Adaptive digital shaping of nuclear pulse based on real-time tracking of system transfer function and xLSTM

• Published in: Applied radiation and isotopes Vol. 225; p. 112048
• Main Authors: Wu, Zhonglin, Huang, Hongquan, Chen, Dejie, Ma, Xingke
• Format: Journal Article
• Language: English
• Published: England Elsevier Ltd 01.11.2025
• Subjects: Digital shaping; Nuclear pulse; Transfer function; xLSTM; Nuclear pulse; Digital shaping; Transfer function; xLSTM
• ISSN: 0969-8043; 1872-9800; 1872-9800
• DOI: 10.1016/j.apradiso.2025.112048

Typically, the radiation signals detected by a nuclear detector are processed through subsequent conditioning circuits and converted into analog pulse signals. Digital shaping techniques are then used to generate standard trapezoidal or Gaussian digital signals for energy spectrum analysis and counting processing. However, detector aging or environmental changes (such as temperature and humidity) can cause variations in the system transfer function. If the digital shaping parameters are not adjusted in a timely and accurate manner, the shaped waveform may become distorted, affecting the accurate extraction of pulse amplitude. To address this issue, this paper proposes a real-time transfer function tracking algorithm based on waveform vector space. This method dynamically captures transfer function variations through a continuously updated mechanism and simultaneously iteratively searches for the optimal pulse signal in a relatively stable state. Furthermore, we innovatively introduce the extended long short-term memory (xLSTM) network into the field of nuclear pulse parameter identification, ensuring adaptive digital shaping optimization in real-time tracking scenarios. Experimental results show that this method can keep the relative error of digital shaping parameters within 0.3 %. •The transfer function of a nuclear detection system typically changes during testing.•Digital shaping algorithms are sensitive to variations in transfer function parameters.•A pulse waveform space update mechanism is designed to track the changing trend of the pulse transfer function.•A deep learning model, xLSTM, is introduced to identify these variations.