Predicting PV system electricity output is necessary for daily operational management and annual power system planning when integrating solar collector-based photovoltaic (PV) stations into micro grids. Tilting the panels at the ideal angle to maximize solar energy capture is necessary to maximize PV station production. This optimal tilt angle (OTA) must be predicted as it is a nonlinear function of the total solar radiation, diffuse solar radiation, and direct solar radiation. This research explores the use of feature selection-based artificial neural networks (ANN) with various machine learning algorithms to predict the OTA for PV systems at specific locations, aiming to maximize PV output in micro grids. The study identifies global solar radiation, diffuse solar radiation, clarity index, and global solar radiation on inclined surfaces as the most critical inputs for predicting OTA, while extraterrestrial radiation is deemed the least significant. Implementing the appropriate input variables significantly enhanced prediction accuracy from 38.59 % to 90.72 %. Among the neural networks evaluated, the Elman neural network demonstrated the greatest improvement.