【作者】 郑华文；

【导师】 杨将新； 曹衍龙；

【作者基本信息】 浙江大学 ， 机械制造及自动化， 2010， 博士

【摘要】 本文结合国家高技术研究发展计划项目“核电站松动件监测系统关键技术研究(863计划,2007AA04Z426)”,就核电站松动件冲击信号的预处理与识别、松动件定位等关键技术开展了研究,提出了“核电站松动件定位方法研究”的博士论文研究课题。第1章：阐述了松动件监测系统(Loose Part Monitoring System,LPMS)的研究背景和意义。介绍了国内外松动件监测系统的发展进程、系统结构以及关键技术,综述了国内外松动件定位方法的研究现状,最后给出了本论文的研究背景和总体框架。第2章：通过分析核电站背景噪声和松动件冲击信号的性质,提出了一种松动件冲击信号的预处理和识别方法。该方法采用自适应AR模型实现检测信号的白化处理,然后以白化信号的线性预测系数(Linear Predictive Coefficients, LPC)作为信号特征,通过支持向量机(Supporter Vector Machine, SVM)识别所测信号是否包含冲击信号；最后通过仿真实验验证了该方法的有效性。第3章：研究了三角形定位方法中时差估算误差和波速测量误差对松动件定位精度的影响规律,提出了一种基于总体平均经验模态分解(Ensemble Empirical Mode Decomposition, EEMD)和希尔伯特变换的松动件定位方法。首先分析了基于EEMD分解的主能量弯曲波提取方法；然后将主能量弯曲波信号的希尔伯特包络线第一个峰值所对应的时刻作为该弯曲波的到达时间,并基于主能量弯曲波信号的频谱计算弯曲波波速；进而根据信号波速和到达时间差实现松动件定位；仿真实验结果表明,该方法能够实现快速准确定位,且具有良好的抗噪声性能。第4章：改进了圆交叉定位法,研究了基于信号时频分析的波传播距离估计方法,在此基础上提出了一种基于平滑伪D-Flandrin分布的松动件定位方法。该方法通过分析弯曲波的频散特性,建立了时频平面上弯曲波频率和传播时延的关系；根据该关系和仿射类时频分布的定义可知平滑伪D-Flandrin分布较适合于对弯曲波信号进行时频分析；时频分析后采用非线性最小二乘法求解信号频散模型的参数以估计弯曲波传播距离,最后应用圆交叉定位法实现定位。仿真实验分析证明了该定位方法的有效性。第5章：研究了圆柱壳和球壳表面的定位分析方法。针对由于各个方向波速不一致造成的圆柱壳表面定位难问题,分析了冲击信号在圆柱壳上的传播模型,并由此提出了一种应用于圆柱壳定位的改进扫描定位方法。该方法通过优化多个目标实现定位：圆柱壳上弯曲波的周向传播速度方差最小、轴向传播速度方差最小以及初始碰撞时间的方差最小；最后以粒子群多目标优化算法进行目标寻优,增加了定位的速度。进行了仿真实验验证,结果表明了所提方法的有效性。第6章：在理论研究的基础上,采用LabVIEW开发了核电站松动件监测(LPMS)原型系统。详细分析了系统架构及主要功能模块,设计了软件模块结构以及软硬件接口,实现了松动件的在线监测、定位分析等功能,并在钢板上进行了验证分析。第7章：概括了全文的研究内容,并对核电站松动件监测研究方面作出展望。更多还原

【Abstract】 Based on the High Technology Research and Development Program of China "Research on the Key Technology of Loose Part Monitoring in Nuclear Power Plant (No.2007AA04Z426)", loose parts’impact signal preprocessing and identification and impact localization methods of loose parts are studied.In chapter one:The importance and background of setting Loose Part Monitoring System (LPMS) in Nuclear Power Plants (NPPs) are expounded. Introduced the key technology, system structure and develop progress of LPMS. The impact localization methods now available are summarized. Finally, the study background and overall frame of the dissertation are given.In chapter two:a signal preprocessing and impact signal identification method is proposed by analyzing the noise signal of NPP and impact signal of loose parts. The signal is whitened using adaptive AR model, and then the SNR of signal can be improved. Extract the Linear Predictive Coefficients (LPC) of the whitened signal as its characteristic, and then using Supporter Vector Machine (SVM) to identify whether the recorded signal contains impact signal. Simulation and experiment results show that the method can identify the signal in low SNR.In chapter three:The time arrival difference error and wave speed measurement error influence on the positioning accuracy of triangle positioning method are studied. A loose part positioning method based on Ensemble Empirical Mode Decomposition (EEMD) and Hilbert Transform was proposed. The method uses EEMD to decompose the signal into some singal vibration mode, and then calculate the envelope of the main vibration mode using Hilbert Transform, find the first peak of the envelope as the arrival time of signal. According the arrival time and wave propagating velocity, the location of loose part can be estimated. Experiment results show that the method can estimate the location of loose part fast and accurately.In chapter four:Improved the circle intersection positioning method to get the position of loose part. A wave propagation distance estimation method based on time-frequency analysis is studied, and then a loose part positioning method based on smoothing pseudo D-Flandrin distribution is proposed. Through the analysis of the dispersive characteristic of bending wave, build the relationship between the group delay and the frequency of wave, and then find that the D-Flandrin distribution is a suitable time-frequency analyzing method for bending wave in plate. After the time-frequency analysis, the Non-linear Least Square Method are taken to estimate the distance between impact point and sensors, then the position of loose part can be estimated using circle intersection positioning method. Experimental results prove the validity of the method.In chapter five:The positioning method in cylindrical shell and spherical shell is studied. It’s hard to positioning the loose part in cylindrical shell as wave axial and circumferential propagation velocity is different. An improved scan positioning method is proposed based on the analysis of impact signal’s propagation model in cylindrical shell. This method estimates the position of loose part by calculate the axial velocity, circumferential velocity, and the initial collision time. The particle swarm optimization algorithm is applied to accelerate scanning speed, which can greatly save search time. Experimental results show the effectiveness of the proposed method.In chapter six:Based on the theory study, a nuclear loose part monitoring system protype is developed using Lab VIEW. Detailed analysis of the system architecture and main modules are taken, and the software module structure and system interfaces are designed, online monitoring, location analysis and other functions are achieved, and the system is simulated on plate.In chapter seven:The main content of the dissertation research is summarized, and the prospect of loose part monitoring research is presented.更多还原