Accurate solar irradiance forecasting is essential for reliable integration of solar PV generation into modern power systems. This review analyzes various Machine Learning (ML) techniques such as ANN, SVR, Random Forest, Decision Tree, and ensemble models for predicting solar irradiance. The study examines the use of historical irradiance and meteorological parameters such as temperature, humidity, cloud conditions, and wind speed. ML techniques can effectively capture nonlinear relationships and improve forecasting accuracy. Accurate irradiance prediction supports PV power forecasting, energy storage scheduling, grid stability, and renewable energy management. The review also discusses key challenges and future directions for developing robust ML-based solar irradiance forecasting systems.