Vision-based multi-level synthetical evaluation of seismic damage for RC structural components: a multi-task learning approach

Recent studies for computer vision and deep learning-based,post-earthquake inspections on RC structures mainly perform well for specific tasks,while the trained models must be fine-tuned and re-trained when facing new tasks and datasets,which is inevitably time-consuming.This study proposes a multi-task learning approach that simultaneously accomplishes the semantic segmentation of seven-type structural components,three-type seismic damage,and four-type deterioration states.The proposed method contains a CNN-based encoder-decoder backbone subnetwork with skip-connection modules and a multi-head,task-specific recognition subnetwork.The backbone subnetwork is designed to extract multi-level features of post-earthquake RC structures.The multi-head,task-specific recognition subnetwork consists of three individual self-attention pipelines,each of which utilizes extracted multi-level features from the backbone network as a mutual guidance for the individual segmentation task.A synthetical loss function is designed with real-time adaptive coefficients to balance multi-task losses and focus on the most unstably fluctuating one.Ablation experiments and comparative studies are further conducted to demonstrate their effectiveness and necessity.The results show that the proposed method can simultaneously recognize different structural components,seismic damage,and deterioration states,and that the overall performance of the three-task learning models gains general improvement when compared to all single-task and dual-task models.

Numerical simulation study on monoblock casting process of ultra-slender structural components and experimental validation

Substrate, a typical ultra-slender aluminum alloy structural components with a large aspect ratio and complex internal structure, was traditionally manufactured by re-assembly and sub-welding. In order to realize the monoblock casting of the substrate, the Pro/E software was utilized to carry out three-dimensional(3D) modeling of the substrate casting, and the filling and solidification processes were calculated, as well as the location and types of casting defects were predicted by the casting simulation software Anycasting. Results of the filling process simulation show that the metal liquid is distributed into each gap runner evenly and smoothly. There is no serious vortex phenomenon in the mold cavity, and the trajectory of the virtual particles is clear. Results of the solidification process simulation show that shrinkage cavities mainly appear at the junction of gap runners and the rail surface of the substrate. The average deformation is 0.6 mm in X direction, 3.8 mm in Y direction, and 8.2 mm in Z direction. Based on the simulation results, the casting process of the substrate was optimized, and qualified castings were successfully produced, which will provide a reference for the casting process design of other ultraslender aluminum alloy structural components.

Bridging finite element and deep learning: High-resolution stress distribution prediction in structural components

Finite-element analysis(FEA)for structures has been broadly used to conduct stress analysis of various civil and mechanical engineering structures.Conventional methods,such as FEA,provide high fidelity results but require the solution of large linear systems that can be computationally intensive.Instead,Deep Learning(DL)techniques can generate results significantly faster than conventional run-time analysis.This can prove extremely valuable in real-time structural assessment applications.Our proposed method uses deep neural networks in the form of convolutional neural networks(CNN)to bypass the FEA and predict high-resolution stress distributions on loaded steel plates with variable loading and boundary conditions.The CNN was designed and trained to use the geometry,boundary conditions,and load as input to predict the stress contours.The proposed technique’s performance was compared to finite-element simulations using a partial differential equation(PDE)solver.The trained DL model can predict the stress distributions with a mean absolute error of 0.9%and an absolute peak error of 0.46%for the von Mises stress distribution.This study shows the feasibility and potential of using DL techniques to bypass FEA for stress analysis applications.

Structural Analysis of Platforms with the Effect of Joint Flexibility

A modified space beam element is presented in this paper to consider the local joint flexibility of T, Y tubular joints subjected to axial forces and in-plane bending moments for analysis of platforms. Two numerical examples are shown to verify the efficiency and validity of the method presented here.

