In response to the significant fluctuations in PV power generation due to weather changes, this paper proposes a time-series modeling approach. The proposed method integrates random weather sampling with fluctuation characteristic simulation techniques, effectively characterizing the uncertainty of PV power. Initially, the characteristics and the fluctuation of PV power under different weather conditions are analyzed, and the impact of weather factors on PV power fluctuations is quantified using the “no-shade coefficient”. Subsequently, based on statistical analysis of historical weather data, a Markov chain-based weather type transition model is constructed to accurately capture the transition patterns between weather types. On this foundation, combined with the Gaussian Mixture Model (GMM) and the Normal Distribution, mean value and fluctuation value models of the PV power no-shade coefficient are established, respectively. These two models collectively provide a detailed depiction of PV power fluctuations. Finally, the validity and accuracy of the proposed modeling method are verified using actual operational data from a PV power station in Wuqing, Tianjin. This method provides a modeling basis and algorithmic support for the new electricity system planning and production process simulation.
Read full abstract