Identification of moving loads based on the information fusion of weigh-in-motion system and multiple camera machine vision

• Published in: Measurement : journal of the International Measurement Confederation Vol. 144; pp. 155 - 166
• Main Authors: Dan, Danhui, Ge, Liangfu, Yan, Xingfei
• Format: Journal Article
• Language: English
• Published: London Elsevier Ltd 01.10.2019; Elsevier Science Ltd
• Subjects: Bridge; Bridge decks; Bridge loads; Bridges; Cameras; Concentrated loads; Data integration; Fuel consumption; Information fusion; Machine vision; Motion systems; Moving load identification; Moving loads; Multisensor fusion; Piezoelectricity; Structural engineering; Traffic control; Traffic flow; Video data; Video monitoring system; Weight; Weight-in-motion system; Moving load identification; Machine vision; Weight-in-motion system; Bridge; Video monitoring system; Information fusion
• ISSN: 0263-2241; 1873-412X
• DOI: 10.1016/j.measurement.2019.05.042

•A Real-time solution on monitoring traffic flow loads on the full bridge deck.•The load values of all vehicles are measured in advance by weigh-in-motion system.•Machine vision technology is used to identify the real-time locations of vehicles.•Specific new technologies are developed to support the data fusion of two systems.•The accuracy of the proposed approach is verified by simulated and the field data. Accurately identifying moving loads is of significance for the health monitoring of bridges. However, since the existing identification techniques can only realize load identification in one direction or for part of bridges, it is still a challenge to simultaneously identify transverse and longitudinal loads on the full deck of bridge. This paper proposed an information-fusion-based method for the load identification to be applied to bridges of different lengths. In this method, the pavement-based weigh-in-motion system (WIMs) laid out at the beginning of the bridge is used to obtain the weight of vehicles captured by cameras. The videos of traffic flow acquired by multiple cameras arranged along the bridge are employed to calculate the vehicle’s trajectory and location. The weight and location data are matched when the vehicle in the video crosses the piezoelectric sensor of WIMs for the same time as the WIMs records a weight information. Further, since the vehicles are equivalent to concentrated loads, values and locations of all moving loads on the whole bridge are identified in real time. The reliability and accuracy of the proposed approach is verified by multi-view 3D simulation video data and the field data from a ramp bridge.