Architecture Design and Implementation of the Nuclear Reactor Big Data System (Ruilong System)
-
摘要: 主要介绍了核反应堆大数据系统的数据架构设计，通过识别业务对象，构建了一套灵活、稳定的数据业务模型，形成了完整、规范、准确的数据体系，实现了多源头、异构数据的逻辑集成。在标签信息缺失的情况下，采用半监督学习方法开发了高性能的标签标注工具，并应用于非结构化数据和时序数据中。针对不同类型的数据缺失，使用基于机理、统计和机器学习的方法进行数据修复和异常筛选。通过关键领域分析和架构设计，建立了层次分明、扩展性强、灵活性高的软件视图，有效避免了设计反复，确保了开发质量。进一步提高核反应堆大数据技术的成熟度，提升数据治理、数据价值挖掘和智能运维技术能力，确保核反应堆装置的长期安全可靠运行。Abstract: This paper introduces the data architecture design of the nuclear reactor big data system. By identifying business objects, a flexible and stable data business model has been constructed, forming a complete, standardized and accurate data system that logically integrates multi-source and heterogeneous data. In cases where label information is missing, a high-performance label annotation tool has been developed using semi-supervised learning methods and applied to unstructured data and time-series data. For different types of data missing, methods based on mechanisms, statistics, and machine learning are used to repair and filter anomalies in the data. Through critical domain analysis and architectural design, a software view characterized by clear layers, strong scalability and high flexibility has been established. This effectively avoids design iterations and ensures development quality. To further enhance the maturity of nuclear reactor big data technology, the next phase will focus on improving data governance, data value mining and intelligent operation and maintenance capabilities, ensuring the long-term safe and reliable operation of nuclear reactor facilities.
-
Key words:
- Nuclear reactor /
- Big data /
- Architecture design
-
表 1 基于机理拟合融合的数据修复算法
Table 1. Date Repair Algorithm Based on Mechanism Fitting Fusion
输入 不完整测点序列x1, x2, x3, $\cdots $, xn，测点$ {t_1} $时刻采集值为$ x_{{t_1}}^1 $, $ x_{{t_1}}^2 $, $ x_{{t_1}}^3 $, $\cdots $, $ x_{{t_1}}^n $ 输出 完整测点序列x1, x2, x3, $\cdots $, xn 过程 开始 如果$ x_{{t_1}}^i $无值，且xi有冗余测点x2, x3, $\cdots $, xn(n≠i)： 如果冗余测点均有数据，则： $ \qquad\qquad\qquad x_{{t_1}}^i = \dfrac{{\displaystyle\sum\limits_{k = 1,k \ne i}^n {x_{{t_1}}^n} }}{{n - 1}} $ 如果冗余测点均无数据，且相关设备和系统也处于未运行状态，则： $x_{{t_1}}^i $={}，该测点对应的设备未运行 如果$ x_{{t_1}}^i $无值，且xi无冗余测点： 如果数据大段缺失，则： 利用具有因果关系的相关测点间的强时序相关性进行修复 如果数据小段缺失，则： 通过数据平滑及拟合对数据进行补全 结束 
下载: 导出CSV
-
[1] 马明旭,范玉顺,尹朝万. 基于产品结构的全生命周期形式化建模[J]. 机械工程学报,2006, 42(9): 83-90. doi: 10.3321/j.issn:0577-6686.2006.09.014 [2] 孙强,王英林. 基于UML的复杂配置产品数据模型研究[J]. 计算机工程,2007, 33(14): 51-53. doi: 10.3969/j.issn.1000-3428.2007.14.018 [3] 顾巧祥,祁国宁,纪杨建,等. 基于元数据的产品数据本体建模技术[J]. 浙江大学学报(工学版),2007, 41(5): 736-741. doi: 10.3785/j.issn.1008-973X.2007.05.007 [4] 谭云杰. 大象: Thinking in UML[M]. 北京: 中国水利水电出版社,2009. -
图(7) / 表(1)
计量
- 文章访问数: 37
- HTML全文浏览量: 19
- PDF下载量: 9
- 被引次数: 0
