Short-term Forecast-Augmented Surrogate Lyapunov-based Real-Time Dispatch of Hybrid Power Plants in Island Power Systems

Authors:
Preprint
Publisher website
Google Scholar
Abstract
Hybrid power plants (HPPs), which combine variable renewable generation and battery energy storage systems (BESSs), are emerging as critical flexibility resources for renewable-rich grids. Yet HPP real-time dispatch remains difficult in island power systems, given both renewable power uncertainty and BESS state-of-charge (SoC) management, where day-ahead (DA) forecast errors can invalidate DA schedules and purely observation-driven online control can be myopic. This paper proposes a short-term-forecast-augmented surrogate Lyapunov drift-plus-penalty algorithm for operator-oriented HPP dispatch. The method embeds imperfect look-ahead information directly into the Lyapunov decision rule by replacing the conditional-expectation drift-plus-penalty objective with a tractable forecast-based surrogate under a bounded-first-moment forecast error condition. The algorithm is embedded in a system-level dispatch framework that couples day-ahead unit commitments, real-time observations, and short-term forecasts. The framework also integrates expert knowledge informed operational flexibility and a generic incidence-matrix representation of heterogeneous HPP portfolios. Numerical studies on a realistic 30-day island power-system case with photovoltaic (PV) uncertainty show that the framework eliminates loss of load, PV curtailment, reserve shortage, and transmission-line violations under imperfect forecasts. Its average operating cost is only 0.37% above an ideal perfect forecast benchmark. Robustness tests show penalty-free operation for short-term forecast error standard deviations up to 40%. These results indicate that the proposed forecast-aware Lyapunov control can bridge online optimization and reliable real-time operation of renewable-rich island grids.

This publication has 21 references indexed in Scilit: