Structural fault diagnosis and isolation using neural networks based on response-only data

摘要

To diagnose faults in engineering structures in the situations where the excitation signals are unavailable or inaccessible, response-only data, transmissibility function, are utilised to train neural networks. The technique is verified with two examples based on two different structural systems. The neural network classifiers clearly deliver the diagnostic indications of the faults introduced into the structural systems, which suggests that the transmissibility function is a sensible response-only data source for structural fault diagnosis.