Probabilistic neural networks for seismic damage mechanisms prediction

The procedure commonly employed to assess the seismic vulnerability of buildings uses simplified qualitative and quantitative observations obtained from the measured data entered into report forms. In Italy, the data sheets adopted by the National Defence Group against Earthquakes (Gruppo Nazionale per la Difesa dai Terremoti—GNDT) play a unifying and reference role. This paper proposes a method for the processing of the data contained in such report forms which is based on probabilistic neural networks producing a Bayesian classification. The final goal is to exploit the fundamental learning and generalization capabilities of neural networks to obtain an estimate of the vulnerability of structural systems. In particular, the aim is to be able to predict the damage mechanisms which may be triggered in the macro-elements of public worship buildings. Copyright © 1999 John Wiley & Sons Ltd.     

Non-local effects on dynamic damage accumulation in brittle solids

This paper presents a non-local analysis of the dynamic damage accumulation processes in brittle solids. A non-local formulation of a microcrack-based continuum damage model is developed and implemented into a transient dynamic finite element computer code. The code is then applied to the study of the damage accumulation process in a concrete plate with a central hole and subjected to the action of a step tensile pulse applied at opposite edges of the plate. Several finite element discretizations are used to examine the mesh size effect. Comparisons between calculated results based on local and non-local formulations are made and non-local effects are discussed. Copyright © 1999 John Wiley & Sons, Ltd.     

Field-based description of near-surface crustal deformation in a high-strain shear zone: A case study in southern Kyushu,Japan

This study investigated geological evidence for near-surface crustal deformation in a high-strain shear zone that has been geodetically identified but which is not associated with obvious tectonic landforms. Fieldwork was conducted in the east–west-trending southern Kyushu high-strain shear zone (SKHZ), Japan, focusing mainly on occurrences of fracture zones, which are defined by a visible fracture density of >1 per 10 cm² and are commonly associated with cataclasite, fault breccia, and gouge. The area in which east–west-trending fracture zones are dominant is restricted to the east–west-trending, ~2-km-wide aftershock area of the 1997 Northwestern Kagoshima Earthquakes. Analysis of slip data from minor faults using the multiple inverse method, irrespective of whether the faults are in fracture zones, reveals that the area where the calculated main stress field is consistent with the current stress field estimated from focal-mechanism solutions of microearthquakes is restricted to the east–west-trending aftershock area. This finding for the SKHZ contrasts with the case of the Niigata–Kobe Tectonic Zone, which is a major strain-concentration zone with many exposed active faults in central Japan and for which the stress field estimated using fault-slip data is considered to be uniform and coincides with the current stress field. The cumulative amount of displacement estimated from the areal density of fracture zones in the SKHZ study area is smaller than that estimated from geodetically measured strain rates. Investigations based on slip data from minor faults and fracture-zone occurrence could help to identify concealed faults that are too small to generate tectonic landforms but which are sufficiently large to trigger major earthquakes.     

THE N2 METHOD FOR THE SEISMIC DAMAGE ANALYSIS OF RC BUILDINGS

A comprehensive, though relatively simple, non-linear method for the seismic damage analysis of reinforced concrete buildings (the N2 method) has been elaborated. The basic features of the method are: the use of two separate mathematical models, application of the response spectrum approach and of the non-linear static analysis, and the choice of a damage model which includes cumulative damage. The method yields results of reasonable accuracy provided that the structure oscillates predominantly in the first mode. Three variants of a seven-storey building have been used as illustrative examples for the application of the method. Four different types of the analysis, with different degrees of sophistication, have been performed in order to estimate the influence of several assumptions and approximations used in the N2 method.     

