Day-ahead scheduling strategy of virtual power plant under uncertainties

Abstract

When designing a day-ahead scheduling strategy, virtual power plant (VPP) faces various uncertain factors, including price uncertainty, forecasting error of renewables, load fluctuation, etc. In order to obtain the optimal tradeoff between economy and reliability, this paper presents a chance constrained programming (CCP) approach to the scheduling problem, where reserve requirement is satisfied with a certain confidence level (CL). Different from a pre-given CL in most literature, this paper proposes a method to judge the optimal CL, hoping to provide references for operator in optimizations involving CCP. A satisfaction function is introduced to depict VPP's satisfaction degree under different CLs. Meanwhile, the function reflects VPP's distinct attitudes toward economy and reliability. A matrix real-coded genetic algorithm combined with Monte Carlo simulation is used to solve the model. Numerical tests are performed in a VPP system, and the best CL is determined through comparing VPP's satisfaction degree under different cases.