FORECAST OF RESERVOIR SEDIMENT TRAP EFFICIENCY USING ARTIFICIAL NEURAL NETWORKS

Qamar Sultana, M. Gopal Naik

Abstract: The reservoirs all over the world experience the sedimentation problem. There are a number of models available for the estimation of reservoir sedimentation and the simple way to estimate it is through the knowledge of the trap efficiency of the reservoir. In this study an artificial neural network (ANN) model developed using Matlab software is used to estimate the trap efficiency of the reservoir. One of the large reservoir in Telangana State , Sriramsagar reservoir located at Pochampadu village in Nizamabad district is taken as a case study. The input parameters used are annual inflow, annual rainfall and age of the reservoir and the output parameter considered is the trap efficiency(Te) of the reservoir for twenty six years i.e. from 1987 to 2012. A conventional regression analysis is conducted, relating the output parameter (Te) to the input parameters. From the values of the performance statistical indicators, it is found that the ANN model predicted the trap efficiency of the reservoir with better accuracy and less effort than that of the conventional method. Further since the forecast of the hydrologic data is very important for the engineers for the management of the available water resources, the time series forecasting of annual reservoir inflows and annual rainfall has been done and subsequently trap efficiency is predicted for the same series data for the next 26 years i.e. from 2013 to 2038 using ANN technique.

Keywords: Forecast, Sedimentation of reservoir, Trap Efficiency method, Artificial Neural Network technique , Conventional Method.

DOI: https://doi.org/10.15623/ijret.2016.0514001