A novel damage detection method is applied to a 3-story frame structure, to obtain statistical quantification control criterion of the existence, location and identification of damage. The mean, standard deviation, an...A novel damage detection method is applied to a 3-story frame structure, to obtain statistical quantification control criterion of the existence, location and identification of damage. The mean, standard deviation, and exponentially weighted moving average (EWMA) are applied to detect damage information according to statistical process control (SPC) theory. It is concluded that the detection is insignificant with the mean and EWMA because the structural response is not independent and is not a normal distribution. On the other hand, the damage information is detected well with the standard deviation because the influence of the data distribution is not pronounced with this parameter. A suitable moderate confidence level is explored for more significant damage location and quantification detection, and the impact of noise is investigated to illustrate the robustness of the method.展开更多
文摘煤矿长时间开采会导致上覆岩层在重力因素下产生刚性弯曲、断裂等,进而造成一系列安全隐患。如何高效地对煤矿区进行地表形变监测,对煤矿安全、自然生态有着十分重要的意义。文章结合小基线集干涉测量(small baseline subset InSAR,SBAS-InSAR)技术与瞬变电磁测深法对煤矿区进行沉降监测。此外,采用标准差椭圆分析了沉降体空间形变规律与特征,利用长短时记忆网络(long short term memory network,LSTM)模型预测了沉降体形变趋势。分析结果表明:通过瞬变电磁法与干涉测量结合,地表形变范围基本与瞬变电磁反演的采空区吻合,标准差椭圆分析出形变中心向西南方移动376 m,沉降形态也在向似圆状过渡;LSTM模型可以预测两个月的形变,误差为2 mm,可以实现煤矿区形变趋势的短期预测。
基金Natural Natural Science Foundation of China Under Grant No 50778077 & 50608036the Graduate Innovation Fund of Huazhong University of Science and Technology Under Grant No HF-06-028
文摘A novel damage detection method is applied to a 3-story frame structure, to obtain statistical quantification control criterion of the existence, location and identification of damage. The mean, standard deviation, and exponentially weighted moving average (EWMA) are applied to detect damage information according to statistical process control (SPC) theory. It is concluded that the detection is insignificant with the mean and EWMA because the structural response is not independent and is not a normal distribution. On the other hand, the damage information is detected well with the standard deviation because the influence of the data distribution is not pronounced with this parameter. A suitable moderate confidence level is explored for more significant damage location and quantification detection, and the impact of noise is investigated to illustrate the robustness of the method.
基金Project supported by China Postdoctoral Science Foundation (20100481488), Key Fund Project of Advanced Research of the Weapon Equipment (9140A33040512JB3401).