Surface grain refinement mechanism of SMA490BW steel cross joints by ultrasonic impact treatment

Ultrasonic impact treatment (UIT) is a postweld technique for improving the fatigue strength of welded joints. This technique makes use of ultrasonic vibration to impact and plastically deform a weld toe and can achieve surface grain refinement of the weld toe, which is considered as the main reason for the improvement of fatigue strength. In this paper, the microstructure of the surface of a treated weld toe was observed by metallographic microscopy and transmission electron microscopy (TEM). The results show that UIT could produce severe plastic deformation on the surface layer of the weld toe and the maximum depth of plastic deformation extended to approximately 260 μm beneath the treated surface. Repeated processing could exacerbate the plastic deformation on the surface layer, resulting in finer grains. We can conclude that the surface grain refinement mechanism of SMA490BW welded joints is related to the high density of dislocation tangles and dislocation walls. © 2017, University of Science and Technology Beijing and Springer-Verlag Berlin Heidelberg.

THEORY OF ELASTICITY OF AN ANISOTROPIC BODY FOR THE BENDING OF BEAMS

A theory of elasticity for the bending of orthogonal anisotropic beams was developed in this paper by analogy with the special case, which can be obtained by applying the theory of elasticity for bending of transversely isotropic plates to the problems of two dimensions. The authors also presented a method to solve the problems of bending of orthogonal anisotropic beams and a new theory of the deep-beam whose ratio of depth to length is larger. It is pointed out that Reissner's theory which takes into account the effect of transverse shear deformation is not suitable for the components of stress in our case.

Seawater Corrosion Fatigue of Welded Joints Under Random Variable Amplitude Loading

Offshore platforms are always subjected to wave action which is random variable amplitude cyclic loading. In order to simulate the stressing condition at the 'hot spot' of the tubular joints and the marine environment, random variable amplitude fatigue tests have been carried out on welded plate joints in sea water. The tests have been conducted under the conditions of loading frequency of 0.2 Hz/, stress ratio of -1, seawater temperature of about 20°C and cathodic protection with the potential about -850 mV, SCE. The test results have been compared with the seawater corrosion fatigue life under constant amplitude loading. Miner's linear cumulative damage summation rule has been used to predict the corrosion fatigue life under variable amplitude loading. The predicted life is in good agreement with the test data.

Regression Formulations for the Stress at Hot Spot of Multiplanar Tubular DT Joints

Stress concentration analysis of multiplanar tubular DT joints plays an important role in the fatigue design of offshore platforms. A semi-analytic method for stress analysis under the condition of any loads is briefly introduced in the paper. Nineteen multiplanar tubular DT joints with one of two braces of the same dimension subjected to axial loads and out- of- plane bending moments are computed for parametric stress analysis by using the present method. The influence of geometrical parameters on the stresses of multiplanar tubular DT joints is discussed and compared with corresponding uniplanar T joints. The regression formulae for the stress at hot spot of multiplanar DT joints are found by modification of SCF of corresponding uniplanar T joints. The parametric formulae for the maximum stress by superposition. Finally, the influences of stiffening effect and load-interaction effect on the maximum stress of DT joints are discussed.

PERTURBATION SOLUTION FOR THERMAL EXPANSION BUCKLING OF NO EXPANSION JOINT SLOPE PAVEMENT PLATE

The mechanical mechanism of thermal expansion buckling of no expansion joint slope pavement undergoing the action of a temperature field was analyzed. By using the regular perturbation method, the formula of perturbation solution for this problem was derived, the relationship between critical laying temperature difference of slope pavement and of level straight pavement was studied, and the unified solution as well as its numerical results was also obtained. In terms of this research, the reasonable laying temperature of no expansion joint slope pavement was given.

P-FEMOL ANALYSIS OF PLATE BENDING PROBLEMS

In this paper, the p- version of the finite element method of lines (FEMOL) for the analysis of the Mindlin-Reissner plate bending problems is presented and a class of p-FEMOL elements with polynomial degrees as high as nine is developed. Numerical examples given in this paper show tremendous performance of the present method: namely, rapid convergence rate, high accuracy for both displacements and stress resultants, removal of shear-locking trouble, capability of dealing with difficult problems such as the boundary layer behavior near a free edge and stress concentration around a hole.

