基于大数据深度学习的输电塔结构抗风易损性评估

为了有效预测输电网系统在强风作用下的响应，并开展高效精准的性能评估，文章提出了基于深度学习模型的风致易损性评估框架。以某具备健康监测系统的输电塔结构为例，首先对监测数据进行清洗和重构，通过大数据深度学习建立荷载输入和响应输出的等效映射模型，然后通过数值模拟生成灾害强度均匀的风场数据并由深度学习模型预测输电塔关键杆件响应，计算不同性能水准的易损性曲线。研究结果表明，经训练的深度学习模型可以涵盖实际工程中存在的各类不确定性因素，有效映射复杂风环境下输电塔结构的动力响应。提出的框架方法可以避免单纯通过数值模型制备大量动态响应数据，更高效地进行输电网系统风致易损性评估。     

Inverted pendulum stabilization: Characterization of codimension-three triple zero bifurcation via multiple delayed proportional gains

The paper considers the problem of stabilization of systems possessing a multiple zero eigenvalue at the origin. The controller that we propose, uses multiple delayed measurements instead of derivative terms. Doing so, we increase the performances of the closed loop in presence of system uncertainties and/or noisy measurements. The problem formulation and the analysis is presented through a classical engineering problem which is the stabilization of an inverted pendulum on a cart moving horizontally. On one hand, we perform a nonlinear analysis of the center dynamics described by a three dimensional system of ordinary differential equations with a codimension-three triple zero bifurcation. On the other hand, we present the complementary stability analysis of the corresponding linear time invariant system with two delays describing the behavior around the equilibrium. The aim of this analysis is to characterize the possible local bifurcations. Finally, the proposed control scheme is numerically illustrated and discussed.     

Study on maximum power point tracking of photovoltaic array in irregular shadow

Traditional maximum power point tracking (MPPT) methods can hardly find global maximum power point (MPP) because output characteristics curve of photovoltaic (PV) array may have multi local maximum power points in irregular shadow, and thus easily fall into the local maximum power point. To address this drawback, Considering that sliding mode variable structure (SMVS) control strategy have such advantages as simple structure, fast response and strong robustness, and P&O method have the advantages of simple principle and convenient implementation, so a new algorithm combining SMVS control method and P&O method is proposed, besides, PI controller is applied to reduce system chattering caused by switching sliding surface. It is applied to MPPT control of PV array in irregular shadow to solve the problem of multi-peak optimization in partial shadow. In order to verity the rationality of the proposed algorithm, the experimental circuit is built, which achieves MPPT control by means of the proposed algorithm and P&O method. The experimental results show that compared with the traditional P&O algorithm, the proposed algorithm can fast track the global MPP, tracking speed increases by 60% and the relative error decreased by 20%. Moreover, the system becomes more stable near the MPP, the fluctuations of output power is greatly reduced, and thus make full use of solar energy.     

基于MSA法的高心墙堆石坝地震沉降易损性分析

多条带分析法(MSA)是新近提出的一种针对离散强度因子的易损性分析方法,为结构安全评估提供了有效路径,但目前在土石坝工程领域应用较少。本文针对高心墙堆石坝,选取坝顶震陷率为性能指标,建议了轻度、中度和重度破坏三个等级;考虑地震动的随机性,采用MSA法开展了易损性研究,得到了不同破坏等级的易损性曲线和概率,为大坝的强震性能评价提供准则和参考;不同条带数量下地震易损性分析结果表明,适当减少水平条带数量,对易损性结果影响不大,这有利于计算工况的减少。     

