Failure load prediction and optimisation for adhesively bonded joints enabled by deep learning and fruit fly optimisation

Adhesively bonded joints have been extensively employed in the aeronautical and automotive industries to join thin-layer materials for developing lightweight components. To strengthen the structural integrity of joints, it is critical to estimate and improve joint failure loads effectually. To accomplish the aforementioned purpose, this paper presents a novel deep neural network (DNN) model-enabled approach, and a single lap joint (SLJ) design is used to support research development and validation. The approach is innovative in the following aspects: (i) the DNN model is reinforced with a transfer learning (TL) mechanism to realise an adaptive prediction on a new SLJ design, and the requirement to re-create new training samples and re-train the DNN model from scratch for the design can be alleviated; (ii) a fruit fly optimisation (FFO) algorithm featured with the parallel computing capability is incorporated into the approach to efficiently optimise joint parameters based on joint failure load predictions. Case studies were developed to validate the effectiveness of the approach. Experimental results demonstrate that, with this approach, the number of datasets and the computational time required to re-train the DNN model for a new SLJ design were significantly reduced by 92.00% and 99.57% respectively, and the joint failure load was substantially increased by 9.96%.     

Specification uncertainty in modeling internet adoption: A developing city case analysis

Internet adoption fosters economic growth and development. Specifying policy control drivers is particularly relevant for developing countries. However, there is no consensus on the most relevant variables. We explored 33.6 million potential models to identify the most important determinants of household internet adoption using stochastic search variable selection and socioeconomic data from Medellín, Colombia. We found that monthly income, the head of household education and voting, and having a computer and cable television at home are the most relevant variables.     

Enriching Context Information for Entity Linking with Web Data

Journal of Computer Science and Technology - Entity linking (EL) is the task of determining the identity of textual entity mentions given a predefined knowledge base (KB). Plenty of existing...     

Microstructure evolution and cooperative reinforcement mechanisms of Al2O3/Al2O3 joints brazed by low-melting borosilicate glass

The Al₂O₃/SiO₂–B₂O₃–Al₂O₃–Na₂O glass/Al₂O₃ joints reinforced cooperatively by glass matrix and in-situ Al₄B₂O₉ whiskers were obtained via a low-melting borosilicate glass braze. The composition of glass seam transformed from SiO₂–B₂O₃–Na₂O to SiO₂–B₂O₃–Al₂O₃–Na₂O due to continuous diffusion and dissolution of Al₂O₃. An appropriate amount of [AlO₄] units introduced into the glass braze played a vital role in strengthening the glass network structure resulting to considerably improved mechanical strength of the glass seam. Meanwhile, plenty of in-situ Al₄B₂O₉ whiskers growing from the Al₂O₃/glass braze interface to the center of glass seam in various directions generated. Three-dimensional crisscross structures were fabricated at the Al₂O₃/glass braze interface domains, where were enhanced by crack-bridging and pull-out effect of the whiskers. Generally, ascribed to the cooperative reinforcement of the glass matrix in the seam and in-situ Al₄B₂O₉ whiskers at Al₂O₃/glass braze interface domains through reactions of Al₂O₃ and borosilicate glass braze, strength of the as-brazed joints was promoted prominently. The shear strength of the joints reached a maximum of 61 MPa brazed at 1050 °C for 60 min.     

遗传算法和模矢法结合的概率积分法参数反演方法

遗传算法在反演概率积分法预计参数时从参数取值范围内的串集开始搜索,并使用弹性策略来维持群体的多样性,使得算法可以跨过局部收敛的障碍,向全局最优解方向进化;但这种概率化的寻优算法存在局部探索能力差、结果不稳定的缺陷,只能获得问题的近似最优解。模矢法是一种降梯度算法,算法局部探索能力强、收敛快;但这种算法对初值选取敏感,初值选择不当易陷入局部极值。本文提出并实现了一种模矢法与遗传算法结合的组合算法:先使用遗传算法求得参数的全局近似最优解,然后将近似最优解作为探索初值,使用模矢法获得参数的稳定、精确最优解。研究结果表明:组合算法反演概率积分法预计参数的精确度高、收敛快、稳定性好,综合性能较遗传算法和模矢法有明显优势。     

