Brian Mukeswe's research
Brian researched supercomputing techniques for managing power flows on electricity grids
Scalability Analysis of a Parallelized Particle Swarm Optimization Algorithm for Calculating Optimal Power Flow
Several approaches have been proposed for solving Optimal Power Flow (OPF) problems. Because such problems are non-convex and non-linear, numerical methods including heuristic search algorithms like Genetic Algorithms (GA), Simulated Annealing (SA) and Particle Swarm Optimization (PSO) have been widely applied to these problems. In this project we parallelized the original PSO algorithm and executed it on a supercomputer. We showed that a more optimal result could be obtained by increasing the number of particles in the PSO-OPF algorithm. Additionally, we showed that increasing the number of particles slows down the algorithm if more computational resources are not allocated. Furthermore, we quantified the parallel scalability of our implementation of the PSO-OPF algorithm using the “Speed Up” metric. The results suggest that parallelization of the PSO-OPF algorithm shows promise in eliminating the obstacle of long computation time when using the PSO technique to optimize the power grid for real-time or hour-ahead electricity markets.
This dissertation project gave me the unique experience of working directly with a supercomputer, and I gained crucial skills in designing and executing computation experiments involving large amounts of data. I would like to thank my supervisor Dr. Adam Carter and the Mastercard Foundation Scholars Program for their guidance and support.