This is the authors' abstract. We don't add key points for this paper.
This paper presents a critic-only Q -learning framework for optimal energy management in systems with stochastic renewable energy inputs and dual battery storage. The methodology addresses a fundamental challenge in renewable energy integration: maintaining stable energy balance under highly uncertain and nonlinear power generation dynamics. We formulate the problem as a tracking Hamilton-Jacobi-Bellman (HJB) equation around a dynamic equilibrium point, where the critic network learns to approximate the optimal value function through iterative policy evaluation. The framework’s distinguishing feature is its ability to handle continuous state-action spaces without requiring an explicit actor network, thereby simplifying the learning architecture while preserving optimality guarantees. To demonstrate the approach, we consider wave-powered autonomous surface vehicles as a representative case study, where mechanical energy harvesting introduces significant stochasticity. Comprehensive simulations under various energy input profiles validate the algorithm’s convergence properties and adaptive performance. The proposed framework offers a systematic solution for energy management problems across diverse renewable energy applications, from marine systems to terrestrial microgrids with intermittent generation sources.
Answers are grounded in this paper and cite their sources.
No takes yet. Share an insight, caveat, or question.
Dongdong Zhang (2025) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: