Recognition of low-active alpha and beta radiation in a closed air volume using open-air gas-discharge counter and artificial neural network

Abstract

To date, artificial neural networks have become firmly established in those areas where the use of conventional algorithmic solutions is limited or impossible. Detection of alpha-active and beta-active nuclide contamination is important in radiation ecology medicine, because such contamination produces strong biological effects on humans. An interesting application of artificial neural networks is the recognition of the type of ionizing radiation based on air ion method. Negative ions from the alpha particle tracks are detected by the open-air gas-discharge counter. Ion clusters being transferred from the particle tracks to the detector volume by an air flux. Avalanche multiplication of detached electrons from air ions allows to achieve a large gain. Background radiation can be characterized as irregular, with a small frequency of pulses of small amplitude (small power of the amplitude spectrum), while ionization from alpha particles is characterized by irregularity (low-active alpha particles), accuracy of pulses (very large number of them at a certain interval), large amplitude (large power of the amplitude spectrum). The ionization of beta particles is characterized by the regularity (due to the greater path length and closed air volume) and medium amplitude spectrum power.