Identification of intermediate debonding damage in FRP-strengthened RC beams based on a multi-objective updating approach and PZT sensors

Identification of the intermediate crack debonding in FRP strengthened reinforced concrete structures at its earliest stages is a very challenging task due to its complexity, its local nature and to a limited control over changing operational conditions which will undoubtedly affect the predictions. Impedances measured at high frequencies are sensitive enough to detect local damage. In this paper, we present an ensemble framework with classification using bagging and based on multi-objective particle swarm optimization (PSO) in order to efficiently evolve a damage identification in response to high frequency impedance measurements captured from different PZT sensors. An ensemble of classifiers has been proved to be an effective way to improve classification accuracy in difficult pattern recognition environments. Furthermore, in contrast to the conventional finite elements, spectral elements provide very accurate solutions at high frequencies and, therefore, they will be used to cooperate with the use of PZT sensors in order to develop an FRP debonding prediction technique. The framework is evaluated on a real-scale FRP-strengthened RC beam instrumented with several PZT sensors and tested with different levels of damage. The analysis of the results indicate that the proposed strategy provides a higher level of accuracy when compared to that using mono-objective optimization.