Innovative Methods of Tomographic Image Reconstruction Based on Machine Learning to Improve Monitoring and optimization in Industrial Processes

摘要

The article presents machine learning methods for acquiring, processing and reconstructing images from measurement data. The industrial tomography enables observation of physical and chemical phenomena without the need of internal penetration and allows real-time monitoring of production processes. The solution includes specialized devices for tomographic measurements and dedicated algorithms for solving the inverse problem. The work focuses on industrial tomography and image reconstruction using machine learning. The researches were carried out for synthetic data and laboratory measurements. The main advantage of the proposed system is the possibility of spatial data analysis and their high processing speed. The presented research results show that the process tomography gives the possibility to analyze the processes taking place inside the facility without disturbing the production, analysis and detection of obstacles, defects and various anomalies. Knowing the characteristics of a given solution, the application allows you to choose the appropriate method to reconstruct the image.