A kNN algorithm for locating and quantifying stiffness loss in a bridge from the forced vibration due to a truck crossing at low speed

This paper proposes a k-Nearest Neighbours (kNN) algorithm for locating and quantifying bridge damage based on the time-varying forced frequencies due to a moving truck. Eigenvalue analysis of a simplified vehicle-bridge coupled system, consisting of a three-axle rigid truck model and a simply supported finite element beam model, shows how the eigenfrequencies of the coupled system vary with the locations of the vehicle and with the damage represented by a stiffness loss. The computational efficiency of eigenvalue analysis is exploited to generate a vast sample of patterns for training a kNN algorithm. In the field, acceleration due to the crossing of a test vehicle would be measured and analysed using a time–frequency signal processing tool to obtain the instantaneous frequencies. The crossing must take place at a low speed to achieve sufficiently high resolution and to minimise deviations from the eigenvalue solution. Then, the kNN algorithm searches for the patterns of forced eigenfrequencies that are closest to the on-site instantaneous frequencies to determine the location and severity of the damage. For theoretical testing purposes, field measurements are simulated here using coupled equations of motion and dynamic transient analysis.