近断层脉冲型地震动作用下面板堆石坝的动力响应

越来越多的高土石坝正在西部地区建设或规划中,其中许多位于发震断裂带附近,遭遇近场强震的概率随着大坝数量的增加而增加,近断层脉冲型地震动作用下大坝地震响应特性亟待研究。本文选取10组台湾集集地震实测加速度时程,结合200 m级的面板堆石坝开展动力有限元分析,研究脉冲和非脉冲地震动作用对面板堆石坝的加速度、水平位移以及面板顺坡向应力的影响。结果表明,与非脉冲型地震动相比,脉冲型地震动对大坝堆石体的水平位移和面板顺坡向应力的影响较大,影响程度随着PGV/PGA(地面峰值速度/地面峰值加速度)的增大而增大。因此,在高地震区修建高土石坝时应对脉冲型地震进行专门研究,综合评价大坝的地震安全性和极限抗震能力。     

Improved response surface method based on sample point selection strategies

In the reliability analysis of complex structures,response surface method(RSM)has been suggested as an efficient technique to estimate the actual but implicit limit state function.A set of sample points are needed to fit to the implicit function.It has been noted that the accuracy of RSM depends highly on the so-called sample points.However,the technique for point selection has had little attention.In the present study,an improved response surface method(IRSM)based on two sample point selection techniques,named the direction cosines projected strategy(DCS)and the limit step length iteration strategy(LSS),is investigated.Since it uses the sampling points selected to be located in the region close to the original failure surface,and since it needs only one response surface,the IRSM should be accurate and simple in practical structural problems.Applications to several typical examples have helped to elucidate the successful working of the IRSM.     

Sensor placement for structural damage detection with modal data

A new method is presented for prioritizing sensor locations for structural health monitoring (SHM). In view of the needs of SHM and damage detection,sensor locations are optimized for the purpose of both sensitivity for local damages and independence of the target mode. However,the two different optimization criterions lead to an inconsistency of the optimal result. Considering the structural response changes that result from damage,the relationship between the structural response and damage is deduced from the structural motion equation by a quasi-analytical mode. Based on the harmony between damage identifiability and mode observability,an object function is set up,including the information of mode independence and damage sensitivity. Utilizing the technique of singular value decomposition,an interior algorithm for the optimum sensor placement is proposed with the multiple objective criterions of minimizing the condition number of coefficient matrix and maximizing the fisher information matrix. A numerical example shows that this approach can effectively avoid the contradiction between the two different optimization criterions. Comparing with the result of single object,the result of damage detection from the optical sensor locations is much more accurate.     

Evaluation of group effect of pile group under dragload embedded in clay

A simple semi-empirical analysis method for predicting the group effect of pile group under dragload embedded in clay was described assuming an effective influence area around various locations of pile group. Various pile and soil parameters such as the array of pile group, spacing of the piles (S), embedment length to diameter ratio of piles (L/D) and the soil properties such as density (γ), angle of internal friction (φ) and pile-soil interface friction coefficient (μ) were considered in the analysis. Model test for dragload of pile group on viscosity soil layer under surface load consolidation conditions was studied. The variations of dragload of pile, resistance of pile tip and the layered settlement of soil with consolidation time were measured. In order to perform comparative analysis, single pile was tested in the same conditions. The predicted group effect values of pile group under dragload were then compared with model test results carried out as a part of the present investigation and also with the values reported in literatures. The predicted values were found to be in good agreement with the measured values, validating the developed analysis method. The model test results show that negative skin friction of pile shaft will reach 80%–90% of its maximum value, when pile-soil relative displacement reaches 2 mm. Foundation item: Project(50679015) supported by the National Natural Science Foundation of China     

The electrical properties and snow melting of graphite slurry infiltrated steel fiber concrete

The conductivity property of graphite slurry infiltrated steel fiber concrete (GSIFCON) was investigated by the four-probe method. The experimental results show that the electrical resistivity of GSIFCON decreases significantly at a specific concentration of graphite, i e, the percolation concentration. A model was accordingly developed to explain its conductive mechanism, where graphite particles could be served as bridge for conductive network between unconnected fiber chains. For an application, snow melting test of GSIFCON was carried out and a simple heating power analysis was performed. The results reveal that the thermal energy produced by GSIFCON makes snow melting effectively and the electrothermal efficiency can reach approximately 20%.