Dual-penalty optimization framework for voltage and frequency regulation in virtual power plant

Abstract

To address the energy crisis and reduce environmental pollution, Electric Vehicles (EVs) have experienced rapid development on a global scale. Simultaneously, Virtual Power Plants (VPPs), which aggregate diverse renewable energy sources and loads, are emerging as a vital strategy to enable EVs to engage in the power market effectively. This paper proposes a novel integrated system that combines a Deep Convolutional Neural Network (Deep CNN) model for accurate forecasting of VPP operating costs and market demand with a Hybrid Kookaburra-based Zebra Optimization Algorithm (Hybrid KZOA) for optimal bidding strategies. The proposed Hybrid KZOA is developed with the integrated version of the Kookaburra optimization Algorithm (KOA) and the Zebra Optimization Algorithm (ZOA). The proposed system incorporates two distinct penalty functions, one addressing voltage regulation to ensure stable voltage levels, and the other focusing on frequency regulation to maintain grid frequency within permissible limits. Bidding strategies are optimized across day-ahead, real-time, and balance markets using the integrated system. Finally, the effectiveness of the VPP bidding model was assessed in MATLAB/Simulink through various scenarios with minimum generation cost and maximum revenue cost.