基于深度学习的文本自动摘要方案

针对自然语言处理（NLP）生成式自动摘要领域的语义理解不充分、摘要语句不通顺和摘要准确度不够高的问题，提出了一种新的生成式自动摘要解决方案，包括一种改进的词向量生成技术和一个生成式自动摘要模型。改进的词向量生成技术以Skip-Gram方法生成的词向量为基础，结合摘要的特点，引入词性、词频和逆文本频率三个词特征，有效地提高了词语的理解；而提出的Bi-MulRnn+生成式自动摘要模型以序列映射（seq2seq）与自编码器结构为基础，引入注意力机制、门控循环单元（GRU）结构、双向循环神经网络（BiRnn）、多层循环神经网络（MultiRnn）和集束搜索，提高了生成式摘要准确性与语句流畅度。基于大规模中文短文本摘要（LCSTS）数据集的实验结果表明，该方案能够有效地解决短文本生成式摘要问题，并在Rouge标准评价体系中表现良好，提高了摘要准确性与语句流畅度。     

Removal of copper(II) from aqueous solution using chemically modified fruit peels as efficient low-cost biosorbents

Fruit peels, which are common agricultural byproducts, have been extensively used as abandoned or low-cost biosorbents to remove heavy metals. In this study, dragon fruit peel (DFP), rambutan peel (RP), and passion fruit peel (PFP) were used to remove Cu(II) ions from an aqueous solution. Concentrations of the adsorbed metal ions were determined using the atomic absorption spectroscopic method. Adsorption experiments were performed with different adsorbent dosages, pH values, contact times, and initial copper concentrations. The optimum set of conditions for biosorption of Cu(II) ions was found to be an adsorbent dosage of 0.25 g, a contact time of 180 min, an initial concentration of 100 mg/L, a pH value of 4 for RP and PFP, and a pH value of 5 for DFP. The adsorption conformed with the pseudo-second-order kinetic model. The adsorption data were consistent with the Langmuir and Freundlich isotherm models, but the best fit was with the Langmuir model. The Langmuir monolayer adsorption capacity values of DFP, RP, and PFP were calculated to be 92.593, 192.308, and 121.951 mg/g, respectively. RP showed a higher adsorption capacity of Cu(II) ions than PFP and DFP for all parameters. The results indicate that these biosorbents might be used to effectively adsorb Cu(II) ions from wastewater treatment plants.     

基于PMU的电力系统监控体系结构优化设计

文中提出了一种利用有限数量的相量测量单元(PMU)和相量数据集中器(PDC)设计最优监控结构的方法。通过在大量的设定值场景下,使电力系统可观测性曲线的期望值最大化,同时使通信基础设施成本最小化,最终确定PMU和PDC的最佳位置。提出了一种非线性动态扩展卡尔曼滤波(EFK)状态观测器。这种状态观测器可以将暂态行为转换为由代数微分方程描述的广域电力系统,而无需非线性反演技术。最后以IEEE-5电力系统为例,说明了该方法的有效性。     

Synthesis of strontium (Sr) doped hydroxyapatite (HAp) nanorods for enhanced adsorption of Cr (VI) ions from wastewater

The development of high-efficiency adsorbents for heavy metal ion removal from wastewater is highly desirable and challenging due to their synthesis complexity and low adsorption capacities. Herein, we reported the synthesis of strontium (Sr) doped hydroxyapatite (HAp) for the increased Cr (VI) adsorption. The effects of pH, temperature, and time on adsorption performances were studied. As a result, the Sr-HAp nanorods can achieve a Cr (VI) adsorption capacity of 443 mg/g, which is significantly higher than that of HAp nanorods (318 mg/g). To better understand the adsorption mechanism, the Langmuir isotherm model was established. The modeling results indicated that Langmuir monolayer chemical adsorption contributed to the efficient Cr (VI) ion removal for Sr-HAp nanorods adsorbents. The surface area and surface functional groups (O–H) contributed to the different Cr (VI) adsorption capacities between HAp and Sr-HAp.     

Property uniformity of thick nickel-based alloy plate for nuclear power steam-generator divider plate

This paper introduces a thick 690 nickel-based alloy plate produced by the former Baosteel Special Steel Co.,Ltd.used as the steam-generator divider plate in the pressurized water reactor nuclear power plant.According to the product characteristics and design requirements of the thick nickel-based alloy plate, multidimensional sampling and testing were conducted to investigate its microstructure and mechanical properties.The results show that all the property indexes of the thick hot-rolled nickel-based alloy plate meet the design requirements, and there is good uniformity in the microstructure and mechanical properties in different dimensions.These findings indicate that China has mastered the core manufacturing technology of thick nickel-based alloy plates for their use as divider plates in nuclear power steam generators.     

