Structural Health Monitoring (SHM) has recently emerged as a useful tool for tracking theperformance parameters of a structure such as strain, deflection, and acceleration through a series ofsensors installed on them. For effective monitoring, interpretation of the monitoring data and isolating theunusual or novel events from high volume of sensor data, both Artificial Neural Networks (ANN) andWavelet Transform (WT) are found to be very useful. In this study the sensor data from Canadian bridgehave been utilized to develop ANN and WT-based methods for processing the SHM data and assessingthe structural conditions. The neural network is constructed with sixteen input nodes accepting data fromfifteen strain gauges and one temperature gauge, and the data from the remaining gauge is used as thetarget. The data collected at different time periods are tested against the trained network to find thepattern of difference in the inter-relation between the input and output data series’. The wavelet transformtechnique has been used for de-noising the sensor data before using them in the neural networks. Thepaper gives an overview of the study and presents the key results demonstrating the feasibility andusefulness of the proposed methods in interpreting SHM data.
展开▼
