Transmissibility based structural assessment using deep convolutional neural network

摘要

This article presents novel deep Convolutional Neural Networks (CNN) approach for transmissibility based damage assessment in structures. The proposed approach operates on images that are generated from the structure's transmissibility functions so to exploit the CNNs' image processing capabilities and thus to automatically extract and select relevant features to the structure's degradation process. These feature maps are then fed into a multi-layer perceptron to achieve damage assessment. The proposed approach is validated and exemplified by means of two case studies involving a mass-spring system and a structural beam where training data is generated from finite element models that have been calibrated on experimental data. The results indicate that the proposed CNN based approach delivers satisfactory damage localization and quantification and thus being an appropriate tool for accurate damage assessment.