Damage identification for composite structures with a Bayesian network

Recent development on the application of distributed sensor networks, in structural health monitoring (SHM) for large structural areas has resulted in more complicated system identification techniques, particularly for those with multiple information sources. This paper presents an application of Bayesian inference network to detection of hole-type damages on a composite plate using multiple sensing data streams from a distributed sensor network. Representative damage features from 50 damage scenarios were used for the learning process. The Bayesian net is found to be promising when correctly diagnosing the damage's location and size for a validation case.