In this work, development of solar and micro hydro power short-term prediction model is described. The first stage of the process represents assortment and analysis of impact meteorological variables on prediction model. In the objective to reduce impact of meteorological variables the higher proportion of prediction is systematically calculated via additional variable, in both prediction models. Following that, the deviation to final result is predicted using artificial neural networks. Improved prediction model is tested on different weather conditions and in different seasonal periods. The analysis of prediction values are made, the results are graphically illustrated for representative days and estimated using the standard evaluation criteria. Satisfactory results are obtained by both prediction models.