A local and global statistics pattern analysis method and its application to process fault identification

Traditional principal component analysis （PCA） is a second-order method and lacks the ability to provide higherorder representations for data variables. Recently, a statistics pattern analysis （SPA） framework has been incorporated into PCA model to make full use of various statistics of data variables effectively. However, these methods omit the local information, which is also important for process monitoring and fault diagnosis. In this paper, a local and global statistics pattern analysis （LGSPA） method, which integrates SPA framework and locality pre- serving projections within the PCK is proposed to utilize various statistics and preserve both local and global in- formation in the observed data. For the purpose of fault detection, two monitoring indices are constructed based on the LGSPA model. In order to identify fault variables, an improved reconstruction based contribution （IRBC） plot based on LGSPA model is proposed to locate fault variables. The RBC of various statistics of original process variables to the monitoring indices is calculated with the proposed RBC method. Based on the calculated RBC of process variables＇ statistics, a new contribution of process variables is built to locate fault variables. The simula- tion results on a simple six-variable system and a continuous stirred tank reactor system demonstrate that the proposed fault diagnosis method can effectively detect fault and distinguish the fault variables from normal variables.

gscaLCA in R: Fitting Fuzzy Clustering Analysis Incorporated with Generalized Structured Component Analysis

Clustering analysis identifying unknown heterogenous subgroups of a population(or a sample)has become increasingly popular along with the popularity of machine learning techniques.Although there are many software packages running clustering analysis,there is a lack of packages conducting clustering analysis within a structural equation modeling framework.The package,gscaLCA which is implemented in the R statistical computing environment,was developed for conducting clustering analysis and has been extended to a latent variable modeling.More specifically,by applying both fuzzy clustering(FC)algorithm and generalized structured component analysis(GSCA),the package gscaLCA computes membership prevalence and item response probabilities as posterior probabilities,which is applicable in mixture modeling such as latent class analysis in statistics.As a hybrid model between data clustering in classifications and model-based mixture modeling approach,fuzzy clusterwise GSCA,denoted as gscaLCA,encompasses many advantages from both methods:(1)soft partitioning from FC and(2)efficiency in estimating model parameters with bootstrap method via resolution of global optimization problem from GSCA.The main function,gscaLCA,works for both binary and ordered categorical variables.In addition,gscaLCA can be used for latent class regression as well.Visualization of profiles of latent classes based on the posterior probabilities is also available in the package gscaLCA.This paper contributes to providing a methodological tool,gscaLCA that applied researchers such as social scientists and medical researchers can apply clustering analysis in their research.

STRUCTURE OF MAIN COMPONENT IN PERFUME "ISO-E-SUPER"

Addition of myrcene and 3-methyl-3-penten-2-one followed by cyclization afforded perfume 'Iso-E-Super' with 2(?)-acetyl-2β,3β,8,8-tetramethyl-1,2,3,4,5,6,7,8-octalin as main component on the basis of its spectroscopic data.

Arrangement for Capital Structure in the Framework of Corporate Governance

Using the method of principal component analysis, the paper conducts a systematic study on the issue of how corporate governance influences capital structure. The study manifests the results that the proportion of circulation shares, the ability that other big shareholders contend with the first biggest shareholder, the proportion of corporation-owned shares, and the frequency of directorate meetings all have a positive relationship with the liability level. Meanwhile, the concentration degree of owners＇ equity, the proportion of state-owned shares, the phenomenon that one person serves as both chairman of directorate and general manager, and the intensity of competition in product market are all negatively related to the level of debt. Finally, the scale of directorate, the proportion of independent directors as well as the percentage of management-owned shares have no significant relationship with the capital structure. The statistic analysis also shows that the proportion of independent directors of some Chinese listed companies does not meet the regulation of the CSRC. In addition, the paper tests the impacts of corporate operating characteristics on capital structure.

Structural elucidation of sex pheromone components of the Geometridae Semiothisa cinerearia(Bremer et Grey)in China

An extract from the female sex gland of Semiothisa cinerearia attracted conspecific males in field tests.A major active component was isolated from the extract and identified by GC-MS,GC-IR and microchemical reactions as cis-3,4-cpoxy-(Z,Z)-6,9-heptadecadiene,which showed strong EAG response.Another minor yet important component was identified as (Z,Z,Z)-3,6,9-heptadecatriene.

