A new background discrimination method using support vector machine (SVM) with gaussian kernel in low-level ³H liquid...

Area of Science:

• Nuclear Science and Technology
• Analytical Chemistry
• Machine Learning Applications

Background:

• Low-level tritium (³H) detection is crucial for environmental monitoring and nuclear safety.
• Traditional liquid scintillation counting methods face challenges in accurately discriminating low-activity samples from background noise.
• Advanced signal processing and machine learning offer potential solutions for improved background discrimination.

Purpose of the Study:

• To develop and evaluate a novel background discrimination method for low-level ³H liquid scintillation measurements.
• To enhance the sensitivity and accuracy of detecting low-activity tritium samples.
• To compare the performance of the proposed method against traditional techniques.

Main Methods:

• Utilized Support Vector Machine (SVM) with a Gaussian kernel for event classification.
• Applied dimensionality reduction techniques to process full waveform data.
• Employed offline processing for systematic parameter optimization.
• Generated high-quality labeled datasets using blank and medium-activity ³H samples.
• Prepared low-activity ³H samples (1-5 Bq) via quantitative dilution.

Main Results:

• The proposed SVM-based method demonstrated significantly improved background discrimination capability.
• Effectively minimized background levels across the entire energy spectrum.
• Preserved detection efficiency for ³H signals.
• Achieved enhanced measurement capabilities for low-activity ³H samples.

Conclusions:

• The developed machine learning approach offers a superior alternative for background discrimination in low-level tritium measurements.
• This method enhances the reliability and sensitivity of environmental and nuclear sample analysis.
• The findings pave the way for more accurate monitoring of tritium in various applications.