PHOTOVOLTAIC ENERGY YIELD PREDICTION USING AN IRRADIANCE FORECAST MODEL BASED ON MACHINE LEARNING FOR DECENTRALIZED ENERGY SYSTEMS

摘要

Over the past few years electricity generation costs for PV technology have dropped massively. Since, at the same time, PV module efficiencies have increased significantly, the market for building-applied PV systems has dramatically changed and in many countries it has become a de facto standard to use PV as the main source for the building's energy needs. Because the power output of PV systems is fluctuating along with solar irradiation, advanced energy storage and management systems are necessary to cover the building energy demand on a stable basis. This paper presents a novel 'gray-model' approach to the estimation the forecast of PV energy systems. It is based on machine learning for solar irradiance forecasting and physical-mathematical models to simulate the PV system itself. The paper presents a comparison between simulated and real-life energy production data of a sample PV system.