This document outlines a research project to develop improved models for forecasting reservoir inflows in the incremental basin of Itaipu using data-driven and hybrid techniques. The objectives are to create semi-distributed hydrological, artificial neural network, and hybrid models and evaluate their performance on short and medium term predictions. Methodologies include wavelet decomposition of inputs to ANNs, sensitivity analysis of ANN structures, and data preprocessing. Results will be made available through an internet-based platform to aid operational forecasting and allow further testing. Limitations include increased computational time for wavelet-ANN models and data availability challenges.