Enhancing Optimization Algorithms for Controlling Large-Scale Solar and Storage Participation in Awarded Energy Market Bids

This paper investigates the development and improvement of optimization algorithms for large-scale solar photovoltaic (PV) and battery energy storage systems (BESS) participating in competitive energy markets. While traditional methods (e.g., linear programming (LP), dynamic programming (DP), and mixed-integer linear programming (MILP) effectively handle many scheduling challenges, they often struggle under rapidly changing prices, solar intermittency, battery cycling, temperature, and high dimensionality. To address these limitations, we propose a hybrid framework that integrates reinforcement learning (RL) with classical optimization to dynamically optimize bids and dispatch. We implement and test Deep Q-Network (DQN), Proximal Policy Optimization (PPO), and Soft Actor-Critic (SAC) agents in a simulation environment driven by public datasets from CAISO, PJM, NREL NSRDB and SAM, and EIA. Simulation results demonstrate significant improvements in profitability, bid accuracy, and flexibility compared to baseline methods, with the RL-based strategies achieving up to 98 % of the ideal cost reduction from advanced predictive solutions. This work paves the way for more adaptive, resilient, and profitable renewable energy participation.