Loading...
PSO method for optimising the demand side management system for peak clipping and load shifting
Author(s)
Mpaka, Abuyile
Date Issued
2026
Type
doctoral thesis
Abstract
Demand Side Management (DSM) faces increasing challenges due to rapid growth in global electricity consumption, volatile energy prices, and transmission network congestion. These issues are especially evident in developing regions such as South Africa, where urbanisation and industrial expansion continue to drive rising energy demand. To address these pressures, utilities require intelligent DSM approaches that can adapt dynamically to changing operational needs. Traditional DSM strategies, often characterised by static control logic and limited predictive capability, struggle to meet these modern requirements. This research introduces advanced, intelligent DSM frameworks that integrate metaheuristic optimisation with predictive modelling to achieve flexible, cost-effective, and sustainable energy management solutions. The first phase of the study focuses on a hybrid DSM model that integrates Particle Swarm Optimization (PSO) with a Machine Learning-Informed Prediction (MLIP) model. The MLIP enhances short-term load forecasting accuracy, enabling the PSO algorithm to optimise power allocation more effectively under time-varying tariff conditions. Applied to industrial users within Western Cape municipalities, the hybrid PSOMLIP system performs load clipping and reshaping by leveraging real-time pricing and historical consumption data to shift non-critical loads dynamically. This approach improves the load factor and reduces overall operational costs. The second phase develops a DSM model based on the Grey Wolf Optimizer (GWO), which utilises social hierarchy and cooperative hunting strategies to enhance search efficiency and solution quality. This model was implemented across residential, commercial, and industrial sectors using real-world consumption data from large power users in Paarl. The GWO-based DSM framework effectively reduces peak demand and energy costs without compromising total energy consumption. Simulation results reveal that the proposed PSO-MLIP, GWO, and the newly developed Barber Optimization Algorithm (BaOA) outperform traditional DSM techniques. The GWO framework achieved an 18.5% reduction in peak demand and a 9.7% reduction in energy costs, surpassing the PSO-MLIP model, which achieved a 14.2% peak reduction. The GWO also exhibited superior robustness and faster convergence when solving nonlinear, multiobjective optimisation problems. Building on these outcomes, the DSM-BaOA framework was designed with a structured architecture: an input layer capturing load and tariff data, an optimisation layer executing BaOA, and an output layer delivering optimised load profiles, cost savings, and peak reductions. Beyond the technical contributions, this research underscores the strategic importance of intelligent DSM in South Africa’s energy transformation. The developed frameworks enhance operational efficiency, economic viability, and environmental sustainability for municipalities and large consumers. They also align with the nation’s vision for smart grid implementation, promoting resilience, demand flexibility, and renewable energy integration. In summary, this study contributes both methodologically and practically by demonstrating how hybrid intelligent algorithms can optimise DSM. Situated at the intersection of predictive analytics and optimisation theory, the findings affirm that adaptive, data-driven DSM approaches can significantly reduce peak demand, enhance energy efficiency, and support sustainable power system management across municipal and industrial sectors.
Additional information
Thesis (DEng (Electrical Engineering))--Cape Peninsula University of Technology, 2026
File(s)![Thumbnail Image]()
Loading...
Name
Mpaka, A_214118738 (1).pdf
Size
4.43 MB
Format
Adobe PDF
Checksum
(MD5):749cdd9c48d89704ed8a7f10e29973a1