An Automatic Bridge Damage Diagnostics Method Using Empirical Mode Decomposition Based Health Indicators and Neuro‐fuzzy Classification

Large amounts of data are generated by structural health monitoring systems continuously. Data-driven methods can transform the available data into valuable information for decision makers. However, these methods for structural health monitoring of bridges are usually developed and tested by analysing a finite element model of the bridge, where the uncertainties affecting an in-field bridge are usually omitted. Modal parameters of the bridge are usually used to monitor the health state of the bridge, but it can be a difficult and time-consuming process to extract these parameters from the bridge vibration data in a reliable manner. Conversely, when the raw vibration behaviour of the bridge is monitored, promising results for bridge condition monitoring and damage diagnostics can be obtained in a fast way. In this paper, we propose a data-driven methodology to assess the health state of bridges, by analysing their vibration behaviour. The aim of the first step of the method is to extract statistical, frequency-based and vibration-based features from the measured bridge vibration. The second step is used to define a set of bridge Health Indicators by assessing the trend of these extracted features over time. The main novelty of this work lies in the use of the empirical mode decomposition method to assess the trend of the extracted features over time, rather than to analyse the dynamic behaviour of the structure directly. Finally, a Neuro-Fuzzy classifier, which is trained using a supervised process, is used to assess the health state of the bridge automatically. The proposed method is validated and tested by monitoring the vibration behaviour of an in-field bridge, which is subjected to a progressive damage process.