PhD student Sofie Claridge is working with AI to detect quench in superconducting materials
In scientific discovery, Artificial Intelligence (AI) is emerging as a new way to perceive and interpret the vast ocean of data that traditional methods might miss or misinterpret. Sofie Claridge's PhD research applies this transformative capability of AI to the phenomenon known as "quenching" in high-temperature superconducting (HTS) magnets.
At their core, HTS magnets are components that exhibit zero electrical resistance under specific conditions, an ideal attribute for high-efficiency applications ranging from MRIs to fusion reactors. However, these magnets are prone to quenching—a rapid loss of their superconducting properties triggered by minute thermal hotspots that, if undetected, can lead to catastrophic failure.
Typically, these hotspots are elusive, masked by the noise and complexity of the data collected during magnet operation. “Traditional sensors often struggle to differentiate between normal operational data and the critical signals that herald a quench,” explains Sofie the problem.
Sofie has used sophisticated machine-learning techniques to outpace traditional quench detection methods. For a layperson, this is AI at real work. – "According to my literature review, no other research has been done from this angle," Sofie explains her unique approach.
The study utilises a type of sensor called Fiber Bragg Grating (FBG), which is adept at measuring strain and temperature changes through shifts in light wavelength reflected back by the sensor. While this method is sensitive to the minute changes indicative of hotspots, the sheer complexity of interpreting the data requires more than a human touch—it requires AI.
"By employing machine learning algorithms, we can sift through the noise and detect those crucial, abnormal changes in temperature before they result in a quench," clarifies Sofie. The AI systems are trained on datasets crafted from experimental setups that simulate real-world conditions, allowing them to learn and predict thermal events with astonishing accuracy.
Imagine AI as a detective with an uncanny sense of intuition, capable of noticing the subtlest clues within chaotic environments. In this case, the AI analyses the spectral data from the FBGs, pinpointing anomalies that human observers might miss. "It's not just looking at the data; it's understanding it, predicting potential failures before they happen," Sofie elaborates on her goals. "It's almost like predicting the future accurately," she jokes.
This approach enhances the safety and reliability of using HTS magnets in critical applications and exemplifies how AI is becoming indispensable in fields where precision and reliability are paramount. The potential extends beyond superconductors—any field dealing with large-scale data that require real-time monitoring and analysis could benefit from this AI-driven methodology.
As AI continues to integrate into various facets of scientific research, its role evolves from a passive tool to an active participant, a 'sixth sense' that provides insights unreachable by humans alone. This synergy of human and artificial intelligence paves the way for breakthroughs once thought impossible, reaffirming that in the quest for knowledge, sometimes looking at the data in a new way can be just as important as the data itself.
"I have always been curious and interested in mathematics and solving problems with numbers. My family, who encouraged me to pursue these scientific dreams, has always supported me. My master's degree at the Paihau–Robinson Research Institute led to this PhD project, which is a dream come true for a young scientist."
The Paihau Robinson Research Institute is not only on the cutting edge of superconductivity, materials science, and magnetism—it also attracts brilliant young minds like Sofie to do groundbreaking research using the latest machine learning and AI capabilities. “You cannot find a better place to do research and an opportunity to collaborate and be in touch with the leading edge researchers globally,” concludes Sofie.