PCA-SDG在反应堆冷却剂系统典型故障诊断中的应用  

Application of PCA-SDG to typical fault diagnosis in the reactor coolant system

作　　者：温志斌 刘永阔 段智勇 彭敏俊 

机构地区：[1]哈尔滨工程大学核安全与仿真技术国防重点学科实验室,黑龙江哈尔滨150001

出　　处：《应用科技》2016年第5期82-87,共6页Applied Science and Technology

基　　金：黑龙江省自然科学基金面上项目(A2016002;E2015053)

摘　　要：随着我国核能应用规模的扩大,核电厂安全成为核能发展中的重要研究课题;为了保障核电厂的安全运行,国内外提出了各种方法对核电厂进行状态监测、故障诊断和故障预报。符号有向图(SDG)能够简洁有效地对各种故障模式进行描述,并且在解释故障传播路径方面有一定优势,但存在节点阈值确定困难的问题,为此提出结合主元分析(PCA)与符号有向图进行研究,PCA通过分析残差检测故障的发生,然后SDG对PCA得到的残差确定节点状态,进行推理,得出故障的类型。通过模拟器PCTRAN的数据测试,验证该方法能及时准确地检测到故障并诊断出故障的类型。With the scale expansion of nuclear power application in China,the safety of a nuclear power plant has become a significant issue. To ensure safe operation of the nuclear power plant,a variety of measures have been applied to condition monitoring,fault diagnosis and forecasting. A signed directed graph（ SDG） is capable of representing various fault patterns concisely,having an advantage in interpreting the propagation path of faults; nevertheless,it has difficulty in deciding the thresholds of each node. Consequently,principal component analysis（ PCA）incorporated with SDG method was proposed. By residuals analysis,the fault was detected by PCA,and the state of nodes in SDG model was determined,identifying the fault types via inferring from SDG model. The applications in PCTRAN testify effectiveness of the proposed approach in fault detection and diagnosis.

关 键 词：核动力装置 故障诊断 主元分析 符号有向图 反应堆冷却系统 

分 类 号：TL36[核科学技术—核技术及应用]