A NOTE ON DAMAGE-BASED INELASTIC SPECTRA

When a structure is subjected to moderate to severe ground motions, a few excursions of the response yield level may take place following the reductions usually enforced in the design forces. These excursions are associated with progressive damage in the structure. Thus, a choice of the design level has to be suitably based on the maximum damage to be allowed in the structure. In this paper, a stochastic technique of developing damage-based non-linear spectra has been proposed for the aseismic design of those structures which can be idealized through Single-Degree-Of-Freedom (SDOF) oscillators. The proposed technique has been illustrated by obtaining the non-linear spectra which can possibly be used for a damage-based design. Along with the spectra, allowable ductility demand which should be supplied through proper sizing and detailing of the members and is compatible with the damage has also been specified. The non-linear SDOF oscillators have been approximated for this purpose by equivalent linear oscillators using a new stochastic linearization technique. The proposed linearization technique has been validated through simulation results in the case of an idealized, non-hysteretic, Elasto-Plastic (EP) model.     

PROPAGATION OF TORSIONAL SURFACE WAVES IN VISCOELASTIC MEDIUM

It is well known that an elastic homogeneous half-space does not allow torsional surface waves to propagate. The present paper attempts to find out the possibility of propagation of such waves in a viscoelastic half-space. The study reveals that although the homogeneous elastic half-space does not allow torsional surface waves to propagate, a viscoelastic half-space does so. The wave is damped due to the viscoelastic parameter. It has also been found that as the viscoelastic parameter decreases, the medium becomes elastic and the torsional surface waves ceases to propagate.     

Embankment deformation analyzed by elastoplastic damage model coupling consolidation theory

The deformation of embankment has serious influences on neighboring structure and infrastructure. A trial embankment is reanalyzed by elastoplastic damage model coupling Biot's consolidation theory. With the increase in time of loading, the damage accumulation becomes larger. Under the centre and toe of embankment, damage becomes serious. Under the centre of embankment, vertical damage values are bigger than horizontal ones. Under the toe of embankment, horizontal damage values are bigger than vertical ones.     

基于轴向位移的钢支撑疲劳损伤评估方法

在罕遇地震下,框架-中心支撑结构中的钢支撑常因局部屈曲位置的低周疲劳开裂而过早退出工作。本文在焊接工字形钢支撑低周疲劳试验研究基础上,提出了一种可用于框架-中心支撑体系非线性动力时程分析中钢支撑疲劳损伤评估的经验方法,并给出了相关步骤的算法流程。研究结果表明:本文方法以钢支撑轴向位移为损伤参量,能实时估算在随机位移荷载下钢支撑的低周疲劳累积损伤发展,并与试验结果吻合较好。     

基于BP神经网络的空间索杆结构节点损伤识别研究

针对某实际空间索杆结构的节点损伤现象,采用BP神经网络与基于振动的损伤识别两步法对其进行了识别研究,即首先确定可能发生节点损伤的子区域,在此基础上利用对应子区域的子网络识别出具体的损伤位置和程度。识别过程中采用两个杆单元模拟发生节点损伤的杆件,用抗弯刚度降低的端部短杆单元模拟节点损伤。研究表明,虽然空间索杆结构的动力性能较为复杂,但基于结构固有频率和模态位移的组合指标对节点损伤仍较为敏感,利用它们进行节点损伤识别是有效的。     

基于小波包分解和模糊聚类的网格结构损伤诊断方法

本文针对网格结构的特点,利用其在快速正弦扫频激励下的动力响应,基于小波包分解频带能量分析和模糊聚类法,提出了一种适合于网格结构的损伤诊断方法。利用两个传感器损伤信息的联合诊断,有效地解决了使用一个传感器无法检测出对称损伤的难题。利用有限元模拟,应用上述方法对一单层球面网壳结构进行了损伤诊断。结果表明,对网壳杆件较小程度的损伤,子频带能量成分变化敏感,用其构造诊断向量和建立损伤样本库,并进行模糊聚类和目标识别,基本可以正确地识别出网壳杆件的损伤位置和程度。同时该方法测试时间短,使用设备少,有可靠的理论基础,具有一定的工程应用参考价值。