基于SOM算法的爆炸物识别方法分析  

Analysis of Explosive Identification Method Based on SOM Algorithm

作　　者：李振榕 LI Zhenrong(School of Mechanical and Electronic Engineering,Donghua University of Technology,Jiangxi 330013,China)

机构地区：[1]东华理工大学机械与电子工程学院,江西330013

出　　处：《电子技术(上海)》2024年第12期360-361,共2页Electronic Technology

摘　　要：阐述一种基于PGNAA技术的爆炸物检测装置的设计和优化,结合SOM模型处理伽马能谱,实现待检测样品的定性分析。建立爆炸物检测的PGNAA模型,根据信噪比优化方法对PGNAA模型进行优化。然后,使用MCNP获得伽玛能谱。使用SOM模型对伽马能谱进行分析。选择ANN和SVM模型与SOM模型进行比较。结果显示,相对于ANN和SVM而言,SOM模型在识别爆炸物方面具有明显的优势,其识别爆炸物的准确率、精确度、召回率和F1分数分别为0.9844、0.9751、0.9668、0.9836。This paper expounds that an explosive detection device based on PGNAA technology is designed and optimized to process the gamma energy spectrum in combination with the SOM model to realize the qualitative analysis of the samples to be detected.First,a PGNAA model for explosives detection was established,and the PGNAA model was optimized according to the signal-to-noise ratio optimization method.Then,the gamma energy spectrum was obtained using MCNP.The gamma energy spectrum was analyzed using the SOM model.ANN and SVM models were selected for comparison with the SOM model.The results show that the SOM model has obvious advantages in recognizing explosives compared to ANN and SVM,and its accuracy,precision,recall and F1 score for recognizing explosives are 0.9844,0.9751,0.9668,and 0.9836,respectively.

关 键 词：SOM模型 PGNAA MCNP 信噪比 爆炸物识别 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] TP18[自动化与计算机技术—计算机科学与技术]