Cross validation-based novelty detection of bridge group

Abstract

Novelty detection based on cross validation was proposed to detect the damage of the cluster of medium and minor bridges under the effect of environmental factors. The effectiveness of this method was tested through the numerical example of one bridge group that was composed of five similar bridges. Meanwhile, advantages and disadvantages of this method were discussed by comparing the effectiveness between proposed method and traditional novelty detection method. The numerical results showed that the cross validation-based novelty detection of bridge group method effectively mitigated the influence of environmental factors on damage detection.