New method for visualizing the dose rate distribution around the Fukushima Daiichi Nuclear Power Plant using artificial neural networks

Abstract

This study proposes a new method of visualizing the ambient dose rate distribution using artificial neural networks (ANNs) from airborne radiation monitoring results. The method was applied to the results of the airborne radiation monitoring which was conducted around the Fukushima Daiichi Nuclear Power Plant by an unmanned aerial vehicle. Much of the survey data obtained in the past were used as the training data for building a network. The number of training cases was related to the error between the ground and converted values by the ANN. The quantitative evaluation index (the root-mean-square error) between the ANN-converted value and the ground-based survey result converged at 200 training cases. This number of training case was considered a rough criterion of the required number of training cases. The reliability of the ANN method was evaluated by comparison with the ground-based survey data. The dose rate map created by the ANNs method reproduced ground-based survey results better than traditional methods.

Artificial intelligence and radiation protection. A game changer or an update?

“…In France, Darley et al has used NN to learn the calibration function of public-available detectors to cosmic radiation in planes (Darley et al, 2020). In other countries, a prospective research has used a commercially-available NN to fit the ground dose rate to the count rate measured by flying drones (most easily acquired) over Fukushima Daichi power plant (Sasaki et al, 2021) and radon dispersion based on environmental parameters has been predicted accurately (Duong et al, 2021).…”

Exploring Advanced Computational Tools and Techniques with Artificial Intelligence and Machine Learning in Operating Nuclear Plants

“…Cho et al (Cho et al 2021) proposed a reproduction strategy using CNNs for radiation maps to compensate for the loss of radiation detection data. Sasaki et al (Sasaki et al 2021) applied ANNs to develop a new method of visualizing the ambient dose-rate distribution around the Fukushima Daiichi NPPs. Sun et al (Sun et al 2020) developed a methodology for optimizing the monitoring locations of long-term radiation air dose-rate monitoring near the Fukushima Daiichi NPPs.…”

Improvement of training data for dose rate distribution using an artificial neural network

“…The ARS can quickly perform measurements in a wide range, but the distribution resolution of the ambient dose rate (air dose rate) by the ARS is rough compared with a ground-based survey because the detector position is far from the radiation source (ground). The inverse problem analysis and machine learning methods have been proposed to realize a more accurate visualization of the ARS data [2][3] [4]. In the Fukushima environmental field, providing an improved position resolution of the dose rate distribution map is helpful in estimating the dose exposure of residents and in making decisions as regards lifting the evacuation zone.…”

“…In this study, we visualize the air dose rate distribution at 1 m above the ground level (agl) based on artificial neural networks (ANNs). The basic methodology of the ANNs for converting the ARS data is performed based on a previous study [4]. To improve the practical map, topography, and photographic color data excluded in the previous studies are added to the input variables.…”