PROPOSAL OF AN ALGORITHM FOR REALTIME AUTOMATIC DETECTION OF AXLE PASSAGE TIME USING ACCELERATION RESPONSE

Abstract

 Fatigue crack is one of the most critical damages in the maintenance of steel bridges. Since the main cause of fatigue cracks is high strain caused by live load, it is essential to understand the actual loading condition, such as the vehicle weight and the number. Although there are various methods to measure the number of traveling vehicles, this study proposes a method to detect the axle passing time using acceleration sensors. The observed acceleration response includes not only the pulse-wave-like peak response that occurs when an axle passes directly over the sensor but also responses transmitted from the opposite pier side and adjacent lanes. The F value, which is the harmonic mean of the recall and the precision, was 54.4 % by a detection method using only the threshold, while the proposed method improved the F value to 94.1 %, indicating that axle detection can be performed with high accuracy.