A scenario-based two-stage stochastic optimization approach for multi-energy microgrids

With the increase in renewable energy penetration, the impact of uncertain factors on the efficient operation of multi-energy microgrids (MEMGs) is becoming more and more prominent. Considering the source-load uncertainties of MEMGs, a two-stage stochastic optimization approach based on scenario analysis is proposed in this paper. First, mixed distribution and conditional distribution were used to fit the forecast errors of wind power and multiple loads respectively, so as to provide basic data for scenario generation. Then, an improved K-means clustering algorithm based on relative entropy was used for scenario reduction. This algorithm ensured the scenario reduction speed while maintaining the probability distribution characteristics of the generated scenarios. Taking the day-ahead forecast and scenario analysis results of the sources and loads as inputs, a two-stage stochastic optimization model of MEMG based on random fluctuation stabilization was constructed. In the first stage, equipment outputs are formulated by deterministic optimization based on day-ahead forecasts. In the second stage, the forecast errors are regarded as fluctuations and combined with scenarized source and load variables, energy storage equipment are given priority to stabilize scenario fluctuations. At the same time, based on the conditional value at risk (CVaR), the outputs of energy supply and storage equipment can be flexibly adjusted under different risk preferences to realize the efficient operation of MEMG. The example simulation showed that the proposed stochastic optimization approach make better use of energy storage equipment, make scheduling plans according to different risk preferences to deal with uncertainty flexibly.