多点高密度配置した振動センサと部分構造の出力誤差による局所損傷検出

Abstract

Information about possible existence and its extent of structural damage is extremely important to diagnose structural health. The damage detection methods based on change in the building characteristics before and after the earthquake have been studied for a long time. However, the characteristics of the whole building system such as natural frequency have problems with ambiguity of relationship between characteristic change and damage state, and also sensitivity to local damages. To solve these problems, some authors have proposed a local damage detection method focusing on the characteristic change of each substructure, and have been doing analytical and experimental studies. A major obstacle to practical application of the method is that a large number of measurement points (sensors) are required, but recent innovative advancements of sensing technologies will allow us a large number and high density sensor deployment to the building. In order to experimentally verify the proposed substructure-based local damage detection method, it was applied to the large shaking table test of 1/3 scaled 18-story steel high-rise building in December 2013 at the E-Defense. A multi point time synchronous data measurement system with 152 MEMS-type six-axis vibration sensors was developed. The sensors were deployed to each beam-to-column joint of the test building to measure its translational acceleration and rotational velocity. The proposed local damage detection method was applied to the measured data for the existence and extent of local structural damages in each test stage to be estimated. The damage detection results were compared with the actual status of the test building to evaluate the performance of the proposed method. As the conclusion of this paper, the following three points were shown from the experimental results. 1) The multi point time synchronous measurement system experimentally demonstrated that six-axis vibration at all beam-to-column joints of the 18-story steel high-rise building (152 sensors × 6 directions = 912 ch.) can be synchronously measured at 500 Hz. It was confirmed that vibration in the rotational direction can be measured with enough accuracy and its amplitude is sufficiently large with respect to the resolution of the MEMS sensor (0.0125 deg./s) 2) The practical damage detection method based on the output error of the substructure was presented as a damage index Di which is the normalized amplitude of output error increase to an initial state of the substructure into consideration of the influence of observation noise and modeling error. As a result of the verification based on the experimental data, the damage detection could be performed using a single threshold value for all substructures even to the data set with different amplitude levels. 3) It was verified that the damage detection result of the proposed method at each stage of the experiment is in good correspondence relationship with the distribution of flange fractures at the steel beam ends as the actual damage status of the test building. As a result, it was proved that the local damage distribution of a steel rigid frame structure can be estimated by the proposed method.