Application of ANN for Reservoir Inflow Prediction and Operation

Abstract

Artificial neural networks (ANNs) are new computing architectures in the area of artificial intelligence. The present study aims at the application of ANNs for reservoir inflow prediction and operation. The Upper Indravati multipurpose project, in the state of Orissa, India, has been selected as the focus area. The project has primarily two objectives: To provide irrigation to 128,000,000 ha of agricultural land and to generate 600 MW of electric power. An autoregressive integrated moving average time-series model and an ANN-based model were fitted to the monthly inflow data series and their performances were compared. The ANN was found to model the high flows better, whereas low flows were better predicted through the autoregressive integrated moving average model. Reservoir operation policies were formulated through dynamic programming. The optimal release was related with storage, inflow, and demand through linear and nonlinear regression and the ANN. The results of intercomparison indicate that the ANN is a powerful tool for input-output mapping and can be effectively used for reservoir inflow forecasting and operation.