New insights into the thermodynamics and kinetics of triacylglycerols crystallization

Understanding the transition from an isotropic liquid state to an organized crystalline state of triacylglycerols has great significance when designing lipid-based materials in general and food products in particular. In the current research, the balance between the demand for the tuning-fork (Tf) conformation, which is the suitable conformation for most crystalline organizations, and the supply of these molecules was discussed. Based on the negative dependency of the nucleation rates and the induction times on temperature, it was suggested that the limiting factor for the nucleation process is the conversion from the Tf conformation to any other conformation. On the other hand, based on the positive dependency of the melting temperatures and enthalpies on crystallization temperature, and the organized morphologies obtained at higher temperatures, it was suggested that the limiting factor during the crystal growth process is the conversion to the Tf conformation from another conformation.     

基于拼接缝自适应消除和全景图矫直的快速图像拼接算法

针对图像拼接存在色差过渡不均匀、图像倾斜扭曲及拼接效率低的现象，提出一种基于拼接缝自适应消除和全景图矫直的快速图像拼接算法。首先，用尺度不变特征变换（SIFT）提取图像指定区域特征点并用双向K近邻（KNN）算法进行图像配准，有效提高算法效率；其次，利用动态规划思想提出自适应公式找到最优拼接缝并用图像融合算法对其自适应消除，解决拼接缝色差过渡不均匀问题；最后，针对累积拼接误差形成全景图倾斜的现象，利用边缘检测算法提出自适应拟合四边形矫直模型，把原始全景图矫直为一个全新的全景图。所提算法与分块图像拼接和二叉树图像拼接算法相比，图像质量提升了5.84%~7.83%，拼接时间仅为原来的50%~70%。实验结果表明，该算法不仅通过自适应更新机制减少不同图像背景下拼接缝色差过渡不均匀的现象，从而提高了图像质量；而且提高了拼接效率，降低了全景图倾斜扭曲程度。     

基于时频域的卷积神经网络运动想象脑电信号识别方法

针对目前运动想象脑电（EEG）信号识别率较低的问题，考虑到脑电信号蕴含着丰富的时频信息，提出一种基于时频域的卷积神经网络（CNN）运动想象脑电信号识别方法。首先，利用短时傅里叶变换（STFT）对脑电信号的相关频带进行预处理，并将多个电极的时频图组合构造出一种二维时频图；然后，针对二维时频图的时频特性，通过一维卷积的方法设计了一种新颖的CNN结构；最后，通过支持向量机（SVM）对CNN提取的特征进行分类。基于BCI数据集的实验结果表明，所提方法的平均识别率为86.5%，优于其他传统运动想象脑电信号识别方法；同时将该方法应用在智能轮椅上，验证了其有效性。     

Detection and control of the morphology of TIG-metal fused coating additive manufacturing

Journal of Mechanical Science and Technology - Metal fused-coating additive manufacturing composite variable polarity tungsten inert gas (MFCAM & TIG) has the advantages of low cost and...     

Evaluation model of green supply chain cooperation credit based on BP neural network

Neural Computing and Applications - More and more enterprises hope to achieve cooperation and win–win. However, many companies often have problems such as insufficient partner credit, which...     

High carrier mobility and remarkable photovoltaic performance of two-dimensional Ruddlesden–Popper organic–inorganic metal halides (PA)2(MA)2M3I10 for perovskite solar cell applications

