost 0.7 for all blocks. Therefore, the FSL is able to distribute the water more equitably among the consumers in water shortage conditions.
|