Multidimensional signal processing for ultrasonic signal classification

Neural network based signal classification systems are being used increasingly in the analysis of large volumes of data obtained in NDE applications. One example is in the interpretation on ultrasonic signals obtained from inspection of welds where signals can be due to porosity, slag, lack of fusion and cracks in the weld region. Standard techniques rely on differences in individual A-scans to classify the signals. This paper proposes an ultrasonic signal classification technique based on the information in a group of signals and examining the statistical characteristics of the signals. The method was 2-dimensional signal processing algorithms to analyze the information in B- and B^′-scan images. In this paper, 2-dimensional transform based coefficients of the images are used as features and a multilayer perceptron is used to classify them. These results are then combined to get the final classification for the inspected region. Results of applying the technique to data obtained from the inspection of welds are presented.