Renewable Energy Scenario Generation Method Based on Clustering of Fluctuation Characteristics and Optimizing the Number of States in MCMC Method

The increasing adoption of renewable energy sources necessitates power systems to effectively manage the uncertainty of renewable energy output. To address this issue, rational scenario generation methods are employed to depict future output scenarios. Markov chain Monte Carlo (MCMC) methods have gained popularity for generating renewable energy output scenarios. This study focuses on the generation of renewable energy output scenarios using MCMC methods. This paper first clustered daily output sequences based on shared fluctuation characteristics, utilizing historical data from a specific region. Thereafter, it proposed the state preference principle from the MCMC method to overcome the subjectivity of determining the number of states in traditional approaches. Then, it empirically analyzed the proposed method using actual PV output data from a province in southeast China. The findings indicate that scenarios generated with the preferred number of states better fit the fluctuation characteristics and probability distributions of the original data. Additionally, these scenarios enhance the accuracy of the MCMC model in portraying renewable energy uncertainty.