252Cf源驱动核材料裂变中子脉冲信号的压缩感知理论应用基础研究

Nuclear proliferation is one of the key security issues that the world is facing today. To prevent the spread of nuclear weapons, a lot of nuclear arms control verification technologies have sprung up. However, the cost of nuclear arms control verification technology is very high and the identification process is non-real time, which have become a major bottleneck of its field application. Based on the special "0,1" sparse structure of 252Cf-source -driven neutron signal, this project will use the compressed sampling reconstruction and processing technologies to construct a kind of low cost, high accuracy and real-time nuclear arms control verification technology. First, in order to reduce data acquisition costs effectively, we will establish the compressive sampling and reconstruction algorithm for neutron signals and point out the relationship between the compressive sampling rate, the measurement matrix and the reconstruction probability. Second, we will propose a new denoising algorithm based on compressive sensing for neutron signal so as to improve the verification accuracy. At last, compressive sampling identification technology will be constructed to reveal the properties of nuclear material and improve the performance of real-time of nuclear arms control verification. This project aims to reveal the application prospect of compressed sensing in nuclear arms control verification technology. This provides a theory and technique basis for the reduction of nuclear arms control verification cost, improvement of the verification accuracy and online identification of nuclear material properties.

核扩散是当今世界面临的核心安全问题之一，核军控核查是防止核扩散的有效手段。然而，现有的核军控核查技术成本较高、实时性欠佳，已成为制约该技术发展与现场应用的主要瓶颈。本项目针对252Cf源驱动核材料裂变中子脉冲序列之特殊的“0,1”稀疏结构，采用压缩采样重构与压缩处理识别技术，研究一种低成本、高准确率的实时在线核查技术方法：（1）建立裂变中子信号压缩采样与重构方法，明确压缩采样率、观测矩阵与重构概率之间的关系，有效降低数据采集成本；（2）研究基于压缩感知的裂变中子信号降噪技术，为提高核查准确率奠定基础；（3）探索核材料属性的压缩采样识别技术，以期改善核军控核查的实时性。本项目旨揭示压缩感知理论在核军控核查技术中应用前景，从而为降低核军控核查成本、提高核查准确率、在线判读核材料属性提供理论依据和技术基础。

当今世界国际政治热点话题之一的伊朗核问题与朝核问题，无不牵动着核扩散潜在危险这一敏感神经，这使得核武器的存在与核技术的发展所产生的一系列不确定因素严重地威胁着人类的安全与世界和平，无疑，屡屡引发了当今国际安全关系中反复涉及的一项重大政治课题，即核军控核查。本项目从252Cf源驱动核材料裂变的核军控核查理论、仪器及方法技术出发，突破传统的Nyquist采样定理，运用压缩感知技术和分形理论，率先在国内开展面向核军控核查的252Cf源驱动核材料裂变中子脉冲序列信号压缩感知技术研究，尤其致力于开展降低高速数据采集成本的研究，涉及裂变中子脉冲序列信号压缩感知重构方法、压缩降噪以及压缩信号处理等关键技术研究工作。本项目主要取得的成果如下：（1）建立了252Cf 源驱动核材料裂变中子脉冲序列信号压缩采样与处理理论模型；（2）提出了适合252Cf源驱动裂变中子信号“0,1”特殊稀疏结构的压缩采样重构方法，将采样率降低90%。（3）研究了裂变中子信号压缩降噪技术，信噪比提高约6dB。（3）建立了压缩信号处理的方法，无需重构原始信号，提升了核军控核查的实时性。（4）研究了中子成像理论及方法，提高了核军控核查的准确性。（5）研究了基于深度学习的核材料浓度识别方法，丰富了核军控核查的技术理论。本项目从以压缩感知理论为核军控核查技术研究发展的新原理，围绕252Cf源驱动裂变中子信号压缩采样重构、降噪和压缩信号处理等关键技术问题，深入、系统地开展了一系列创新性的、且卓有成效的研究工作，为降低核军控核查成本、提高核查准确率、在线判读核材料属性提供理论依据和技术基础，这对提高我国在核材料识别技术领域的研究水平，突破国外技术封锁，具有重要而深远的意义。