近年来，时间投影室(TPC)技术因其在立体角、能量分辨以及多观测量关联测量等方面的显著优势而越来越多地应用于原子核裂变的实验研究中。但现有的技术手段也存在理论局限，即对于裂变碎片径迹的分析，其精度仍受制于重离子能损理论，无法准确提取碎片的核素种类。此外，低能重离子径迹显著的横向歧离也使得径迹分析不适宜用简单的直线模型。. 本项目针对上述局限，提出借助人工神经网络算法不依赖于具体数学表达形式的独特优势，建立全新的裂变碎片径迹分析技术。在该技术中神经网络训练样本选用结合缓发γ能谱的实验数据，从而避免了大部分人工神经网络方法对计算机模拟数据的依赖性，结果更加可信。. 本项目的成功实施不仅能提高TPC技术对低能重离子的测量水平，其研究思路也为在裂变核数据评价工作中运用机器学习技术总结一套行之有效的范式。更有望为中国散裂中子源的其他交叉学科应用提供先进的核探测技术支撑和可以借鉴的范例。

In recent years, the technique of Time Project Chamber (TPC) is more and more employed in the experimental research of nuclear fission due to several advantages such as solid angle, energy resolution, and correlated measurement of multi observables. However, in current techniques there is a shortage that the tracking precision of fission fragment is limited due to the theory of heavy ion energy loss. It thus eliminates the capability of identifying the isotopes of fragments. Furthermore, the significant transverse struggling in the low-energy heavy-ion track also suggests a straight-line model is not compatible for the tracking algorithm. ..Aiming at these issues, this proposal plans develop a new method of tracking analysis for the fission fragments, taking the unique advantage of an artificial neural network (ANN) that its response does not rely on any explicit mathematical form. In this method the learning samples are selected from the measured data with the delayed gamma spectrum as reference. It will eliminate the dependence of simulation which is used in most ANN applications and offers a more reliable results...Not only can the implements of the proposal improve the TPC technique for heavy ion measure, but the method may also inspire the usage of ANN in fission data evaluation as an effective example. Furthermore, this work would offer a technical support of the advance nuclear detection for other cross-area application in CSNS.