Abstract:

In order to explore the influence of miners' alertness on risk perception ability, miners' alertness tests and risk perception experiments were designed and implemented. During the experiments, fNIRS, behavioral data, and peripheral physiological signals were collected. Methods such as the normality test and one-way analysis of variance (ANOVA) were applied to investigate the differences in risk perception ability among miners with different alertness levels. Thirteen significantly different indicators were selected as feature variables. Thirteen significant differential indicators were selected as feature indicators, and Sine-SSA-BP was introduced to construct a classification and recognition model for miners' risk perception ability. The results show that miners' alertness significantly affects their risk perception ability. With increasing alertness, the correct rate of risk perception improves notably. As the alertness level rises, significant differences appear in the activation index β values of the dorsolateral prefrontal cortex and frontopolar areas. The mean skin conductance (SC_mean) in electrodermal activity (EDA) increases significantly, while the mean inter-beat interval (Mean_IBI), standard deviation of normal to normal R-R intervals(SDNN), and root mean square of successive differences (RMSSD) in heart rate variability (HRV) decrease significantly, and mean heart rate (Mean_HR) increases. The constructed miners' risk perception ability classification and identification model based on the Sine-SSA-BP achieves an accuracy of 92.30%, demonstrating excellent overall performance and robustness.

Key words: alertness, miners, risk perception ability, identification model, functional near-infrared spectroscopy (fNIRS), sine chaotic mapping sparrow search algorithm-back propagation neural network (Sine-SSA-BP)