Mexico provides fertile ground for AI in power generation

Mexico is a global middleweight in artificial intelligence use for business. But the technology is also becoming increasingly important in the national energy sector.

SMART GRIDS AND RENEWABLES

The renewables sector holds much potential for AI applications in Mexico. In contrast to AI use in oil and gas, mining or construction, where the technology is mostly used for risk modeling and monitoring, in the renewable sector it helps to deal with the variable weather problem.

This problem leads to uneven energy production, while electricity usage relies on managing peaks to prevent overloads.  

The government has sent mixed signals on wind and solar, setting lofty goals for renewable power generation but also canceling renewable auctions earlier this year.

Recently though, clean energy agency INEEL prepared the ground for stronger support of AI in wind and solar power generation. Earlier this month, INEEL described in a blog post how neural networks, decision trees, and classification algorithms could help with machine learning to forecast conditions for wind farms.    

Meanwhile, AI already helps to reduce electricity usage, especially for tech companies.  

“As our dependence on cloud computing centers grows, novel methods of forecasting are needed to handle peaks of energy, given the demands on such centers,” Andrés Méndez Vázquez, a professor at the advanced research center (Cinvestav), where he helms a machine learning research group, told BNamericas.

Known as Green IT, AI applications can help reduce energy usage while also optimizing energy output from renewables. An example, according to Méndez, are Google‘s datacenters. He said the company cut the energy use of these centers by half since 2014 through a combination of machine learning and IoT. 

Another Google project, DeepMind, could even forecast wind energy consumption 36 hours in advance using a neural network with weather and energy records.

And at Cinvestav, researchers investigate how AI could turn smart grids into the control system for alternative energy sources.

DEMOCRATIZING AI

In June, Microsoft awarded a climate change grant under its “AI for Earth” initiative to preservation group CEDO Intercultural, based in Sonora state, one of only five Latin American recipients and 124 globally.

According to Microsoft, CEDO Intercultural is using the grant for developing AI-enabled search of media publications about combatting climate change. 

Large datasets are the lifeblood of AI and as these are used to monitor climate change and enable researchers to design smart grids, they also enable individual energy solutions. 

“Several blogs have codes to use for deeplearners to start doing their own energy forecasts,” Méndez said. “Actually you only need an IoT device at the charge center to start getting your own data and forecast the usage of your own energy.”

CHALLENGES

But “Mexico still depends in a great manner on the technology being produced in the US and Canada,” Méndez said, adding that AI often just includes data analysis or business intelligence tools.

On the upside, Méndez told BNamericas that thanks to sustained public investment several Mexican universities and research institutes already have the background for advanced statistical analysis that forms the basis of AI. “We have some research centers doing some work in what is known as new AI, machine learning, data sciences and deep learning,” he said.

These elements have put Mexico in a strong position for AI in business. Over the past few years, Mexico has attracted foreign investment or nursed startups geared toward AI applications in IT, banking, and healthcare. 

In 2018 Mexico launched a national AI strategy, becoming the first Latin American country to do so. 

Though the strategy has yet to yield results, President Andrés Manuel López Obrador (AMLO) is likely to embrace it as he has refrained from criticism, unlike with other initiatives developed by the previous administration. 

Notably, the AI strategy dovetails with AMLO’s own policy goals. 

Comparing Mexico’s AI strategy with those of Canada, France and South Korea, technology consultancy Oxford Insights found that Mexico’s plan “is unique for its focus on the social impacts of AI.”

But Méndez cautions that many AI applications are still in their early stages, focusing on AI-enabled control neural networks. “Full-blown AI,” Méndez said, remains in the distant future.