基于先进神经网络的PGNAA元素成分高精确度分析方法研究

PGNAA technology is a kind of typical nuclear analysis technology. Higher precision and accuracy of material element composition analysis technology based on PGNAA are required by modern industrial production enterprises. In order to improve the precision and accuracy of the elements composition analysis based on the technology of PGNAA, this subject will study the physical mechanism of single element characteristics influence each other. Then single element standard spectrum without background effect is got to remove the background of the multi-element composition spectrum. Subsequently an advanced neural network technology is used in full spectrum resolution and the analysis of element component content for the corrected multi-element composition spectrum. This new method of material element composition analysis based on PGNAA constructed in the subject will improve precision and accuracy of the elements composition analysis. The results of this project will provide support of physical mechanism, complete physical model and rapid data analysis method with high precision and accuracy for improving the level of PGNAA industrial material composition test.

PGNAA技术是一种具有代表性的核分析技术，现代化的工业生产企业对基于PGNAA技术的物料元素成分分析的精密度与准确度提出了更高的要求。为了提高基于PGNAA技术的元素成分分析精密度及准确度，本课题在探究PGNAA技术用于元素分析时所面对的单元素特征谱相互影响的物理机制的基础上，通过构建去除背景影响的单元素标准能谱，对多元素组分能谱进行去除背景影响修正，而后利用先进的神经网络技术对修正后的多元素能谱进行进行全谱的解谱和元素成分含量分析，从而构造基于PGNAA技术的工业物料元素成分分析新方法，提高元素分析的精密度及准确度。该项目的研究结果将为提高PGNAA工业物料成分检测水平提供物理机制的支撑、完整的物理模型以及简单快速高精密度及准确度的数据分析方法。