Research on the Application of Lightweight Neural Network Models for Pulse Parameter Prediction

In nuclear radiation measurement, pulse distortion is inevitable due to the interference of the measurement system itself and the measurement environment. If the parameters of such pulses cannot be accurately estimated, the resolution performance of the energy spectrum will be reduced. [Purpose]: In order to accurately estimate the height of distorted pulses, this article proposes using six lightweight neural network models for parameter prediction of distorted pulses, including pulse amplitude parameters and distortion time parameters. [Methods]: Based on the distorted pulses generated by predefined mathematical models, the dataset required for model training is obtained through digital triangulation. [Results]: When evaluating the parameter prediction performance of six neural network models, the UNet model achieved the lowest relative error on the test set, with a relative error of approximately 0.57% for amplitude parameters and 3.51% for time parameters. In the signal-to-noise ratio experiment, noise was superimposed on the test set to obtain noise test sets with different signal-to-noise ratios. [Conclusions]: The results show that the proposed models can achieve accurate estimation of the parameters of distorted pulses.

在核辐射测量中，受测量系统自身及测量环境的干扰，脉冲畸变现象难以避免。若无法对这类畸变脉冲的参数进行精准估计，将导致能谱的分辨率性能下降。
【研究目的】为实现畸变脉冲幅度的精准估计，本文提出采用六种轻量级神经网络模型开展畸变脉冲的参数预测任务，涵盖脉冲幅度参数与畸变时间参数。
【研究方法】基于预定义数学模型生成的畸变脉冲，通过数字三角测量获取模型训练所需的数据集。
【研究结果】在对六种神经网络模型的参数预测性能进行评估时，UNet模型在测试集上取得了最低的相对误差：幅度参数的相对误差约为0.57%，时间参数的相对误差约为3.51%。在信噪比（Signal-to-Noise Ratio）实验中，研究人员向测试集叠加噪声，构建得到不同信噪比的含噪测试集。
【研究结论】结果表明，本文所提出的模型可实现畸变脉冲参数的精准估计。