Automatic fault detection using ultrasonic techniques: expert system vs. signal processing

In general, ultrasonic sensors are used in environments where there can be hostile conditions which impede the successful use of other methods such as artificial vision. When this occurs, ultrasonic sensors can be used in order to check for the existence of fabrication defects in foundry pieces before allowing them to be machined. This check is carried out by studying the echoes coming from the piece under analysis. This article is based on the comparison of the results obtained from two analysis techniques which use these echoes. Both systems work by differentiating the valid pieces from the defective ones, using the high part of the ultrasonic frequency range (215 kHz). In the first case, the process of separation of valid and defective pieces is based on expert systems, applied directly on the temporal characteristics of the echoes. For the second defect detection method, signal processing techniques are used on these high-frequency echoes; firstly, digitally filtering the signal and then, once the part of the signal pertaining to the echoes coming from the piece is separated, it is decomposed into its frequential components. The techniques used in these two working methods are explained in this article