In this work we present a computational procedure applied to ultrasound based inspection methods, making the diagnosis of metallic welding faults less dependent on the experience of the professional who carries out the evaluation, and, at the same time, less susceptible to errors. The proposed approach uses digital signal processing techniques, with emphasis to the neural network technique, in order to classify and identify faults.

Extensive laboratory experiments were carried out in order to inspect a large number of specimen and collect samples for verifying the robustness of the proposed approach. The results of this work prove the efficiency of the proposed method in identifying the presence of faults. They also indicate its efficiency in classifying the most important types of metallic welding faults, and consequently, in obtaining a satisfactory performance in an automatic diagnosis system.

The mathematical tools developed in this work for welding inspection and automatic diagnosis in metallic plates can be extended to other welding geometries involving different metallic alloys.

Key words: artificial intelligence, computer processing and simulation