Improving predictive reliability and automation of smart grids using the StarNet ensemble model.

Area of Science:

• Electrical Engineering
• Computer Science
• Artificial Intelligence

Background:

• Smart grids require high predictive reliability and automation for stable electricity distribution.
• Existing methods may lack robustness and generalization across diverse grid conditions.

Purpose of the Study:

• To introduce the StarNet Ensemble Model, a web-based intelligent framework for enhancing smart grid stability.
• To develop an automated, real-time monitoring and prediction system for grid performance.
• To validate the model's effectiveness on both synthetic and real-world benchmark datasets.

Main Methods:

• Developed a stacking-based ensemble learning framework (StarNet Ensemble Model).
• Integrated a machine learning-driven graphical user interface (GUI) for automated operations.
• Utilized CatBoost, AdaBoost, Random Forest, SVM, and KNN as base learners with a Random Forest meta-model.
• Employed stratified 10-fold cross-validation for rigorous model evaluation.

Main Results:

• Achieved 99.43% accuracy on a synthetic dataset.
• Attained 98.94% accuracy on the UCI Smart Grid Stability Dataset.
• Reached 97.83% accuracy on the IEEE 14-Bus Test System.
• Demonstrated a cross-dataset transfer accuracy of 95.41%.

Conclusions:

• The StarNet Ensemble Model exhibits robustness and strong generalization capabilities.
• The framework effectively enhances predictive reliability and automation in smart grids.
• The model offers a viable solution for improving smart grid operational stability and efficiency.