Two-dimensional Ruddlesden–Popper (2DRP) metal halides have attracted extensive attention in photovoltaic applications due to their high stability, low self-doping levels and long-lived free carriers. Among them, (PA)₂(MA)₂Pb₃I₁₀ presents itself as a superior candidate, demonstrating greater moisture resistance and improved heat and light stability over many other 2DRP metal halides. This study takes on the opportunity to search for lead-free alternatives by investigating the optoelectronic and carrier transport properties, as well as the photovoltaic performance of such (PA)₂(MA)₂M₃I₁₀ type metal halides as the photovoltaic absorber, where M = Pb, Cd, Cr, Cu, Ge, Mn, Ni, Sn, Yb, Zn. Our results indicate that the bandgap of (PA)₂(MA)₂M₃I₁₀ can be tuned to the optimum photovoltaic application range of 0.9–1.6 eV, along with improved optical and enhanced photo-response capacity, when Sn, Cd, Mn, Ge, and Zn are used to replace Pb. In particular, (PA)₂(MA)₂Zn₃I₁₀ possesses the largest Stokes shift and Huang-Rhys factor, while showing the best photoluminescence tendency and broadest emission nature. (PA)₂(MA)₂Ge₃I₁₀ displays the most excellent of carrier transport capacities with high mobilities of 73 cm² V⁻¹ s⁻¹ and 43 cm² V⁻¹ s⁻¹ for electron and hole carriers, respectively, which are even comparable to that of 3D counterparts. Furthermore, (PA)₂(MA)₂Zn₃I₁₀ is predicted to have the highest power conversion efficiency of 23.36% based on an empirical energy loss (0.5 eV), which is quite close to the Shockley–Queisser limit, thereby featuring it as a suitable absorber for photovoltaic applications. These findings shed light on new strategies for designing and developing lead-free 2DRP metal halides targeted at future applications in photovoltaic solar cell devices.     

A time-aware searchable encryption scheme for EHRs

Despite the benefits of EHRs (Electronic Health Records), there is a growing concern over the risks of privacy exposure associated with the technologies of EHR storing and transmission. To deal with this problem, a time-aware searchable encryption with designated server is proposed in this paper. It is based on Boneh's public key encryption with keyword search and Rivest's timed-release cryptology. Our construction has three features: the user cannot issue a keyword search query successfully unless the search falls into the specific time range; only the authorized user can generate a valid trapdoor; only the designated server can execute the search. Applying our scheme in a multi-user environment, the number of the keyword ciphertexts would not increase linearly with the number of the authorized users. The security and performance analysis shows that our proposed scheme is securer and more efficient than the existing similar schemes.     

MIMO-GFDM系统中低复杂度动态禁忌搜索检测算法的改进

针对多输入多输出的广义频分复用（MIMO-GFDM）系统的等效信道矩阵维度极大，传统的MIMO检测算法复杂度高且性能不佳的问题，将大规模MIMO系统中的动态禁忌搜索（RTS）检测算法运用到MIMO-GFDM系统中，并解决了RTS算法初始值的求解复杂度高的问题。首先利用最小均方误差（MMSE）检测算法所用到矩阵的正定对称性将矩阵Cholesky分解，并结合Sherman-Morrison公式迭代计算初始值，降低了初始值求逆的高复杂度；然后以改进的MMSE检测结果作为RTS算法的初始值，从初始值逐步全局搜索最优解；最后通过仿真，对不同算法的迭代次数和误码率（BER）性能进行了研究。理论分析与仿真结果表明：在MIMO-GFDM中，所提改进RTS信号检测算法误码率远低于传统信号检测算法。在4QAM时，RTS算法大约有低于MMSE检测6 dB的信噪比性能增益（误码率在10^-3时）；在16QAM时，RTS算法大约有低于MMSE检测4 dB的信噪比性能增益（误码率在10^-2时）。与传统RTS算法性相比，所提改进算法在不影响误码率性能的同时降低了算法复杂度。     

基于特征融合的中文简历解析方法研究

针对基于规则和统计的传统中文简历解析方法效率低、成本高、泛化能力差的缺点，提出一种基于特征融合的中文简历解析方法，即级联Word2Vec生成的词向量和用BLSTM（Bidirectional Long Short-Term Memory）建模字序列生成的词向量，然后再结合BLSTM和CRF（Conditional Random Fields）对中文简历进行解析（BLSTM-CRF）。为了提高中文简历解析的效率，级联包含字序列信息的词向量和用Word2Vec生成的词向量，融合成一个新的词向量表示；再由BLSTM强大的学习能力融合词的上下文信息，输出所有可能标签序列的分值给CRF层；再由CRF引入标签之间约束关系求解最优序列。利用梯度下降算法训练神经网络，使用预先训练的词向量和Dropout优化神经网络，最终完成对中文简历的解析工作。实验结果表明，所提的特征融合方法优于传统的简历解析方法。