An Improvement of Load Flow Solution for Power System Networks using Evolutionary-Swarm Intelligence Optimizers

• Published in: Covenant Journal of Engineering Technology Vol. 8; no. 1
• Main Authors: Otonye E. Ojuka, Christopher Ahiakwo, Dikio Idoniboyeobu, Sepiribo Braide, Biobele Wokoma
• Format: Journal Article
• Language: English
• Published: Covenant University 01.01.2024
• Subjects: Accuracy; Comparative; Convergence; Evolutionary-Swarm; Intelligence; Optimizers
• ISSN: 2682-5317; 2682-5325

Load flow report which reveals the existing state of the power system network under steady operating conditions, subject to certain constraints is being bedeviled by issues of accuracy and convergence. In this research, five AI-based load flow solutions classified under evolutionary-swarm intelligence optimizers are deployed for power flow studies in the 330kV, 34-bus, 38-branch section of the Nigerian transmission grid. The evolutionary-swarm optimizers used in this research consist of one evolutionary algorithm and four swarm intelligence algorithms namely; biogeography-based optimization (BBO), particle swarm optimization (PSO), spider monkey optimization (SMO), artificial bee colony optimization (ABCO) and ant colony optimization (ACO). BBO as a sole evolutionary algorithm is being configured alongside four swarm intelligence optimizers for an optimal power flow solution with the aim of performance evaluation through physical and statistical means. Assessment report upon application of these standalone algorithms on the 330kV Nigerian grid under two (accuracy and convergence) metrics produced PSO and ACO as the best-performed algorithms. Three test cases (scenarios) were adopted based on the number of iterations (100, 500, and 1000) for proper assessment of the algorithms and the results produced were validated using mean average percentage error (MAPE) with values of voltage profile created by each solution algorithm in line with the IEEE voltage regulatory standards. All algorithms proved to be good load flow solvers with distinct levels of precision and speed. While PSO and SMO produced the best and worst results for accuracy with MAPE values of 3.11% and 36.62%, ACO and PSO produced the best and worst results for convergence (computational speed) after 65 and 530 average number of iterations. Since accuracy supersedes speed from scientific considerations, PSO is the overall winner and should be cascaded with ACO for an automated hybrid swarm intelligence load flow model in future studies. Future research should consider hybridizing ACO and PSO for a more computationally efficient solution model.