面向电信承载网的监控系统设计与实现

面对电信承载网连接的日益增长的海量终端设备，运营商需要结合网络拓扑对终端设备产生的数据进行高效的汇聚统计、异常分析、故障定位处理等操作。针对已有系统存在的操作困难、分析效率低等问题，设计与实现了一个面向电信承载网的高效监控系统，提供实时与离线数据分析和多维可视化分析的能力。对网管、认证、终端等系统及设备采集的数据进行结构化存储，对采集的数据进行拓扑相关性和时间序列方法分析，根据分析结果实现基于动态阈值控制的异常实时告警、定位等操作，并提供多维度可视化分析对网络状态进行实时监控。实际应用结果表明，该系统性能优异，具有良好交互性，能较好地满足承载网运维人员业务分析需求。     

Design and validation of robotic bionic eye with multiple flexible ropes parallel mechanism inspired by oculomotor law

Producing a stable and agile bionic eye for visual image acquisition in robotics is a challenging task. In this paper, we design a bionic eye with mirror-symmetric distribution and cross-connection of flexible ropes. This mechanism is based on oculomotor law and the physiological structure of the extraocular muscles (EOMs). Specifically, the basic structural parameters are determined by Listing’s law, and the unique connection of the flexible ropes can realize the functions of the recti and oblique muscles. Furthermore, to mimic the trochlea structure, a pulley mechanism is constructed to permit the free movement of the flexible ropes. Through simulation and physical experiments, it is demonstrated that the bionic eye mechanism can move with agility under the structural parameters. The experimental results indicate that the proposed bionic eye mechanism has a superior motion accuracy of 2.798 mm, which is 6.7% of the maximum motion distance, and the repeatable accuracy of the mechanism can up to 0.210 mm.     

基于时空的弱点状运动目标检测技术

如何在复杂背景下持续有效地检测目标位置,一直是研究者们需要面对的主要挑战。本文在研究红外点状移动目标特征的基础上,根据目标无纹理,无形状的特性,提出一种改进的形态学目标增强算法,并利用目标连续时空不变性检测目标。首先,建立多尺度的图像金字塔,在每层上采用改进的形态学算法快速、粗糙定位小目标。然后进一步的根据目标在时空上的位置相关性,提出基于目标运动特征分析的精确检测方法。得到精确稳定的检测结果。最后实验结果表明与经典的形态学检测算法及其他算法相比,该技术能更有效地检测弱小目标,具有更高的鲁棒性。     

Synergistic effects of modified TiO2/multifunctionalized graphene oxide nanosheets as functional hybrid nanofiller in enhancing the interface compatibility of PLA/starch nanocomposites

The aim of this study is to investigate the synergistic effects of modified TiO₂/multifunctionalized graphene oxide nanosheets at different ratios on the interface compatibility between starch and poly(lactic acid) (PLA). To this end, silanylated nano-TiO₂ (MTiO₂, 1 and 2%) and alkylated maleic anhydride grafted graphene oxide (f-GO, 0.1, 0.2, and 0.4%) at different combinations are blended with the PLA-starch composites using solution blending technique. Then, the synergistic effects of MTiO₂ and f-GO on PLA/starch matrix are investigated in terms of the morphology, crystallinity, structural characterization, thermal stability, dynamic mechanical, and antiaging properties, and the related mechanisms. The Raman and Fourier transform infrared spectroscopy spectra verify the successful synthesis of the two modified nanofillers (f-GO and MTiO₂) and the formation of strong hydrogen bond within the PLA-starch nanocomposites. Due to the strong interfacial interaction and the synergistic effect from the combination of 1% MTiO₂ and 0.2% f-GO, obvious improvement was observed in PLA-starch versus other nanocomposites in terms of morphology, thermal stability, surface hydrophobicity, storage modulus, ultraviolet-shielding capacity, and aging-resistance. Furthermore, differential scanning calorimeter (DSC), isothermal crystallization kinetic, and X-ray diffraction analysis demonstrate that f-GO and the M-TiO₂ significantly synergize in enhancing the crystallization rate and crystallinity of PLA/starch matrix. These results provide novel insights for constructing high-performance nanocomposites and facilitate their applications in food packaging.     

