兵工自动化
兵工自動化
병공자동화
Ordnance Industry Automation
2015年
11期
86-91
,共6页
神经网络%快速识别%γ谱%主成分分析
神經網絡%快速識彆%γ譜%主成分分析
신경망락%쾌속식별%γ보%주성분분석
neural network%fast identification%γ spectrum%PCA
针对传统的方法解析γ谱数据实现核素识别存在步骤多、精度低、专业知识要求高、识别速度慢等缺点,提出了一种基于人工神经网络的核素识别分析方法。在对全谱γ数据进行主成分分析的基础上,提取出γ谱的主要特征,将此特征信息输入人工神经网络,利用BP网络算法和RBF网络算法可快速地完成γ谱解析。分析结果表明:该方法降低了对探测器能量分辨率的要求,同时避免了寻峰、能量刻度与效率刻度等问题,简化了核素识别的过程,有效地提高了放射性核素的快速识别能力。
針對傳統的方法解析γ譜數據實現覈素識彆存在步驟多、精度低、專業知識要求高、識彆速度慢等缺點,提齣瞭一種基于人工神經網絡的覈素識彆分析方法。在對全譜γ數據進行主成分分析的基礎上,提取齣γ譜的主要特徵,將此特徵信息輸入人工神經網絡,利用BP網絡算法和RBF網絡算法可快速地完成γ譜解析。分析結果錶明:該方法降低瞭對探測器能量分辨率的要求,同時避免瞭尋峰、能量刻度與效率刻度等問題,簡化瞭覈素識彆的過程,有效地提高瞭放射性覈素的快速識彆能力。
침대전통적방법해석γ보수거실현핵소식별존재보취다、정도저、전업지식요구고、식별속도만등결점,제출료일충기우인공신경망락적핵소식별분석방법。재대전보γ수거진행주성분분석적기출상,제취출γ보적주요특정,장차특정신식수입인공신경망락,이용BP망락산법화RBF망락산법가쾌속지완성γ보해석。분석결과표명:해방법강저료대탐측기능량분변솔적요구,동시피면료심봉、능량각도여효솔각도등문제,간화료핵소식별적과정,유효지제고료방사성핵소적쾌속식별능력。
The traditional method of nuclide identification by analyzing the data inγ spectrum takes various steps and suffers low accuracy and identification speed. In addition, it requires much more professional knowledge and work. This paper proposed a novel method of nuclide identification and analysis using neural network. Based on the principal component analysis (PCA) of all the data inγ spectrum, it extracts the main features ofγ spectrum and then imports the extracted information into the neural network. It can implement fast analysis of the data inγ spectrum with high accuracy by using BP and RBF network algorithms. The experimental results show that our proposed method does not require high energy resolution of detector as well as solves the problems of peak searching, energy calibration and efficiency calibration. Therefore, it simplifies the process of nuclide identification and significantly improves the ability of fast identification of radioactive nuclide.