Digital twin modeling method based on biomimicry for machining aerospace components

High-performance aerospace component manufacturing requires stringent in-process geometrical and performance-based quality control. Real-time observation, understanding and control of machining processes are integral to optimizing the machining strategies of aerospace component manufacturing. Digital Twin can be used to model, monitor and control the machining process by fusing multi-dimensional in-context machining process data, such as changes in geometry, material properties and machining parameters. However, there is a lack of systematic and efficient Digital Twin modeling method that can adaptively develop high-fidelity multi-scale and multi-dimensional Digital Twins of machining processes. Aiming at addressing this challenge, we proposed a Digital Twin modeling method based on biomimicry principles that can adaptively construct a multi-physics digital twin of the machining process. With this approach, we developed multiple Digital Twin sub-models, e.g., geometry model, behavior model and process model. These Digital Twin sub-models can interact with each other and compose an integrated true representation of the physical machining process. To demonstrate the effectiveness of the proposed biomimicry-based Digital Twin modeling method, we tested the method in monitoring and controlling the machining process of an air rudder.     

Mathematical model and discrete artificial Bee Colony algorithm for distributed integrated process planning and scheduling

With the development of the globalization of economy and manufacturing industry, distributed manufacturing mode has become a hot topic in current production research. In the context of distributed manufacturing, one job has different process routes in different workshops because of heterogeneous manufacturing resources and manufacturing environments in each factory. Considering the heterogeneous process planning problems and shop scheduling problems simultaneously can take advantage of the characteristics of distributed factories to finish the processing task well. Thus, a novel network-based mixed-integer linear programming (MILP) model is established for distributed integrated process planning and scheduling problem (DIPPS). The paper designs a new encoding method based on the process network and its OR-nodes, and then proposes a discrete artificial bee colony algorithm (DABC) to solve the DIPPS problem. The proposed DABC can guarantee the feasibility of individuals via specially-designed mapping and switching operations, so that the process precedence constraints contained by the network graph can be satisfied in the entire procedure of the DABC algorithm. Finally, the proposed MILP model is verified and the proposed DABC is tested through some open benchmarks. By comparing with other powerful reported algorithms and obtaining new better solutions, the experiment results prove the effectiveness of the proposed model and DABC algorithm successfully.     

Re-optimization of the binary Sb − Se system aided by ab-initio calculations

Based on the ab-initio calculations the thermodynamics of the Sb₂Se₃ intermediate phase was modeled and used in the calculations of the Sb–Se phase diagram together with the thermodynamic properties of liquid phase elaborated by the association model. The modeled heat capacity of Sb₂Se₃ phase shows excellent agreement with the experiment data available in the literature in the wide temperature range. The calculated equilibrium lines of the Sb–Se phase diagram good correlate with the experimental ones. The determined thermodynamic parameters can be applied in future calculations of the high-ordered systems and for determination of the Sb₂Se₃ thermoelectric properties.     

Third-generation CALPHAD description of pure GeO2 at 1 atm

The thermodynamic properties of crystalline and liquid/amorphous phases of germanium dioxide have been evaluated using DFT calculations and third-generation CALPHAD models. The values for the standard entropies of crystalline and amorphous phases were reassessed through use of the Planck-Einstein approach. This allowed a more physically-based representation of the set of thermodynamic properties of germanium dioxide to be obtained.     

Customer entrepreneurship on digital platforms: Challenges and solutions for platform business models

Drawing on the mixed methods of qualitative research and agent‐based simulation, this study examines: (a) how end‐users use digital platforms to become customer–entrepreneurs undertaking commercial activities on platforms; and (b) how platform providers can convert this customer entrepreneurship into a revenue stream. Considering that end‐users have traditionally been defined as passive and uncharged actors in platform business models, an in‐depth understanding of their commercial activities and the viable revenue model to monetize this emerging customer practice is warranted. Our qualitative study reveals that customer–entrepreneurs make substantial use of platform offerings to advertise their products; communicate with end‐consumers; and accept payments. These commercial activities are largely exercised for free on platforms, even though they could otherwise serve as a source of revenue. On this point, our simulation results identify two pricing models achieving the generation of nearly identical revenues over time. First, platform providers may charge both advertising and transaction fees, which maximize the survival of professional customer–entrepreneurs. Second, platform businesses may levy advertising fees only, which maximizes the survival of informal customer–entrepreneurs operating on a micro‐scale and part‐time basis. This study offers theoretical, methodological, and managerial implications for platform studies.     

