In this study, we compare the effectiveness of fuzzy logic, neural networks, and decision trees for
optimizing energy consumption in smart grids. We evaluate these techniques based on their forecasting
accuracy, load balancing efficiency, computational time, and flexibility. Our results reveal that neural
networks exhibit the highest forecasting accuracy at 92.4% and superior load balancing efficiency of
82.1%, though they require significant computational time (25 minutes). Fuzzy logic provides a balanced
performance with a forecasting accuracy of 85.2%, a load balancing efficiency of 78.5%, and a moderate
computational time of 12 minutes. It also scores highly in flexibility, demonstrating strong adaptability
to changing conditions. Decision trees, while the most computationally efficient with a processing time
of 8 minutes, show the lowest forecasting accuracy (80.1%) and load balancing efficiency (74.6%),
indicating limitations in handling complex energy management tasks. This comparison highlights the
strengths and trade-offs of each technique, offering insights into their suitability for real-time energy
optimization in smart grids.
