Dennis Schroeder/National Renewable Energy Laboratory
From ‘The Terminator’ to ‘The Matrix’, Hollywood has taught us to be wary of artificial intelligence. But rather than sealing our doom on the big screen, algorithms could be the solution to at least one issue presented by the climate crisis.
Researchers at the ARC Centre of Excellence in Exciton Science have successfully created a new type of machine learning model to predict the power-conversion efficiency (PCE) of materials that can be used in next-generation organic solar cells, including ‘virtual’ compounds that don’t exist yet.
Unlike some time-consuming and complicated models, the latest approach is quick, easy to use and the code is freely available for all scientists and engineers.
The key to developing a more efficient and user-friendly model was to replace complicated and computationally expensive parameters, which require quantum mechanical calculations, with simpler and chemically interpretable signature descriptors of the molecules being analysed. They provide important data about the most significant chemical fragments in materials that affect PCE, generating information that can be used to design improved materials.
The new approach could help to significantly speed up the process of designing more efficient solar cells at a time when the demand for renewable energy, and its importance in reducing carbon emissions, is greater than ever. The results have been published in the Nature journal Computational Materials.
After decades of relying on silicon, which is relatively expensive and lacks flexibility, attention is increasingly turning to organic photovoltaic (OPV) solar cells, which will be cheaper to make by using printing technologies, as well as being more versatile and easier to dispose of.
A major challenge is sorting through the huge volume of potentially suitable chemical compounds that can be synthesised (tailor-made by scientists) for use in OPVs.
Researchers have tried using machine learning before to address this issue, but many of those models were time consuming, required significant computer processing power and were difficult to replicate. And, crucially, they did not provide enough guidance for the experimental scientists seeking to build new solar devices.
Now, work led by Dr Nastaran Meftahi and Professor Salvy Russo of RMIT University, in conjunction with Professor Udo Bach’s team at Monash University, has successfully addressed many of those challenges.
“The majority of the other models use electronic descriptors which are complicated and computationally expensive, and they’re not chemically interpretable,” Nastaran said.
“It means that the experimental chemist or scientist can’t get ideas from those models to design and synthesise materials in the lab. If they look at my models, because I used simple, chemically interpretable descriptors, they can see the important fragments.”
Nastaran’s work was strongly supported by her co-author Professor Dave Winkler of CSIRO‘s Data 61, Monash University, La Trobe University, and the University of Nottingham. Professor Winkler co-created the BioModeller program which provided the basis for the new, open source model.
By using it, the researchers have been able produce results that are robust and predictive, and generate, among other data, quantitative relationships between the molecular signatures under examination and the efficiency of future OPV devices.
Nastaran and her colleagues now intend to extend the scope of their work to include bigger and more accurate computed and experimental datasets.
The Latest Updates from Bing News & Google News
Go deeper with Bing News on:
Clean energy generation
- Pacific 'increasingly investing in renewable energy', resulting in lower costs - briefing paper
The Asian Development Bank's policy paper looks at renewable energy investments in Fiji, Palau, Papua New Guinea, Samoa, Solomon Islands, and Tonga.
- New Energy Storage Solutions Test Lithium-Ion’s Reign
With the rise of intermittent renewable energy generation, the need for long-duration energy storage is rising fast. Lithium-ion batteries currently dominate the market, with record-low prices fueled ...
- Asset-level generation forecasts? 15-day solar predictions? AI has made it a reality
Those who buy and sell electricity can now view solar generation forecasts individually or in aggregate for portfolio management.
- Dominion solar farm powering 20,000 homes in Halifax County, one of many in move to clean energy
They have over 20 solar farms in development in Virginia that will supply power to 2.4 million homes once complete.
- Amperon Launches Asset-Level Renewable Energy Generation Forecasts
Amperon, the leading provider of AI-powered electricity forecasting and analytics services, announced today the launch of their Asset-level Renewable Generation Forecasts.
Go deeper with Google Headlines on:
Clean energy generation
[google_news title=”” keyword=”clean energy generation” num_posts=”5″ blurb_length=”0″ show_thumb=”left”]
Go deeper with Bing News on:
New type of machine learning model
- Machine learning models teach each other to identify molecular properties
This new technique could make it easier for researchers to use machine learning algorithms to identify and ... lack key information or include too much of one data type, introducing bias into the ...
- NIST releases new tool to check AI models’ security
has released a new open source software package, Dioptra, that allows developers to determine what type of attacks would make an AI model perform less effectively. “Testing the effects of adversarial ...
- Identification of novel toxins associated with the extracellular contractile injection system using machine learning
Toxins delivered by the extracellular contractile injection system (eCIS) of 950 microbial genomes were identified using XGBoost classifier algorithm, four of which were experimentally confirmed and ...
- Earth's plate tectonics fired up hundreds of millions of years earlier than we thought, ancient crystals reveal
New research hints that plate tectonics began earlier than 4 billion years ago — not long after Earth had formed.
- Deep learning models can be trained with limited data: New method could reduce errors in computational imaging
Deep learning models, such as those used in medical imaging to help detect disease or abnormalities, must be trained with a lot of data. However, often there isn't enough data available to train these ...
Go deeper with Google Headlines on:
New type of machine learning model
[google_news title=”” keyword=”new type of machine learning model” num_posts=”5″ blurb_length=”0″ show_thumb=”left”]