Hybrid EIT Method Combines Math and Machine Learning for Improved Building Structure Analysis
Researchers from the Tokyo University of Science (TUS) have developed a novel hybrid approach that combines traditional mathematical techniques with cutting-edge machine learning methods to improve the analysis of building structures. The method, called AND, combines the benefits of the iterative Gauss-Newton (IGN) algorithm and the one-dimensional convolutional neural networks (1D-CNN) algorithm.
Electrical impedance tomography (EIT) is a non-destructive imaging technique used to visualize the interior of materials. It involves injecting an electric current between two electrodes and measuring distortions caused by foreign objects inside the material. EIT has the advantage of being low cost and less cumbersome compared to other imaging methods such as X-ray imaging or magnetic resonance imaging.
However, accurately reconstructing the obtained information as images has been a challenge in EIT. Existing mathematical algorithms used for reconstruction often result in inaccuracies. To overcome this issue, researchers have started exploring the use of machine learning algorithms such as 1D-CNN. But these algorithms have limitations when it comes to handling new or unseen data, reducing their effectiveness.
The innovative AND method developed by the researchers performs 2D logical operations on multiple EIT images to detect small foreign objects inside materials. In their study, the team tested their method on actual cement samples using both simulation and experimental data. They found that the AND method outperformed both the IGN and 1D-CNN methods in accurately reconstructing the position and size of foreign objects, especially when the size of the objects decreased.
Additionally, the researchers identified another method to improve the accuracy of EIT by changing the current injection pattern. By changing the spatial distribution of the electric field and combining it with other non-destructive evaluation techniques, the resolution for detecting foreign particles can be further improved.
The proposed EIT reconstruction method has advantages in terms of equipment size and cost, making it a valuable tool for non-destructive foreign object detection. It can contribute to preventing building collapses and be deployed for rapid safety screening after natural disasters or explosions. This method is also expected to be easy to train inspectors and personnel to use.
This development marks a significant step forward in EIT technology, with the potential to become an important detection technique for preventing building collapse in the future. The researchers are now focused on further improving the accuracy of EIT by exploring different current injection patterns and combining EIT with other non-destructive evaluation techniques.
The study was financially supported by the AGC Foundation. The findings of the research were published in Volume 14, Issue 1 of the journal AIP Advances.
Tokyo University of Science (TUS) is a prestigious private research university in Japan. It has four campuses and has made significant contributions to science and technology since its establishment in 1881. Associate Professor Takashi Ikuno from TUS led the research team. His research interests include developing electronic devices with nanocarbon and low-dimensional nanomaterials.
In conclusion, the hybrid EIT method developed by the researchers combines mathematical approaches with machine learning, providing improved analysis of building structures. With its potential for non-destructive testing and foreign object detection, it has great significance in preventing building collapses and ensuring building health and safety.
