Structural Condition Classification of Railway Bridge KW51 Before, During, and After Retrofitting

Abstract

A linear classification algorithm for various structural states (before, during, and after retrofitting) of railway steel arch bridge KW51 based on Linear Discriminant Analysis (LDA) of principal components is proposed. The technique is typically employed in other fields, such as genetics, but its use in civil engineering is still limited. The method is applied in two steps, first, the original vibration measurements are projected to a lower dimensional subspace using Principal Component Analysis (PCA). The significant principal components are provided as inputs to LDA to constitute the newly transformed subspace. Joining PCA and LDA improves the performance aptitude of LDA when only first major principal components hold the main signature of the features. The acceleration datasets that represent bridge conditions under train passage on the steel arch railway bridge in Belgium, mentioned as bridge KW51 have been utilized for validation of the algorithm. The outcome of the visualization process is further analyzed using three clustering methods (kmeans, Fuzzy C means, and Gaussian Mixture Modelling). The confusion matrix is used to examine the performance of the clustering methods. The proposed algorithm provides promising results that outperformed other classification methods.