The development of novel materials is critical to the future of nuclear energy, which can generate power without emitting hazardous pollutants. An Argonne scientist is using computer vision to weed out the finest applicants from a crowded field.
Imagine the frame-by-frame tale that can be derived from a single movie if a picture can speak a thousand words. Five minutes of video at 200 frames per second may provide 60,000 pictures, creating a visual “Moby Dick.” Does it seem to be time-consuming to digest and catalogue? That is, which is why scientists seldom examine the recordings of their experiments in such depth.
Wei-Ying Chen, a lead materials scientist in the nuclear materials division at the DOE’s Argonne National Laboratory, is experimenting with artificial intelligence (AI) developments to alter that. The deep learning-based multi-object tracking (MOT) method he used to extract data from films, as revealed in a new paper published in Scientific Reports, intends to assist the United States in improving advanced nuclear reactor designs. In result, updated nuclear power would provide more safe and dependable energy while emitting less greenhouse emissions.
Nuclear energy now generates more power on less land than any other renewable energy source. Several commercial nuclear reactors, which generate roughly 20% of total power in the United States, employ outdated materials and technology. Newer materials and sophisticated designs, according to scientists and engineers, might significantly enhance the amount of clean energy provided by nuclear power plants.
“We want to create improved reactors that can operate at greater temperatures, so we need to find materials that can withstand higher temperatures and irradiation doses,” Chen said. “We’re on pace to gather all of the data we need from all of the video frames using computer vision capabilities.”
Chen consults users and performs experiments at Argonne’s Intermediate Voltage Electron Microscope (IVEM) facility, a national user facility and a DOE Nuclear Science User Facility partner facility (NSUF). The IVEM, which is half transmission electron microscope and part ion beam accelerator, is one of roughly a dozen equipment across the globe that allow researchers to see material changes generated by ion irradiation as they occur (in situ). This implies that scientists like Chen will be able to investigate the impact of various energy on materials suggested for use in future nuclear reactors.
Knowing why, where, and when materials degrade and display faults under harsh circumstances during their lifespan is crucial for determining a material’s fitness for use in a nuclear reactor. The earliest indicators that a material may corrode, become brittle, or break are very small flaws. Defects occur within a picosecond, or one trillionth of a second, during experiments. These flaws emerge and dissipate in tens of milliseconds at high temperatures. Chen is a specialist in IVEM studies, and he admits that even he has difficulty plotting and interpreting such fast-moving data.
The transient nature of flaws during experiments explains why scientists have historically recorded just a sprinkling of data points along critical lines of measure.
Chen has spent the last two years at IVEM developing computer vision to detect material changes in recorded tests. In one research, he looked at 100 frames per second from videos ranging from one to two minutes in length. In another, he recovered one frame every second from recordings ranging from one to two hours in length.
IVEM’s computer vision detects material faults and structural holes in the same way that face recognition software can spot and follow individuals in surveillance film. Instead of creating a library of faces, Chen creates a massive, dependable database of data on temperature resistance, irradiation resilience, microstructural flaws, and material lifetimes. This data may be displayed to improve models and design better experiments.
Chen emphasises that saving time—a widely touted advantage of computer-enabled work—is not the only benefit of IVEM’s use of AI and computer vision. With a better understanding of and control over ongoing studies, IVEM users may make on-the-fly modifications to make the most of their time at IVEM and gather essential data.
“Videos look extremely great, and we can learn a lot from them, but they are generally exhibited once at a conference and then are rarely utilised again,” Chen added. “We can really learn a lot more about seen occurrences using computer vision, and we can translate video of phenomena into more relevant data.”
DefectTrack demonstrates its accuracy and dependability.
Chen and co-authors from the University of Connecticut (UConn) unveiled DefectTrack, a MOT capable of retrieving complex defect data in real time when materials were irradiated, in a study published in Scientific Reports.
DefectTrack was used in the research to monitor up to 4,378 individual defect clusters in only one minute, with lifetimes ranging from 19.4 to 64 milliseconds. The results were much better than those of human equivalents.
“Our statistical analyses revealed that the DefectTrack is more accurate and quicker than human specialists in assessing the defect lifetime distribution,” stated co-author and Ph.D. candidate Rajat Sainju of UConn.
Improved speed and accuracy are two of the many benefits of computer vision.
“We urgently need to accelerate our knowledge of nuclear materials deterioration,” said Yuanyuan Zhu, assistant professor of materials science and engineering at UConn, who headed the university’s team of co-authors. “Dedicated computer vision models have the potential to transform analysis and help us better understand the nature of nuclear radiation damage,” says the study’s lead author.
Chen believes that computer vision tools like DefectTrack would enhance nuclear reactor design.
“Computer vision may reveal knowledge that was previously inaccessible in practise,” Chen said. “It’s thrilling that we now have access to such a large amount of raw data with remarkable statistical significance and consistency.”
You might also be interested in reading, For a greener ocean journey, an electric boat takes to the air
