A neural network approach for structural identification and diagnosis of a building from seismic response data

This work presents a novel procedure for identifying the dynamic characteristics of a building and diagnosing whether the building has been damaged by earthquakes, using a back‐propagation neural network approach. The dynamic characteristics are directly evaluated from the weighting matrices of the neural network trained by observed acceleration responses and input base excitations. Whether the building is damaged under a large earthquake is assessed by comparing the modal parameters and responses for this large earthquake with those for a small earthquake that has not caused this building any damage. The feasibility of the approach is demonstrated through processing the dynamic responses of a five‐storey steel frame, subjected to different strengths of the Kobe earthquake, in shaking table tests. Copyright © 2002 John Wiley & Sons, Ltd.     

Comparative study of the inelastic response of base isolated buildings

This article presents a numeric comparative study of the inelastic structural response of base isolated buildings. The comparative study includes the following isolation systems: laminated rubber bearings, New Zealand one, pure friction and the frictional pendulum ones. The study is based on obtaining non‐linear response spectra for various design parameters using six earthquake records. Usually the base isolation of a new building seeks to maintain the structure in the linear elastic range. The response of old weak buildings or the response of new ones subjected to extreme earthquakes may not be, necessarily, in the aforementioned ideal elastic range. Consequently, it is important to characterize the response of isolated buildings responding inelastically. A conclusion from this research is that the isolators affect significantly the structural response of weak systems. Rubber isolators seem slightly less sensitive to plastification that may occur in the structure compared to friction isolators. Ductility demands in the structure are affected significantly by friction and neoprene protected systems, in particular sliding ones where larger demands are obtained. Copyright © 2002 John Wiley & Sons, Ltd.     

Simplified vibratory characterization of alluvial basins

For the analysis of seismic wave amplification, modal methods are interesting tools to study the modal properties of geological structures. Modal approaches mainly lead to information on such parameters as fundamental frequencies and eigenmodes of alluvial basins. For a specific alluvial deposit in Nice (France), a simplified modal approach involving the Rayleigh method is considered. This approach assumes a set of admissible shape functions for the eigenmodes and allows a fast estimation of the fundamental frequency of the basin. The agreement with other numerical results (Boundary Element Method) and experimental ones is fully satisfactory. The simplified modal method then appears as an efficient mean for the global vibratory characterization of geological structures towards resonance. To cite this article: J.-F. Semblat et al., C. R. Geoscience 335 (2003).     

Analysis of the Chiufengershan landslide triggered by the 1999 Chi-Chi earthquake in Taiwan

A disastrous earthquake rocked Taiwan on September 21, 1999, with magnitude M_L=7.3 and an epicenter near the small town of Chi-Chi in central Taiwan. The Chi-Chi earthquake triggered landslide on the dip slope at the Chiufengershan. In this study, a review of the topography and geology of this area was followed by field investigations. Laboratory testing was applied to understand the geomaterial composing the slope. Then, based on a series of limit equilibrium analyses, the failure mechanism of this landslide and the risk of the residual slope were studied.

According to the stability analyses, the pre-quake slope is quite stable, with factor of safety of 1.77 (dry) to 1.35 (full groundwater level); explaining why there is no written record of a landslide here for the past 100 years. In contrast, a back analysis shows that the Chi-Chi earthquake-induced dynamic loading is far more than the dip slope can sustain, due in part to the short distance to the epicenter. A Monte Carlo type probability analysis suggests that the residual slope is more dangerous than the pre-quake slope and needs more attention.     

Local site effects in Latur, India: an empirical study based on the aftershocks of the Killari earthquake of September 29, 1993

The site amplification is estimated at five seismic stations of the Latur region using the horizontal to vertical spectral ratios of 33 aftershocks of the main Killari earthquake of September 29, 1993 (UTC). Spectral amplifications, ranging from a factor of 2–6 are found to vary with frequency at different places. Significant amplification is found at four sites within the Latur region, at Basavakalyan, Kasgi, Killari, and Mudgad Eakoji villages. Our results show a positive correlation between the site amplification and the damage pattern in area. The pattern and the nature of the site amplification estimated in the present study corroborates also with the analytical models and the borehole data indicating alternating layers of unconsolidated sediments and basaltic rocks.     

