Machine learning optimization for hybrid electric vehicle charging in renewable microgrids.

Area of Science:

• Electrical Engineering
• Computer Science
• Artificial Intelligence

Background:

• Renewable microgrids are crucial for enhancing power system security, reliability, and power quality.
• Integrating solar and wind energy sources in microgrids helps reduce greenhouse gas emissions.
• Managing energy demand, particularly from hybrid electric vehicles (HEVs), is a key challenge in microgrid operation.

Purpose of the Study:

• To propose a machine learning-based energy management system for renewable microgrids.
• To model and manage the impact of hybrid electric vehicle (HEV) charging demand.
• To optimize microgrid operation through coordinated and intelligent charging strategies.

Main Methods:

• Utilized Gaussian Process (GP) for modeling HEV charging demand.
• Developed a novel optimization method inspired by the Krill Herd Algorithm (KHA) for energy management.
• Implemented a self-adaptive modification within the KHA for tailored solutions.
• Simulated the proposed methods on an IEEE microgrid.

Main Results:

• Achieved a low Mean Absolute Percentage Error (MAPE) of 1.02381 for predicting total HEV charging demand.
• Demonstrated the efficiency of both coordinated and intelligent charging scenarios.
• Showcased a reduction in microgrid operation costs with the intelligent charging strategy compared to coordinated charging.

Conclusions:

• The proposed machine learning approach effectively manages energy in renewable microgrids with reconfigurable structures.
• Intelligent HEV charging offers significant operational cost savings compared to coordinated charging.
• The integration of GP and KHA provides a robust solution for microgrid energy management challenges.