In the rapidly evolving energy sector, the integration of renewable energy sources (RESs) like wind and solar power is transforming power systems worldwide. However, this shift brings a significant challenge: maintaining frequency stability. As traditional power plants are replaced by inverter-interfaced renewables, the system’s inertia—the ability to resist frequency changes—declines, making it more complex to estimate and manage.
Enter Bingzhang Liu, a researcher from The State Key Laboratory of Alternate Electrical Power System with Renewable Energy Sources at North China Electric Power University in Beijing. Liu and his team have developed a novel approach to estimate area inertia using deep learning, a method that could revolutionize how power systems are operated and controlled.
The team’s work, published in the International Journal of Electrical Power & Energy Systems, introduces a time-series-based residual neural network (TS-ResNet). This deep learning model is designed to estimate inertia in power systems with a mix of synchronous and virtual inertia, a common scenario in modern grids with high shares of renewables.
“Most existing methods rely on simplified models of system frequency dynamics,” Liu explains. “But with the increasing penetration of virtual inertia from renewables, these models struggle to maintain accuracy. Our approach extracts dynamic features from nodal frequencies and tie-line powers, providing a more precise estimation.”
The TS-ResNet model uses probing signals that do not disrupt system stability, making it a practical solution for real-world applications. Moreover, it demonstrates robustness under various noise levels and can generalize to unseen scenarios, such as different time delays of virtual synchronous generators (VSGs) and variations in DC power transmission.
The commercial implications of this research are substantial. Accurate inertia estimation is crucial for power system operators to maintain grid stability and prevent blackouts. As the share of renewables continues to grow, the ability to estimate inertia in real-time will become even more critical. Liu’s method could enable operators to make more informed decisions, improve grid reliability, and ultimately, facilitate the transition to a cleaner energy future.
The research also highlights the potential of data-driven approaches in power systems. As Liu notes, “Our findings suggest that deep learning models can offer a fresh perspective on inertia estimation, paving the way for more advanced and adaptive control strategies.”
In the future, this research could shape the development of smarter grids that can seamlessly integrate high shares of renewables. It could also drive the development of advanced control technologies that can respond in real-time to changes in system inertia, ensuring grid stability and reliability.
As the energy sector continues to evolve, the work of researchers like Bingzhang Liu will be instrumental in navigating the challenges and opportunities that lie ahead. Their innovative approaches could very well light the way to a more sustainable and stable energy future.