Resolving Power Outages with the help of Artificial Intelligence

  • published : 2022/12/26
  • Comments : no comment

Using artificial intelligence, China has built a new power grid that can fix disruptions and outages within seconds.

State company tested how artificial intelligence could minimise electricity disruptions
The Chinese power grid has used AI mainly for backbone networks and industrial users but now looks to support residential communities

China’s state-owned power grid company launched the nation’s most powerful AI for electricity distribution on Tuesday, according to Science and Technology Daily.
Following a community blackout it usually takes six to 10 hours to fix the problem and resume power supply, the report said, but artificial intelligence cuts that down to three seconds.

A month-long test run at Qitailu, a residential community with more than 200 families in Urumqi in Xinjiang region shows the system’s effectiveness, according to the report.

The Qitailu community power grid has more sensors than anywhere else in the country, and each sensor has its own “brain” that can decide the routes of power supply without human intervention.

When an equipment failure occurs in the low-voltage distribution network in the area, the edge computing terminals will “instantly start the self-healing function, including automatic fault location, fault isolation and power restoration,” said the report.

“Now the outage is almost nothing. The power comes as soon as it stops, not affecting life and work in any way,” the state-owned newspaper quoted an unnamed resident as saying.

The application of AI-driven power grid technology would quickly expand from Qitailu to other communities, local authorities said.

China produces a third of the electricity in the world, more than that generated by the United States, India, Russia and Japan combined.

The Chinese power grid has been using AI in its daily operations for years, especially after the massive construction of wind farms and solar plants which had unstable energy output.
However, these applications have mostly restricted to backbone networks or to prevent outages for industrial users, according to a senior engineer working at the headquarters of the State Grid Corporation of China in Beijing.
“In a residential community, the grid connects to a large number of end users. The environment is quite complex. Many different kinds of problems can happen. Most issues need to be located and identified manually,” said the engineer who asked not to be named because he was not authorised to speak to the media.
Blackouts rarely happen in China, especially in cities. “And Chinese power grid workers are probably the most efficient in the world. But when it happens, it still takes hours to fix the problem,” he said.
The AI works differently from existing automated systems in the power grid, according to the engineer.
A traditional power grid management computer generatse a large number of error codes when a blackout happens. A human operator has to make an assessment based on the information and devise a solution, such as a re-routing plan based on his or her experience.
The AI uses natural language processing to “understand” the text explanation behind each error codes, said the engineer.
It then learns from a large number of previous blackout cases to figure out hidden patterns that would help find and solve the problem.
But to improve its accuracy, the AI requires a huge amount of data collected by a wide range of sensors, according to the engineer.
“In the past these sensors and related data transmission technology were only available for high-voltage power lines. Now they are making ways into neighbourhoods,” he said.