A Smart Deep Convolutional Neural Network for Real-Time Surface Inspection

A proper detection and classification of defects in steel sheets in real time have become a requirement for manufacturing these products, largely used in many industrial sectors. However, computers used in the production line of small to medium size companies, in general, lack performance to attend real-time inspection with high processing demands. In this paper, a smart deep convolutional neural network for using in real-time surface inspection of steel rolling sheets is proposed. The architecture is based on the state-of-the-art SqueezeNet approach, which was originally developed for usage with autonomous vehicles. The main features of the proposed model are: small size and low computational burden. The model is 10 to 20 times smaller when compared to other networks designed for the same task, and more than 700 times smaller than general networks. Also, the number of floating-point operations for a prediction is about 50 times lower than the ones used for similar tasks. Despite its small size, the proposed model achieved near-perfect accuracy on a public dataset of 1800 images of six types of steel rolling defects.     

基于BIM超高层密集布筋区埋入式型钢混凝土柱脚施工技术

以漳州市某超高层项目为例,针对超高层密集布筋区埋入式型钢柱脚与钢筋的深化设计、地脚螺栓的定位预埋、型钢柱脚支撑体系设计、节点钢筋穿插连接施工、埋入式型钢柱脚安装等方面,提出切实可行的施工方法,解决了埋入式型钢混凝土柱脚的地脚螺栓精准预埋、与底板密集钢筋穿插连接施工、悬空支撑稳定等一系列施工技术难题,既满足了设计方案的要求,又可为类似工程提供借鉴。     

Prediction models for multiphase-flow-induced corrosion of API X80 steel in CO2/H2S environment

In the offshore oil and gas industry, mainly focusing on the use of rigid or flexible pipes of subsea infrastructure applied to risers or flowlines, one of the greatest difficulties is the interpretation of the combined effects of the various correlated phenomena (hydrodynamic effects of intermittent flow, the effects of corrosivity of the environment in addition to variations in pressure, temperature, and dynamic loading). On the basis of this scenario, defining the degree of severity of each of the correlated system variables becomes of fundamental importance for establishing reliable criteria for selecting materials for subsea application. The established flow pattern directly affects the corrosion rate (or the pipe material mass loss), but the balance of other variables including possible changes in the physical and transported fluid chemical properties may increase the damage up to an order of magnitude, which is a piece of information normally not foreseen in design criteria. Therefore, to improve the understanding of the corrosion study influenced by multiphase flow, a testing loop was designed and assembled at the Corrosion and Protection Laboratory of the Institute for Technological Research, in which API X80 steel coupons were positioned in locations with a 0° and 45° inclinations. Tests were conducted by varying the partial pressure of the gaseous phase containing blends of CO₂ and H₂S with N₂ balance, mixed with the liquid phase containing light oil and heavy oil in water with salinity (NaCl)-simulating oil well conditions with 80% water cut. The main objective of this study is to establish models that can predict the corrosion intensity in conditions close to those obtained experimentally. To achieve results, the multiple regression and Box–Cox transformation methods were applied. These models will make possible damage prediction and optimization of matrix parameters for the multiphase-loop test.     

Studies on waterline corrosion processes and corrosion product characteristics of carbon steel in 3.5 wt% NaCl solution

The waterline corrosion behaviors of carbon steel partially immersed in a 3.5 wt% NaCl solution were investigated using the wire beam electrode technique, and the effects of corrosion products on the processes of waterline corrosion were analyzed. The results demonstrated that the initial stage and development stage of waterline corrosion were mainly controlled by the concentration and diffusion of dissolved oxygen, respectively, and the deceleration stage of waterline corrosion was mainly affected by corrosion products. The main component of the yellow corrosion products was γ-FeOOH, and γ-FeOOH that exhibited a high reduction reactivity could be involved in the cathodic reaction. The black corrosion products were mainly composed of Fe₃O₄ with strong thermodynamic stability and the processes of dissolved oxygen diffusion and ion transports were obviously affected due to the continuous accumulation of Fe₃O₄ on the surface of the electrodes. Polarity reversals were observed on the single electrodes below the waterline, but the reasons for the phenomena were different from each other.     