X射线干涉光刻光束线关键光学元件热-结构耦合分析

为了保证上海光源X射线干涉光刻光束线的稳定性,减小热变形对实验结果的影响,对X射线干涉光刻光束线的3个关键光学元件——偏转镜、聚焦镜和精密四刀狭缝进行热-结构耦合分析。首先,计算偏转镜、聚焦镜和精密四刀狭缝所承载的功率密度;然后,建立其有限元模型;最后,获得光学元件的温度场和热变形的结果。结果表明,偏转镜和聚焦镜采用间接水冷方式可有效抑制热变形,冷却后的最大面形误差分别为7.2μrad和9.2μrad。精密四刀狭缝未冷却时,刀片组件温度介于271.56~273.27℃,刀口热变形为0.19 mm,直线导轨热变形为0.08 mm;经过铜辫子冷却后,刀片组件温度降至22.24~23.94℃,刀口热变形降至0.2μm,直线导轨热变形降至0.1μm;采用影像法和接触探头法测试后,刀口直线度、平行度和重复精度均满足技术要求。偏转镜、聚焦镜和精密四刀狭缝的热变形通过间接水冷和铜辫子的冷却方式可以得到很大程度的抑制,进而保证光斑质量。     

PIXE study on recovery of making-technology of Chinese Longquan celadon made in the Southern Song Dynasty (1127–1279 CE)

Proton induced X-ray emission (PIXE) technique and factor analysis were used to study the recovery of making-technology of Chinese Longquan celadon made in the late Southern Song Dynasty (1127–1279 CE). Based on factor analysis, for the raw material for body and the recipe for glaze, there is a close but slightly different relationship between these Longquan celadons unearthed from the core area of Dayao and the non-core area, despite all the Longquan celadons belong to the high-calcium-glaze system. The chemical compositions of most of the tentative body and glaze samples are very close to those of Longquan celadons unearthed from the core area of Dayao and the non-core area. This study can provide a guidance for recovery of making-technology of Longquan celandon.     

Strength of pure alumina ceramics above 1 GPa

Up to now, commercially available alumina ceramics were claimed to have strength between 400 and 550 MPa. However, our study shows strength ~ 2 times higher for commercially available alumina than commonly believed. The average and characteristic strength, measured on 31 pure alumina ceramic discs by ball on three balls (B3B) test, were 1205 ± 93 MPa and 1257 MPa, respectively, with a Weibull modulus of m = 11.8. Tested specimens were in form of discs with a diameter of 5 mm and thickness 0.5 mm. The grain size distribution of the alumina is bimodal with an average grain size of ~ 850 nm measured at the surface. The fracture reveals a mixed transgranular / intergranular failure mode. To avoid incorporation of additional flaws, the discs were tested as sintered. The characteristic flexural strength measured in B3B was recalculated according to Weibull theory for standard 4-point bending bars of size 3 × 4 × 45 mm as bend 856 MPa. The measured strength of nearly 900 MPa shows the potential of strength for high purity alumina ceramics.     

A Mobile Cloud-Based eHealth Scheme

Mobile cloud computing is an emerging field that is gaining popularity across borders at a rapid pace. Similarly, the field of health informatics is also considered as an extremely important field. This work observes the collaboration between these two fields to solve the traditional problem of extracting Electrocardiogram signals from trace reports and then performing analysis. The developed system has two front ends, the first dedicated for the user to perform the photographing of the trace report. Once the photographing is complete, mobile computing is used to extract the signal. Once the signal is extracted, it is uploaded into the server and further analysis is performed on the signal in the cloud. Once this is done, the second interface, intended for the use of the physician, can download and view the trace from the cloud. The data is securely held using a password-based authentication method. The system presented here is one of the first attempts at delivering the total solution, and after further upgrades, it will be possible to deploy the system in a commercial setting.     

Operational time-series data modeling via LSTM network integrating principal component analysis based on human experience

Today’s information technologies involve increasingly intelligent systems, which come at the cost of increasingly complex equipment. Modern monitoring systems collect multi-measuring-point and long-term data which make equipment health prediction a “big data” problem. It is difficult to extract information from such condition monitoring data to accurately estimate or predict health statuses. Deep learning is a powerful tool for big data processing that is widely utilized in image and speech recognition applications, and can also provide effective predictions in industrial processes. This paper proposes the Long Short-term Memory Integrating Principal Component Analysis based on Human Experience (HEPCA-LSTM), which uses operational time-series data for equipment health prognostics. Principal component analysis based on human experience is first conducted to extract condition parameters from the condition monitoring system. The long short-term memory (LSTM) framework is then constructed to predict the target status. Finally, a dynamic update of the prediction model with incoming data is performed at a certain interval to prevent any model misalignment caused by the drifting of relevant variables. The proposed model is validated on a practical case and found to outperform other prediction methods. It utilizes a powerful deep learning analysis method, the LSTM, to fully process big condition monitoring series data; it effectively extracts the features involved with human experience and takes dynamic updates into consideration.