Abstract
The deterioration of bridge should be adequately evaluated to use it safely and to develop an appropriate maintenance strategy. Therefore, the present research proposed a method to quantify damage severity and existence of cracks by use of multipoint acceleration measurement and random forest which is one of the supervised machine learning method. Lots of previous research only used natural frequencies to detect damages, but the accuracy is not necessarily enough. This research therefore increases the amount of information including the maximum and standard deviation of accerelation and logarithmic decrement. Finally, the accuracy of the developed method is verified by leave-one-out cross validation and the vibration test results of actual damaged specimens.
