To investigate the nonlinear chaotic fractal characteristics of high-pressure gas rock-breaking vibration signals, a new rock-breaking gas generator was developed and gas rock breaking experiments were conducted in a stable limestone open-pit mine. High-speed photography was used to capture the rock breaking process and monitor the vibration effects. The nonlinear characteristics of the collected rock-breaking vibration signals were finely extracted. The results show that, compared with emulsion explosive blasting, high-pressure gas rock breaking has high-frequency and low-amplitude characteristics, with smaller fractal box dimension values and more limited frequency domain distributions. The opening width of the multifractal spectrum is narrower and the singularity of energy distribution is smaller; As the frequency decreases, the dominant mode component chaotic attractor contained in the gas rock-breaking signal exhibits an elliptical trajectory in the two-dimensional phase plane with a gradually decreasing ratio of the long axis to the short axis, ultimately converging near the stable point at the center of the ellipse, exhibiting typical chaotic dynamic characteristics. The research results provide exploratory ideas for safe and efficient mining and disaster assessment of mineral rocks under dual carbon targets.

Aiming at the problem of insufficient time-frequency resolution of tunnel blasting vibration signal, a time-frequency image enhancement algorithm based on convolutional neural network is applied, through the time-frequency image enhancement of the measured tunnel blasting signal, the aggregation range of the blasting signal energy in the time-frequency domain is captured, and the real signal reflecting the blasting characteristics is reconstructed; according to the real signal, the initiation time of detonator in blasting network is accurately distinguished, and the characteristics of tunnel blasting detonator disaster source are identified.The analysis shows that the time-frequency image enhancement algorithm based on convolutional neural network can effectively suppress the cross-terms in the signal, retain the auto-terms of the signal to the greatest extent, and improve the energy aggregation and time-frequency resolution of the blasting signal; The mixed use of different batches of detonators is the main disaster causing factor of tunnel safety.Supervision should be strengthened to realize safe and efficient tunnel construction.