The rising demand for grid resilience in the face of unpredictable renewable energy integration and cyber threats has caught the attention of leaders around the world.
With power systems becoming more dynamic and susceptible to transient instabilities, experts are urging for innovative, data-driven approaches to ensure stability and efficiency.
One of such experts, Israel Olugbemi, has shed light on a groundbreaking method that combines artificial intelligence and semi-supervised learning to transform how power grids are assessed and stabilized.
Speaking on the development, Israel Olugbemi, an electrical engineer with a Master’s degree and extensive field experience in power and control systems, emphasized the urgency of adopting smarter models.
“Traditional methods for transient stability assessment (TSA) are no longer fast or scalable enough for modern power systems,” he said.
“We need approaches that can process both labeled and unlabeled data efficiently while minimizing computational costs — and that’s where Semi-Supervised Active Learning (SSAL) comes in.”
In his co-authored research paper titled “Review of Transient Stability Assessment of Power Systems Based on Semi-Supervised Active Learning,” Olugbemi explores how SSAL can enhance the reliability and scalability of TSA models. The study addresses challenges of data scarcity, high labeling costs, and the limitations of conventional simulations in dynamic grid environments — especially with increasing renewable energy penetration.
At the core of the SSAL methodology is the ability to identify high-value data samples using K-means clustering and entropy-based uncertainty sampling.
This significantly reduces the amount of labeled data required to train predictive models, while still achieving a high level of accuracy — up to 94.01%, according to tests conducted using the IEEE 39-bus system.
“SSAL leverages the power of machine learning to not only reduce dependency on labeled data but also to improve model performance over time through self-training loops,” Olugbemi explained.
“This is especially crucial for developing countries, where data labeling resources and computational capacity can be limited.”
The study introduces a smart feedback mechanism where high-confidence, pseudo-labeled samples are incorporated into the model’s training dataset. This continuous loop of learning and retraining allows for TSA models that adapt to evolving grid conditions without requiring manual intervention at every step.
Olugbemi’s professional background lends weight to his research. With a career spanning both the public and private sectors — including roles at the Federal Capital Development Authority (FCDA) and Warri Refining and Petrochemical Company — he has managed complex electrical infrastructure projects and improved power system uptime by 24% through preventative maintenance initiatives.
Among his achievements is leading a $1.2 million project to completion on schedule, and reducing control system repair times by 15% through advanced diagnostics and fault isolation. His resume further includes the successful overhauling of a burnt 30MW gas turbine and implementing energy-saving protocols that cut consumption by 20%.
He credits his hands-on experience with real-world power systems as a major influence behind the need for smarter, AI-powered diagnostics.
“I’ve seen firsthand how fragile power infrastructures can become under stress. Our systems need to be not just reactive, but predictive,” he noted.
Olugbemi also pointed out that SSAL holds immense potential for grid modernization efforts in regions like Sub-Saharan Africa.
“We don’t need to reinvent the wheel. We need to apply data intelligently. SSAL offers a practical, scalable pathway for countries struggling with aging infrastructure and growing energy demand,” he asserted.
As nations invest heavily in renewable energy and smart grid technologies, Olugbemi’s research offers a timely contribution to global discourse on sustainable energy management.
He urged policy makers and grid operators to prioritize investment in AI-driven tools for grid stability, stressing that waiting for failure before acting is no longer a viable strategy.
“Our power systems must evolve — and SSAL provides the framework for doing so with both intelligence and efficiency,” he concluded.
and then