Improving the Temporal Resolution of Magnetic Induction Tomography for Molten Metal Flow Visualization

Magnetic induction tomography (MIT) is a nondestructive monitoring technique for imaging the electromagnetic properties of an object by mutual induction data of pairs of excitation and sensing coils. MIT has potential application in monitoring of molten metal flow in continuous casting. The main advantage of the MIT for this application is the contactless nature of the measurement process and its potential to generate images with high temporal resolution. Traditionally, the MIT image is formed using one frame of measured data that includes data from all excitation coils. In this paper, we study the improvement of the temporal resolution by reconstructing MIT images using data from one excitation at a time and using the Kalman-filtering approach. The images are presented using experimental data of the MIT system tested in a continuous casting unit. The MIT system used in this study can produce images 8 times faster than before by using new dynamical image reconstruction technique.