December 18, 2019
Categorized: Artificial Intelligence
Share this post:
Advances in computing power are enabling utility companies to better manage problematic vegetation and more effectively predict and respond to disruptions caused by weather.
Detailed weather forecasting and artificial intelligence are being used to curtail power outages caused by severe weather amid concerns that extreme weather events are becoming more frequent due to climate change.
One new solution developed by IBM and The Weather Company predicts where outages are likely to occur during heat waves, storms and other extreme conditions.
It works by first analysing the past three years of an electricity distribution company’s weather-based outage information, looking for signals that preceded the outages and what the weather was doing at the time.
Jamie Azzopardi, Australia and New Zealand head of The Weather Company.
“It’s not just wind, as many people think, it’s temperature, humidity and dust also,” says Jamie Azzopardi, Australia and New Zealand head of The Weather Company.
The process also relies on the resolution and frequency of The Weather Company’s forecasts, which update every 15 minutes and drill down to within 500m of a specific location, anywhere across Australia.
Senior meteorologist at The Weather Company Michael Wang says that five to 10 years ago, only about 10 kilometres resolution was possible when forecasting.
“Computing power is one of the things that has limited weather forecasting,” he says. “But in recent years more computing power has really enabled us to forecast in greater resolution.”
Once the outage prediction model has been built based on historical data, it’s loaded into IBM’s app-based solution. It can then be used by the utility company to identify up to three days in advance where there might be outages in the grid.
“What that allows them to do is a bunch of different things,” says Azzopardi. “They can start pushing resources into those locations. If it’s a company that covers regional and remote areas the distances can be quite vast, so they can start moving their crews into place; they can have people working after hours, and they can schedule call centres and be much more responsive.”
The tool continually updates, and is loaded with the distributor’s policies and processes.
“It doesn’t just say ‘we think there’s an outage in this location’ or ‘there’s potential for an outage’, Azzopardi says. “It will also say how we believe, based on your processes, you should respond.”
The other new tool developed by IBM for utility companies is designed to reduce the incidence of outages by optimising vegetation management.
“While vegetation is not the only cause of outages, it is a significant one,” Azzopardi says. “This is an area that the utility companies invest a lot of time and a lot of cost in, particularly when they’re operating in regional and remote areas.”
Currently, most companies monitor the vegetation surrounding their grids by either driving out under the wires and manually checking, or with helicopters flying over. Some are now also looking at using drones.
IBM’s solution analyses various data sources, mostly satellite imagery and other high-resolution imagery. It then creates a visual map of the utilities grid. The AI model will identify vegetation around and encroaching onto the grid. Once this base is established, every five days it updates with lower-resolution imagery — to keep costs down — and changes in the state of vegetation, with risks identified provided to the power company.
“It won’t just pick up a branch that has grown overnight, for instance,” says Azzopardi. “But it will pick up a tree that has shifted, or half fallen over or something like that. It will allow the utility company to plan and predict better and earlier, and to respond faster if there’s a problem — not wait for a quarter-year or half-year cycle.
Earlier this year, New Zealand power company Vector was the first to adopt both solutions.
Vector, which has used artificial intelligence to manage electricity demand and network data since 2017, said: “Ultimately, we are confident this technology will make extreme weather events more manageable and less disruptive for our customers.”
Technological advances in the cloud is what’s really enabling these kinds of solutions, says Azzopardi.
“The cloud is becoming more cost-effective and easier for more organisations to work with,” he says. “It’s really allowing a lot of these business transformation processes.”
Hyperlocal weather forecasting combined with AI models has applications across a wide range of industries, according to Azzopardi, including retail, agriculture, emergency services and transport.
For retailers, as an example, people are likely to buy salads and fruits if it’s hot and comfort food if it’s cold, he says, “so our ability to provide that 500m grid across the country is a significant value proposition. When you can build those insights, you can identify patterns and let retailers know what we believe the impacts on that store will be.”
Originally published by News Corp on The Australian.