Risk-averse multi-stage stochastic programming for real-time optimal market strategy of virtual power plant

This paper implements a multi-stage stochastic programming method for the operational cost reduction of the virtual power plant (VPP). Thinking over the uncertainty of future prices at the current time interval, nonanticipativity constraints are implemented to ensure that the VPP decision-making process remains a continuous sequence. Additionally, risk management plays a crucial role in formulating optimal market strategies for the VPP considering the risk-averse. In view of the Conditional-Value-at-Risk (CVaR), the original objective of the scheduling problem is revised to minimize a risk-related function associated with the VPP’s operational costs. This approach enables the risk-averse VPP to integrate a non-differentiable risk index and various operation constraints. Experimental results indicate that the proposed method effectively accounts for the influence of future uncertainties on market strategy and also explains the effects of risk aversion.