Revamping SiO2 Spheres by Core–Shell Porosity Endowment to Construct a Mazelike Nanoreactor for Enhanced Catalysis in CO2 Hydrogenation to Methanol

Beyond the catalytic activity of nanocatalysts, the support with architectural design and explicit boundary could also promote the overall performance through improving the diffusion process, highlighting additional support for the morphology-dependent activity. To delineate this, herein, a novel mazelike-reactor framework, namely multi-voids mesoporous silica sphere (MVmSiO₂), is carved through a top-down approach by endowing core-shell porosity premade Stöber SiO₂ spheres. The precisely-engineered MVmSiO₂ with peripheral one-dimensional pores in the shell and interconnecting compartmented voids in the core region is simulated to prove combined hierarchical and structural superiority over its analogous counterparts. Supported with CuZn-based alloys, mazelike MVmSiO₂ nanoreactor experimentally demonstrated its expected workability in model gas-phase CO₂ hydrogenation reaction where enhanced CO₂ activity, good methanol yield, and more importantly, a prolonged stable performance are realized. While tuning the nanoreactor composition besides morphology optimization could further increase the catalytic performance, it is accentuated that the morphological architecture of support further boosts the reaction performance apart from comprehensive compositional optimization. In addition to the found morphological restraints and size-confinement effects imposed by MVmSiO₂, active sites of catalysts are also investigated by exploring the size difference of the confined CuZn alloy nanoparticles in CO₂ hydrogenation employing both in-situ experimental characterizations and density functional theory calculations.     

Engineering 2D Multifunctional Ultrathin Bismuthene for Multiple Photonic Nanomedicine

2D monoelemental nanomaterials (Xenes) have shown tremendous potential for versatile biomedical applications. Bismuth, as a heavy element in pnictogens, has acquired massive research interest due to its unique optical performance, high biocompatibility, stability, and relatively low cost. However, the utilization of 2D bismuthene in nanomedicine has not been achieved because of the difficulty in engineering bismuthene with crucial structural/compositional characteristics for satisfying strict biomedical requirements. Herein, to address this Gordian knot, a facile strategy to intercalate and delaminate Bi bulk for generating mass few-layered 2D bismuthene with high yield by employing a water molecule mediated freezing–thawing process and sodium borohydride-triggered reduction treatment is proposed. The resulting 2D bismuthene displays good optical performance in the near-infrared (NIR) biowindow and can be excited via red light for reactive oxygen species generation, enabling applications in multiple photonic cancer nanomedicine settings, including photothermal hyperthermia and photodynamic therapy. Utilizing the intrinsic desirable optical absorbance and strong X-ray attenuation of bismuthene, dual photonic therapy can be conducted under the supervision of photoacoustic/computed tomography guided multimodal imaging. This research not only offers a potential mass-production ready, cost-effective, and eco-efficient methodology for engineering 2D Xenes, but also exploits an innovative 2D bismuthene based photonic cancer nanomedicine.     

Heat capacities and standard entropies and enthalpies of some compounds essential for steelmaking and refractory design approximated by Debye-Einstein integrals

Heat capacity data for compounds located in the binary CaO–SiO₂, CaO–Al₂O₃ and MgO–Al₂O₃ systems are fitted by Debye-Einstein integrals. Starting from the fitted heat capacities, the standard values of the thermodynamic functions of these compounds are calculated. In almost all cases investigated, the derived standard entropies are within the uncertainties of the values provided in literature. The Debye-Einstein coefficients obtained in this thermodynamic assessment can be used to approximate the heat capacities, enthalpies and entropies of these compounds in the temperature range from 0 to 298.15 K.     

Chemical Bonding by the Chemical Orthogonal Space of Reactivity

The fashionable Parr–Pearson (PP) atoms-in-molecule/bonding (AIM/AIB) approach for determining the exchanged charge necessary for acquiring an equalized electronegativity within a chemical bond is refined and generalized here by introducing the concepts of chemical power within the chemical orthogonal space (COS) in terms of electronegativity and chemical hardness. Electronegativity and chemical hardness are conceptually orthogonal, since there are opposite tendencies in bonding, i.e., reactivity vs. stability or the HOMO-LUMO middy level vs. the HOMO-LUMO interval (gap). Thus, atoms-in-molecule/bond electronegativity and chemical hardness are provided for in orthogonal space (COS), along with a generalized analytical expression of the exchanged electrons in bonding. Moreover, the present formalism surpasses the earlier Parr–Pearson limitation to the context of hetero-bonding molecules so as to also include the important case of covalent homo-bonding. The connections of the present COS analysis with PP formalism is analytically revealed, while a numerical illustration regarding the patterning and fragmentation of chemical benchmarking bondings is also presented and fundamental open questions are critically discussed.