Statistical Pattern Recognition for Railway Bridge Structural Damage Detection

Concerning the features of railway bridge structural damage, the method of combining the step-by step damage detection method with statistical pattern recognition is adopted to identify the railway bridge structural damage. Damage detection of railway bridges is divided into three identification steps, namely damage early warning, damage location and damage extent diagnosis. Binary-class pattern classification, multi-class pattern classification methods and support vector regression of statistical pattern recognition is used to realize the damage early warning, damage location and damage extent diagnosis respectively. The proposal method is verified by the measured data from one certain railway continuous girder bridge model test. It is shown that the innovative approach has a good ability to identify the damage, can be applied to detect the damage in real bridge structure. In comparison with the optimization method, the proposal method is apparently different in anti-noise capability, solution methods and ideas, so it is highly innovative.