AI Tools Applied to Power System Analysis
Faculty Mentor Information
Dr. Mary Everett, University of Idaho; and Dr. John Shovic, University of Idaho
Presentation Date
7-15-2026
Abstract
This research focused on developing a computational model that uses data from PowerWorld, a power system simulation software, to optimize power grid parameters for improved efficiency, reliability, and performance. Because power grids are highly complex systems, this project explored the use of a Genetic Algorithm (GA) to analyze system behavior, optimize operating conditions, and predict potential future problems, including blackouts and other grid failures.
The methodology began by manually running PowerWorld simulations to create a baseline dataset. This dataset was then used as input for the Genetic Algorithm, which searched for more efficient and accurate solutions. The results were evaluated through graphical analysis, allowing comparisons between the baseline simulations and the optimized solutions generated by the algorithm.
The experimental results showed that adjusting the reactive power (Mvar) output of generators connected to different buses helped reduce the likelihood of blackouts and power outages on transmission lines. While the Genetic Algorithm successfully identified improved operating conditions, the optimization process required a significant amount of computation time because it evaluated many possible solutions. Additionally, when manually validating the algorithm's recommendations, testing was limited to the removal of only one transmission line at a time rather than multiple simultaneous line outages.
The findings indicate that the Genetic Algorithm was effective at optimizing the system for single-line contingency scenarios but was not designed to analyze multiple simultaneous line failures. Furthermore, some optimized solutions did not consistently maintain bus voltages within the desired range of 1% to 3% of the target voltage, indicating opportunities for further refinement of the optimization process. Overall, this research demonstrates that Genetic Algorithms can improve power grid analysis and optimization while highlighting areas for future work, including faster optimization techniques, multi-line contingency analysis, and improved voltage regulation.
AI Tools Applied to Power System Analysis
This research focused on developing a computational model that uses data from PowerWorld, a power system simulation software, to optimize power grid parameters for improved efficiency, reliability, and performance. Because power grids are highly complex systems, this project explored the use of a Genetic Algorithm (GA) to analyze system behavior, optimize operating conditions, and predict potential future problems, including blackouts and other grid failures.
The methodology began by manually running PowerWorld simulations to create a baseline dataset. This dataset was then used as input for the Genetic Algorithm, which searched for more efficient and accurate solutions. The results were evaluated through graphical analysis, allowing comparisons between the baseline simulations and the optimized solutions generated by the algorithm.
The experimental results showed that adjusting the reactive power (Mvar) output of generators connected to different buses helped reduce the likelihood of blackouts and power outages on transmission lines. While the Genetic Algorithm successfully identified improved operating conditions, the optimization process required a significant amount of computation time because it evaluated many possible solutions. Additionally, when manually validating the algorithm's recommendations, testing was limited to the removal of only one transmission line at a time rather than multiple simultaneous line outages.
The findings indicate that the Genetic Algorithm was effective at optimizing the system for single-line contingency scenarios but was not designed to analyze multiple simultaneous line failures. Furthermore, some optimized solutions did not consistently maintain bus voltages within the desired range of 1% to 3% of the target voltage, indicating opportunities for further refinement of the optimization process. Overall, this research demonstrates that Genetic Algorithms can improve power grid analysis and optimization while highlighting areas for future work, including faster optimization techniques, multi-line contingency analysis, and improved voltage regulation.