摘要
核部件属性检测是核材料检测中的热点与难点问题,针对铀部件检测系统中丰度信息易受探测器基本属性、几何位置及噪声的影响,通过分析铀部件探测信号互相关函数的双谱特性,提取双谱中的特征量作为铀部件丰度检测的数据集,消除了本底噪声对探测信号的影响。此外,基于BP神经网络,构建了一种同质量球形铀部件的丰度判定方法,实现了对测试样本的丰度检测。测试结果表明,该方法在存在高斯白噪声情况下可有效判定同质量球形铀部件的丰度,其测试误差的平均值为0.005115,最大误差为0.009927。
Attribute detection of uranium components is a hot and difficult problem in nuclear material detection.Because of the abundance information of uranium components is easily affected by the basic properties,geometry and noise.The bispec-trum characteristics of cross-correlation function are analyzed,and the characteristics of bispectrum are extracted as the basis for determining the abundances of uranium components,the background noise to the detected signal is eliminated,the abundances of same mass spherical uranium components are determined by bp neural network,the test results show that the method can effectively determine the abundance of same mass spherical uranium components in the presence of white gaussian noise,the average value of test error is 0.005115 and the maximum error is 0.009927.
作者
任立学
刘知贵
付聪
REN Lixue;LIU Zhigui;FU Cong(Southwest University of Science and technology,School of Defense Science and technology,Mianyang Sichuan 621010,Chian;Southwest University Of Science And Technology,Graduate School,Mianyang Sichuan 621010,China)
出处
《自动化与仪器仪表》
2020年第2期4-7,共4页
Automation & Instrumentation
基金
国家自然科学基金NSAF联合基金项目(No.11175031)
关键词
铀部件
丰度检测
双谱估计
BP神经网络
高斯白噪声
uranium components
abundance detection
bispectrum estimation
bp neural network
white gaussian noise
