Two-stage stochastic scheduling of virtual power plant based on transactive control

Thermostatically controlled loads (TCLs) have huge thermal inertia and are promising resources to promote the consumption of renewable energy sources (RESs) for carbon reduction. Thus, this paper employs the virtual power plant (VPP) to regulate TCLs to address the problems caused by RESs. Specifically, a two-stage VPP scheduling framework based on the multi-time scale coordinated control of TCLs is proposed to address the forecast errors of variable RES power output. In the first stage (hour time scale), TCLs are controlled as virtual generators to mitigate the forecast errors between the hour-ahead and day-ahead RES power. In the second stage (minute time scale), TCLs are regulated as virtual batteries to mitigate the forecast errors between the intra-hour and hour-ahead RES power. To respect the wills and preferences of end-users, a transactive energy (TE) market within VPP is built to guide TCL behaviors via the price mechanism. Moreover, a stochastic VPP schedule using the Wasserstein-metric-based distributionally robust optimization method is developed to consider RES power uncertainties, and its solution process is transformed into a computationally tractable mixed-integer linear programming problem based on the affine decision rule and the duality theory. The proposed method is effectively validated by the comparison with robust optimization and stochastic optimization. Simulation results demonstrate that the proposed two-stage VPP scheduling method employs TCL flexibilities more comprehensively to mitigate RES output power forecast errors in VPP operations.