Optimal schedule for virtual power plants based on price forecasting and secant line search aided sparrow searching algorithm

Abstract

With a growing focus on the environment, the power system is evolving into a cleaner and more efficient energy supply infrastructure. Photovoltaic (PV) and storage are key assets for the power industry’s shift to sustainable energy. PV generation has zero carbon emission, and the integration of a substantial number of PV units is fundamentally important to decarbonize the power system. However, it also poses challenges in terms of voltage stability and uncertainty. Besides, the daily load and real-time price are also uncertain. As a prosumer, energy storage demonstrates the capacity to enhance accommodation and stability. The adoption of Virtual Power Plants (VPPs) emerges as a promising strategy to address these challenges, which allows the coordinated orchestration of PV systems and storage to participate power dispatch as a virtual unit. It further augments the flexibility of the power distribution system (PDS). To maximize the profit of VPP, a data-driven price forecasting method is proposed to extract useful information from historical datasets based on a novel LSTM-Transformer-combined neural network. Then, an improved sparrow searching algorithm (SSA) is proposed to schedule VPPs by combining the secant line search strategy. The numerical results, obtained from testing the model on IEEE 13-node and 141-node distribution systems, demonstrate the effectiveness and efficiency of the proposed model and methodology.