Evaluating Model Fit for a Time Series Solar Irradiance and DC Power Data to Forecast Accuracy Measures

Abstract

Solar-powered photovoltaic system provides an unpolluted energy solution to current global warming. The solar input parameters and weather data are important to configure the desired power output. Installation at site, system degradation, and lifetime reliability analysis are crucial for long-term power optimization. Statistical analysis of solar data is crucial to forecast the behavioral trend of photovoltaic system. Mean absolute percentage error predicts accuracy for forecasting the trend in statistics which is widely used to due to scale independency and interpretability. Different models are considered to fit the actual curve and to reduce the mean absolute percentage error, mean square deviation, and mean absolute deviation.