ITEM: AI could help fight climate change by making renewable energy solutions like solar and wind power more affordable and reliable. One big barrier: hardly anyone is working on it – at least, not yet.
Wind and solar farms are costly to operate and maintain, in large part because the technology is complex and prone to malfunction, which is also one reason why adoption of such technologies has been slow – so slow in fact that at the current adoption rate, it will be impossible to meet climate goals of pushing global emissions into sustained decline towards net zero by 2050. The International Energy Agency reckons that investment in renewable energy will have to be trebled between now and 2031 to hit those targets.
Joyjit Chatterjee, a data scientist at the University of Hull, and Nina Dethlefs, director of computer science research (also at the University of Hull), say that a potential solution is predictive analytics powered by AI to anticipate equipment failures, reports IEEE Spectrum:
Wind turbines and solar panels on utility-scale farms are loaded with sensors that allow operators to remotely monitor their power production and health status. These sensors include vibration sensors, temperature sensors, accelerometers, and speed sensors. The data they generate presents an opportunity. AI models trained on historical power production and failure data could predict unexpected failure in a wind turbine gearbox or a solar panel inverter, helping operators prepare for power outages and plan routine maintenance.
Other industry sectors are also looking to apply predictive analytics in similar ways. The barrier for applying it to renewable energy that the current work is focused on use cases such as manufacturing, healthcare and network O&M. To be fair, one reason for that is that AI researchers have very little data to work with, the report says:
For the AI community, a big barrier in creating better models is the limited amount of openly available data given the commercially sensitive nature of the wind and solar industry. Besides the industry being unwilling to share data openly, the lack of standards can impair AI model development, Chatterjee says. “Wind farm operators in different parts of world manage data differently so it’s really challenging for researchers to use resources collectively.”
Chatterjee and Dethlefs published an opinion paper in the data-science journal Patterns earlier this month outlining the hurdles limiting AI’s impact on renewables, and how to surmount them using established and emerging AI methods.
One point the paper makes is that AI solutions for renewables should ideally be simple. Advanced AI/ML models leveraging neural networks could be very useful, but the trade-off is that neural networks require power-hungry high-performance computing infrastructure which would not only be inaccessible to developing nations, but also counterproductive to the whole point of adopting renewable energy in the first place:
“Often neural networks are not needed for every problem,” says Chatterjee. “Why increase carbon emissions by training and developing more computationally complex neural networks? Future research needs to be on models that are less resource-hungry and carbon-intensive.”
Full story here.