Deep Learning Based Muti-Objective Reactive Power Optimization of Distribution Network with PV and EVs.

Area of Science:

• Electrical Engineering
• Power Systems Engineering
• Renewable Energy Integration

Background:

• Increasing integration of photovoltaic (PV) and electric vehicle (EV) charging stations causes significant line loss and voltage deviation in distribution networks.
• Traditional reactive power compensation methods are insufficient for modern grid demands, impacting operational safety and cost-efficiency.

Purpose of the Study:

• To propose novel reactive power regulation methods utilizing the potential of PV and EV systems.
• To establish accurate reactive power regulation models for PV and EV systems.
• To evaluate the dynamic adjustable capacity of reactive power for these distributed energy resources.

Main Methods:

• Development of reactive power regulation models for PV and EV systems.
• Introduction of dynamic evaluation methods for assessing reactive power adjustable capacity.
• Optimization of the proposed models using five distinct intelligent algorithms.
• Approximation of optimal solutions using deep learning, focusing on minimizing line loss and voltage deviation.

Main Results:

• Deep learning models demonstrated a remarkable ability to approximate the Pareto front generated by intelligent optimization algorithms.
• The proposed models effectively address the challenges of reactive power optimization in distribution networks with high PV and EV penetration.

Conclusions:

• Leveraging the reactive power regulation capabilities of PV and EV systems offers a viable solution for enhancing distribution network performance.
• Deep learning provides an efficient and accurate approach for predicting optimal reactive power dispatch strategies in power grids.