Fang Shijie, Chen Yifan, Wu Xianwei, Zhao Nuo, Liu Yong
College of Physics and Optoelectronic Engineering, Shenzhen University, Shenzhen, 518060, Guangdong, China.
College of Physics and Optoelectronic Engineering, Shenzhen University, Shenzhen, 518060, Guangdong, China; Shenzhen Key Laboratory of Nuclear and Radiation Safety, Shenzhen, 518060, Guangdong, China.
J Environ Radioact. 2025 Jan;281:107593. doi: 10.1016/j.jenvrad.2024.107593. Epub 2024 Dec 2.
To improve the safety and reliability of radon exhalation rate monitoring systems, this study introduces an early warning method that integrates a VMD-GRU prediction model with a similar day analysis. Initially, radon exhalation rate data are decomposed into components with different informational content using the Variational Mode Decomposition (VMD) algorithm. Each component is forecasted by using the Gated Recurrent Unit (GRU) algorithm, and these forecasts are aggregated to estimate the overall radon exhalation rate. The effectiveness of the VMD-GRU model is validated through comparisons with ELMAN, LSTM, GRU,VMD-ELMAN and VMD-LSTM models. Finally, by combining the VMD-GRU model's outcomes with the similar day analysis, the system performs real-time monitoring and anomaly detection of radon exhalation rates. The results demonstrate that the proposed model effectively identifies and early warnings to abnormal radon fluctuations, significantly enhancing the precision of anomaly early warnings and providing robust decision support for radon monitoring and control, thus paving new paths for similar early warning systems.
