Subjects: Astronomy >> Astrophysical processes submitted time 2023-01-18 Cooperative journals: 《天文研究与技术》
Abstract： With the rapid development of astronomical observation technology, astronomical data processing software has become increasingly complex, and it is challenging to deploy the software environment. Container technology was used for packaging the pulsar data processing environment to images by applying hierarchical encapsulation. According to the data processing requirements, a private library of images has been established, and the users can select the images according to the data processing mode. The hardware environment, the traditional virtual machine environment, and the container environment performed coherent dedispersion for pulsar baseband data. The resource utilization and data processing efficiency of different platforms are compared. Experimental results show that the performance of the container is similar to the physical machine. In multi-threaded parallel processing, the resource allocation of the container is more reasonable and can improve the efficiency of service resource utilization. Implemented container-based data processing architecture deployment on Xinjiang Astronomical Observatory pulsar data processing server, designed and developed the container management graphical user interfaces. Through optimizing functions such as multi-user login, authentication, and data volume mount to improve the efficiency of astronomical data processing using container technology.
Subjects: Astronomy >> Astrophysical processes submitted time 2022-04-14 Cooperative journals: 《天文研究与技术》
Abstract：针对射电天文观测过程中的射频干扰(RFI)问题，详细分析了国内外台站RFI抑制策略。根据各天文台站实际观测过程中遇到的RFI问题，分别从主动预防阶段、预相关阶段、后相关阶段、机器学习和深度学习等方面研究了RFI的预防策略和抑制方法。详细分析了主动预防阶段可采取的方法，预相关阶段的自适应滤波，和空间滤波方法；后相关阶段的VarThreshold，SumThreshold和奇异值分解等方法。探讨了基于机器学习的主成分分析、支持向量机、全卷积神经网络、卷积神经网络、U-Net等相关技术和方法在RFI信号处理方面的应用。
Subjects: Astronomy >> Astrophysical processes submitted time 2021-09-29 Cooperative journals: 《天文研究与技术》
Abstract：瞬时射频干扰（RFI）信号是最难以识别并消除的一类干扰。针对射电天文观测过程中遇到瞬时RFI问题，本文提出了一种基于RFI信号峰值的特征提取方法，利用该方法生成了RFI信号的特征模板，并基于动态时间规整（DTW）算法和打分策略设计了特征识别算法，实现了RFI信号识别以及分类功能。利用ARRL官网提供的RFI数据完成了特征模板的交叉验证对比，测试结果显示RFI特征模板与同类别RFI相似度高，表明该算法能够对RFI进行识别和分类，该算法为瞬时RFI的特征提取以及识别提供了一种新思路。
Subjects: Astronomy >> Astrophysical processes submitted time 2021-03-30 Cooperative journals: 《天文研究与技术》
Abstract：针对超宽带及多波束接收系统海量天文信号实时高效传输与处理问题，对基于FPGA+GPU的主流终端设备软件系统进行了测试分析，超宽带接收设备要求终端系统软件能够在更宽带宽，更高时间、频率分辨率条件下，实现实时数据流传输与处理。结合大口径射电观测设备未来发展方向提出了利用高速并行环形缓冲区实现数据流缓存、基于GPU集群实现数据流实时处理、基于BeeGFS实现分布式并行数据存储，模块化构建射电天文信号传输管道软件的设计思路。
Subjects: Astronomy >> Astrophysical processes submitted time 2020-12-21 Cooperative journals: 《天文研究与技术》
Abstract： 射电脉冲信号在传输过程中受到星际介质的影响会导致轮廓展宽和变形，在研究过程中需要对信号进行消色散处理。本文设计并实现了基于零拷贝的脉冲星数据GPU相干消色散算法，采用设备内存映射以消除主机到设备的拷贝开销，利用CUDA的cuFFT库进行多BATCH傅里叶变换以提高DFT效率，同时采用多线程实现了传递函数的加速计算。实验结果表明，与传统CPU及GPU算法相比，本文提出的算法在大数据量时表现良好。
Subjects: Astronomy >> Astrophysical processes submitted time 2019-12-04 Cooperative journals: 《天文研究与技术》
Abstract： 面向天文信号实时处理需求，搭建了基于ROACH2的射电天文数字终端实验平台。利用MATLAB、Xilinx等开发环境进行仿真，取得了原始实验数据；利用CASPER硬件平台实现了信号控制和预处理，并通过高速以太网将数据包传输至计算服务器进行后处理。搭建的实验平台实现了仿真、编译、运行的完整流程，为天文信号实时处理研究提供了良好的实验环境。
Subjects: Astronomy >> Astronomical Instruments and Techniques submitted time 2018-06-21 Cooperative journals: 《天文研究与技术》
Abstract： After years of observation, Xinjiang Astronomical Observatory(XAO) Nanshan 26 meters radio telescope (referred to as NSRT) had accumulated massive scientific data. A remote backup system was established for the online data storage of NSRT, this redundant storage system contains two storage clusters, one cluster was in XAO headquarters and another one located in Nanshan station, the real-time synchronization of NSRT data can be realized between two storage clusters. Based on the object storage technology, centralized Luster storage system was created for storage clusters, and the I/O performance test of luster systems was finished. Redundant storage system solved the online archive and data safety issue for NSRT data.
Subjects: Astronomy >> Astrophysical processes submitted time 2018-05-15 Cooperative journals: 《天文研究与技术》
Abstract：新疆天文台Taurus高性能计算系统由1个登陆节点、16个计算节点、2个I/O节点、100TB高速存储组成。其CPU理论双精浮点计算能力6.7584Tflops，通过Linpack测试实际峰值为6.289Tflops,可提供计算能力为理论值的93.06%；GPU理论计算能力18.72 Tflops，实际测试计算峰值为14.882Tflops,计算效率为理论能力的79.5%。系统计算与存储节点均采用56Gb Infiniband交换互连，通过IOZone测试存储系统，单节点写可达460MB/s,多节点写可达800MB/s。系统已在多相滤波及消干扰GPU算法加速、蒙特卡罗模拟等领域得到了应用。
Hits
Hits
Downloads
Comment