Damage Detection in Railway Bridges Under Moving Train Load

In vibration based structural health monitoring, measured response data is used to detect structural damage. This study considers monitoring of railway bridges using response data under moving train loads. The effect of train-bridge interaction, especially with heavy trains, makes the train-bridge system time-varying, and modal identification challenging. The problem becomes even more complex when only the bridge response data is available, and characteristics of the train load (mass, speed etc.) are unknown. To avoid this complexity we engage into a strictly data based technique. Signal energies of the measured responses from healthy and damaged systems are compared statistically to detect the damage in the system. This comparison accounts for: (a) operational variability from different train masses and speeds, and (b) uncertainty from limited instrumentation and unknown input. The technique is validated though numerical simulations and the results promise faster detection of damage.