In the pursuit of a greener future, the power industry is undergoing a significant transformation, with new energy sources taking center stage. As countries worldwide strive to meet “dual carbon” goals—reducing carbon emissions and achieving carbon neutrality—the construction of a new power system with renewable energy as its backbone has become a top priority. However, this shift brings with it a unique set of challenges, particularly in ensuring the stability and reliability of the power grid.
Enter a team of researchers led by Dr. Li Xiang from the School of Electrical Engineering and Automation at Wuhan University, who have developed a groundbreaking method for assessing transient power angle stability in power systems dominated by grid-forming new energy sources. Their work, published in the journal *Shanghai Jiaotong Daxue xuebao* (translated as *Journal of Shanghai Jiao Tong University*), introduces a novel approach that combines physics-informed sequence-to-sequence (PI-seq2seq) neural networks and cascaded convolutional neural networks to predict and evaluate the stability of power systems.
The team’s method addresses a critical need in the energy sector: the ability to quickly and accurately assess the transient stability of power systems. Traditional time-domain simulations, while thorough, can be time-consuming and impractical for real-time decision-making. The researchers’ solution leverages the power of machine learning to predict future power angle trajectories, using a loss function with physical loss terms to guide the model training process. This ensures fast and accurate evaluations without the need for long-duration simulations.
“Our method not only speeds up the evaluation process but also provides a confidence level for the stability assessment,” said Dr. Li Xiang. “This is crucial for operators who need to make quick decisions to maintain the stability of the power grid.”
The researchers also overcame the challenge of fixed evaluation lengths, which can impact the accuracy of stability assessments. By configuring a threshold judgment mechanism for the evaluation confidence level, they achieved transient stability judgments of non-fixed evaluation lengths. This flexibility is essential for handling the dynamic and unpredictable nature of power systems with high levels of renewable energy integration.
The team’s method was verified in the Kundur system, a widely used benchmark for power system studies, and the results were promising. The simulation demonstrated the method’s effectiveness in both power angle curve prediction and stability evaluation, paving the way for its potential application in real-world power systems.
The implications of this research for the energy sector are significant. As the world moves towards a future dominated by renewable energy, ensuring the stability and reliability of the power grid becomes paramount. The ability to quickly and accurately assess transient stability can help operators prevent blackouts, minimize downtime, and maintain the overall health of the power system.
Moreover, the researchers’ method can be integrated into existing power system monitoring and control systems, providing operators with real-time insights and enabling proactive decision-making. This can lead to improved grid reliability, reduced operational costs, and enhanced customer satisfaction.
As the energy sector continues to evolve, the need for innovative solutions to ensure grid stability will only grow. The work of Dr. Li Xiang and his team represents a significant step forward in this direction, offering a powerful tool for operators to navigate the complexities of modern power systems.
In the words of Dr. Li Xiang, “Our method is a testament to the power of combining physics-based models with advanced machine learning techniques. It’s an exciting time for the energy sector, and we’re proud to contribute to the development of a more stable and reliable power grid.”