Chaotic characteristics and nonlinear prediction of radon exhalation rate in granitoid materials with single fissure roughness variations

This study examines the relationship between cumulative radon concentration and fractal dimension of single fissure in synthetic granite materials, motivated by global radiation concerns stemming from radon emanation in underground geological disposal laboratories. Three analogous materials with distinct fissure fractal dimensions (1.05, 1.15, and 1.25) were synthesized and subjected to time series analysis on radon exhalation rates. The findings revealed chaotic characteristics of the radon exhalation rate time series, characterized by maximal Lyapunov exponents of 0.1306, 0.1452, and 0.1581, respectively. An optimal embedding dimension of 4 was identified for all three materials. The analysis further showed that dissipative behavior intensified with increasing fissure fractal dimensions, resulting in cumulative radon concentration amplification. The effectiveness of an RNN-LSTM deep learning network in accurately predicting radon exhalation rates in granitoid materials is demonstrated. The model successfully captured the chaotic characteristics of the time series data and made precise short-term predictions, spanning a predicted period of 44 min. This achievement facilitates the implementation of early warning mechanisms and control strategies to ensure operator safety through effective radiation protection measures.