Acoustic Leak Detection at Complicated Geometrical Structures Using Fuzzy Logic and Neural Networks

Methods of acoustic leak monitoring are of great practical interest for the safety of pressure vessels and pipe lines not only at the primary circuit of nuclear power plants. In this report some aspects of acoustic leak localization at complicated three-dimensional topologies for the case of leakage monitoring at the reactor vessel head of a VVER-440 are discussed.
An acoustic method based on pattern recognition is being developed. During the learning phase, the localization classifier is trained with sound patterns that are generated with simulated leaks at all locations endangered by leak. After training unknown leak positions can be recognized through comparison with the training patterns.
The sound patterns of the simulated leaks are simultaneously detected with an AE-sensor array and with high frequency microphones measuring structureborne sound and airborne noise, respectively.
The initial results show the used classifiers principally to be capable of detecting and locating leaks, but they also show that further investigations are necessary to develop a reliable method.