TL;DR:
- EIT, a cost-effective imaging technique, gains prominence in material inspection.
- Associate Professor Takashi Ikuno and team introduce a groundbreaking hybrid approach called “AND.”
- AND method combines IGN and 1D-CNN algorithms to enhance EIT accuracy.
- A significant reduction in size errors was observed with AND method.
- Research findings were published in AIP Advances on January 12, 2024.
- Potential applications include disaster prevention and structural health assessment.
- AND method outperforms IGN and 1D-CNN in detecting foreign objects within materials.
- Opportunity for future research lies in altering current injection patterns to improve resolution.
- EIT offers cost-effective and portable non-destructive testing for building health assessment.
- Promising applications in safety screening post-earthquake or explosion incidents.
Main AI News:
Electrical Impedance Tomography (EIT), a powerful non-destructive imaging technique, has emerged as a cost-effective and versatile solution for visualizing the inner workings of various materials. This method involves the injection of an electric current between two electrodes, generating an electric field, while additional electrodes capture distortions caused by foreign objects within the material. Unlike conventional imaging methods such as X-rays, computed tomography, and magnetic resonance imaging, EIT offers the advantages of affordability and portability, making it a promising tool for structural health monitoring of complex cement-based building materials.
Nevertheless, the challenge with EIT lies in the precise reconstruction of acquired information into meaningful images. Existing algorithms like one-step Gauss-Newton, primal dual interior point method, and iterative Gauss-Newton (IGN) are commonly employed for this purpose. However, due to the inherent nature of EIT, these mathematical methods often yield results marred by inaccuracies.
Recent attempts to address this issue have turned to machine learning algorithms, particularly one-dimensional convolutional neural networks (1D-CNN). While these algorithms exhibit promise, their Achilles’ heel lies in their ability to handle unforeseen data, diminishing their effectiveness.
In a groundbreaking development, Associate Professor Takashi Ikuno, in collaboration with researchers Keiya Minakawa, Keigo Ohta, and Hiroaki Komatsu from Tokyo University of Science (TUS), along with Associate Professor Tomoko Fukuyama from Ritsumeikan University, Japan, has introduced a revolutionary hybrid EIT approach named “AND.” This innovative technique seamlessly merges the strengths of IGN and 1D-CNN to overcome existing challenges.
In a scenario involving a very small foreign body with a cross-sectional area ratio of 5×10-4, the AND method drastically reduces size errors to less than 1/6th of those associated with conventional EIT methods. Their groundbreaking findings, published in the prestigious journal AIP Advances on January 12, 2024, have far-reaching implications.
Associate Professor Ikuno emphasizes the significance of their work: “From a disaster prevention standpoint, assessing the structural integrity of buildings constructed during periods of high economic growth is paramount. Our novel approach has the potential to elevate EIT as a non-destructive testing method and contribute significantly to averting building collapses.”
The AND method leverages 2D logical operations on multiple EIT-derived images to detect minute foreign objects embedded within materials. The research team conducted comprehensive testing of the AND method on real cement samples, employing both simulation and experimental data. The results unequivocally highlight the superior performance of the AND method compared to IGN and 1D-CNN methods in accurately reconstructing the position and size of foreign objects.
Furthermore, when dealing with experimental data, the AND method, along with the 1D-CNN method, outperforms IGN in terms of accuracy. The study also sheds light on a potential avenue for enhancing EIT further – by altering the current injection pattern and optimizing the spatial distribution of the electric field. Combining this approach with the present method and other Non-Destructive Evaluation (NDE) techniques could substantially enhance the resolution in detecting foreign particles’ size and position, paving the way for future research.
Dr. Ikuno envisions the practical implications of their proposed EIT reconstruction method: “While it may not match other NDE techniques in terms of resolution, it excels in terms of equipment size and cost-effectiveness. This breakthrough promises improved non-destructive detection of foreign objects, facilitating more consistent assessments of building health. Its potential applications extend to rapid safety screening post-earthquake or explosion incidents. Furthermore, the ease of training inspectors and personnel in utilizing this technology adds to its appeal.“
Conclusion:
The introduction of the innovative AND method signifies a major breakthrough in the field of Electrical Impedance Tomography (EIT). This advancement not only enhances the accuracy of non-destructive imaging but also opens up new avenues for applications in disaster prevention, structural health assessment, and safety screening. With its cost-effectiveness and portability, EIT is poised to become a valuable asset in the market for non-destructive testing and building health assessment technologies.
