Analysis of ceramic pellets using ultrasound: time of flight and frequency spectrum with artificial neural network

One of the objectives of this work is to analyze the velocity of ultrasonic signals at ceramic pellets, establishing a correlation between the velocity from the ultrasonic wave and porosity. It is also be proposed, a correlation between ultrasonic frequency spectrum and pellet's porosity. At a first step, patterns of ultrasonic frequency spectrum using different samples with porosity previously defined are obtained. Then, such patterns are used to train an artificial neural network (ANN) which is used to analyze new signals and predict material's porosity. A motivation for this work is to develop a non-destructive technique which is able characterize the UO2 ceramic pellets, which are generally fabricated by sintering pressed powdered UO2 or mixed oxide at high temperature. (author)