Evaluation of false alarm characteristics and majority-voting logic integration of spectral-based discrimination methods for continuous air monitors

Continuous air monitors (CAMs) are essential devices for the rapid detection of accidental releases of artificial radionuclides at nuclear facilities. However, false alarms are frequently observed even in the absence of transuranium elements (TRUs) posing a challenge for radiation safety management. In this study, three discrimination methods – the ROI-1/ROI-2 ratio method (where ROI stands for region of interest), the correlation-based method, and the α/β ratio method – were evaluated simultaneously under identical conditions using a CAM equipped with a silicon semiconductor detector. The performance tests revealed distinct characteristics and vulnerabilities for each algorithm. The ROI-1/ROI-2 ratio method was prone to false alarms during periods of low 218Po concentration and low total acquisition counts. On the other hand, the correlation-based and α/β ratio methods were affected by fluctuations in the concentration ratios of naturally occurring background radionuclides. Although ideal discrimination cannot be achieved using any single method alone, the results suggest that combining multiple methods based on majority-voting logic can effectively reduce false alarm events and improve system reliability. These findings will contribute to the development of more reliable atmospheric radiation monitoring technologies.