Integrated Field Data Collection System to Determine Bridge Dynamic Impact Factor

Bridges experience dynamic amplification of structural responses during vehicle crossings, especially at approach-deck transitions where abrupt surface irregularities increase the dynamic impact factor (IM). This study presents a rapid method to estimate IMs using surface profiles, vehicular loading, and bridge characteristics. Field data were collected from the three major types of bridges commonly found in Oklahoma—concrete, steel, and prestressed concrete—each including both single- and multi-span configurations. The field data collection employed strain gauges, inertial measurement units, portable weigh-in-motion systems, high-definition cameras, an inclinometer-based profiler, and a submillimeter highway speed laser imaging system. IMs were computed as the ratio of dynamic to static strain using a customized smoothing approach, in which the strain signals were transformed into the frequency domain using a Fourier transform, high-frequency noise was removed through a power spectral density analysis, and static strain components were converted using a Butterworth low-pass filter. Multiple linear and random forest regression predictive models were evaluated, with the random forest ( R 2 = 0.589) proving a better fit because of predictors’ multicollinearity and nonlinearity. A feature importance analysis identified that surface roughness metrics were the most influential predictors, followed by vehicle attributes, such as gross vehicle weight, speed, and axle configuration, and bridge-related parameters including structure length and natural frequency. The results demonstrate the feasibility of estimating bridge impact factors from surface roughness, vehicle characteristics, and bridge parameters. Future work will focus on expanding datasets and exploring advanced models, such as neural networks and long short-term memory, to enhance predictive accuracy.