Image Reconstruction in Manufacturing Processes Using Deep Learning

Image reconstruction for industrial application-based Electrical Capacitance Tomography (ECT) has been widely used. ECT techniques gained a growing acceptance for many process applications, especially online monitoring, and control, due to their non-invasive nature, fast data acquisition speed, low construction cost, and safety. ECT was also found to be appropriate for various vessel sizes. Image reconstruction-based ECT aims to find the permittivity distribution of dielectric material over the cross-section based on the collected capacitance measurements. There are three main problems associated with the image reconstruction process using ECT (1) the mathematical model which describes the relationship between the permittivity distribution, and capacitance is nonlinear (2) the number of capacitance measurements collected during the reconstruction process is inadequate due to the limited number of used electrodes and (3) the reconstruction process is susceptible to noise leading to the ill-posed problem. Therefore, constructing an accurate and fast algorithm for real-time images and overcoming these restrictions is critical. Currently, we are exploring a deep neural network to handle the problem of reconstructing images of a Lost Foam Casting (LFC) process-based ECT. This process is known as the image reconstruction process. The goal is to accurately predict the image pixels such that a better understanding of the process of metal filling is achieved [1-3]. Dr. Sheta and his former student explored the significant advantages of deep learning in handling the image reconstruction problem and published several articles including recent research at the IEEE Access Journal [4].

Publications

1. A. Al-Afeef, A. F. Sheta, and A. Al-Rabea, “Image reconstruction of a metal fill industrial process using genetic programming,” in the 10th International Conference on Intelligent Systems Design and Applications, ISDA 2010, November 29 - December 1, 2010, Cairo, Egypt, pp. 12–17, 2010.

2. Alaa Sheta, Peter Rausch and Alaa Al-Afeef, Quality Management Using Electrical Capacitance Tomography and Genetic Programming: A new Framework, In the Proceedings of the SGAI International Conference on Artificial Intelligence, Cambridge, UK, pp. 211-216, December 2011. Paper received the Best Poster Award.

3. A. F. Sheta, P. Rausch, and A. Al-Afeef, “A monitoring and control framework for lost foam casting manufacturing processes using genetic programming,” International Journal of Bio-Inspired Computation (IJBIC), vol. 4, no. 2, pp. 111–118, 2012

4. W. Deabes, A. Sheta, and M. Braik, “An electrical capacitance tomography model-based long short-term memory recurrent neural networks for conductive materials,” IEEE Access, vol. 9, pp. 76325–76339, 2021.