基于深度负相关学习的室内WiFi指纹定位技术

本文提出了一种新的基于深度负相关学习的WiFi室内定位模型.通过负相关学习约束,多个学习器能学习到不同的表征特性,从而有效降低模型的过拟合,并极大地提升其泛化能力.同时该模型将负相关学习方法应用到去噪自编码器和回归预测器上,并利用深度学习方法使其很好地适应随环境和时间变化的RSSI信号,提高了在长时间间隔内的定位性能.利用负相关学习方法使定位模型在初始时候的平均定位误差从1.57 m下降为0.77 m,60 d平均定位误差也仅为0.89 m,误差仅仅只增加0.12 m,验证了负相关学习能够削弱环境变化对定位的影响.     

天然橡胶生物合成分子调控机制的研究进展

天然橡胶(NR)生物合成的精确调控机制解析是尚未攻克的重要理论问题。综述产胶植物基因组和NR生物合成过程相关基因的调控机制,重点阐述小橡胶粒子蛋白质基因在产胶植物的橡胶合成过程中作用的研究进展,为NR生物合成分子机制以及分子育种研究提供理论参考。指出NR生物合成调控机制的研究与收获技术和提取工艺技术相结合,有助于解析产胶植物的NR合成机制和提高其产胶量。     

Ultra-small PbSe Quantum Dots Synthesis by Chemical Nucleation Controlling

Ultra-small PbSe quantum dots (QDs) were synthesized using conventional hot-injection method.A small amount of Sn was used as a nucleation promotion agent to control nucleation and growth during the QDs synthesis process.The average diameter of the QDs is about 1.6 nm,of which absorption peak centers at 550 nm and photoluminescence peak centers at 750 nm under 350 nm laser excitation with power as low as 500 μW.Transmission electron microscopy images confirm that the QDs size well matches with t...     

加快煤矿智能化建设推进煤炭行业高质量发展

引入图像语义分割技术,对矿井次光照环境中的目标物进行分割,将图像分成原始清晰图像和次光照图像两类,采用基于深度学习的图像增强方法对次光照条件下拍摄的图像增强细节后替换,再利用单应变换算法对数据集进行扩充,进而构建矿井巷道图像语义分割标准数据集。提出一种基于自注意力机制的轻量级编码—解码结构网络：以DeepLab V3+编码—解码网络为基础网络,在编码结构中,提取矿井图像深、浅层语义特征信息,将深层语义特征信息经由轻量级自注意力机制模块进行特征激活,而浅层语义特征信息直接送入解码器中,在解码结构中拼接深、浅层语义特征信息,恢复原始图像尺寸,输出分割结果。与传统算法就图像预测进行对比实验,结果表明：在网络复杂度方面,对于3通道512×512像素大小的图像,算法的网络理论计算量FLOPs仅48.80 G,参数量仅11.90 M;在网络分割精度方面,平均交并比76.50%,平均像素精度87.75%,领先其他主流网络;在速度方面,推理一张图像的速度能达到0.032 s,可满足轻量级网络的要求。

     

南海发展第二水厂自控系统优化整合

通过采用新型驱动器及网络通讯协议软件等方法,分别采用Controller Link网络和工业以太网的自控系统对南海发展第二水厂进行整合和优化,实现了全厂自控系统的统一调度和协同工作,提高了突发事件的反应速度。     

Synergistic effects of α/β‐nucleating agents on propylene‐ethylene random copolymer: New clues for the pipe extrusion

Adding β‐nucleating agents (β‐NAs) has been generally considered as an effective approach to induce the formation of β‐crystals in propylene‐ethylene random copolymer (PPR). However, it is still difficult to obtain the expected percentage of β‐crystal under normal processing conditions, due to the temperature control is always a key factor no matter with or without adding β‐NAs. We assumed that simultaneously adding both α‐ and β‐NAs to PPR in rational ratios would facilitate the growth of β‐crystal. The effects of adding various amount of α‐ and/or β‐NAs on the α‐ and β‐crystallization behavior and morphology were investigated. More importantly, the influence of temperature control on inducing β‐crystal in the presence of different NAs was parallel explored by utilizing sample bars prepared via injection molding and sample pipes produced from a typical industrial extrusion line. Different crystallization situations were carefully discussed in response to various nonisothermal crystallization processes in laboratory experiments and in actual pipes extrusion. At last, the mechanical properties of PPR samples were examined to provide new clues for further study of PPR pipes. POLYM. ENG. SCI., 56:1089–1095, 2016. © 2016 Society of Plastics Engineers     

钼碚砂氨浸过程中钾的释放行为与控制研究

低钾钼产品在电子、医疗等领域具有广泛的应用前景,而氨浸对低钾钼产品的生产具有十分重要的影响。为了更好地控制氨浸过程中钾的释放量以减少后续工艺中的降钾压力,通过对钼焙砂氨浸过程中的释钾影响因素进行了单因素模拟实验,分析了钾元素的释放规律,并给出了其动力学模型和控制方法。结果表明,温度、固液比、钼焙砂粒度等对钾元素的释放具有较大的影响,最佳的氨浸条件为：温度30℃、固液比1∶3,钼焙砂粒径范围150～178μm,氨水浓度20%。此外,Ca元素更利于氨浸过程除钾,而Na元素不利于氨浸过程除钾。熟石灰对水洗除钾有一定作用,碳酸钠对水洗除钾的效果并不明显,故在钼焙砂水洗过程中加入熟石灰有利于低钾钼酸铵的生产。钼焙砂氨浸过程前期释钾速度较快,后期速度变慢,得到Avrami方程可以较好地模拟氨浸前期和后期的释钾过程,确定了氨浸过程的动力学方程。     

不锈钢保温杯旋压模具设计

为了降低劳动强度,提升企业综合竞争力,提高不锈钢原材料利用率,增加社会经济效益,以不锈钢保温杯的生产制造为研究对象,通过焊接不锈钢管状坯料,采用旋压模具生产不锈钢保温杯.采用变薄旋压技术生产不锈钢保温杯,能够节省时间、材料和成本,保温杯零件精度高,综合性能好.生产实践证明,设计的旋压模具结构合理,实用性、经济性强.     

Formation mechanism and microstructure evolution of Ba2Ti9O20 ceramics by reaction sintering method

Ba₂Ti₉O₂₀ single-phase ceramics were prepared by reaction sintering method using TiO₂ and BaCO₃ as raw materials after heat treating at 1150°C for 10 hours. Furthermore, the formation mechanism and microstructure evolution of Ba₂Ti₉O₂₀ ceramics prepared by reaction sintering method were investigated. The formation behavior of Ba₂Ti₉O₂₀ phase was analyzed from the perspective of diffusion, where the reaction activation energy required for the process was calculated to be about 386.17 kJ/mol. Combined with the scanning electron microscopy and the energy dispersive spectrometer, it was revealed that the pores on Ba₂Ti₉O₂₀ grains in the process of reaction sintering might be caused by the absence of oxygen element. Meanwhile, the reason for the roughness of ceramic surface was that the local inhomogeneous distribution of barium on the surface of Ba₂Ti₉O₂₀ grain leaded to the enrichment of titanium.