Search for Author, Title, Keyword
RESEARCH PAPER
Improving the reliability of industrial reactors by using differential neural network architecture in ultrasonic tomography
 
More details
Hide details
1
WSEI University, Poland
 
2
Research and Development Center, Netrix S.A., Poland
 
3
Lublin University of Technology, Poland
 
 
Submission date: 2025-06-29
 
 
Final revision date: 2025-07-18
 
 
Acceptance date: 2025-08-01
 
 
Online publication date: 2025-08-02
 
 
Publication date: 2025-08-02
 
 
Corresponding author
Tomasz Rymarczyk   

WSEI University, Poland
 
 
 
HIGHLIGHTS
  • Innovative neural network structure with a differential layer.
  • A universal method suitable for various types of neural networks.
  • Achieved sharper reconstructions with enhanced structural detail and fidelity.
  • Validated on real-world reactor data using a 16-transducer UST system.
  • Suitable for industrial monitoring under limited and noisy measurement conditions.
KEYWORDS
TOPICS
ABSTRACT
Ultrasonic tomography (UST) represents a powerful non-invasive diagnostic technique for monitoring and analyzing internal processes within industrial reactors. Despite its potential, UST-based reconstructions are often challenged by the ill-posed nature of the inverse problem, limited measurements, and the presence of noise. To address these limitations, this study introduces a novel differential neural network architecture that enhances conventional deep learning models by incorporating a specialized differential layer. This layer processes two parallel input streams and operates on their residuals, thereby amplifying subtle variations in the data critical for accurate tomographic reconstructions. This study aims to empirically validate the concept of the efficacy of differentiated architecture. Reconstruction performance was evaluated using established quantitative metrics. Results demonstrate that models incorporating the differential layer consistently outperform their standard counterparts, delivering higher resolution, and superior noise robustness.
eISSN:2956-3860
ISSN:1507-2711
Journals System - logo
Scroll to top