Southern Power Grid: The Role of Artificial Intelligence Behind A Million Lights

In the mountains of western Guangdong, drones autonomously launch from their nests to conduct routine inspections of power transmission lines. Meanwhile, the Guangdong Power Grid's drone management center, located 500 kilometers away, receives real-time images from these inspections, capturing details such as bird nests on towers, insulators, and even tiny lost bolt caps that might otherwise go unnoticed. The intelligent analysis system automatically identifies faults, issuing alerts for any detected equipment anomalies.

At the Southern Power Grid's AI Infrastructure Operations Center, Hu Rong, Deputy General Manager of the Digital Grid Research Institute of Southern Power Grid Co., emphasizes the impact of digitalization on the construction of new power systems. Intelligent inspections have significantly enhanced fault localization accuracy and visualized the state of transmission lines, improving inspection efficiency by 80 times. With over a million kilometers of overhead lines to cover, and approximately 35 images needed per tower for detailed inspections, the volume of data generated is staggering.

Dong Zhaojie, a leading technical expert at Southern Power Grid's AI Technology Co., explains that previously, data analysts relied on visual inspection, often leading to eye strain. Now, with advanced AI technology, machines can analyze complex environments, addressing the challenges of visual inspections in the power sector. Drone-assisted inspections are just one example of how AI is transforming the grid.

Inside the AI Infrastructure Operations Center, various AI applications are apparent across the electricity generation and distribution processes. On September 26, 2023, Southern Power Grid launched an AI Innovation Platform for the power industry, showcasing China's first controllable large-scale power model, "Da Watt." This model can process 100 problem images per minute and identify 20 types of defects — ten times more efficient than traditional AI algorithms. It also generates automated response plans in seconds for grid anomalies, aligning with market regulation demands.

Currently, "Da Watt" is widely applied in over 80 scenarios across five provinces concerning electricity generation, transmission, transformation, distribution, and usage. The rapid development of renewable energy, particularly solar and wind, faces challenges from weather dependencies. Implementing advanced digital technologies such as AI, big data, and cloud computing in power forecasting significantly enhances prediction accuracy and supports the entire renewable energy supply chain.

To address these challenges, Southern Power Grid developed the "Kua Fu" high-precision power forecasting system, akin to a "weather forecasting" tool for the power sector. This system utilizes high-frequency updates, high-resolution data, and small-scale numerical weather forecasts to deliver accurate, cost-effective predictions. It reduces the uncertainty of renewable energy generation, minimizes the grid's reserve capacity, supports decision-making for dispatch, and enhances the security and economy of grid operations. Notably, the forecasting accuracy for wind and solar sites exceeds national standards by 5% to 7%.

Hu Rong highlights Southern Power Grid's commitment to leveraging digital developments and historical industry opportunities, aiming to deepen the synergy between digitalization and green initiatives as it drives the evolution of new power and energy systems through concentrated resources and accelerated transformations.