Fiducial generalized p-values for testing zero-variance components in linear mixed-effects models

Linear mixed-effects models are widely used in analysis of longitudinal data. However, testing for zero-variance components of random effects has not been well-resolved in statistical literature, although some likelihood-based procedures have been proposed and studied. In this article, we propose a generalized p-value based method in coupling with fiducial inference to tackle this problem. The proposed method is also applied to test linearity of the nonparametric functions in additive models. We provide theoretical justifications and develop an implementation algorithm for the proposed method. We evaluate its finite-sample performance and compare it with that of the restricted likelihood ratio test via simulation experiments. We illustrate the proposed approach using an application from a nutritional study.

SIMULATION OF THE INTERFACED STRUCTURAL COMPONENT BASED ON A MULTIPLE-TIME-SCALE ALGORITHM

A multiple-time-scale algorithm is developed to numerically simulate certain structural components in civil structures where local defects inevitably exist. Spatially, the size of local defects is relatively small compared to the structural scale. Different length scales should be adopted considering the efficiency and computational cost. In the principle of physics, different length scales are stipulated to correspond to different time scales. This concept lays the foundation of the framework for this multiple-time-scale algorithm. A multiple-time-scale algorithm, which involves different time steps for different regions, while enforcing the compatibility of displacement, force and stress fields across the interface, is proposed. Furthermore, a defected beam component is studied as a numerical sample. The structural component is divided into two regions： a coarse one and a fine one; a micro-defect exists in the fine region and the finite element sizes of the two regions are diametrically different. Correspondingly, two different time steps are adopted. With dynamic load applied to the beam, stress and displacement distribution of the defected beam is investigated from the global and local perspectives. The numerical sample reflects that the proposed algorithm is physically rational and computationally efficient in the potential damage simulation of civil structures.

A plating method for metal coating of fiber Bragg grating

We present a method for metal coating optical fiber and in-fiber Bragg grating. The technology process which is based on electroless plating and electroplating method is described in detail. The fiber is firstly coated with a thin copper or nickel plate with electroless plating method. Then, a thicker nickel plate is coated on the surface of the conductive layer. Under the optimum conditions, the surfaces of chemical plating and electroplating coatings are all smooth and compact. There is no visible defect found in the cross-section. Using this two-step metallization method, the in-fiber Bragg grating can be well protected and its thermal sensitivity can be enhanced. After the metallization process, the fiber sensor is successfully embedded in the 42CrMo steel by brazing method. Thus a smart metal structure is achieved. The embedding results show that the plating method for metallization protection of in-fiber Bragg grating is effective.

Continuous-wave and Q-switched performance of an Yb:YAG/YAG composite thin disk ceramic laser pumped with 970-nm laser diode

Using front face-pumped compact active mirror laser （CAMIL） structure, we have demonstrated an Yb：YAG/YAG composite ceramic disk laser with pumping wavelength at 970 nm. The laser has been operated in both continuous-wave （CW） and Q-switching modes. Under CW operation, laser output power of 1.05 W with 2% transmission output coupler was achieved at the wavelength of 1031 nm. Q- switched laser output was gotten by using an acousto-optic Q-switch. The repetition rate ranged from 1 to 30 kHz and the pulse width varied from 166 to 700 ns.

X-ray photon-counting detector based on a micro-channel plate for pulsar navigation

The pulse time of arrival （TOA） is a determining parameter for accurate timing and positioning in X-ray pulsar navigation. The pulse TOA can be calculated by comparing the measured arrival time with the predicted arrival time of the X-ray pulse for pulsar. In this study, in order to research the measurement of pulse arrival time, an experimental system is set up. The experimental system comprises a simulator of the X-ray pulsar, an X-ray detector, a time-measurement system, and a data-processing system. An X-ray detector base is proposed on the basis of the micro-channel plate （MCP）, which is sensitive to soft X-ray in the 1–10 keV band. The MCP-based detector, the structure and principle of the experimental system, and results of the pulse profile are described in detail. In addition, a discussion of the effects of different X-ray pulse periods and the quantum efficiency of the detector on pulse-profile signal-to-noise ratio （SNR） is presented. Experimental results reveal that the SNR of the measured pulse profile becomes enhanced as the quantum efficiency of the detector increases. The SNR of the pulse profile is higher when the period of the pulse is smaller at the same integral.