Abstract
Early judgment of damage level of infrastructures is required after the earthquake. Human-based inspections are carried out mainly, however it takes long time and is expensive. On the other hand, accelerometer is relatively inexpensive and is installed easily. But a method to detect the state-change of structures from only the acceleration information in real-time has not been established. In this study, by applying wavelet transform to acceleration data of several large shaking table tests, we tried to detect the state change of structures. First, in the analysis of the acceleration data measured on the top of the RC column which had damage by shaking table test, it was possible to detect singularity points of the acceleration waveform by wavelet transform. To confirm whether the singularities were caused by damage, we checked strain of rebar, hysteresis and video records. As a result, it was shown that most of singularities in acceleration waveform occurred when RC columns had damages such as cracks and rebar yielding. We usually measure acceleration with sampling frequency 100 ∼ 200 Hz in seismic monitoring. However, there is a possibility that the occurrence of damage affects the higher frequency components of acceleration, so we conducted measurement with high frequency sampling (10 kHz). As a result, on the points which have been detected as singularities in 200 Hz sampling, we found high frequency components existed. In addition, it was suggested that such as the frequency of these high-frequency components varies due to the damage conditions of RC columns, there is a difference among the detected singularities due to damage. From these results, it is considered to be possible to detect damage from singularities which are extracted by wavelet transform. We consider that these results give us merits for developing efficient inspection techniques of actual RC structures.
