Work

Optimal Scheduling of a Wind–Solar-Coupled Regional Grid-Side Power System with CAES–BESS Hybrid Energy Storage Based on Reinforcement Learning

Year: 2026
Type: article
Abstract: Large-scale integration of wind and photovoltaic generation increases uncertainty in regional grid-side power systems and raises new requirements for coordinated hybrid energy storage scheduling. This study develops a reinforcement-learning-based scheduling framework for a wind–solar-coupled regional grid-side power system with compressed air energy storage and battery energy storage (CAES–BESS) hybrid storage. The model represents electricity-side operation, including renewable generation, regional electric load, CAES and BESS operation, grid exchange, and time-of-use electricity pricing. The scheduling task is formulated as a Markov decision process in which the agent observes CAES SOC, BESS SOC, time, renewable output, load, net load, electricity price, and outputs continuous CAES/BESS charging and discharging actions. DSAC is integrated with storage operating constraints and normalized reward terms for grid-interaction cost, grid-power deviation, storage throughput, and reverse power. Its performance is examined using DDPG, PPO, a no-storage reference, and a deterministic optimization benchmark. The cost-smoothing trade-off is assessed using separate engineering metrics.
Source: Energies
Language: English
Open Access status: gold
Estimated APC paid: $3,137
OpenAlex ID: W7235751750
FWCI:
Cites: 36
Cited by:
Related to: 10