Data-efficient learning of robotic clothing assistance using Bayesian Gaussian process latent variable model

ABSTRACT

Motor-skill learning for complex robotic tasks is a challenging problem due to the high task variability. Robotic clothing assistance is one such challenging problem that can greatly improve the quality-of-life for the elderly and disabled. In this study, we propose a data-efficient representation to encode task-specific motor-skills of the robot using Bayesian nonparametric latent variable models. The effectivity of the proposed motor-skill representation is demonstrated in two ways: (1) through a real-time controller that can be used as a tool for learning from demonstration to impart novel skills to the robot and (2) by demonstrating that policy search reinforcement learning in such a task-specific latent space outperforms learning in the high-dimensional joint configuration space of the robot. We implement our proposed framework in a practical setting with a dual-arm robot performing clothing assistance tasks.     

Novel dynamic semiempirical proton exchange membrane fuel cell model incorporating component voltages

This paper introduces a novel dynamic semiempirical model for the proton exchange membrane fuel cell (PEMFC). The proposed model not only considers the stack output voltage but also provides valid waveforms of component voltages, such as the no‐load, activation, ohmic, and concentration voltages of the PEMFC stack system. Experiments under no‐load, ramping load, and dynamic load conditions are performed to obtain various voltage components. According to experimental results, model parameters are optimised using the lightning search algorithm by providing valid theoretical ranges of parameters to the lightning search algorithm code. In addition, the correlation between the vapour and water pressures of the PEMFC is obtained to model the component voltages. Finally, all component voltages and the stack output voltage are validated by using the experimental/theoretical waveforms mentioned in previous research. The proposed model is also compared with a recently developed semiempirical model of PEMFC through particle swarm optimisation. The proposed dynamic model may be used in future in‐depth studies on PEMFC behaviour and in dynamic applications for health monitoring and fault diagnosis.     

Conditional Preference Networks with User's Genuine Decisions

User's choices involve habitual behavior and genuine decision. Habitual behavior is often expressed using preferences. In a multiattribute case, the Conditional Preference Network (CP-net) is a graphical model to represent user's conditional ceteris paribus (all else being equal) preference statements. Indeed, the CP-net induces a strict partial order over the outcomes. By contrast, we argue that genuine decisions are environmentally influenced and introduce the notion of “comfort” to represent this type of choices. In this article, we propose an extension of the CP-net model that we call the CP-net with Comfort (CPC-net) to represent a user's comfort with preferences. Given that preference and comfort might be two conflicting objectives, we define the Pareto optimality of outcomes when achieving outcome optimization with respect to a given CPC-net. Then, we propose a backtrack search algorithm to find the Pareto optimal outcomes. On the other hand, two outcomes can stand in one of six possible relations with respect to a CPC-net. The exact relation can be obtained by performing dominance testing in the corresponding CP-net and comparing the numeric comforts.     

A Novel Beam Search to Improve Neural Machine Translation for English-Chinese

Neural Machine Translation (NMT) is an end-to-end learning approach for automated translation, overcoming the weaknesses of conventional phrase-based translation systems. Although NMT based systems have gained their popularity in commercial translation applications, there is still plenty of room for improvement. Being the most popular search algorithm in NMT, beam search is vital to the translation result. However, traditional beam search can produce duplicate or missing translation due to its target sequence selection strategy. Aiming to alleviate this problem, this paper proposed neural machine translation improvements based on a novel beam search evaluation function. And we use reinforcement learning to train a translation evaluation system to select better candidate words for generating translations. In the experiments, we conducted extensive experiments to evaluate our methods. CASIA corpus and the 1,000,000 pairs of bilingual corpora of NiuTrans are used in our experiments. The experiment results prove that the proposed methods can effectively improve the English to Chinese translation quality.     

基于深度EPG的智能导航平台建设

随着三网融合的大力推进以及双向互动业务的大力开展,运营商使用原有简单的EPG不足以支撑开展更多的增值业务,不能主动为用户推荐节目,不能为运营商带来新的收入机会。作为重要的创新能力支撑系统,介绍了江苏有线智能导航平台的设计与应用,为相关广电运营商提供一定的设计参考。