| Article ID: | iaor200952663 |
| Country: | Netherlands |
| Volume: | 23 |
| Issue: | 3 |
| Start Page Number: | 509 |
| End Page Number: | 529 |
| Publication Date: | Feb 2009 |
| Journal: | Water Resources Management |
| Authors: | Mehta Rama, Jain Sharad K |
| Keywords: | developing countries, fuzzy sets, neural networks |
Present paper is aimed to develop operation policy for a multi–purpose reservoir using Neuro–Fuzzy technique in an efficient way. Ramganga reservoir behind Ramganga dam, Kalagarh, India has been considered as a study reservoir. The developed policy minimizes the damage due to floods and droughts and determines optimum releases against demands for domestic supply, irrigation and hydropower generation for monsoon and non–monsoon periods. Three Fuzzy Rule Based (FRB) models for monsoon period and three for non monsoon period have been developed and tested. Actual releases have been used to formulate the general operation fuzzy rules. Releases computed from all developed models using Fuzzy Mamdani (FM) and ANFIS (Adaptive Neuro–Fuzzy Interactive System) Grid and Cluster have been compared and it was found that ANFIS–cluster gives the best results but FM is more users friendly. For any expected inflow, reservoir level and demand, release can be calculated using developed GUI windows of the models.