Particle Detector Calibration with Shielding Units and Bayesian Classifier
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current particle detection methods, such as gamma-only sources, time-of-flight data, and Gaussian peak fitting, face limitations in accurately distinguishing between neutrons and gamma rays due to contamination, noise, and distributional assumptions, requiring specialized equipment and expert knowledge, and are not applicable to all classification feature spaces.
Innovation Solution
A system and method for calibrating radiation detectors using a two-unit framework with shielding units to generate training data for a Bayesian classifier, allowing classification of particles without specialized equipment or expert knowledge, and applicable to any source, including those emitting alpha particles, fast neutrons, thermal neutrons, or gamma rays.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional particle detection methods (gamma-only sources, time-of-flight data, Gaussian peak fitting) are used, then particle detection capability is provided, but accuracy in distinguishing between neutrons and gamma rays deteriorates due to contamination, noise, and distributional assumptions
Solution Approach 1:
The patent applies preliminary action by generating training data in advance through simulations that model various contamination and noise conditions. The Bayesian classifier is trained beforehand with this pre-generated data, enabling it to accurately distinguish between neutrons and gamma rays even when contamination and noise are present during actual detection, without requiring real-time complex processing
Solution Approach 2:
The patent changes parameters by using a Bayesian classifier that can adapt to different contamination levels and noise conditions. The classifier's parameters are optimized through training with simulated data that varies in contamination and noise characteristics, allowing the system to maintain high accuracy across different operational conditions without requiring distributional assumptions
2Measurement precision
If specialized equipment and expert knowledge are used for particle classification, then classification capability is provided, but system complexity and operational difficulty increase
Solution Approach 1:
The patent applies self-service by creating a classification system that automatically generates its own training data through simulations and performs self-calibration. The Bayesian classifier is trained with simulated data that reflects actual operational conditions, enabling the system to classify particles accurately without requiring external expert knowledge or specialized equipment for calibration
Solution Approach 2:
The patent uses copying by creating simulated training data that replicates actual detection conditions. Instead of requiring physical reference samples or specialized calibration equipment, the system generates virtual copies of detection scenarios through simulation, which are then used to train the classifier, simplifying the physical system requirements
3Adaptability or versatility
If traditional classification methods with distributional assumptions are used, then classification process is provided, but applicability to different classification feature spaces deteriorates
Solution Approach 1:
The patent applies universality by developing a Bayesian classifier framework that can handle multiple types of particles and various classification feature spaces. The same basic classifier structure and training approach can be applied whether classifying neutrons from gamma rays, or other particle types, making the system versatile without sacrificing accuracy through distributional assumptions
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method provides a reliable and efficient classification of particles by generating substantial training data, reducing the need for separate classification steps and expert knowledge, and is applicable to various classification features, achieving accuracy comparable to traditional methods without distributional assumptions.
Implementation Method 1
each shielding unit configured to block at least a portion of the first type of particles or rays and to allow the second type of particles or rays to traverse substantially unimpeded therethrough
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Techniques for calibration of particle detectors are disclosed. In one aspect, a system for calibrating a particle detector includes a source configured to emit particles including a first and a second type of particles; a first and a second shielding unit configured to be removably positioned in a travel path of the particles and configured to block at least a portion of the first type of particles and to allow the second type of particles to traverse substantially unimpeded therethrough; one or more particle detectors positioned in the travel path of the particles to receive particles after traversing through the first or the second shielding unit and produce electronic pulses in response to the detection of the particles; and a processor coupled to the one or more particle detectors to generate training date and a classifier that allows classification of the first and the second types of particles.