Research on Power System Optimization Dispatching and Power Market Trading Strategies Based on Deep Reinforcement Learning

The integration of renewable energy sources into power systems has made the optimization of dispatching and market trading approaches vital for preserving system dependability and economic efficacy. The research presents optimization tactics for power system dispatch and market trading through Deep Reinforcement Learning (DRL). A proposed model integrates key mechanisms of the power system, including power generation units, energy storage systems, external grids, and market-based instruments. A main challenge in power systems is the uncertainty of renewable energy generation, significant to unproductive dispatching and market volatility. The examination uses the Power System Dispatch and Market Trading dataset, with data preprocessing performed using Min-Max normalization to advance training performance. To address this issue, a dual approach is presented for optimizing power system dispatch, involving direct control of Thermostatically Controlled Loads (TCLs) and indirect control of price-responsive loads. An advanced deep reinforcement learning algorithm, the Optimal Power Reinforced Twin Deterministic Policy Gradient (OPRTDPG), is used to solve the optimization problem in the dynamic and uncertain situation. The OPRTDPG technique adjusts to real-time fluctuations in renewable energy output and market conditions, providing an optimal decision-making approach that reduces costs and improves system dependability. The results demonstrate that the OPRTDPG algorithm enhanced the power supply system stability by decreasing market price variations by 10%, improving energy use by 15%, and decreasing the system cost by 12%. The outcomes the efficiency of OPRTDPG in optimizing power system dispatch and market trading tactics in real-world power systems.