Search

A condenser air removal system plays a crucial role in radiation monitoring within nuclear power plants, specifically for detecting potential leaks of radioactive material from primary-to-secondary systems. This system assesses the leakage rate (LR) by measuring the emission of β particles from noble gases using a thin plastic scintillation (TPS) detector, which is specifically designed for counting purposes. Among the factors used in LR calculation, the detection efficiency (k) must be experimentally determined in advance. However, this is challenging because of the limitations posed by radiation safety regulations. Moreover, simulation-based calculations are difficult to achieve because only a portion of the β -ray energy is transferred in the TPS detectors, resulting in a lack of prior energy information. To overcome these challenges, this study employed a multi-objective parameter optimization technique to simultaneously calculate the parameters required for the β -ray detection response function and energy calibration equation. Subsequently, k values were calculated using the optimization results. The optimization performance was demonstrated by comparing the measured spectra of the TPS detector with the spectra obtained from the Monte Carlo N-particle simulation. Furthermore, we confirmed that the calculated k values agreed with the reference values. • Overcomes challenges of insufficient information in measured spectra for beta radiation using the TPS detector. • Optimizes beta response function with a genetic algorithm and estimates TPS detector efficiency using MCNP simulation. • Results show high agreement between simulated and measured data, with relative errors under 2%. • Findings validate optimization and support using the TPS detector for accurate beta nuclide detection in the CAR system. [ABSTRACT FROM AUTHOR]

We developed an untrained neural network-based unfolding method (UNU) for both qualitative and quantitative analysis of NaI(Tl) scintillation detectors. Unlike existing deep learning approaches, which commonly rely on a supervised learning strategy, UNU is not required to prepare large numbers of training datasets owing to its scheme of grafting the physical principle of the measured gamma spectrum into the parameter optimizing process within the neural network. We designed a fully connected neural network that receives a measured spectrum as its input and produces a latent fluence spectrum through a series of iterative optimization processes. To verify its performance, we simulated spectra contributed by gamma rays with two or more energies and different relative intensities and measured experimental spectra with different combination of four radioisotopes. For both the simulated and measured spectra, UNU was found to identify the energies of gamma rays and determine their relative intensities with acceptable accuracy. We anticipate that UNU can be developed as a solver for various inverse problems in the field of the radiation measurement. • We proposed untrained neural network based unfolding method for gamma spectroscopy. • No training data are required to optimize the parameters of the neural network. • It achieved good performance on both the simulated and measured spectra. • There has been potential to be applied as a solver into other inverse problem. [ABSTRACT FROM AUTHOR]

Dose-rate monitoring instruments measure the ambient dose equivalent and hence are crucial for protecting workers from radiation exposure. Although plastic scintillation detectors (PSDs) are ideal equipment for dosimetry, they are rarely used owing to the lower detection efficiency than other scintillation detectors. In this study, we acquired ten types of G(E) functions to utilize a PSD in spectroscopic dosimetry using the least-square and first-order methods. The energy response of PSD was much improved in terms of dose evaluation. • Measurement of ambient dose equivalent using a plastic scintillation detector. • Detector with an EJ-276G plastic scintillator coupled to a silicon photomultiplier. • G(E) function obtained using the least-square and first-order methods. • Ensemble G(E) function showed the most stable and superior performance. [ABSTRACT FROM AUTHOR]