As the integration of solar photovoltaic (PV) systems into modern power grids expands, accurate forecasting of PV output becomes increasingly crucial for maintaining grid reliability and informed energy planning. This article presents the development of a feedforward neural network (FNN) model, implemented using MATLAB/SIMULINK, to predict solar photovoltaic (PV) power output based on key environmental parameters. The model utilises data sourced from the PVGIS database, with input features optimised using the Minimum Redundancy Maximum Relevance (MRMR) algorithm. The network, trained via the Levenberg–Marquardt backpropagation method, was tested on datasets from Edinburgh, Scotland, and cross-validated with data from Abuja, Nigeria. Results show high prediction accuracy, with regression values near unity and minimal mean squared error during testing. However, limitations emerged in scenarios involving zero and low PV output values, indicating areas for future enhancement. Despite these challenges, the model demonstrates strong potential for real-world solar forecasting applications across varying geographical locations.