太陽光と需要の予測誤差を考慮して地域の電力運用を計画

Day-ahead operation of local energy communities (LECs) is affected by uncertainty in PV generation and electricity demand, which may compromise the feasibility of committed exchanges with the upstream grid. This paper proposes a forecast-residual-based chance-constrained optimization (CCO) framework that characterizes uncertain parameters through their forecasts and associated residual distributions. Under the adopted zero-mean Gaussian assumption, the resulting chance constraints admit an exact deterministic-equivalent reformulation in which uncertainty is embedded through analytical safety margins, preserving the mixed-integer linear structure of the original scheduling problem. The model jointly coordinates PV and BESS operation, low-voltage distribution network constraints, internal energy sharing, and day-ahead grid-exchange commitments. Once the physical schedule is determined, an ex-post allocation stage distributes the available community energy pool among users according to predefined participation coefficients, identifies post-allocation surpluses and deficits, and maximizes their internal matching while preserving the aggregate grid exchanges obtained from the CCO. The framework is evaluated on a reduced 206-node European low-voltage feeder with up to 55 community users and benchmarked against a two-stage stochastic programming formulation. Results show that, for forecast-error levels close to 5\%, the proposed CCO achieves operating outcomes comparable to the stochastic benchmark while retaining computational requirements close to the deterministic formulation. The ex-post allocation results further show that internal matching can reduce gross energy exchanges that require settlement with the upstream grid.