Study on critical slowdown characteristics and early warning model of damage evolution of sandstone under freeze–thaw cycles

Freeze–thaw damage of rock mass poses a great threat to the safety engineering, ground buildings, and low-temperature storage liquefied natural gas (LNG) in cold regions. By collecting acoustic emission (AE) signals sandstone during uniaxial compression failures, this paper analyzed critical slowdown phenomenon different types freeze–thaw failure. According auto-correlation coefficients variance AE under windows steps, precursors were determined warning model engineering failure based on principle was proposed. Then Grey Wolf Optimizer (GWO) algorithm used optimize initial weights thresholds back propagation (BP) neural network, influence factors working conditions input as training sets train network. The results showed that correlation between predicted value real GWO-BP network reached 99.90% 98.81% respectively, indicating accuracy BP prediction improved. This study provides new method for early warning, has theoretical guiding significance enriching improving monitoring technology.