CAJ | 학술논문

针对传统γ能谱核素识别算法在较低放射性水平下的情况下存在核素识别慢，对相干核素识别效果差的问题，基于统计学习理论提出了基于支持向量机的快速核素识别方法。本方法一方面通过对峰位确定方法改进来提高特征量提取的精确度，去除γ能谱中重叠峰对相干核素识别中的影响，另一方面使用支持向量机方法实现核素的快速分类。通过与传统方法的比对，新方法在识别相干核素方面达到了很好的识别效果。
침대전통γ능보핵소식별산법재교저방사성수평하적정황하존재핵소식별만，대상간핵소식별효과차적문제，기우통계학습이론제출료기우지지향량궤적쾌속핵소식별방법。본방법일방면통과대봉위학정방법개진래제고특정량제취적정학도，거제γ능보중중첩봉대상간핵소식별중적영향，령일방면사용지지향량궤방법실현핵소적쾌속분류。통과여전통방법적비대，신방법재식별상간핵소방면체도료흔호적식별효과。
Nuclide identification algorithm for energy spectrum at low level radioactive nuclide identification exits slow, poor re-sults on coherent radionuclide identification problem.based on statistical learning theory proposed based on support vector ma-chine quick radionuclide identification method. By this method one hand, spectrum data smoothing, peak position to determine ways to improve to improve the accuracy of feature extraction, removal of overlapping peaks in spectra for radionuclide identifica-tion of coherent effects, on the other hand using the support vector machine approach to nuclear prime fast classification. By com-parison with the traditional methods, the new method in identifying coherent nuclide has reached a good recognition results.