Adaptive energy loss optimization in distributed networks using reinforcement learning-enhanced crow search...

Area of Science:

• Electrical Engineering
• Computer Science
• Optimization Theory

Background:

• Modern power distribution networks increasingly integrate distributed generation (DG), presenting significant energy management challenges.
• Traditional optimization algorithms like Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and Crow Search Algorithm (CSA) often face limitations such as premature convergence and poor adaptability to real-time network variations.

Purpose of the Study:

• To propose and validate a novel Reinforcement Learning Enhanced Crow Search Algorithm (RL-CSA) for optimizing network reconfiguration and minimizing energy losses in power distribution systems.
• To enhance the efficiency, adaptability, and global search capabilities of existing optimization techniques for smart grid applications.

Main Methods:

• The research introduces the Reinforcement Learning Enhanced Crow Search Algorithm (RL-CSA), which dynamically refines its search trajectory using real-time feedback.
• The RL-CSA model was validated on the IEEE 33 and 69 Bus test systems, evaluating performance metrics including power loss reduction, voltage stability, execution time, DG utilization, and energy cost.

Main Results:

• RL-CSA achieved a significant 78% reduction in energy losses, limiting power loss to 5 kW (IEEE 33-Bus) and 8 kW (IEEE 69-Bus).
• The algorithm demonstrated superior speed, with execution times of 1.4 s (IEEE 33-Bus) and 1.8 s (IEEE 69-Bus), outperforming GA, PSO, and CSA.
• RL-CSA ensured high DG utilization (98%) and improved voltage stability (< 0.005 p.u.).

Conclusions:

• The proposed RL-CSA offers a robust and intelligent solution for real-time energy loss reduction and network optimization in distributed power systems.
• RL-CSA's dynamic adaptability and efficient exploration-exploitation balance make it a scalable and effective alternative for modern smart grid management.
• The algorithm's performance validates its capability to address the complexities introduced by distributed generation in power networks.