基于YOLOv3的安全帽佩戴检测方法研究

在大型锅炉、钢铁和其他制造业工厂中,正确佩戴安全帽是进入现场的前提。但是在现实场景中,由于各种原因,工人伤亡事故中没有正确佩戴安全帽而导致的事故占比一直较高。为了减少该类事故的发生,对深度学习、人工智能展开了研究,采用了卷积神经网络的方法。首先,对相关数据集标定,将数据集总体分为三种类别,针对标定好的数据,使用迁移学习的训练方式训练YOLOv3网络。接着,使用该参数和模型对监控视频数据检测人形和安全帽的类别和位置。最后,对检测的三种类别计算相关交并比,并以此判断工人是否正确佩戴安全帽。该算法创新性地将YOLOv3网络模型和交并比算法结合,对网络输出的类别和位置信息再细化处理,提高了检出的准确性,降低了误识别率。试验结果表明,该算法可以满足安全帽佩戴检测中的实时性要求,同时能够准确检出未佩戴安全帽的人员并通知有关人员。     

融合指针网络的新闻文本摘要模型

本文针对实际党建领域中的新闻标题进行自动生成,提出了一种融合指针网络的自动文本摘要模型-Tri-PCN.相比于传统基于编码器-解码器框架的自动文本摘要模型,党建新闻标题生成模型还需要满足(1)从更长的文本序列提取特征;(2)保留关键的党建信息.针对党建新闻比普通文本摘要任务面临更长文本序列问题,论文使用Transformer模型在解码阶段提取多层次全局文本特征.针对党建新闻标题生成过程中需要保留关键的党建信息,论文引入指针生成网络模型的复制机制在新闻标题生成时可以直接从新闻文本中复制关键词信息.实验采用ROUGE值作为评测指标,结果表明本文提出的Tri-PCN模型在党建新闻领域自动文本摘要任务上效果明显优于基准模型,比其他模型具有更好的效果.     

基于多层局部信息融合的在线论坛用户心理危机识别

心理健康问题已经成为当今社会关注的焦点,它严重威胁着家庭和睦与社会稳定.有心理危机的用户经常通过特定的社区论坛或者社交媒体来求助或倾述,这为用户心理危机识别开辟了一个新的途径.论坛帖子长短不一,但判断心理危机的核心信息往往体现在局部内容上,基于此特点,本文构建了一个结合分层长短记忆网络和卷积神经网络的多层局部信息融合模型(Multi-layer Partial Information Fusion model,MPIF),利用论坛用户发布的帖子,检测用户的心理危机严重程度.模型的特点在于:1)利用预训练语言模型BERT对用户帖子中的句子进行向量化表示,充分考虑词语在不同语境中的不同含义表达;2)分别从词、短语、以及句子层面挖掘反映用户心理危机状态的信息,采用深度分层LSTM网络和注意力机制相结合的方式来获取待分类帖子中词语层面以及句子层面的局部信息,利用CNN网络中多种大小不同的卷积核来提取帖子中短语层面的局部信息;3)采用注意力机制和最大池化层,使得模型不仅能够有效地利用局部信息给出心理危机程度的判断,同时可以将这些局部信息展示给心理专家,辅助专家更快了解患者.基于CLPsych2019 Shared Task评测任务的实验结果显示,与评测时排名第一的模型相比,MPIF模型的官方评测指标All-F1值(自杀风险程度a,b,c,d 4个类别的F1值取平均)高出3.9%.经消融实验发现,去除LSTM词语层、CNN短语层、LSTM句子层,All-F1分别下降4%、4.3%、2.4%.