Acceleration Data from a Laboratory Bridge Excited using a Single Rolling Axle

Published: 09-06-2019| Version 1 | DOI: 10.17632/np8fvycwnx.1
Aaron Ruffels


The aim of this research was to detect damage to a laboratory model bridge using only measured data as an input to an algorithm. The data collected included the bridge accelerations at 7 locations for a number of crossings of a 12.7 kg axle. The velocity of the axle and the air temperature were also recorded. An artificial network (ANN) was trained to predict the recorded bridge accelerations in the healthy structural state. Damage sensitive features were defined as the root mean squared errors between the measured data and the ANN predictions. A baseline healthy state was then established with which new data could be compared to. By predicting the accelerations from additional unseen data, damage sensitive features could be extracted for these runs. Outliers from the reference state were taken as an indication of damage. Two outlier detection methods were used: Mahalanobis distance and the Kolmogorov-Smirnov test. The method showed promising results and damage was successfully detected for four out of the five single damage cases. A gradual damage case was also detected however for some instances, greater damage did not result in an increase in the damage index. The Kolmogorov-Smirnov test performed best at detecting small single damage cases while Mahalanobis distance was better at tracking gradual damage. The contents of the dataset and relevant material required for its reproduction are described below. 0_index_of_runs: Index of each crossing of the mass over the bridge. Includes temperature data, information regarding the damage case as well as additional comments. 1_acceleration_data: Separate file for each crossing of the mass over the bridge. Each file contains seven channels with acceleration data and one channel with voltage data. The voltage readings were obtained by the mass crossing a copper strip placed at 1 meter intervals along the 5 m bridge deck. This data was recorded using the Catman data aquisition software. 2_bridge_drawings: Drawings used to manufacture the model steel bridge. 3_accelerometer_locations: Locations for where the accelerometers were placed on the bridge. 4_damage_locations: Locations of which members were removed or damaged. See index of runs for further information. 5_gradual_damage: Figure showing the various levels of gradual damage applied to the structure. 6_rolling_mass: Dimensions of the steel rolling mass used to excite the structure. 7_bridge_ramp: Overview of the bridge and ramp set-up. 8_ramp_drawings: Drawings of the ramp used to accelerate the mass.


Steps to reproduce

1. Construct bridge according to "2_bridge_drawings". 2. Construct ramp to launch the mass from according to the figure "7_bridge_ramp" and the drawings "8_ramp_drawings". 3. Construct rolling steel mass (12.7 kg) according to "6_rolling_mass". 3. Place 7 accelerarometers according to "3_accelerometer_locations". 4. Place 6 pairs of copper tape strips at 1 meter intervals along the bridge deck. These are to act as switches. When the steel mass crosses a pair it should close a circuit connected to a 9 volt battery. 5. Connect the accelorometers and circuit connected to the copper strips to a signal amplifier such as a Spider8. 6. Set up the accelerators to record at 400 Hz using a data acquisition system (DAQ) such as Catman. 7. Start recording a run using the DAQ. 8. Release the mass from a height of 0.399 m above the bridge deck and allow to roll off the end of the bridge. 9. Stop recording for the run. 10. Repeat for each run described in "0_index_of_runs", with reference to "4_damage_locations" and "5_gradual_damage" for the relevant modifications to be made for each run.