Application of Linear Discriminant Analysis to Ultrasound Signals for Automatic Microstructural Characterization and Classification

During the gas tungsten arc welding of nickel based superalloys, the secondary phases such as Laves and carbides are formed in final stage of solidification. But, other phases such as gamma(boolean AND)" and delta phases can precipitate in the microstructure, during aging at high temperatures. Nevertheless, choosing the appropriate conditions of welding can minimize the formation of the Nb-rich Laves phases and thus reduce the susceptibility to solidification cracking. This study proposed and evaluated the performance of an automated processing system to microstructurally characterizing the kinetics of phase transformations on a Nb-base alloy, thermally aged at 650 and 950 degrees C for 10, 100 and 200 h, using Linear Discriminant Analysis (LDA) on Background echo and Backscattered ultrasound signals at frequencies of 4 and 5 MHz. The main goal of this work is to design a more practical processing system in terms of the accuracy and the speed of processing. This system is composed of three methodologies: the first methodology uses LDA coefficients of normalized ultrasound signals, the second methodology uses LDA coefficients of error signals of the third-order linear prediction model of normalized ultrasound signals and the third methodology uses LDA coefficients of Discrete Cosine Transform. In all three methods, the Probabilistic Neural Network was used as a classifier. The highest accuracies were provided by the third methodology with average classification accuracies of 94.50 and 75.50 %, respectively for thermal aging at 650 and 950 degrees C. Indeed, LDA proved to be an efficient processing step for microstructural classification tasks.