Photovoltaic Time Series Aggregation Method Based on K-means and MCMC Algorithm

Establishment of photovoltaic power time series model is of great significance for long-term planning, annual dispatching and operation of power system. This paper proposes a photovoltaic power time series data aggregation method based on K-means clustering and Markov chain Monte Carlo (K-means-MCMC). First, the historical photovoltaic power data is clustered, and the optimal clustering number of clustering scenario samples is selected by calculating the DBI index. Then, a Markov chain Monte Carlo (MCMC) method is used to construct a state transition probability matrix between various scenario clusters. Finally, according to the state transition probability matrix, samples in the scenario cluster are sequentially extracted and a simulated photovoltaic time power aggregation series is generated. A provincial power grid was chosen as an example to compare the multiple evaluation index of the K-means-MCMC method and typical daily methods. The results show that the aggregation series generated by K-means-MCMC method can reflect the characteristics of photovoltaic power more accurately, which has substantial engineering value.