Experimental study on prediction model of wet gas pressure drop across single-orifice plate in horizontal pipes in the low gas phase Froude number region

The pressure drop prediction of wet gas across single-orifice plate in horizontal pipes had been solved satisfactorily under an annular-mist flow in the upstream of orifice plates. However, this pressure drop prediction is still not clearly determined when the upstream is in an intermittent flow or stratified flow, which is corresponding to a region of low Fr_G (gas phase Froude number) in the flow pattern map of wet gases. In this study, the wet gas pressure drop across a single-orifice plate was experimentally investigated in the low Fr_G region. By the experiment, the flow pattern transition in the downstream of single-orifice plates, as well as the effects of Fr_G and Fr_L (liquid phase Froude number) on Φ_G (gas phase multiplier), were determined and compared when the upstream is in the flow pattern transition and the stratified flow region, respectively. Prediction performances were examined on the available pressure drop models. It was found that no model could be capable of jointly predicting the wet gas pressure drop in the low Fr_G region with an acceptable accuracy. With a new method of correlating Fr_G and Fr_L simultaneously, new correlations were proposed for the low Fr_G region. Among which the modified Chisholm model shows the best prediction accuracies, with the prediction deviations of Φ_G being within 7% and 3% when the upstream is in flow pattern transition and stratified flow region, respectively.     

基于改进Faster R-CNN及YOLO的大角度车牌识别算法

随着大数据时代的到来,智慧城市等一系列概念的提出,人工智能开始在城市各个角落得到广泛应用。其中,车牌识别作为城市道路交通重要的一环,也取得了重大突破。基于深度学习的车牌识别一经提出,各类算法应用于车牌检测及识别上,对比于传统的车牌识别,大大提高了识别的速率和准确率。然而,在非限制条件下,如大角度车牌识别准确率仍有提升的空间。提出了一种改进的FasterR-CNN与YOLO的深度学习车牌识别算法,将对大角度下的车牌识别准确率达到了99.7%。由于是轻量型的框架,为后续的移动设备部署工作提供了便利。     

对大中型钢结构企业生产管理体系的研究分析

以某大型钢结构制造企业为例,通过对产量分布、质量控制、劳动效率等基本数据的分析,研判外协生产管理体系方面存在的问题,从组织形态、管理模式、职能管理等多个维度提出了可行性建议,可为同类型钢结构制造企业提供经验借鉴。     

An Exemplar Imidazoline Surfactant for Corrosion Inhibitor Studies: Synthesis,Characterization, and Physicochemical Properties

An optimized one-pot recipe has been developed to synthesize a surfactant molecule, referred to as OMID, consisting of an imidazoline head group and aliphatic tail, which is an exemplar corrosion inhibitor for carbon steel in acidic solutions. As evidenced by gas chromatography, ¹H and ¹³C nuclear magnetic resonance, and Fourier-transform infrared data, a high-purity product was achieved without the use of either a solvent or catalyst. Critical micelle concentration values and corrosion inhibition efficiencies ( η %) were determined in aqueous solutions of hydrochloric acid and sulfuric acid using surface tensiometry and linear polarization resistance measurements, respectively. Hydrolysis of the imidazoline head group as a function of pH (0–11) was explored with ultraviolet–visible absorption spectroscopy. In addition, N 1s and C 1s X-ray photoelectron spectroscopy data were acquired from both surface-adsorbed OMID and a multilayer of the imidazoline head group of OMID. These latter data are highly relevant to those attempting to understand OMID inhibition chemistry.     

铌对取向硅钢冷轧组织织构的影响

采用光学显微镜(OM)、电子背散射衍射(EBSD)等研究了铌对取向硅钢冷轧过程中的组织与织构的影响。结果表明:含铌正火板和不含铌正火板经冷轧后都是条带状组织,但含铌冷轧板在宽度方向更窄。铌元素的存在对织构也具有优化作用,经过中间退火后,含铌冷轧板二次冷轧织构得到明显改善,相较于不含铌二次冷轧板拥有强度较低的旋转立方织构以及强度较高的{111}<112>织构,其取向密度最高可达12.460。     

氮含量对HPD-1双相不锈钢组织及力学性能的影响

通过显微组织观察与力学性能测试研究了氮含量(0.08%~0.22％,质量分数)对HPD-1双相不锈钢硬度、拉伸性能、低温冲击性能及疲劳性能的影响。结果表明,氮含量变化可显著影响试验钢γ/α相比例,当氮含量由0.08％升高到0.22％,γ相含量由38.1％提高至56.5％。α相的硬度由273 HV10提高到343 HV10,γ相的硬度由238 HV10提高到299 HV10,试验钢强度明显提升。氮元素对两相比例和奥氏体相韧性的双重影响导致试验钢低温冲击性能呈先上升后下降的趋势;更高的氮含量抑制疲劳裂纹萌生与拓展,是影响HPD-1双相不锈钢室温疲劳性能的主要因素。撕裂棱是疲劳断口的显著特征。