Estimation of causative fault parameters of the Rudbar earthquake of June 20, 1990 from near field SH-wave data

We analyze the strong motion accelerograms recorded for the large (M_S=7.7, M_W=7.3, m_b=6.4) Rudbar earthquake of June 20, 1990. The earthquake had a complex source process. We have identified the imprints of rupture of three localized asperities on the major causative fault on the accelerograms. These asperities are interpreted to correspond to (i) the main shock that initiated the rupture process and was located in the domino block between the Kabateh and Zard Goli faults, (ii) a foreshock that occurred about 10 s earlier in the Kabateh fault and (iii) a later shock, on the western end of the Baklor fault, which terminated the bilateral rupture process at the western end. We estimate the strike, dip and slip of these causative sub-event rupture planes using the SH spectral amplitudes, based on a point source representation of sub-events and a non-linear least square formulation for inversion of the amplitude data. The results of our inversion of the near field data are comparable to other studies based on teleseismic data.     

慢地震慢前兆的机制研究

通过岩石高温高压破裂实验与有机玻璃破裂试验的应力—应变曲线对比、地震波形记录(观测资料)与试样破裂波形(实验资料)对比、地震前后定点应变与主破裂前后应变形态变化的对比，认为慢地震是临界或亚临界破裂或预先滑动所致，是低频应变波动。它是材料屈服、弱化或塑性变化的必然结果。而慢前兆则是临近大地震(大破裂)前出现的诸如形变、低频地震波(破裂弹性波)等的短期及临震前兆现象，它比较可靠。但由于许多岩石主破裂前并无明显的临界破裂或预滑动现象，也并非所有岩石都一定出现明显屈服，所以也并非每个地震前都有慢地震，因此，也就不一定都出现低频波动。从而，也并不一定都出现相同的短期、临震前兆现象。     

昆仑山口西8.1级地震前青藏块体边界断层异常活动

系统分析了青藏块体边界断层的形变资料，研究了断层活动的动态过程及空间分布。结果表明，昆仑山口西8．1级地震前孕震影响范围达到青藏块体的周边地区；发震断层所在的构造带震前断层活动最为剧烈；加强对构造块体断层整体活动的宏观动态比较和分析，有助于判定未来强震发生的危险地段；震后应力将转移并集中到西秦岭构造带及其邻近地区。     

丽江7.0级地震重力前兆模式研究<

为了研究1996年丽江MＳ7.0地震前地震孕育过程或前兆表现，利用滇西地震实验场内高精度重复重力观测数据，结合地质调查和地球物理推断结果，并考虑到观测数据和模型粗差， 采用稳健或抗差-贝叶斯最小二乘算法和多断层位错模型，首次初步反演获得了研究区主要活动断裂滑动的时间变化分布．结果表明，1990～1997年断层运动的时间变化，较好地反映了1996年丽江MＳ7.0地震孕育过程．其主要前兆模式图象具有主震余震型特征，遵循地壳内部密度和地壳形变耦合运动模式（简称DD耦合运动模式）．      

丽江地震前后重力场变化的有限矩形位错模型分析

研究了用有限矩形位错模型计算地壳形变引起的地面重力场变化的方法．以丽江MＳ7.0地震为例，讨论了确定用于模型计算的断层面参数的原则，并给出了结果．计算和分析了不同类型位错引起的重力场变化图象特征，并与丽江地震前后观测到的重力场变化进行比较．结果表明，在发震断层有限范围内模型可解释同震重力场的变化，但模型对于更大空间范围上的重力场变化并不能给出很好地解释.