核电站瞬态运行参数数据压缩与复原方法  

Data compression and recovery method for transient operation parameters of nuclear power plants

作　　者：李翔宇 程坤 黄涛 余霖 谭思超 LI Xiangyu;CHENG Kun;HUANG Tao;YU Lin;TAN Sichao(Heilongjiang Provincial Key Laboratory of Nuclear Power System & Equipment, Harbin 150001, China;College of Nuclear Science and Technology, Harbin Engineering University, Harbin 150001, China;Science and Technology on Reactor System Design Technology Laboratory, Nuclear Power Institute of China, Chengdu 610213, China)

机构地区：[1]黑龙江省核动力装置性能与设备重点实验室,黑龙江哈尔滨150001 [2]哈尔滨工程大学核科学与技术学院,黑龙江哈尔滨150001 [3]核反应堆系统设计技术重点实验室,四川成都610213

出　　处：《哈尔滨工程大学学报》2021年第12期1720-1725,共6页Journal of Harbin Engineering University

基　　金：核反应堆系统设计重点实验室基金项目(HT-WDZC-02-202000);中央高校基本科研业务费专项(3072020CFT2403).

摘　　要：针对当前因核电站瞬态运行测量参数种类较多导致数据在网络传输、数据存储和算法训练过程中存在传输速度慢、算法拓扑结构复杂等问题,本文通过改进特征工程算法建立了一种核电站瞬态运行参数数据压缩和数据复原方法。该方法利用主成分分析方法提取瞬态运行参数的特征向量,减少瞬态运行参数的数据维度;同时配合高斯过程回归方法可对降维后的瞬态运行数据实现高精度的复原。结果表明:瞬态运行数据先压缩再复原后,复原值与实际值最大误差不超过0.000002。同时压缩后的数据特征也可为核电站故障诊断提供判断依据,进一步提高核电站安全与可靠性。Currently,because of many types of transient operation parameters of nuclear power plants,some problems are encountered in network transmission,data storage,and algorithm training processes,such as slow transmission speed,large storage space,slow algorithm training speed,and complex structure.In this study,a method for data compression and restoration of transient operation parameters of nuclear power plants is established by improving the characteristic engineering algorithm.The method uses principal component analysis to extract the feature vector of transient operation parameters,and combined with the Gaussian process regression method,the reduced dimension transient operation data can be restored with high precision.The results show that,after the transient operation data are compressed and restored,the maximum error between the restored and actual values does not exceed 0.000002.The compressed data characteristics can also provide a basis for the judgment of fault diagnosis of nuclear power plants and further improve the safety and reliability of nuclear power plants.

关 键 词：核动力装置与设备 瞬态运行参数分析 数据挖掘 机器学习 特征工程 数据降维 主成分分析 高斯过程回归 

分 类 号：TL363[核科学技术—核技术及应用]