A data-driven approach for drive-by damage detection in bridges considering the influence of temperature change

This paper proposes the use of a new data-driven approach for drive-by monitoring of bridge condition. The proposed algorithm uses an Artificial Neural Network (ANN) which is trained to predict bridge behavior using acceleration measurements from multiple passes of a traversing vehicle. A simple formulation is presented which allows the response at the point of contact between the tire and the pavement to be inferred from the measurements in the vehicle. The frequency content of this contact-point (CP) response is then used as the primary input to the ANN, along with the vehicle speed. The ANN is also trained to recognize the influence of temperature on the response and a new damage indicator is proposed, allowing the progression of damage over time to be visualized. Results show that using the CP-response in the ANN provides improved performance over the traditionally used axle-response. It is demonstrated that the proposed algorithm is capable of detecting mid-span and quarter-span cracking at all vehicle speeds considered and also in the presence of a rough pavement and varying temperature conditions. The proposed algorithm demonstrates clear benefits over existing drive-by bridge inspection techniques and provides a suitable approach for long term bridge condition monitoring.

Automated summary

This paper proposes a new data-driven approach for drive-by monitoring of bridge condition using an Artificial Neural Network (ANN). The ANN is trained to predict bridge behavior based on acceleration measurements from a passing vehicle. The response at the tire-pavement contact point is inferred and used as the primary input to the ANN, along with vehicle speed. The proposed algorithm demonstrates improved performance in detecting bridge cracking at different speeds, pavement conditions, and temperature variations compared to traditional methods, making it a suitable approach for long term bridge condition monitoring.